General Reading iGAS Guidelines - Published January 2012

CLICK HERE

Educational Interim UK guidelines for management of close community contacts of invasive group A streptococcal disease. Health Protection Agency, Workshops 2012 Group A Working Group. Communicable Disease and Public Health 2004; 7(4):354-361.

CLICK HERE

Keynote Presentation: Diagnosis and Complicated infections of and skin structures: when the infection is more than skin deep. DiNubile MJ, Lipsky, B. Journal of treatment Antimicrobial Chemotherapy, 2004, 53, Suppl. S2, ii37-ii50 of skin and soft CLICK HERE Practice guidelines for the diagnosis and management of skin and tissue infections soft tissue infections. Stevens DL et al. Clinical Infectious Disease 2005; 41:1373–1406

CLICK HERE

Infections of skin and soft tissue: Outcomes of a classifi cation scheme. Eron J. Clinical Infectious Diseases 2000;31:287(A432).

CLICK HERE

Occurrence and antimicrobial susceptibility patterns of pathogens isolated from skin and soft tissue infections: report from the SENTRY READING Antimicrobial Surveillance Program (United States and Canada, 2000). Rennie RP et al. Diagn Microbiol Infect Dis. 2003 Apr; 45(4):287-293. LIST CLICK HERE Comparison of community and health care associated methicillin resistant aureus infection. Naimi TS, et al. JAMA 2003; 290: 2976-2984

CLICK HERE

Methicillin resistant S. aureus infections amoung patients in the emergency department. Moran GJ et al. The New England Journal of Medicine 2006

CLICK HERE

HPR 2011;5(7): News CLICK HERE

Polyclonal multiply antiobiotic-resistant methicillin-resistant Staphylococcus aureus with Panton-Valentine leucocidin in England. JAC 2009; doi: 10.1093/jac/dkp386; CLICK HERE

Eff ect of on Staphylococcus aureus producing panton- valentine leukocidin. Dumitrescu O, et al. Antimicrobial Agents and Chemotherapy. 2007, 1515–1519 CLICK HERE

Centers for Disease Control and Prevention, Skin & Soft Tissue Infections in Returned Travelers - Chapter 5 - 2012 Yellow Book - Travelers’ Health CLICK HERE Fever and the returning traveller. N Kumar, DJ Lewis. BMJ Gottlieb SL, Kretsinger K, Tarkhashvili N, et al. 2012;344:e2400 Published April 2012 Long-term outcomes of 217 botulism cases in CLICK HERE the Republic of Georgia. Clin Infect Dis 2007; 45:174 Severity assessment of skin and soft tissue infections: CLICK HERE cohort study of management and outcomes for hospitalised patients. Marwick et al. Journal of Botulism, Sobel J. Clin Infect Dis 2005 October Antimicrobial Chemotherapy, doi:10.1093/jac/dkq362, 15;41(8):1167-73 2010 CLICK HERE CLICK HERE The GAS men Guidelines for UK practice for the diagnosis and The prevalence of beta-haemolytic streptococci management of methicillin-resistant Staphylococcus in throat specimens from healthy children and aureus (MRSA) infections presenting in the community. adults. Scand J Prim H Care 1997, 15: 149

Nathwani D, Morgan M, Masterton R, Dryden M, CLICK HERE Cookson B, French G, Lewis D. Journal of Antimicrobial Chemotherapy. 2008 doi:10.1093/jac/dkn096 “Cloud” health-care workers. Sherertz RJ. CLICK HERE (Emerging Infectious Diseases 2001, 7:241) CLICK HERE Intravenous immunoglobulin therapy for streptococcal --a comparative observational study. case report Kaul R, McGeer A, Norrby-Teglund A, Kotb M, Schwartz B, Intravenous immunoglobulin G therapy in streptococcal toxic O’Rourke K, Talbot J, Low DE. Clinical Infectious Diseases shock syndrome: a European randomised, double blind, placebo 1999 Apr;28(4):800-7. controlled trial. CID 2003;37:333-340

CLICK HERE CLICK HERE

Diagnosis and management of cellulitis, Phoenix G et al, . Bellapianta JM, Ljungquist K, Tobin E, Uhl R. J Am BMJ 2012;345:e4955 Acad Orthop Surg 2009 17(3):174-82

CLICK HERE CLICK HERE

An infected insect bite? Group A streptococcus peri-partum infection- following the Health Protection Agency Centre for Infections, Duty guidelines Doctor Botulism Protocol, November 2011 Global emm type distribution of group A streptococci: systematic

CLICK HERE review and implications for vaccine development. Steer AC et al. Lancet 2009;9:611-16 Werner SB, Passaro D, McGee J, et al. Wound botulism in CLICK HERE California, 1951-1998: recent epidemic in heroin injectors. Clin Infect Dis 2000; 31:1018 Painful calf

CLICK HERE Streptolysin S and necrotising infections produced by group G strep- tococcus. Humar, D., V. Datta, D. J. Bast, B. Beall, J. C. De Azavedo, and Passaro DJ, Werner SB, McGee J, et al. Wound botulism V. Nizet. 2002. Lancet 359:124-129. associated with black tar heroin among injecting drug CLICK HERE users. JAMA 1998; 279:859 CLICK HERE Invasive group A, B, C and G streptococcal infections in Denmark 1999–2002: epidemiological and clinical aspects, Ekelund, K., P. Sam AH, Beynon HL. Images in clinical medicine: Wound Skinhoj, J. Madsen, and H. B. Konradsen. Clinical Microbiology and botulism. N Engl J Med 2010; 363:2444 Infection 2005 11:569-576.

CLICK HERE CLICK HERE

Yuan J, Inami G, Mohle-Boetani J, Vugia DJ. Recurrent Clinical characteristics of necrotizing fasciitis caused by group G wound botulism among injection drug users in California. Streptococcus: Case report and review of the literature. Sharma, M., Clin Infect Dis 2011; 52:862 R. Khatib, and M. Fakih. 2002. Scandinavian Journal of Infectious Diseases 34:468-471. CLICK HERE CLICK HERE Journal of Infection (2012) 64,1e18

www.elsevierhealth.com/journals/jinf

PRACTICE GUIDELINES Guidelines for prevention and control of group A streptococcal infection in acute healthcare and maternity settings in the UK

Jane A. Steer a, Theresa Lamagni b, Brendan Healy c, Marina Morgan d, Matthew Dryden e, Bhargavi Rao b, Shiranee Sriskandan f, Robert George g, Androulla Efstratiou g, Fiona Baker h, Alex Baker i, Doreen Marsden j, Elizabeth Murphy k, Carole Fry l, Neil Irvine m, Rhona Hughes n, Paul Wade o, Rebecca Cordery p, Amelia Cummins q, Isabel Oliver r, Mervi Jokinen s, Jim McMenamin t, Joe Kearney u,v,* a Department of Microbiology, Derriford Hospital, Plymouth, UK b Healthcare-Associated Infection & Antimicrobial Resistance Department, Health Protection Agency, London, UK c Department of Microbiology, Public Health Wales, Cardiff, UK d Department of Microbiology, Royal Devon and Exeter Hospital, Exeter, UK e Department of Microbiology, Royal Hampshire County Hospital, Winchester, UK f Centre for Infection Prevention & Management, Department of Infectious Diseases, Imperial College, London, UK g Respiratory & Systemic Infections Department, Health Protection Agency, London, UK h Infection Prevention & Control Department, North Devon District Hospital, Barnstaple, UK i Communications, Health Protection Agency, London, UK j Lee Spark NF Foundation, Preston, UK k Occupational Health Department, NHS Grampian Occupational Health Service, Aberdeen, UK l Infectious Diseases and Blood Policy, Department of Health, London, UK m Public Health Agency, Northern Ireland, UK n Obstetrics & Gynaecology, Royal Infirmary, Edinburgh, UK o Directorates of Pharmacy and Infection, Guy’s & St. Thomas’ NHS Foundation Trust, London, UK p North East and North Central London Health Protection Unit, Health Protection Agency, London, UK q Essex Health Protection Unit, Health Protection Agency, Witham, UK r Health Protection Agency, South West, Gloucester, UK s Development Department, Royal College of Midwives, UK t Health Protection Scotland, Glasgow, UK u Health Protection Agency, East of England, Witham, UK

Accepted 1 November 2011 Available online 17 November 2011

* Corresponding author. Tel.: þ44 0845 241 2266; fax: þ 44 0 1376 302278. E-mail address: [email protected] (J. Kearney). v On behalf of the GAS Guideline Development Working Group.

0163-4453/$36 Crown Copyright ª 2011 Published by Elsevier Ltd on behalf of The British Infection Association. All rights reserved. doi:10.1016/j.jinf.2011.11.001 2 J.A. Steer et al.

KEYWORDS Summary Hospital outbreaks of group A streptococcal (GAS) infection can be devastating Group A streptococcus; and occasionally result in the death of previously well patients. Approximately one in ten cases Infection control; of severe GAS infection is healthcare-associated. This guidance, produced by a multidisciplin- Midwifery; ary working group, provides an evidence-based systematic approach to the investigation of sin- Disease outbreaks; gle cases or outbreaks of healthcare-associated GAS infection in acute care or maternity Great Britain settings. The guideline recommends that all cases of GAS infection potentially acquired in hospital or through contact with healthcare or maternity services should be investigated. Healthcare workers, the environment, and other patients are possible sources of transmission. Screening of epidemiologically linked healthcare workers should be considered for healthcare-associated cases of GAS infection where no alternative source is readily identified. Communal facilities, such as baths, bidets and showers, should be cleaned and decontaminated between all pa- tients especially on delivery suites, post-natal wards and other high risk areas. Continuous sur- veillance is required to identify outbreaks which arise over long periods of time. GAS isolates from in-patients, peri-partum patients, neonates, and post-operative wounds should be saved for six months to facilitate outbreak investigation. These guidelines do not cover diagnosis and treatment of GAS infection which should be discussed with an infection specialist. Crown Copyright ª 2011 Published by Elsevier Ltd on behalf of The British Infection Associa- tion. All rights reserved.

Introduction associated, most (58%) being post-surgical infections.5 Be- tween 2 and 11% of all severe GAS infections are associated with recent childbirth, a rate of approximately 0.06 per The overriding trend over the last century has been one of 1000 births.5e7 Findings from the 2006e08 triennial report dramatic decline in severe GAS infections. However, the on maternal deaths identified an increase in the numbers last three decades have witnessed periodic upsurges in of maternal deaths associated with GAS genital tract Europe and beyond.1 The reasons for these changes are not from around 1 death per annum in 2000-02 to 4 per annum understood, but might represent evolutionary shifts in cir- in 2006e08.8 Several of these deaths were in women with culating strains, driven by population immunity. Current es- a recent respiratory tract infection or women with family timates of annual incidence of severe GAS infection range members with recent history of sore throats. Infection in from 2 to 5 per 100,000 population in developed countries, e the mother carries a further immediate risk of infection with case fatality rates ranging from 8 to 23%.1 4 Data col- in the baby.9,10 lected in 2003-04 as part of a European project recorded a rate of 3.33 cases per 100,000 population in England, Wales and Northern Ireland.5 Outbreaks of GAS in acute care settings

Incidence of healthcare-associated and postpartum A review of healthcare-associated invasive GAS infections GAS infection in Ontario between 1992 and 2000 identified one in 10 cases as being linked to an outbreak.6 Hospital outbreaks of GAS infection can escalate rapidly, be prolonged and Between 5 and 12% of cases of severe GAS infection are e result in both patients and healthcare workers (HCWs) be- found to be healthcare-associated.1,3 6 UK data in 2003-04 ing infected.6 The national reporting system for significant identified 9% of severe GAS infections as being healthcare- health protection incidents in England (HPA Incident Re- porting Information System) identified 10 outbreaks of the GAS infection in hospital settings during 2008 and Glossary 2009 combined. Surgical, obstetrics and gynaecology, and burns units are most commonly involved in hospital out- breaks, although outbreaks have been seen in a wide GAS group A streptococcus 6 HPA Health Protection Agency range of different hospital settings. Investigation of these iGAS invasive group A streptococcus outbreaks has identified a range of transmission routes: IPCT infection prevention and control team (or colonised HCWs to patients, environmental sources to pa- equivalent) tients, and patient-to-patient transmission. Patients with HCW healthcare worker both community and healthcare-associated GAS infection OH Occupational Health have initiated hospital outbreaks with secondary cases typically arising within one month of the index case al- PPE personal protective equipment 6 SIGN Scottish Intercollegiate Guidelines Network though longer intervals have been documented. In STSS streptococcal toxic shock syndrome HCWs, throat colonisation is the most common source, al- SUI serious untoward incident though skin, vaginal and rectal colonisation have also been linked to outbreaks.6,11 UK guidelines on prevention and control of GAS in healthcare settings 3

Methods was made with leading streptococcal researchers across the world. Relevant papers were reviewed and graded using Search strategy the Scottish Intercollegiate Guidelines Network (SIGN) method by a minimum of two independent members of the working group.12 The working group made recommendations A literature review was undertaken in November 2009 which on the basis of this evidence. included case reports, outbreak/cluster investigation re- ports, retrospective and prospective surveillance studies Case definitions and national guidelines. The following sources were searched: Medline (1950 onwards), the Cochrane Library Invasive GAS (iGAS) infection and The National Health Service Centre for Reviews and Invasive GAS infection is illness associated with the iso- Dissemination. Reports from working groups, expert com- lation of GAS from a normally sterile body site, such as mittees and the Royal Colleges were also included. The key blood, cerebrospinal fluid, joint aspirate, pericardial/peri- word search used the following individual terms and com- toneal/pleural fluids, bone, endometrium, deep tissue or bined the terms using AND/OR: infection control, healthcare at operation or post mortem. For the purposes of associated infection; nosocomial; maternity; health care these guidelines it also includes severe GAS infections, workers; clusters; surgical; outbreaks; transmission; puer- where GAS has been isolated from a normally non-sterile peral sepsis; group A, C and G and beta-haemolytic strep- site in combination with a severe clinical presentation, such tococcus; Streptococcus pyogenes; invasive; as streptococcal toxic shock syndrome (STSS) or necrotising prophylaxis; carriage. The search was not restricted accord- fasciitis. ing to language of publication; the only restriction was to hu- man studies. Relevant studies identified from the electronic Peri-partum GAS infection search were reviewed for relevance by title and abstract. For the purposes of these guidelines, peri-partum GAS The full text of studies of potential relevance was retrieved. infection is defined as isolation of GAS up to 7 days post All studies identified also had their references checked for discharge or delivery in the mother in association with relevant articles. To identify national guidelines that might a clinical infection, such as endometritis, STSS, wound not be published in the scientific literature, direct contact infection, or isolation from a sterile site.

Algorithm 1 Management of a single case of GAS infection. 4 J.A. Steer et al.

Healthcare-associated GAS infection Initial investigations A healthcare-associated GAS infection is defined as a GAS infection that is neither present nor incubating at the time of Initial investigations should establish if the infection or admission but considered to have been acquired following colonisation with GAS is community or healthcare- admission to hospital or as a result of healthcare interven- associated. It should be established if the patient had tions in other healthcare facilities. Typically, onset of GAS symptoms or signs consistent with GAS infection such as > infection is 48 h after admission, or postoperatively at any sore throat or on or just prior to admission time after admission and for up to seven days post discharge. or childbirth. Intra-familial spread of GAS is common and enquiries should be made as to whether close personal Outbreak contacts or visitors are suffering from any illness that An outbreak should be considered if there are two or more could be attributable to GAS. Identification of a close cases of suspected GAS infection related by person or place. personal contact with symptoms or signs of GAS infection These cases will usually be within a month of each other but reduces the likelihood that the infection was acquired the interval may extend to several months. It should be noted from a healthcare source. Symptomatic close contacts that the interval between cases in published outbreak reports should seek advice from their GP. The infection should for GAS has, on occasion, extended to more than a year. be considered to be healthcare-associated if symptoms Reference laboratory typing from culture-proven cases is and signs of infection were not present on admission and needed to confirm that cases are related. they have developed during a hospital stay or within 7 days of discharge from hospital or post delivery, with no Infection prevention and control of GAS other obvious source of transmission. In this case, infection screening of HCWs as a possible source should be considered - see Transmission from healthcare worker The successful management of every case of GAS is to patient. important, not only to prevent spread and possible serious Contacts of community-acquired cases of invasive GAS infection should be managed according to the existing infections, but also to investigate if transmission is occur- 9 ring from an ongoing and preventable source. All GAS community guidelines. infections suspected of being healthcare-associated should be investigated further (see Algorithm 1).

Reporting cases Recommendations

All cases of suspected GAS infection identified in acute care IPCT should establish whether the case is community settings or maternity units, including stand-alone midwife or healthcare-associated. led units, and any cases identified within seven days of Further investigation of potential sources of infec- discharge or delivery that could have been healthcare- tion is warranted for any case of GAS infection con- associated should be reported to the infection prevention sidered to be healthcare-associated. and control team (IPCT) or equivalent. SIGN GRADING Good practice points Invasive GAS infection is a notifiable disease in England, Wales and Scotland.13 All iGAS infections should be dis- cussed with the local health protection specialist so that contact assessment can be initiated according to existing Prospective and retrospective surveillance national guidance.9 Outbreaks of GAS infection and deaths in patients with The interval between identified cases in published outbreak healthcare-associated GAS infection should be reported as reports for GAS has, on occasion, extended up to one or serious untoward incidents via normal reporting routes. more years,14 and as such the IPCT should maintain ongoing In the event of a death due to confirmed or suspected GAS infection surveillance where a case of healthcare- GAS - see Communication with, and advice to, mortuary associated GAS infection has been identified. The IPCT and pathology staff. should review surveillance records for the past six months at a minimum to establish if the new case is sporadic or Recommendations part of a possible outbreak of healthcare-associated GAS infection. Following a case of healthcare-associated GAS infec- All cases of suspected GAS infection identified in the tion the IPCT should consider prospective enhanced acute care setting or maternity units and stand alone surveillance which may include, for example, sampling midwife led units and any cases identified within seven infected wounds of patients in the vicinity of the index days of discharge or delivery that could have been case or who are being cared for by the same HCWs. In healthcare-associated should be reported to the IPCT. addition, the IPCT should be informed of any cases which All iGAS cases should be discussed with and notified may be caused by GAS, e.g. cases of puerperal sepsis to the local health protection specialist by the rele- treated empirically. Post-discharge surveillance, if re- vant clinician and microbiologist. quired, would help identify healthcare-associated cases SIGN GRADING Good practice points presenting after discharge. UK guidelines on prevention and control of GAS in healthcare settings 5

Personal protective equipment (PPE) Recommendations Whilst the patient is considered infectious, HCWs must use IPCT should undertake a retrospective analysis of mi- personal protective clothing including disposable gloves crobiology and surveillance records to identify possi- and aprons when in contact with the patient and their ble linked cases of healthcare-associated GAS equipment or immediate surroundings. Facial protection, infection arising in the past 6 months. such as a fluid repellent surgical mask and eye shield or IPCT should maintain GAS continuous alert organism visor, is recommended where a risk of transmission from surveillance to identify outbreaks which may arise droplets is identified; examples include bronchoscopy, over prolonged periods of time. suctioning or dressing wounds that are producing a large Following a case of healthcare-associated GAS infec- amount of exudate. Fluid repellent surgical masks with tion the IPCT should consider prospective enhanced visors must be used at operative debridement/change of surveillance which may include, for example, sam- dressings for cases of necrotising fasciitis. If an HCW has pling of infected wounds of patients in the vicinity any break in skin integrity e.g. a cut or skin lesion, this of the index case or who are being cared for by the must be covered with a waterproof dressing. In the event same HCWs. of failure to comply with PPE or needlestick - see SIGN GRADING Good practice points Transmission from patient to healthcare worker. Visitors must be given information about how to pre- vent the transmission of infection, and shown how to use appropriate PPE when visiting the affected individual. The Patient isolation PPE required by visitors will depend on risk assessment of the factors affecting transmission (e.g. if there is a high Patients diagnosed with or clinically suspected of having risk of droplet transmission) and also the visitor’s level of GAS infection should be isolated in a single room, with direct contact and involvement in the affected person’s a self contained toilet and its own hand basin. Breast care. feeding should be supported where possible. Mother and baby should not be separated unless the mother or baby requires admission to an ICU. Notes and charts should be kept outside the room and patients should have dedicated Recommendations equipment where possible. It is frequently cited that isolation should continue for 24e48 h after commencement of appropriate antibiotic HCWs should wear PPE including disposable gloves therapy. Studies suggesting that exclusion for 24 h of and aprons when in contact with the patient or their effective therapy is appropriate, have primarily been equipment and their immediate surroundings. performed in children with pharyngitis or Breaks in the skin must be covered with a waterproof (Padfield, personal communication). However, case reports dressing. show that GAS can be isolated from superficial sites beyond Fluid repellent surgical masks with visors must be 24 h of antibiotic treatment, including the drying umbilical used at operative debridement/change of dressings cord.14,15 In a recent case report of transmission from a pa- of necrotising fasciitis and for procedures where tient with necrotising fasciitis to an HCW, this occurred 50 h droplet spread is possible. after initiation of appropriate antimicrobial therapy.16 Visitors should be offered suitable information and The working party felt that although there were some relevant PPE following a risk assessment of the visi- instances when patients should be isolated until culture tor’s level of direct contact/involvement in the af- negative, 24 h of effective therapy was appropriate for the fected person’s care. majority of cases seen in hospitals; examples include SIGN GRADING Good practice point necrotising fasciitis where there is significant discharge of potentially infectious body fluids, patients with infected eczema where there is a high risk of shedding, mothers and neonates on maternity units, and patients on burns units. Hand hygiene

Semmelweis identified the importance of hand washing in Recommendations preventing the spread of puerperal sepsis on maternity units.17 HCWs must adhere to strict hand hygiene policy us- ing an effective technique i.e. hand washing with soap and Patients with GAS should be placed in isolation for water or decontamination with alcohol hand rub before and a minimum of 24 h of effective antibiotic therapy. after contact with the patient and/or their environment, Cases of necrotising fasciitis and other cases where regardless of the use of gloves and other protective there is significant discharge of potentially infected measures.18 body fluids or high risk of shedding, mothers and ne- Where appropriate the patient and their visitors must be onates on maternity units and patients on burns offered suitable information and facilities to encourage units, should be isolated until culture negative. their own adherence to standard infection control practice SIGN GRADING D/Good practice points including effective hand hygiene practice. 6 J.A. Steer et al.

facility is not recommended unless unavoidable or essential Recommendations for the individual’s clinical care. Isolation dictates that the movement of patients for non-clinical reasons should be HCWs must adhere to strict hand hygiene policy. minimised. Details of the risk of infection must be effec- Visitors should be offered suitable information and tively communicated to the ambulance service, the receiv- facilities to be able to adhere to standard infection ing ward/department or facility, and the receiving IPCT control practice, including good hand hygiene. must be informed using the inter-healthcare transfer SIGN GRADING Good practice points form. If it is found that a case of GAS could have acquired the infection in another hospital, that information should be relayed to the referring hospital. Environmental cleaning

The isolation room, furniture and equipment must be cleaned daily as a minimum and terminal cleaning un- Recommendations dertaken. Detergent and water followed by hypochlorite at 1000 ppm, or a combined product, is recommended for all Transfer only if unavoidable or essential for the pa- environmental and equipment cleaning where a patient is tient’s care. known to have an infection, healthcare associated or Details of the risk of infection must be effectively otherwise.19,20 communicated to the ambulance service, the receiv- Communal facilities such as baths, bidets and showers ing facility, IPCT and if appropriate, the referring should normally be cleaned and decontaminated between hospital. patients irrespective of whether they are known to be SIGN GRADING Good practice points infected or not. In the case of delivery suites and early post-natal care this is particularly important because of the high risk of blood and body fluid contamination, the exposed nature of episiotomy wounds and the supporting Infections occurring in mothers and babies evidence that these communal utilities have acted as the source of outbreaks - see Environment as source of 21e23 Although peri-partum GAS infection is typically acquired at outbreak. the time of or after childbirth from both exogenous and endogenous sources,28,29 pregnant women who are found to Recommendations be infected with or carrying GAS earlier in pregnancy should be treated at the time and have this clearly documented in 30 The isolation room, furniture, and equipment should the maternity notes. be cleaned with detergent and water followed by hy- Babies born to infected or colonised mothers may pochlorite at 1000 ppm daily (or combined detergent become colonised and this can be detected by swabbing of the umbilicus, ears and nose. Occasionally the baby may hypochlorite product). 31e36 Communal facilities such as baths, bidets and develop infection including invasive disease. Maternal showers should be cleaned and decontaminated be- and neonatal infection tend to be closely related in terms tween all patients especially on delivery suites, of timing. Mother and baby should not be separated unless post-natal wards and other high risk areas, such as the mother or baby requires admission to an ICU. e burns units. Following the identification of infected mother baby SIGN GRADING D/Good practice points. pairs in the UK, interim guidance for their management was published in 2004.9 Antibiotics should be administered to mother and baby if either develops suspected or confirmed Linen and waste invasive GAS disease in the neonatal period (first 28 days of life). Of note, one neonatal sepsis and one necrotising fas- Whilst the patient is considered infectious, linen and waste ciitis of the scalp have been reported in association with e must be handled as hazardous.24 27 the use of foetal scalp electrodes.37

Recommendation Recommendations

Whilst the patient is considered infectious, linen and Antibiotics should be administered to mother and waste must be handled as hazardous. baby, if either develops suspected or confirmed inva- SIGN GRADING Good practice points sive GAS disease in the neonatal period (first 28 days of life). SIGN GRADING C Transferring patients Pregnant women infected or colonised with GAS prior to admission should be treated and have this clearly documented in the maternity notes. In order to minimise the risk of cross-infection, the transfer SIGN GRADING Good practice points of any patient with an infection to another healthcare UK guidelines on prevention and control of GAS in healthcare settings 7

Transmission from patient to close personal antibiotics. The working party recommends HCWs receive contacts a 3 day course of amoxicillin (500 mg, orally, three times a day) in these instances, unless there is evidence of active Antibiotics should not be routinely administered to contacts infection in the HCW, where a full course should be given. of GAS cases. Close personal contacts of a case of invasive GAS infection should receive written information outlining Recommendations the signs and symptoms of invasive GAS infection and ad- vised to seek medical attention if they develop such symp- HCWs working without appropriate PPE whilst a pa- toms within 30 days of a diagnosis in the index case in tient is infectious should be advised about the signs accordance with previous guidance.9 This is the responsibil- and symptoms of GAS infection for 30 days after ity of the local health protection specialist, although, local the diagnosis in the index patient and if symptomatic arrangements should be made so that patient information is seek urgent medical advice. available and can be given to the relatives in the acute care Any such exposures should be referred to occupa- setting - see Appendix 3. Close personal contacts are de- tional health. Antibiotic prophylaxis should be consid- fined as the same as for , that is ered for HCWs who sustain a needlestick injury or sharing a household or kissing contacts within the seven direct contamination of mucous membranes or breaks days prior to the onset of the illness.38 in the skin with potentially infectious material. SIGN GRADING Good practice points Recommendations Transmission from patient to patient Antibiotics should not be routinely administered to all contacts of GAS cases. The local health protection specialist should be noti- Transmission from patient to patient is minimised with fied of all iGAS infections. isolation and full compliance with standard precautions for Close contacts of iGAS cases should receive written infection prevention and control. The IPCTshould establish if information and have a heightened awareness of other recent cases are connected. Patients with both the signs and symptoms of GAS for 30 days after community and healthcare-associated GAS infection and the diagnosis in the index patient. colonised and infected HCWs have seeded hospital out- 6 Close contacts of iGAS cases should seek urgent med- breaks. Antibiotics should not be routinely administered ical advice if they develop such symptoms within 30 to contacts of GAS except in exceptional circumstances - days of a diagnosis in the index case in accordance see Use of chemoprophylaxis. Consideration should be given with previous guidance. to providing information to patients in close contact with the SIGN GRADING Good practice points index case if there has been significant close contact prior to infection control procedures being instituted - see Communi- cation with, and advice to, close contacts and Appendix 3. Transmission from patient to healthcare worker Transmission from healthcare worker to Transmission from patient to HCW has been most frequently patient described in the context of necrotising fasciitis where multiple contacts may become infected or colonised.39,40 Although many healthcare-associated GAS infections will be One HCW with dermatitis developed cellulitis of the arm due to endogenous flora, some patients will have acquired within 48 h of nursing a patient without gloves.41 Transmission their infection from a HCW - see Healthcare workers as has also been described during a post mortem - see Commu- source of outbreak. Depending on the circumstances of nication with, and advice to, mortuary and pathology staff.42 the case in question, and where there is no other obvious Appropriate PPE should be worn - see Personal protec- source of transmission, the IPCT should consider screening tive equipment (PPE). HCWs who have performed direct HCWs in contact with the patient. physical procedures on a patient with GAS infection, e.g. For a single case of healthcare-associated GAS, all HCWs mouth-to-mouth resuscitation, should be advised by the in contact or working in close proximity to the patient IPCT on the signs and symptoms of GAS disease and advised (patient’s bed space, theatre, delivery room) should be to seek medical advice if they develop such symptoms considered as possible sources of healthcare-associated within 30 days of a diagnosis in the index case.9 Any such GAS. The HCWs most likely to have transmitted GAS are exposed HCW should be referred to occupational health. those with direct contact with the patient within seven Antibiotic prophylaxis should be considered for HCWs days of the onset of the infection. In particular, the follow- who sustain a needlestick injury or direct contamination of ing groups should be considered for screening: mucous membranes or breaks in the skin with material potentially infected with GAS. The decision to treat should those present in theatre and performing post-operative be made on a case-by-case basis after discussion between dressing changes for surgical cases6,43 a microbiologist or other infection specialist and an those performing vaginal examinations or dealing with occupational health practitioner, taking into account the episiotomies and those present at delivery for mater- type of exposure and length of time the patient has been on nity cases6 8 J.A. Steer et al.

The IPCT may wish to take a step-wise approach to their investigations accordingly. The IPCT should consider asking Recommendations HCWs to present to Occupational Health for screening if they have been symptomatic with a sore throat or skin infection, In the event of death, the hospital mortuary staff or have had skin lesions/dermatitis/eczema or vaginitis or should be informed of the risk of infection and routes pruritus ani during the week prior to the index patient’s on- of transmission. set of infection - see example letter Appendix 4. The IPCT Pathology staff should be informed when unfixed tis- may decide to screen asymptomatic HCWs in certain circum- sue from a case of necrotising fasciitis is sent for ex- stances e.g. screening theatre staff following a post- amination. operative case of necrotizing fasciitis. The HCWs should be SIGN GRADING Good practice point seen and screened by an occupational health practitioner. Few studies of GAS throat carriage in the healthy adult population have been undertaken, but of those conducted, Communication with, and advice to, close carriage rates of 5% or less are reported, with most studies contacts reporting carriage in less than 1%.44e47 Similarly, studies of GAS vaginal and rectal colonisation, restricted to pregnant It is important that suitable and accurate information is women, report carriage rates of 1% or less.48,49 As such, communicated to any patient with iGAS infection and their a positive screening result should be considered as indica- close personal contacts by the responsible consultant or tive of likely source of transmission and dealt with as a member of their team - see, Appendices 2 and 3. The local such whilst awaiting typing results. Please refer to Screen- health protection specialist in liaison with the IPCT should ing of healthcare workers for further advice on HCW ensure relevant information is given in written form to close screening and section Management of colonised and in- personal contacts in accordance with existing community fected healthcare workers for management of GAS colon- guidancee see Appendix 2 and Transmission from patient ised or infected healthcare staff. to close personal contacts. All HCWs should be fully informed at handover of shifts so that communication with the patient and their family is consistent, accurate, and documented. Recommendations Recommendations All HCWs in contact with the patient, either in direct contact or working in the close vicinity (patient’s Suitable and accurate information should be pro- bed space), should be considered as possible sources vided promptly to the patient and close personal of healthcare-associated GAS. contacts for iGAS infections. HCWs in contact with a case of healthcare- Effective hand over between health care teams associated GAS should be considered for screening should ensure communication with the patient with if they have suffered a sore throat or skin infection, iGAS infection and their close personal contacts is or have had skin lesions/dermatitis/eczema, vagini- consistent, accurate and documented. tis or pruritus ani within seven days of the onset of SIGN GRADING Good practice points the infection in the patient. If so, the HCW should be seen and relevant swabs taken by occupational health. Isolates from positive swabs should be sent Management of an outbreak of GAS infection for typing along with the patient isolate if not al- ready sent. The investigation and control of single cases of GAS also The IPCT may decide to screen asymptomatic HCW in applies to cases in outbreaks. certain circumstances. SIGN GRADING D Formation of outbreak control team

When a suspected or confirmed outbreak of GAS has been Communication with, and advice to, mortuary identified, interventions to prevent further transmission and pathology staff and further cases should be put in place immediately (see Algorithm 2). The Director of Infection Prevention and Con- There are reports of invasive streptococcal infections trol, infection control doctor or deputy should set up an acquired by healthcare workers from patients, including outbreak control team. The make-up of the team will de- a case of necrotising fasciitis following needlestick injury in pend on the nature of the outbreak, but may include infec- a mortician.42,50 In the event of a patient death the mortu- tion control nurses, a consultant microbiologist, consultant ary staff should be informed of the risk of infection and from the specialty involved, occupational health adviser, routes of transmission such that the necessary precautions local health protection specialist, local commissioning can be undertaken. A cadaver bag should be used. The body lead, cleaning manager, bed manager, appropriate health- can be viewed, but no embalming or other preparation of care manager and communications adviser. A member of the body should take place.51 Pathology staff should also the IPCT should supervise the daily management of the out- be informed when unfixed tissue from a case of necrotising break and oversee the immediate implementation of pre- fasciitis is sent for examination. ventative measures. UK guidelines on prevention and control of GAS in healthcare settings 9

Healthcare workers as source of outbreak Recommendation Surgeons, nurses, anaesthetists, midwives, and wound care An outbreak control team should be convened to teams have all been implicated in transmission of GAS to manage and control an outbreak of GAS infection. patients, either whilst infected or colonised.14,31,40,43,52e60 SIGN GRADING D In Canada, symptomatic HCWs account for 8.2% of outbreaks and colonised, asymptomatic HCWs for 34%.61 Of the outbreaks linked to colonised healthcare workers, throat colonisation is the most common site of colonisation Epidemiological investigation although skin, vaginal and rectal colonisation have also been linked to outbreaks.6,11,54 A case definition should be agreed. Time lines are useful to Rates of transmission can be high with throat car- establish overlap, or not, of hospital stays. Detailed patient riage.6,31,43,52,55,61 The post-operative infection rates from histories and in-patient journeys should be explored to an anaesthetist and surgeon with asymptomatic throat col- establish common exposures and timely investigation of onisation reached 7% and 7.5% respectively in patients not possible sources of infection should be undertaken. Pre- receiving prophylaxis.43,55 In obstetric care, two midwives vious investigations of outbreaks are helpful in identifying found to be positive in the throat transmitted GAS to 11 the likely routes of transmission. Other patients, HCWs and cases in an 11 day period.31 the environment are possible sources of outbreaks, as Colonisation and transmission of GAS from the anal, described below. vaginal, perineal areas, skin and nose, without concurrent A retrospective analysis of all GAS infection diagnosed evidence of throat colonisation is well docu- in hospital patients in the past 6 months should be mented.6,41,43,54-57,59,60 Recent Canadian research has re- performed by the IPCT to find links with other cases. vealed that this is the case in 29.5% of implicated cases.6 Prospective microbiological surveillance for further GAS Seven patients suffered post partum GAS infections over cases should be undertaken. Establish contact with the 14 months, strongly associated with one HCW rectally reference laboratory to agree priority for typing of out- colonised.57 A nurse with colonised atopic dermatitis trans- break isolates. mitted GAS to 1.4% of women who delivered while she was

Algorithm 2 Management of an outbreak of GAS infection. 10 J.A. Steer et al. on duty.60 In both these cases family members of the HCW Environment as source of outbreak were also colonised.57,60 Seven cases of post partum GAS (18%) were infected by an obstetrician found to be anally Baths, bidets and showers have all been implicated in 56 colonised. A surgeon with nasal but no throat carriage transmission of GAS during outbreaks. A review of hospital caused GAS infection in a surgical patient and in a second GAS outbreaks identified an environmental source in 9.8% of 58 patient in whom he had only administered a vaccine. outbreaks (6 of 61).6 There are reports of streptococci sur- Although HCWs are more likely to transmit GAS if they viving in dust, and on fomites, and reports of environmental have direct patient contact, it has been suggested that reservoirs being implicated in outbreaks.63 McKee showed spread can occur from ‘cloud HCWs’, who are colonised that, simply by undressing and moving around, an infected rectally, vaginally, or on the skin, by airborne dispersal as surgeon contaminated the environment with GAS.59 59,62 indicated by volumetric or settle plate air cultures. Environmental sources are particularly prominent in Some of these outbreaks may evolve over several maternity outbreaks. In an outbreak involving six cases of 14,57 months. Twenty-eight cases of GAS infection, including GAS over 10 days, all were found to have had a vaginal four cases of iGAS infection and three deaths over a nine- delivery and a daily bath compared to those who were non- month period, occurred in patients being cared for by a wound infected who had had caesarean sections and had not used 14 care team. Longitudinal surveillance by the IPCT is important. the bath.21 In a further maternity outbreak, where infec- tions were restricted to those using the shower, environ- Screening of healthcare workers mental sampling showed the hand held shower to be heavily contaminated and the toilet seat next to the shower The outbreak control team may choose to screen, depend- to be scantily contaminated.22 On a Scottish maternity unit, ing on the circumstances of the outbreak, those with eight mothers and three infants developed GAS infection closest contact first - see Transmission from healthcare and GAS was isolated from a bidet common to all the in- worker to patient. Less than 5% of the adult population fected mothers. None of the mothers who had exclusively are likely to carry GAS in their throat so a positive screening used a separate toilet/bidet were affected.23 result should be considered as indicative of the likely In addition, bathrooms have been implicated in trans- source of transmission and acted upon accordingly until mission of infection on maternity units of both group G 44e47 typing proves otherwise. streptococcus and group C streptococcus.64,65 For asymptomatic HCWs epidemiologically linked to Communal facilities such as baths, bidets and cases of healthcare-associated GAS infection, swabs of showers should be cleaned and decontaminated between throat and skin lesions (including all exfoliating skin condi- patients. The working group agreed that particularly in the tions) should be taken initially. Samples from dry skin lesions case of high risk areas such as delivery suites and post natal should be taken with a moistened swab. Other sites known to care, baths, showers and bidets should be cleaned and be implicated in transmission are nose, anus and vagina, and decontaminated between each patient use. This is partic- screening of these is advised when a HCW is implicated in ularly important because of the high risk of blood and body transmission and throat and skin lesions are negative. fluid contamination, exposed nature of episiotomy wounds Recommended sites are anus, vagina, and anterior nares, and supporting evidence that these communal facilities are as carriage at these sites have all been linked to outbreaks - sources of outbreaks. This should be after all patients see Transmission from healthcare worker to patient. irrespective of whether they are known to be infected Screening is best carried out by an occupational health or not. practitioner for confidentiality reasons and so that HCWs can be examined for skin lesions and dermatitis. Details on the management of HCWs who screen positive for GAS are Recommendations given in section Management of colonised and infected healthcare workers. The method and frequency of cleaning and decon- Symptomatic HCWs should be managed in liaison with tamination of equipment and relevant ward areas the GP and occupational health practitioner. They should should be reviewed. be advised by an occupational health practitioner regarding Communal facilities such as baths, bidets and workplace adjustments and fitness for work. Alternative showers should be decontaminated between all pa- duties in non-clinical settings may be appropriate. tients especially on delivery suites, post-natal wards and other high risk areas, such as burns units. Recommendations SIGN GRADING C

Initial HCW screening should include throat and skin lesions. Environmental sampling HCWs may need to be examined for skin lesions and dermatitis by an occupational health practitioner. If the epidemiological investigation suggests common expo- Other sites known to be implicated in transmission sure to a potential environmental source, relevant environ- are nose, anus, and vagina, and screening of these mental sampling should be undertaken.66 Large sterile sites is advised when a HCW is implicated in trans- swabs of gauze, moistened in 0.9% sodium chloride, wiped mission and throat and skin lesions are negative. across a large part of the surface of the implicated equip- SIGN GRADING D ment and then placed into broth, is more likely to yield UK guidelines on prevention and control of GAS in healthcare settings 11 positive results than small areas swabbed with standard medical swabs. Decontaminating the equipment is impera- Recommendation tive before further use and if feasible, the equipment should be taken out of use whilst awaiting the results of cultures. Recommendations for chemoprophylaxis should be Decontamination may have taken place before sampling made by the outbreak control team on a case by leading to false negative results. In this instance, a risk as- case basis. sessment of the continued use of the equipment including SIGN GRADING D the type and frequency of decontamination should be reviewed. Communication strategy

Recommendation Patients, close contacts and HCWs should be provided with clear, concise information about the outbreak e see Appendix In a possible outbreak environmental sources of 2 and 3. As an important part of prevention is enhanced sur- transmission should be considered and relevant sam- veillance, information should be provided to hospital medical pling undertaken. or surgical teams and primary care teams, as appropriate, to SIGN GRADING D encourage heightened awareness of the symptoms of GAS, to take specimens from potentially infected wounds, to give early treatment where GAS is suspected, and to promptly no- Screening of patients tify the outbreak control team - see Prospective and retro- spective surveillance. Screening patients can establish the extent of the outbreak It should be noted that iGAS can cause press interest and and identify patients at risk of subsequent GAS disease. The press office advice should be obtained in these circum- extent of contact tracing, including patients who have been stances. It may require an ongoing press briefing in some discharged, should be decided by the outbreak control circumstances. team and will be dependent on the circumstances of the outbreak and whether a source has been identified. The Recommendations normally short incubation period of GAS infection (1e3 days) should be taken into consideration. Many patients will Patients, close contacts and HCWs should be pro- have received antibiotic treatment for other conditions vided with clear, concise information about the which would have covered GAS during the period in outbreak. question. In many instances, information regarding the Information should be provided to relevant HCWs to nature of infections and the likely symptoms can be given encourage heightened awareness of the symptoms of to patients deemed at risk rather than chemoprophylaxis. GAS, to take specimens from symptomatic patients, The disadvantage of screening when a source has not been give early treatment where GAS is suspected, and identified is that acquisition of the outbreak strain may promptly notify the outbreak control team. subsequently occur after screening. The advantage is that Consider active involvement of a press officer to deal it may help to elucidate the source more clearly. with media enquiries. SIGN GRADING Good practice points Use of chemoprophylaxis

Chemoprophylaxis aims to eradicate carriage in those who Management of colonised and infected have newly acquired the invasive strain and who may healthcare workers themselves be at risk of infection. In the community setting, chemoprophylaxis is only Treatment of infection recommended for the entire household if two or more cases of iGAS infection occur in the same household within HCWs with suspected or confirmed GAS infection e.g. phar- a 30-day time period. About 2000 contacts need to be yngitis or infected skin lesions, pose a potential risk to treated to prevent one case, assuming 100% efficacy.9 In patients both through direct transmission of the organism to the healthcare setting, routine chemoprophylaxis is not patients or indirectly through contamination of the hospital recommended following a single case, except in mother environment. The degree of risk to patients will be dependent or baby cases during the neonatal period.9 However, during upon closeness of patient contact involved in the HCW’s duties an outbreak, recommendations for chemoprophylaxis can and HCWs with symptoms of GAS infection pose a greater risk be made by the outbreak control team according to the than asymptomatic carriers.67 All HCWs with symptoms of pos- specific scenario, taking into account the nature of the out- sible GAS infection should inform and seek advice from their break - number of cases, severity of cases, vulnerability of line manager and/or contact occupational health for guidance patients, and source of the outbreak. The length of prophy- on performance of clinical duties on the basis of a risk assess- laxis and the choice of antimicrobial should be agreed lo- ment. Treatment of infection should be undertaken in liaison cally based on the clinical circumstances, whether the with the HCW’s GP.Where a risk assessment has indicated that source has been identified and eliminated, the susceptibil- an HCW’s infection poses a risk to patients, the HCW should be ity of the patients, and the likely site of infection if it excluded from work until 24 h of appropriate treatment and occurred. resolution of symptoms has occurred. 12 J.A. Steer et al.

Eradication of carriage associated risks (such as selection of difficile) outweighed the benefits although they may be appropriate There are no randomised controlled trials that establish the in some cases as directed by an infection specialist. most appropriate first-line therapy in the HCW setting, Decisions regarding antimicrobial therapy should be based where timely eradication of carriage is important. Further on microbiological principles including reliable absorption of transmission following failure of eradication or re-infection the antibiotic, site of colonisation, tolerability, and risk of is well documented, occurring in four out of four cases in side effects. Treatment regimes obtained from the literature one review.6 Eradication of carriage is therefore recom- include combinations of a penicillin, a macrolide or clinda- mended in all cases where onward transmission of GAS mycin with rifampicin or oral vancomycin (Table 1). has occurred (see Algorithm 3 for treatment options). Clearance screens should be taken 24 h after the end of In the general population, penicillin V results in eradi- treatment and again at one, three, six, and twelve weeks cation of GAS in around 80% of cases.68,69 The failure rate in post treatment. If persistent or recurrent GAS colonisation terms of eradication increases with duration of follow up is found in the HCW early consideration should be given to and found to be as high as 65% at 3 months in one study.70 possible sources of re-colonisation within the HCW’s house- Clindamycin eradicates colonisation in 100% of patients hold. Screening throat and skin lesions of the HCWs’ close who have failed penicillin therapy for GAS throat carriage child household contacts and throat, skin lesion, anal and at 4e6 days,69 but eradication success at nine weeks post vaginal swabs of close adult household contacts should be treatment may fall to 85%.71 It is not clear whether this re- considered initially. flects failure of eradication or re-infection. There is no ro- The exact duration of follow up to ensure clearance has bust evidence on which to base recommendations for not been reliably established, particularly as HCWs may 54,59,78 eradication of vaginal and anal colonisation. become re-colonised from a close household contact. Although there is evidence suggesting that cephalospo- In some cases it may be felt appropriate to screen for lon- rins may produce higher clearance rates than pen- ger than that recommended above, particularly if colonisa- 56 icillins,72e75 the working group felt that the evidence did tion has been difficult to eradicate. not justify a recommendation for their use as a standard Whilst eradication is not essential for asymptomatic first line therapy for eradication; it was felt that the HCWs carrying GAS strains different from the case or

Algorithm 3 Management of colonised and infected healthcare workers by occupational health. UK guidelines on prevention and control of GAS in healthcare settings 13

Table 1 Antibiotic regimens used to eradicate healthcare worker carriage in published reports. Antimicrobial used Cases Pharyngeal carriage cleared Non-pharyngeal carriage cleared Penicillin alone 12 3/3 6/9 Penicillin plus rifampicin 3 2/2 0/1 Penicillin plus vancomycin 1 0/0 1/1 Macrolide with/without rifampicin 2 2/2 0/0 Clindamycin with/without rifampicin 2 1/1 1/1 Vancomycin plus Rifampicin 1 0/0 1/1 Not named 3 1/1 2/2 Two HCWs treated with penicillin alone had primary treatment failure; 1 HCW treated with penicillin alone and 1 treated with penicillin and rifampicin initially cleared their carriage but were identified as re colonized with the same strain 4 and 15 months later, respec- tively. Retreatment attempts were ultimately successful in all 4. Successful re-treatment regimens were vancomycin plus rifampicin followed by long-term low-dose penicillin “prophylaxis,” intramuscular penicillin given monthly for 1 year, intramuscular penicillin plus oral vancomycin, and intravenous penicillin plus vancomycin.54,56,76,77 Taken from Daneman et al, surveillance for hospital outbreaks of invasive group A streptococcal infections in Ontario, Canada, 1992 to 2000. Ann Intern Med August 21, 2007 147 with permission from the Annals of Internal Medicine.6 outbreak strain, IPCT and Occupational Health should risk personal contacts, whether the contacts are symptomatic assess return to work. or not, could be considered where failure of eradication/ re-colonisation occurs as carriage in such contacts may thwart attempts at long-term eradication. Recommendations If GAS cannot be eradicated from HCWs, and close personal contacts screen negative, there is some evidence to suggest that pets have been implicated in re-infection HCW contacts who have been screened and found to 79e82 and this should be considered. be positive for GAS should receive eradication In complex cases, such as failure of eradication or therapy. ongoing transmission resulting in lengthy exclusions or re- Clearance screens should be taken 24 h after com- deployment, senior management should perform and doc- pleting treatment, and again at 1, 3, 6, and 12 weeks ument a risk assessment based on supporting evidence from following the end of treatment infection control and occupational health. SIGN GRADING D Pharyngeal carriage: Treatment options include oral penicillin V (500 mg four times a day for 10 days), amoxicillin (500 mg Recommendations three times a day for 10 days), clindamycin (300 mg four times a day for 10 days), or azithromy- Persistent or recurrent GAS colonisation may indi- cin (maximum dose of 500 mg once a day) for 3 days. cate re-colonisation within the household. Screening Clindamycin (300 mg four times a day for 10 days) of household contacts should be considered in such should be used for eradication of throat carriage in circumstances. cases where first-line therapy with penicillin has When considered necessary by the IPCT or occupa- been unsuccessful. tional health physician, the health protection spe- SIGN GRADING D cialist should liaise with GPs regarding screening Non-pharyngeal carriage: and treatment of close household contacts of HCWs infected or colonised with GAS. Penicillin treatment alone may not be sufficient. SIGN GRADING D Treatment options include clindamycin 300 mg four times a day for 10 days, or azithromycin 12 mg per kg per day (maximum 500 mg once a day) for 5 Length of exclusion from work days with some limited reports in literature of com- bining with oral rifampicin or oral vancomycin. Asymptomatic throat carriage and pharyngitis SIGN GRADING D Following a single case of GAS infection, HCWs with throat carriage on screening should stay away from clinical work until at least 24 h of appropriate treatment if asymptomatic, Failure of eradication and, if symptomatic, until at least 24 h of appropriate treatment and resolution of symptoms has occurred. The ma- Close personal contacts can be the source of GAS to HCWs jority of individuals (96%) with pharyngeal carriage will be implicated in healthcare-associated transmission.54,59,63,78 culture-negative 24 h after starting treatment.(Padfield per- If family members of an infected HCW are symptomatic, sonal communication) The rate of successful eradication de- swabbing and antibiotic treatment should ideally be under- creases over time, with some individuals relapsing many taken in liaison with the relevant GP. Swabbing of close weeks after apparent successful eradication.71 14 J.A. Steer et al.

In circumstances where carriage has been linked to an emm-typing is the molecular gold standard for typing GAS outbreak or confirmed transmission, the duration of exclu- and there are currently more than 180 emm types described sion from work should be decided on a case-by-case basis globally. The determination of emm/M type together with and will depend on the clinical situation, the likelihood and the identification of opacity factor and the T are associated risk of further transmission, the site of coloni- the key ‘markers’ for both phenotypic and molecular typ- sation and evidence of previous transmission. ing.83 Further sub-typing may be required to define more Eradication of throat carriage (and therefore follow up clearly a potential outbreak and to monitor the investigation screening) and continued exclusion from work are not essen- of a potential outbreak, including emm sequence sub- tial if typing later excludes the HCW from being implicated in typing, and other methods based on sequencing and gene transmission but should be subject to a local risk assessment. polymorphisms. Whilst results from emm typing of GAS iso- lates from clinical cases and screening isolates forms an es- Skin lesions or other site colonisation sential part of outbreak investigation by providing further HCWs with active skin lesions are at increased risk of evidence of likely transmission routes, broader epidemiolog- colonisation and shedding, and have been particularly asso- ical investigations and control measures, including the issu- ciated with intra-hospital spread, including in the delivery 14,40,53,60 ing of prophylaxis where warranted, should not await typing suite. A longer length of exclusion from work is re- results given the additional delays that might be incurred. quired for HCWs with skin lesions, as time will be required Rapid tests for the diagnosis of GAS are available; for any infected skin lesions to heal, or in the case of derma- however, negative results should be confirmed by culture. titis, for optimal resolution of the . The HCW The working group felt more work was needed to evaluate should be reviewed by an occupational health physician, the advantage over 24 h culture in clinical settings, given advice regarding good skin care, and referred to a der- particularly in the detection of asymptomatic carriage, matologist if required. In general, these HCWs should not do before recommendations could be made. clinical work until eradication is felt to have been effective. Enquiries should be made as to whether any close personal contacts are suffering from conditions that may be related Recommendations to GAS infection, as re-colonisation from a close personal con- tact will frustrate efforts to clear carriage. Occupational GAS isolates from invasive disease should be referred health should liaise with the relevant GP if this is the case. to the reference laboratory for typing. The reference laboratory should be contacted if an outbreak is being investigated. Recommendations SIGN GRADING Good practice points Save all GAS isolates from in-patients, peri-partum patients, neonates, and those from post-operative HCWs with symptomatic GAS pharyngitis should stay wounds for six months. away from clinical work until at least 24 h of appro- SIGN GRADING D priate therapy and resolution of symptoms has oc- curred. Asymptomatic HCWs should stay away from work until 24 h of appropriate therapy. Applicability to other settings A longer period of time may be required for HCWs with skin lesions or in other circumstances where The principles included in these guidelines could be usefully carriage has been linked to an outbreak or confirmed used for the investigation of GAS infection following home transmission. This should be at the discretion of the birth. Elements of this guidance could be usefully applied to IPCT team in liaison with the occupational health other settings such as care homes given the similarities with practitioner and discussed on a case-by-case basis hospitals in terms of vulnerability of the resident popula- after a risk assessment. tions, close proximity of healthcare providers to residents SIGN GRADING Good practice points and existence of communal facilities.88 This is borne out by a study in the USA which found residents of long-term care Microbiological investigation facilities for the elderly have six times the risk of developing iGAS infection than elderly counterparts residing in the com- munity and over one and half times the risk of death.84 Cases All GAS isolates from in-patients, peri-natal patients and arising in nursing or residential homes are more likely to form neonates, or identified as being from the immediate post part of an outbreak, as is the case for hospital settings, given discharge period e.g. post-operative wound swabs from the potential for onward transmission in such settings.6,84 general practice, should be saved by the microbiology The 2004 community guidelines provide further advice on laboratory for at least six months. This is important to contact management in relation to cases of invasive disease capture all healthcare-associated GAS infections, including arising in institutional settings.9 those that may occur after discharge, for retrospective analysis in the event of a potential outbreak. Applicability to group C/G beta haemolytic All invasive isolates of GAS should be sent to the HPA Respiratory and Systemic Infections Laboratory, Streptococcus streptococci and Diphtheria Reference Unit (SDRU) as part of the ongoing surveillance of invasive disease due to GAS. Clinical and Outbreaks of puerperal sepsis caused by Lancefield group C demographic details should be provided on the referral form. and G streptococci beta-hemolytic streptococci have been UK guidelines on prevention and control of GAS in healthcare settings 15 described in the literature, albeit less commonly than for Review of guidance group A streptococci.64,65,85,86 Group C and G outbreaks have also been documented in burns units, general hospital 86 The Working Group will consider updating these guidelines wards, and outpatient clinics. Outbreak investigations three years from publication. The decision on whether a full, have suggested transmission occurs through similar mecha- partial or no update is needed will be made on the basis of nisms as group A streptococci, including from colonized whether new evidence has emerged since the evidence healthcare workers to patients, although with a stronger review was undertaken that would alter the recommenda- emphasis on environmental sources, such as contaminated 64,65,85,86 tions contained within the guidelines, or whether any changes douches, showers and toilet seats. In each in- in healthcare practice or organisation have been imple- stance, the outbreaks were controlled through strict infec- mented which alter the execution of the recommendations. tion control procedures and disinfection. Although there is less information on the incidence, transmission mechanisms Suggestions for further research and control of group G and C streptococcal outbreaks in hospital and maternity settings, given the similarities to group A streptococcal outbreaks and potential conse- Further research is needed on the use of rapid tests for quences of infection, extension of these guidelines to group GAS, particularly for the detection of asymptomatic car- C and G streptococcal strains would be a reasonable riage. The eradication of GAS from chronic carriers de- approach. serves further attention also.

Working group membership

Association of Medical Microbiologists M. Morgan, Consultant Medical Microbiologist, (now British Infection Association) Royal Devon and Exeter Hospital, HPA Regional Microbiology Network British Infection Society S. Sriskandan, Consultant in Infectious Diseases, Centre for Infection Prevention & Management, Department of Infectious Diseases, Imperial College London British Society of Antimicrobial Chemotherapy M. Dryden, Consultant in Microbiology and Communicable Disease Department of Health Carole Fry (Department of Health, Observer) Faculty of Occupational Medicine E. Murphy, Lead Consultant Occupational Health Physician, of the Royal College of Physicians NHS Grampian Occupational Health Service Health Protection Scotland J. McMenamin, Consultant Epidemiologist Health Protection Agency R. Cordery, Consultant in Communicable Disease Control, HPA NENC London Health Protection Unit A. Efstratiou, Unit Head, Streptococcus & Diphtheria Reference Unit, HPA Microbiology Services R. George, Director, Respiratory & Systemic Infections Department, HPA Microbiology Services J. Kearney (chair), Regional Director, HPA East of England T. Lamagni, Senior Epidemiologist, Healthcare Associated Infection & Antimicrobial Resistance Department, HPA Health Protection Services I. Oliver, Regional Director, HPA South West B. Rao, Healthcare Associated Infection & Antimicrobial Resistance Department, HPA Health Protection Services G. Rooke (Project Manager), HPA East of England Hospital Infection Society J. Steer, Consultant Microbiologist, Department of Microbiology, Derriford Hospital, Plymouth Infection Prevention Society F. Baker, Lead CNS Infection Control, North Devon District Hospital Lee Spark NF Foundation D. Marsden, Lee Spark NF Foundation Public Health Agency Northern Ireland N. Irvine, Consultant in Health Protection, Public Health Agency, Northern Ireland Public Health Medicine Environmental Group A. Cummins, CCDC, HPA Essex Health Protection Unit Public Health Wales B. Healy, Consultant in Microbiology & Infectious Diseases, Department of Microbiology, Public Health Wales, Cardiff Royal College of Midwives M. Jokinen, Practice and Standards, Development Co-ordinator, Development Department Royal College of Obstetrics & Gynaecology R. Hughes, Lead Obstetrician for Lothian, Royal Infirmary, Edinburgh United Kingdom Clinical Pharmacy P. Wade, Consultant Pharmacist - Infectious Diseases, Association - Infection Management Group Directorates of Pharmacy and Infection, Guy’s & St. Thomas’ NHS Foundation Trust, London 16 J.A. Steer et al.

Consultation process 8. Cantwell R, Clutton-Brock T, Cooper G, Dawson A, Drife J, Garrod D, et al. Saving mothers’ Lives: Reviewing maternal deaths to make motherhood safer: 2006-2008. The Eighth re- In accordance with the Health Protection Agency Policy and port of the Confidential enquiries into maternal deaths in the Guidance on the Development and Delivery of High Level United Kingdom. BJOG 2011;118(Suppl 1):1e203. Scientific Advice (OP001), these guidelines were open to 9. Health Protection Agency Group A Streptococcus Working public consultation for a three month period (14 May to 6 Group. Interim UK guidelines for management of close commu- August 2010).87 All comments received were shared with nity contacts of invasive group A streptococcal disease. Com- Working Group members with the designated senior respon- mun Dis Public Health 2004;7(4):354e61. sible officer (SRO), Dr Joe Kearney, taking responsibility for 10. Yamada T, Yamada T, Yamamura MK, Katabami K, Hayakawa M, analysing and responding to all comments. Tomaru U, et al. Invasive group A streptococcal infection in pregnancy. J Infect 2010;60(6):417e24. 11. Prevention of Invasive Group A Streptococcal Infections Work- Acknowledgements shop Participants. Prevention of invasive group A streptococcal disease among household contacts of case patients and among postpartum and postsurgical patients: recommendations from The Working Group extends its sincere gratitude to the the Centers for Disease Control and Prevention. Clin Infect following individuals whose input into the development of Dis 2002;35(8):950e9. these guidelines is greatly appreciated e Ms Lynsey Emmet, 12. Scottish Intercollegiate Guidelines Network. SIGN 50: a guide- HPA East of England Regional Epidemiology Unit; Dr Simon line developer’s handbook. Edinburgh: Scottish Intercollegiate Padfield, North Yorkshire & the Humber Health Protection Guidelines Network; 2008. Unit; Prof. Anne Bouvet, Centre National de Re´fe´rence des 13. Health protection (Notification) Regulations 2010 (Statutory Streptocoques, Paris; Dr Ibrahim Hassan, Wythenshawe Instruments 2010 No 659): England 2010. Hospital; Prof. Maria Zambon, HPA Executive Sponsor for 14. Felkner M, Pascoe N, Shupe-Ricksecker K, Goodman E. The the guideline development; Mr Grahame Antony Short, H.M. wound care team: a new source of group A streptococcal nos- ocomial transmission. Infect Control Hosp Epidemiol 2005; Coroner for the County of Hampshire. Finally, we extend 26(5):462e5. our thanks to the many individuals who took the time to 15. Kwantes W, James JR. Haemolytic streptococci on the neona- read and comment on the draft guidelines during the open tal umbilicus. Br Med J 1956;2(4992):576e8. consultation. 16. Lacy MD, Horn K. Nosocomial transmission of invasive group A streptococcus from patient to health care worker. Clin Infect Dis 2009;49(3):354e7. 17. Semmelweis I. Die aetiologie, der begriff und die prophylaxis Appendix. Supplementary material des kindbettfiebers. Pest: C.A. Hartleben’s Verlags-Expedition; 1861. Supplementary material associated with this article can be 18. World Health Organisation. WHO guidelines on hand hygiene in found, in the online version, at doi:10.1016/j.jinf.2011.11. health care 2009. 001. 19. Clean Safe Care. High Impact Intervention No. 8: care bundle to improve the cleaning and decontamination of clinical equipment. Department of Health, http://hcai.dh.gov.uk/ References files/2011/03/2011-03-14-HII-Cleaning-and-decontamination- FINAL.pdf; 2010. 1. Lamagni TL. The epidemiology of severe Streptococcus pyogenes 20. Hoffman P, Bradley C, Ayliffe G. Disinfection in healthcare. 3rd disease in Europe [dissertation]. University of Helsinki, 2008. ed. London: Health Protection Agency; 2004. Available from: http://urn.fi/URN:ISBN:978-951-740-885-1. 21. Wiesenthal AM. A maternal-neonatal outbreak of infections 2. Lamagni TL, Neal S, Keshishian C, Powell D, Potz N, Pebody R, due to an unusual group A beta-hemolytic streptococcus. In- et al. Predictors of death after severe Streptococcus pyogenes fect Control 1984;5(6):271e4. infection. Emerg Infect Dis 2009;15(8):1304e7. 22. Claesson BE, Claesson UL. An outbreak of endometritis in a ma- 3. O’Loughlin RE, Roberson A, Cieslak PR, Lynfield R, Gershman K, ternity unit caused by spread of group A streptococci from Craig A, et al. The epidemiology of invasive group A strepto- a showerhead. J Hosp Infect 1985;6(3):304e11. coccal infection and potential vaccine implications: United 23. Gordon G, Dale BA, Lochhead D. An outbreak of group A States, 2000-2004. Clin Infect Dis 2007;45(7):853e62. haemolytic streptococcal puerperal sepsis spread by the 4. O’Grady KA, Kelpie L, Andrews RM, Curtis N, Nolan TM, Selvaraj G, communal use of bidets. Br J Obstet Gynaecol 1994; et al. The epidemiology of invasive group A streptococcal disease 101(5):447e8. in Victoria, Australia. Med J Aust 2007;186(11):565e9. 24. Safe management of healthcare waste version 1.0. Department 5. Lamagni TL, Neal S, Keshishian C, Alhaddad N, George R, of Health, http://www.dh.gov.uk/en/Publicationsandstatistics/ Duckworth G, et al. Severe Streptococcus pyogenes infections, Publications/PublicationsPolicy AndGuidance/DH_126345; 2011 United Kingdom, 2003-2004. Emerg Infect Dis 2008;14(2): [cited 8/2/2011]. 201e9. 25. NHS Standard Service Level Specifications. Specific service 6. Daneman N, Green KA, Low DE, Simor AE, Willey B, Schwartz B, Specification e Waste management and Disposal. Department et al. Surveillance for hospital outbreaks of invasive group A of Health [serial online], http://www.dh.gov.uk/en/Manag streptococcal infections in Ontario, Canada, 1992 to 2000. ingyourorganisation/NHSprocurement/Publicprivatepartner Ann Intern Med 2007;147(4):234e41. ship/Privatefinanceinitiative/Newstandardoutputspecificatio 7. Cordery R, Efstratiou A, George RC, Cohuet S, Lamagni TL. In- ns/DH_4016183; 2006 [cited 8/2/2011]. vasive group A streptococcal disease in maternity settings: 26. Health and Safety Executive. Managing offensive/hygiene time to reassess case and cluster management? XVII Lance- waste 2009. field. International Symposium on Streptococci and Strepto- 27. Department of Health. Hospital Laundry arrangements for coccal Diseases June 2008 [Porto Heli, Greece]. used and infected linen. London: Department of Health; 1995. UK guidelines on prevention and control of GAS in healthcare settings 17

28. Colebrook DC. The source of infection in puerperal fever due 48. Mead PB, Winn WC. Vaginal-rectal colonization with group A to haemolytic streptococci. London: HMSO. Med Research streptococci in late pregnancy. Infect Dis Obstet Gynecol Council, Spec Rep Series No.205; 1935. 2000;8(5e6):217e9. 29. Colebrook L. The story of puerperal fever; 1800 to 1950. BMJ 49. Hassan IA, Onon TS, Weston D, Isalska B, Wall K, Afshar B, et al. 1956;1(4961):247e52. A quantitative descriptive study of the prevalence of carriage 30. Stefonek KR, Maerz LL, Nielsen MP, Besser RE, Cieslak PR. (colonisation) of haemolytic streptococci groups A, B, C and Group A streptococcal puerperal sepsis preceded by posi- G in pregnancy. J Obstet Gynaecol 2011;31(3):207e9. tive surveillance cultures. Obstet Gynecol 2001;98(5 Pt 1): 50. Hawkey PM, Pedler SJ, Southall PJ. Streptococcus pyogenes: 846e8. a forgotten occupational hazard in the mortuary. Br Med J 31. Gonzalez-Rey C, Belin AM, Jorbeck H, Norman M, Krovacek K, 1980;281(6247):1058. Henriques B, et al. RAPD-PCR and PFGE as tools in the investi- 51. Bakhshi SS. Code of practice for funeral workers: managing in- gation of an outbreak of beta-haemolytic Streptococcus group fection risk and body bagging. Commun Dis Public Health 2001; A in a Swedish hospital. Comp Immunol Microbiol Infect Dis 4(4):283e7. 2003;26(1):25e35. 52. Raymond J, Schlegel L, Garnier F, Bouvet A. Molecular charac- 32. Verboon-Maciolek MA, Krediet TG, van EI, Gerards LJ, Fleer A. terization of Streptococcus pyogenes isolates to investigate an Severe neonatal group A streptococcal disease. Eur J Pediatr outbreak of puerperal sepsis. Infect Control Hosp Epidemiol 2000;159(6):450e2. 2005;26(5):455e61. 33. Shanks GD, Anderson RT, Lazoritz S, Hemming VG. Bilateral 53. McGregor J, Ott A, Villard M. An epidemic of "childbed fever". neonatal group A streptococcal hydrocele infection associated Am J Obstet Gynecol 1984;150(4):385e8. with maternal puerperal sepsis. Pediatr Infect Dis 1986;5(1): 54. Berkelman RL, Martin D, Graham DR, Mowry J, Freisem R, 107e8. Weber JA, et al. Streptococcal wound infections caused by 34. Casenova-Roman M, Rios J, Sanchez-Porto A, Casenova- a vaginal carrier. JAMA 1982;247(19):2680e2. Bellido M. Case report - newborn bacteraemia with GAS and or- 55. Paul SM, Genese C, Spitalny K. Postoperative group A beta- ganism also grown from mother’s locia. Minerva Pediatr 2002; hemolytic Streptococcus outbreak with the pathogen traced 54:161e3. to a member of a healthcare worker’s household. Infect Con- 35. Panaro NR, Lutwick LI, Chapnick EK. Intrapartum transmission trol Hosp Epidemiol 1990;11(12):643e6. of group A streptococcus. Clin Infect Dis 1993;17(1):79e81. 56. Viglionese A, Nottebart VF, Bodman HA, Platt R. Recurrent 36. Bingen E, Denamur E, Lambert-Zechovsky N, Boissinot C, group A streptococcal carriage in a health care worker associ- Brahimi N, Aujard Y, et al. Mother-to-infant vertical transmis- ated with widely separated nosocomial outbreaks. Am J Med sion and cross-colonization of Streptococcus pyogenes con- 1991;91(3B):329Se33S. firmed by DNA restriction fragment length polymorphism 57. Nosocomial group. A streptococcal infections associated with analysis. J Infect Dis 1992;165(1):147e50. asymptomatic health-care workerseMaryland and California, 37. Mahieu LM, Holm SE, Goossens HJ, Van Acker KJ. Congenital 1997. MMWR Morb Mortal Wkly Rep 1999;48(8):163e6. streptococcal toxic shock syndrome with absence of 58. Rutishauser J, Funke G, Lutticken R, Ruef C. Streptococcal against streptococcal pyrogenic exotoxins. J Pediatr 1995; toxic shock syndrome in two patients infected by a colonized 127(6):987e9. surgeon. Infection 1999;27(4e5):259e60. 38. Health Protection Agency Meningococcus and Haemophilus Fo- 59. McKee WM, Di Caprio JM, Roberts Jr CE, Sherris JC. Anal car- rum. Guidance for public health management of meningo- riage as the probable source of a streptococcal epidemic. Lan- coccal disease in the UK. Health Protection Agency, http:// cet 1966;2(7471):1007e9. www.hpa.org.uk/web/HPAwebFile/HPAweb_C/119494 60. Ejlertsen T, Prag J, Pettersson E, Holmskov AA. 7-month out- 7389261; 2011 [cited 8/2/2011]. break of relapsing postpartum group A streptococcal infections 39. Kakis A, Gibbs L, Eguia J, Kimura J, Vogelei D, Troup N, et al. linked to a nurse with atopic dermatitis. Scand J Infect Dis An outbreak of group A Streptococcal infection among health 2001;33(10):734e7. care workers. Clin Infect Dis 2002;35(11):1353e9. 61. Daneman N, McGeer A, Low DE, Tyrrell G, Simor AE, 40. Chandler RE, Lee LE, Townes JM, Taplitz RA. Transmission of McArthur M, et al. Hospital-acquired invasive group A strepto- group A Streptococcus limited to healthcare workers with ex- coccal infections in Ontario, Canada, 1992-2000. Clin Infect Dis posure in the operating room. Infect Control Hosp Epidemiol 2005;41(3):334e42. 2006;27(11):1159e63. 62. Sherertz RJ, Bassetti S, Bassetti-Wyss B. "Cloud" health-care 41. Lannigan R, Hussain Z, Austin TW. Streptococcus pyogenes as workers. Emerg Infect Dis 2001;7(2):241e4. a cause of nosocomial infection in a critical care unit. Diagn 63. Falck G, Kjellander J. Outbreak of group A streptococcal infec- Microbiol Infect Dis 1985;3(4):337e41. tion in a day-care center. Pediatr Infect Dis J 1992;11(11):914e9. 42. Hagberg C, Radulescu A, Rex JH. Necrotizing fasciitis due to 64. Haynes J, Anderson AW, Spence WN. An outbreak of puerperal group A streptococcus after an accidental needle-stick injury. fever caused by group G streptococci. J Hosp Infect 1987;9(2): N Engl J Med 1997;337(23):1699. 120e5. 43. Kolmos HJ, Svendsen RN, Nielsen SV. The surgical team as 65. Teare EL, Smithson RD, Efstratiou A, Devenish WR, Noah ND. An a source of postoperative wound infections caused by. Strepto- outbreak of puerperal fever caused by group C streptococci. coccus pyogenes. J Hosp Infect 1997;35(3):207e14. J Hosp Infect 1989;13(4):337e47. 44. Ditchburn R, Ditchburn J. Rate of carriage of group A beta hae- 66. Health Protection Agency. Examining food, water and environmen- molytic streptococci. BMJ 1995;311(6998):193. tal samples from healthcare environments. Health Protection 45. Spitzer J, Hennessy E, Neville L. High group A streptococcal Agency, http://www.hpa.org.uk/web/HPAwebFile/HPAweb_C/ carriage in the Orthodox Jewish community of north Hackney. 1296681857486; 2010. Br J Gen Pract 2001;51(463):101e5. 67. Wannamaker LW. The epidemiology of streptococcal infec- 46. Hoffmann S. The throat carrier rate of group A and other beta tions. In: McCarty M, editor. Streptococcal infections. New hemolytic streptococci among patients in general practice. York: Columbia University Press; 1953. p. 157e75. Acta Pathol Microbiol Immunol Scand B 1985;93(5):347e51. 68. Gastanaduy AS, Kaplan EL, Huwe BB, McKay C, 47. Stromberg€ A, Schwan A, Cars O. Throat carrier rates of beta- Wannamaker LW. Failure of penicillin to eradicate group A hemolytic streptococci among healthy adults and children. streptococci during an outbreak of pharyngitis. Lancet 1980; Scand J Infect Dis 1988;20(4):411e7. 316(8193):498e502. 18 J.A. Steer et al.

69. Orrling A, Stjernquist-Desatnik A, Schalen C, Kamme C. Clinda- 79. Wilson KS, Maroney SA, Gander RM. The family pet as an un- mycin in persisting streptococcal pharyngotonsillitis after pen- likely source of group A beta-hemolytic streptococcal infection icillin treatment. Scand J Infect Dis 1994;26(5):535e41. in humans. Pediatr Infect Dis J 1995;14(5):372e5. 70. Edmond EW, Cramblett HG, Siewers CM, Crews J, Ellis B, 80. Crowder HR, Dorn CR, Smith RE. Group A Streptococcus in pets Jenkins GR. Comparison of efficacy of phenoxymethyl penicil- and group A streptococcal disease in man. Int J Zoonoses 1978; lin and buffered penicillin G in treatment of streptococcal 5(1):45e54. pharyngitis. J Pediatr 1966;68(3):442e7. 81. Copperman SM. Cherchez le chien: household pets as reservoirs 71. Tanz RR, Poncher JR, Corydon KE, Kabat K, Yogev R, Shulman ST. of persistent or recurrent streptococcal sore throats in chil- Clindamycin treatment of chronic pharyngeal carriage of group dren. N Y State J Med 1982;82(12):1685e7. A streptococci. J Pediatr 1991;119(1 ( Pt 1)):123e8. 82. Mayer G, Van Ore S. Recurrent pharyngitis in family of four. 72. Pichichero M, Casey J. Comparison of European and U.S. results Household pet as reservoir of group A streptococci. Postgrad for cephalosporin versus penicillin treatment of group A strep- Med 1983;74(1):277e9. tococcal tonsillopharyngitis. Eur J Clin Microbiol Infect Dis 83. Efstratiou A. Group A streptococci in the 1990s. J Antimicrob 2006;25(6):354e64. Chemother 2000;45(Suppl):3e12. 73. Casey JR, Pichichero ME. Meta-analysis of cephalosporin versus 84. Thigpen MC, Richards Jr CL, Lynfield R, Barrett NL, penicillin treatment of group A streptococcal tonsillopharyng- Harrison LH, Arnold KE, et al. Invasive group A streptococcal in- itis in children. Pediatrics 2004;113(4):866e82. fection in older adults in long-term care facilities and the com- 74. Casey JR, Pichichero ME. Meta-analysis of cephalosporins ver- munity, United States, 1998-2003. Emerg Infect Dis 2007; sus penicillin for treatment of group A streptococcal tonsillo- 13(12):1852e9. pharyngitis in adults. Clin Infect Dis 2004;38(11):1526e34. 85. Galloway A, Noel I, Efstratiou A, Saint E, White DR. An out- 75. Casey JR, Pichichero ME. The evidence base for cephalosporin break of group C streptococcal infection in a maternity unit. superiority over penicillin in streptococcal pharyngitis. Diagn J Hosp Infect 1994;28(1):31e7. Microbiol Infect Dis 2007;57(3 Suppl):39Se45S. 86. Efstratiou A. Outbreaks of human infection caused by pyogenic 76. Schaffner W, Lefkowitz Jr LB, Goodman JS, Koenig MG. Hospi- streptococci of Lancefield groups C and G. J Med Microbiol tal outbreak of infections with group a streptococci traced to 1989;29(3):207e19. an asymptomatic anal carrier. N Engl J Med 1969;280(22): 87. HPA. Open consultation on new public health guidelines on 1224e5. the management and control of group A streptococcal infec- 77. McIntyre DM. An epidemic of Streptococcus pyogenes puerperal tions in hospitals in the UK. Health Protection Report [serial and postoperative sepsis with an unusual carrier siteethe anus. online] 2010;4(19): news. Available from: http://www.hpa. Am J Obstet Gynecol 1968;101(3):308e14. org.uk/hpr/archives/2010/news1910.htm [cited 14 May 2010]. 78. Kaan JA, van Dijk Y, Mascini EM, van Kessel RP, Schellekens JF. A 88. Cummins A, Millership S, Lamagni T, Foster K. Control measures midwife involved in patients with puerperal fever in three dif- for Invasive Group A Streptococci (iGAS) outbreaks in care ferent hospitals. Ned Tijdschr Geneeskd 2008;152(41):2245e8. homes. J Infect (in press).

Journal of Antimicrobial Chemotherapy (2004) 53, Suppl. S2, ii37–ii50 DOI: 10.1093/jac/dkh202

Complicated infections of skin and skin structures: when the infection is more than skin deep

Mark J. DiNubile1* and Benjamin A. Lipsky2

1Merck Research Laboratories, BL 3–4, PO Box 4, West Point, PA 19486, USA; 2University of Washington School of Medicine and Veterans Administration Puget Sound Health Care System, Seattle, WA, USA

Skin and skin-structure infections are common, and range from minor to severe necrotizing infections. Complicated infections are defined as involving abnormal skin or wounds, occurring in com- promised hosts, or requiring surgical intervention. Classification schemes for these infections are varied Downloaded from and confusing. Distinguishing characteristics include the aetiological agent(s), clinical context and findings, depth of tissue involvement and rate of progression. The most common pathogens are aerobic Gram-positive cocci, but complicated infections frequently involve Gram-negative bacilli and anaerobic bacteria. Initial antibiotic therapy is usually empirical, and later modified by the results of stains and cultures of wound specimens. Broad-spectrum coverage is frequently needed for complicated infections.

Ertapenem is a once-a-day parenteral Group 1 carbapenem antibiotic, recently licensed in the USA and http://jac.oxfordjournals.org/ Europe, which may assume an important role in treating some complicated skin and skin-structure infec- tions. Surgical debridement is important for many complicated infections, and is the critical element in managing necrotizing fasciitis and myonecrosis.

Keywords: cellulitis, necrotizing fasciitis, myonecrosis, gas at :: on August 8, 2012 Introduction or lymphatic drainage, sensory neuropathies, mellitus, pre- vious cellulitis, the presence of a foreign body, accidental or surgical Skin and skin structures are among the most frequent sites of human trauma, obesity, poor hygiene and certain .10 1–5 bacterial infection. They represent one of the most common indi- A second level of classification divides skin and skin-structure cations for antibiotic therapy and account for ∼10% of hospital infections into uncomplicated or complicated, the latter defined as admissions in the USA.6 Furthermore, the incidence of soft-tissue involving abnormal skin or wounds, occurring in a compromised infections appears to be increasing, at least in some populations.6 host, or requiring substantial surgical intervention.14 These infections Such infections are highly diverse in their aetiology, clinical manifes- are often further characterized as being acute (present for days to at tations and severity.1,2,7–10 Bacteria do not cause all skin infections, most a few weeks) or chronic (persisting for many weeks to months). but this article will review only bacterial aetiologies. The pathogen- esis of these infections usually involves direct inoculation of patho- Soft-tissue infections can be localized or focal (e.g. , gens, but infection occasionally spreads to the skin contiguously abscess) or diffuse (e.g. cellulitis, fasciitis). A clinically useful from deeper foci11–13 or haematogenously from distant sites. Severity distinction with important management implications subdivides ranges from minor superficial lesions to invasive, fulminant and even soft-tissue infections into non-necrotizing and necrotizing pro- 1 lethal infections. cesses. The key to treating serious infections successfully is prompt recognition, followed by appropriate antibiotic and, when needed, surgical therapy. Classification of soft-tissue infections Specific infections of skin and skin structures can be grouped The terminology used for infections of skin and skin structures is according to causative organism(s), the soft tissues involved (related often confusing. Primary skin infections occur in otherwise normal to specific layers or depth of invasion) or the clinical syndrome skin and are usually caused by group A streptococci or Staphylo- (setting and presentation).1 Other relevant issues include the coccus aureus. Secondary infections complicate chronic skin con- epidemiology,5,15–17 pathogenesis4,18 and prognosis of the infection.19 ditions (e.g. eczema or atopic dermatitis). These underlying disorders Most proposed organizational schemata are cumbersome and diffi- act as portals of entry for virulent bacteria. Other factors predisposing cult to remember or apply. We believe that a clinically useful system to skin infections include vascular insufficiency, disrupted venous should be based on easily obtainable demographic, historical, physical

...... *Corresponding author. Tel: +1-484-344-3331; Fax: +1-484-344-3404; E-mail: [email protected] ...... ii37 JAC © The British Society for Antimicrobial Chemotherapy 2004; all rights reserved. M. J. DiNubile and B. A. Lipsky

Table 1. Classification of skin and skin-structure infections bation period of <24 h after trauma or surgery is most consistent with infection caused by Streptococcus pyogenes, Clostridium per- 2 Uncomplicated infections fringens or Pasteurella multocida. Other pathogens that character- superficial: impetigo, istically have a short incubation period, abrupt onset and rapid 26 deeper: , cellulitis progression include Aeromonas hydrophila and . hair follicle associated: , furunculosis On the other hand, wound infections caused by S. aureus and Entero- abscess: , other cutaneous bacteriaceae usually incubate for at least 48–72 h (often longer) Complicated infections before clinical manifestations become evident, and they tend to secondary infections of diseased skin advance less quickly. Small inocula may be followed by an incuba- acute wound infections tion period of longer than a week. Indurated developing at traumatic the site of vaccinia inoculation more than a week after immunization bite-related is more probably due to a robust cell-mediated immune response than post-operative to a bacterial superinfection. Patients with devitalized tissue or chronic wound infections immunological deficiencies are susceptible to infections with bacteria diabetic foot infections generally considered non-pathogenic for normal hosts (e.g. coagulase- venous stasis ulcers 27–30 pressure sores negative staphylococci, diphtheroids or Bacillus species). perianal cellulitis ± abscess Necrotizing fasciitis Cellulitis polymicrobial fasciitis (type I) Fournier’s gangrene Most routine cellulitis acquired in the community is caused by S. pyo- synergic necrotizing ‘cellulitis’ with fasciitis and myonecrosis genes and/or S. aureus.31–37 Cellulitis due to S. aureus is more likely Downloaded from streptococcal gangrene (type II) to be bullous and associated with a concomitant skin wound. Group A fasciitis due to V. vulnificus and other Vibrio species (and less often groups B, C and G) β-haemolytic streptococci can Myonecrosis cause particularly aggressive cellulitis.38–42 When the lymphatics crepitant myonecrosis are involved by S. pyogenes, the skin becomes tense and palpably clostridial myonecrosis

thickened to produce a ‘peau d’orange’ feel and appearance. The http://jac.oxfordjournals.org/ traumatic resulting erysipelas has elevated, well demarcated and usually atraumatic gas gangrene synergic necrotizing ‘cellulitis’ with fasciitis and myonecrosis rapidly advancing borders. The extremities, face, breast and perianal 43 non-crepitant myonecrosis area may be involved. Patients often have prominent constitutional streptococcal gangrene with myonecrosis complaints, such as fever, chills and malaise, which may antedate A. hydrophila myonecrosis local signs or symptoms.42 Acute febrile neutrophilic dermatosis (‘Sweet’s syndrome’) may present as a facial cellulitis resembling erysipelas. 44 Recurrent cellulitis is predominantly due to group A and other β-haemolytic streptococci.38,40 Repeated bouts of lower-extremity at :: on August 8, 2012 cellulitis may follow saphenous venectomy or varicose vein strip- examination and laboratory information that allows diagnostically ping;40,45,46 recurrent cellulitis at other sites (e.g. the arm or breast) is 20 important and therapeutically useful distinctions (Table 1). Appro- frequently caused by impaired lymphatic drainage secondary to neo- priate categorization should assist the clinician in recognizing: (i) plasia, radiation therapy, surgery or prior infection.47–50 Patients with which patients need prompt surgical intervention; and (ii) what a persistent break in the cutaneous barrier also experience recurrent empirical antibiotic regimen is most appropriate. infections.51 A common factor predisposing to recurrent cellulitis is A few caveats regarding the nosology of these infections may be tinea pedis, which is present in about half of the reported cases helpful. Although some soft-tissue infections primarily affect deeper involving the leg.5,52 After the acute infection is controlled, the structures, most entities have cutaneous manifestations as part of primary dermatological condition must be addressed to prevent their initial (or sometimes later) presentation. In many cases, the recurrences. Patients with diabetes mellitus, chronic renal failure on critical first step is to recognize that an apparent cellulitis is, in reality, haemodialysis, or who use illicit parenteral drugs may develop recur- a fasciitis or myositis, or a manifestation of an underlying osteo- rent staphylococcal skin infections. This problem has been linked to myelitis or visceral abscess.21–23 Anatomical boundaries are not nec- nasal carriage of S. aureus and eradicating carriage with topical or essarily respected by invasive pathogens.24 For example, although systemic antimicrobials may lower the incidence.53–56 Prophylactic group A streptococcal gangrene is often grouped with necrotizing or symptom-prompted, patient-initiated antibiotic therapy may fasciitis, the process can range from a rapidly progressive cellulitis reduce morbidity from recurrent pyodermal infections.57–60 Unusual with a predilection to involve lymphatic vessels to a frank non-crepi- pathogens can cause recurrent cellulitis in immunocompromised tant myositis.19,25 Although muscle injury may be the direct conse- hosts, as illustrated by Helicobacter cinaedi in HIV-infected patients. quence of infection, it can also result from a para-inflammatory Some patients suffering from repeated episodes of what appears to be process without bacterial invasion; in either case, oedema may lead to cellulitis actually have a non-infectious inflammatory disease.61–67 a serious compartment syndrome. Episodes of ‘pseudoerysipelas’ are particularly difficult to distin- The likelihood that a wound will become infected is directly guish from infection, but may respond more quickly and consistently related to the size of the microbial inoculum and the virulence of to anti-inflammatory than to antimicrobial therapy.67 Lipodermato- the organism(s), and inversely related to local and systemic host sclerosis can occasionally present repetitively as a tender red plaque resistance. An estimate of the incubation period and apparent rate on the lower leg above the medial malleolus in patients with venous of progression of the infection can be diagnostically useful. An incu- insufficiency, mimicking recurrent infection.68,69

ii38 Skin and skin-structure infections

Table 2. Aetiological bacteria of soft-tissue infections by risk factor and setting

Risk factor/setting Expected bacterial pathogen(s)

Cat bite P. multocida and other Pasteurella species Dog bite P. multocida, Capnocytophaga canimorsus, CDC group EF-4 Rat bite Spirillum minora Shark bite V. carchariae Human bite , Fusobacterium, Prevotella, streptococci, etc. Animal hides, carcasses B. anthracis, Francisella tularensis Injection drug use S. aureus, Clostridium spp., E. corrodens, S. pyogenes Hot tub or wading pool P. aeruginosa Body piercing S. aureus, S. pyogenes, P. aeruginosab, Clostridium tetani Medicinal leeches A. hydrophila, Aeromonas sobria, Serratia marcescens, Vibrio fluvialis Salon foot baths Mycobacterium fortuitum Fresh water injury A. hydrophila Salt water injury V. vulnificus Soil contamination Nocardia braziliensis, Clostridium spp. Fishmonger E. rhusiopathiae, Streptococcus iniae Fish tank exposure Mycobacterium marinum Downloaded from aStreptobacillus moniliformis, the other recognized cause of rat-bite fever, does not classically present with prominent cutaneous signs at the site of the bite. bP. aeruginosa infections have generally followed piercing of the ear cartilage, not the ear lobe, and may cause disfiguring auricular chondritis. http://jac.oxfordjournals.org/ Haemophilus influenzae is responsible for a distinctive cellulitis, P. multocida and other Pasteurella species are Gram-negative usually in young children, which typically presents with a purplish coccobacilli found in the oral cavity of many animals.87,88 Bites, discolouration of the cheek.70 It is frequently associated with bacter- scratches or licking of open wounds by cats or dogs may result in aemia or .70,71 This entity has markedly decreased in fre- cellulitis within hours to a few days.4,18,89 Tenosynovitis is the most quency since the introduction of the conjugate H. influenzae type b frequent local complication of Pasteurella soft-tissue infection. vaccine.72–74 Although uncommon, Streptococcus pneumoniae occa- Fever develops in a minority of cases and bacteraemia is uncom- sionally causes cellulitis, especially in patients with systemic lupus mon.88,90 Penicillins are effective treatment, but first-generation erythematosus and other collagen vascular diseases.75,76 cephalosporins are unreliable.88 Broader-spectrum agents that pro- The aetiological agent responsible for cellulitis may be suggested vide more comprehensive coverage of the variety of aerobic and at :: on August 8, 2012 by the specific clinical context (Table 2). Erysipelothrix rhusio- anaerobic organisms associated with bite wounds (e.g. oral amoxi- pathiae is a pleomorphic Gram-positive bacillus responsible for a cillin–clavulanate or parenteral ampicillin–sulbactam) are often generally indolent cellulitis, most commonly involving fingers that administered.4,10,91 come into contact with fresh fish.77,78 The lesion is V. vulnificus typically causes a rapidly advancing cellulitis characteristically painful, well demarcated, slightly raised, faintly associated with haemorrhagic bullae in both healthy persons and violaceous and spreads peripherally. Fever and systemic symptoms compromised hosts.92 The process may begin as, or progress to, are not common,79 and untreated lesions often involute over a few necrotizing fasciitis (see below).92,93 The history typically discloses weeks. Suggested antibiotic therapy is with penicillins, cephalo- recent exposure of a skin abrasion to warm brackish seawater. V. vul- sporins, clindamycin or ciprofloxacin; vancomycin should not be nificus infection can result in disseminated disease culminating in used. septic shock, with or without a cutaneous portal of entry, usually in Bacillus anthracis causes a focal necrotizing cellulitis in persons persons with cirrhosis or iron overload syndromes.26,92 Fulminant having contact with infected animals or contaminated animal prod- disease may quickly follow ingestion of raw oysters or clams in ucts (e.g. hides, goat hair).17,80,81 Unfortunately, its use as an agent of patients with hepatic cirrhosis, even when the liver disease is well bioterrorism has removed the epidemiological constraints of naturally compensated.94–96 Vibrio hollisae soft-tissue infections mimic the occurring infection.82,83 A papule appears at the site less than a week clinical picture of V. vulnificus. Other Vibrio species also cause after inoculation. Within 2 days, small vesicles containing few leuco- serious wound infections, including Vibrio carchariae following cytes but many large Gram-positive bacilli evolve and enlarge, then shark bites, Vibrio alginolyticus and Vibrio damsela. Prompt anti- undergo necrosis to form a painless covered by a black eschar biotic therapy, ideally with doxycycline, and early surgical interven- surrounded by extensive non-pitting oedema. Skin lesions resem- tion are usually indicated. bling different stages of cutaneous anthrax include furuncles and car- buncles, ecthyma, orf, brown recluse spider bites, rickettsial tache noire and ulceroglandular infections (such as tularaemia).84,85 Pre- Complicated skin infections sumptive laboratory identification is based on Gram-stained smears Clinical presentations of fluid or scrapings from the lesion. Oral therapy with ciprofloxacin, doxycycline or amoxicillin for 1–2 weeks is adequate for most cases Simple pyodermas are the most common skin infections, but are unless inhalational exposure is also suspected.86 generally easy to diagnose, involve a limited number of predictable

ii39 M. J. DiNubile and B. A. Lipsky pathogens and respond well to oral antibiotics.31 Complicated infec- Antibiotic therapy tions (e.g. extensive cellulitis, perianal abscess, traumatic or surgical Empirical antibiotic regimens for complicated skin-structure infec- wound infections, and foot infections in diabetic patients) are both tions should always include coverage for aerobic Gram-positive 2,10,14,97 more severe and difficult to treat. Even when limb- or life- cocci, specifically staphylococci and streptococci, and often for anaer- threatening, these infections may be indolent in their pace but inexor- obes, including the Bacteroides fragilis group.7,98,99 Methicillin- ably progressive despite medical therapy. The aetiological microbes resistant S. aureus and Gram-negative bacilli are found in some in complicated infections are predominantly S. aureus and strepto- mixed infections, especially those that occur in the hospital.32 cocci, but often involve mixed Gram-positive and Gram-negative P. aeruginosa is an uncommon soft-tissue pathogen in the commun- aerobic and anaerobic bacteria as well.7,98,99 Enteric Gram-negative ity, but can cause wound infection following fresh water exposure or bacilli and P. aeruginosa tend to be associated with nosocomially piercing of the ear cartilage, hot-tub folliculitis, or deep foot infec- acquired infections as well as infections in compromised hosts and tions after a puncture wound through a sports shoe.111–115 Other injection drug users.32 non-fermentative species (e.g. Acinetobacter) rarely cause skin Several clinical syndromes are recognizable.2,20,21 Crepitant cellu- infections. Gram-negative organisms and methicillin-resistant litis may complicate dirty community-acquired traumatic staphylococci assume greater importance in superinfections after as well as surgical wounds. Nosocomial cases have developed at multiple courses of antibiotics, and in infections associated with profound neutropenia or injection drug use.34,116–118 However, indwelling catheter sites. Clostridial cellulitis is a superficial infec- methicillin-resistant S. aureus (MRSA) has been increasingly recog- tion associated with less systemic toxicity than clostridial myo- nized as a cause of sporadic and epidemic skin infections arising in necrosis (gas gangrene; see section ‘Clostridial myonecrosis’). The the community in those patients with no other risk factors, most process is usually indolent and rarely life threatening. Pain is rela- 119–121

commonly in children and young adults. Community-acquired Downloaded from tively mild, and the bullous and necrotic skin lesions of gas gangrene MRSA often harbour the novel type IV staphylococcal cassette do not develop. Crepitus, however, is more prominent in clostridial chromosome (SCC)mec element, which typically does not confer cellulitis than in gas gangrene. Gram-stained smears of exudate resistance to antimicrobial drugs other than β-lactam antibiotics, and or aspirates disclose abundant large Gram-positive bacilli with sur- sometimes contains the Panton–Valentine virulence gene, which may prisingly few neutrophils, and C. perfringens is usually recovered be involved in the pathogenesis of necrotizing skin or lung infections. from anaerobic cultures. The clinical picture of non-clostridial crepi- Several antibiotic classes have been shown to be effective, alone http://jac.oxfordjournals.org/ tant cellulitis resembles clostridial cellulitis but may be more aggres- or in combination, against complicated soft-tissue infections:122 sive; causative bacteria include combinations of facultative species penicillin–β-lactamase inhibitor combinations (e.g. ampicillin– (e.g. , Klebsiella, various streptococci) and strict sulbactam, piperacillin–tazobactam),123–125 cephalosporins of all anaerobes (e.g. Bacteroides, Peptostreptococcus). generations (e.g. cefazolin, cefixime, cefoxitin),126–128 fluoro- 129–132 A focal but necrotizing infection, termed progressive bacterial quinolones (e.g. levofloxacin, moxifloxacin, clinafloxacin), 133 134 synergic gangrene, presents ∼1–2 weeks after surgery as a necrotic glycopeptides (e.g. vancomycin), quinupristin/dalfopristin and oxazolidinones (e.g. linezolid).135 Clindamycin and metronidazole ulcer with an outer zone of violaceous erythema.1,2 The process is are often added to the regimen to cover anaerobes, depending on the at :: on August 8, 2012 seen most often in peri-colostomy and other postoperative infections clinical context and other antibiotics being used. Reported clinical of the abdominal wall. Meleney’s ulcer denotes a similar lesion response rates are typically ∼80–90%, with similar microbiological associated with multiple fistulous tracts emerging at a distance eradication rates. New agents for treating these infections may be from the infected wound. This process results from co-infection with helpful because of growing antibiotic resistance and to enhance a microaerophilic Streptococcus and either S. aureus or a Gram- convenience. negative bacillus. The involved area progressively enlarges in widen- The traditional carbapenems (e.g. imipenem and meropenem)136 ing circles unless treated appropriately by both antibiotics and wide cover an exceptionally wide spectrum of aerobic and anaerobic path- surgical excision of all infected tissue.100 The differential diagnosis ogens, and have been found to be effective in complicated soft-tissue of these lesions includes gangrenosum, which can develop infections.137,138 However, these older Group 2 carbapenems may at sites of trauma, such as a postoperative wound, and amoebic have a broader spectrum than needed for most skin and soft-tissue ulcers. Supervening myonecrosis or necrotizing fasciitis should be infections,136 and require multiple daily doses. In contrast, ertapenem suspected if the patient develops ecchymoses, bullae, crepitus, wet is a new Group 1 carbapenem antibiotic136 that is given once a day gangrene or anaesthesia of the overlying skin, especially in conjunc- and is active against aerobic and anaerobic organisms generally 85,139 tion with systemic toxicity or laboratory evidence of rhabdomyolysis associated with community-acquired infections. Its spectrum is or disseminated intravascular .20–22 appropriate for many complicated skin and skin-structure infections. The syndrome of purpura fulminans101,102 may complicate sepsis A large multicentre trial for this indication compared intravenous therapy with ertapenem (1 g once a day) with piperacillin–tazo- caused by several different bacterial species, including meningo- bactam (3.375 g every 6 h) in a randomized double-blind study of 540 coccaemia. The sharply margined purpuric lesions are typically 125 103–105 adults. The most common conditions were soft-tissue abscesses symmetrical, often on the distal extremities, and evolve into and diabetic lower extremity infections. Patients with necrotizing bullae filled with serous fluid, and ultimately to cutaneous necro- fasciitis and myonecrosis were excluded. Mean duration of antibiotic 106 sis. Skin changes probably result from disseminated intravascular therapy was 9–10 days. Clinical cure rates exceeded 80% and were coagulation or protein C deficiency.101,102,107 statistically equivalent for the two regimens. Response rates for the most frequently complicates P. aeruginosa bacteraemia, but has two treatment groups were similar when stratified by diagnosis and been associated with septicaemia due to A. hydrophila and other infection severity. The overall frequency and severity of drug-related Gram-negative bacilli.108–110 adverse events were also comparable in both treatment groups. Thus,

ii40 Skin and skin-structure infections

Table 3. Distinguishing features of the major types of deep, diffuse, necrotizing soft-tissue infections requiring prompt surgical intervention

Depth of Incubation Rate of involvement Usual pathogens Predisposing event period progression Characteristic features

Polymicrobial fascia and obligate and wound long (48–96 h) hours to days foul-smelling necrotizing muscle facultative drainage fasciitis (type I) anaerobes Streptococcal skin, fascia, group A>C>G>B minor cut or short (6–48 h) a few hours distinct margins gangrene muscle streptococci abrasion (necrotizing fasciitis type II) Gas gangrene muscle traumatic: contaminated short (6–48 h) a few hours extreme systemic toxicity (clostridial C. perfringens > wound myonecrosis) C. novyi atraumatic: gastrointestinal C. septicum lesion, but no local insult Non-clostridial muscle and obligate and wound variable (12–96 h) hours to days soft-tissue gas when myonecrosis fascia facultative polymicrobial aetiology

anaerobes or Downloaded from A. hydrophila

ertapenem appears to offer an additional option for complicated skin If S. aureus with methicillin resistance or decreased susceptibility to http://jac.oxfordjournals.org/ and skin-structure infections likely to be caused by susceptible mixed vancomycin becomes widespread in the community, the standard flora. empirical approach to antibiotic therapy for serious community- Although the intravenous route has traditionally been used to ini- acquired infections will need to be carefully rethought.145–147 tiate treatment for most complicated infections, oral antibiotics may 130–132,140–142 be adequate under some circumstances. Typically such Necrotizing fasciitis cases involve patients with mild-to-moderate infections for whom there are appropriate agents available that can be tolerated by the oral Necrotizing fasciitis is an uncommon life-threatening infection route. Increasingly, oral therapy has been substituted in a step-down affecting subcutaneous tissue.116,148–150 Onset is often acute and the approach from the initial parenteral therapy.89,123,124 Certain antibiotics course can be rapid.151 The confusing nomenclature is based on the at :: on August 8, 2012 reliably achieve essentially equivalent plasma concentrations when aetiological agent(s), clinical findings, type and level of tissue administered parenterally or orally (e.g. trimethoprim–sulfamethox- involved, and rate of progression (Table 3). The term encompasses azole, metronidazole, doxycycline, linezolid and the fluoroquinolones). two distinct bacteriological entities that share many pathological and Other drugs (e.g. amoxicillin–clavulanate and clindamycin) are less clinical features.20,149 Type I necrotizing fasciitis is a polymicrobial bioavailable after oral administration at standard doses but may still infection caused by facultative bacteria, such as non-groupable achieve therapeutic levels. For patients who do not require intra- streptococci and Enterobacteriaceae, along with strict anaerobes, venous access for other reasons, intramuscular therapy is another frequently including Bacteroides and Peptostreptococcus.152,153 consideration. Ertapenem and ceftriaxone are characterized by a long Obligate aerobic organisms, such as P. aeruginosa, are rarely half-life and little local pain or inflammation after intramuscular involved, but occasional cases are caused exclusively by anaerobes. injection.143 This route of administration is generally not well Type II necrotizing fasciitis, or streptococcal gangrene, is caused by accepted by patients for more than a few injections. Intramuscular group A (less often group B, C or G) streptococci, usually alone but administration may be most beneficial for outpatients who need only sometimes in association with S. aureus. Surgical procedures are a limited number of parenteral doses, or as a stopgap measure paramount in treating these infections; antibiotic therapy plays an between the loss of intravenous access and the re-establishment of important but secondary role. new access.144 In a non-compromised host with a focal superficial abscess, Polymicrobial (type I) necrotizing fasciitis incision and drainage alone (without antibiotic therapy) may be suffi- cient treatment. For most other skin and skin-structure infections, Factors that predispose to type I necrotizing fasciitis include diabetes antibiotic therapy is needed. The optimal duration of treatment for mellitus, morbid obesity, alcoholism and parenteral drug use. The these infections has not been well defined, but antibiotics should most common sites of involvement are the legs, abdominal wall, peri- usually be continued for ∼3 days after all systemic and most local neal area, postoperative wounds and, in the newborn, the umbilical signs and symptoms of infection have subsided. The total duration of stump. The affected area is initially swollen, erythematous with therapy for uncomplicated infections rarely needs to be more than indistinct margins, warm, shiny and exquisitely tender. The process a week in a normal host. More severe infections, especially in a evolves with sequential colour changes from red–purple to patches of compromised patient, can require 3–4 weeks of therapy. Underlying blue–grey. Cutaneous bullae and frank gangrene ultimately develop. osteomyelitis dictates an even longer duration of antibiotic treatment. Soft-tissue gas is often detectable, and the central area may become

ii41 M. J. DiNubile and B. A. Lipsky anaesthetic secondary to destruction of superficial sensory nerves. fasciitis. Clinical distinctions are usually based on the presence or Cutaneous hypoaesthesia and prominent systemic toxicity may ante- absence of crepitus and the presumed identity of the causative organ- date the appearance of skin necrosis and indicate that the process is ism. deeper than a superficial cellulitis. The infection dissects along the tissue plane just superficial to the deep fascia. Marked oedema may Clostridial myonecrosis produce a compartment syndrome with secondary myonecrosis, requiring prompt decompression. Gram-stained smears of exudate Clostridial myonecrosis (gas gangrene) refers to a rapidly pro- usually reveal a mixture of Gram-positive and Gram-negative gressive, life-threatening, toxaemic infection of skeletal muscle 175 organisms. Specimens of pus and deep tissue should be cultured for caused by clostridial species (principally C. perfringens). It aerobes and anaerobes; some of the pathogens may be recovered usually occurs after deep trauma with gross contamination or, less 176,177 from blood cultures. often, surgery. Rare cases have occurred after minor trauma (e.g. parenteral injections)16,178–180 Despite the high frequency of Fournier’s gangrene. Fournier’s gangrene refers to necrotizing clostridial contamination of open wounds, the low incidence of gas fasciitis involving the male genital region. The process may be gangrene attests to the prerequisite for devitalized tissue or foreign confined to the scrotum or extend to the perineum, penis, buttock and bodies in the pathogenesis of this infection. Clostridium species have 154 been recovered from wounds and blood cultures in patients without abdominal wall. Predisposing factors include diabetes mellitus, 181,182 local trauma, paraphimosis, periurethral extravasation of urine, peri- evidence of infection due to these organisms. rectal or perianal infections, and genitourinary surgery.155 The infec- Atraumatic gas gangrene designates clostridial myonecrosis tion commonly begins as a cellulitis adjacent to the portal of entry. developing in the absence of an obvious external insult. ‘Spon- Scrotal oedema and crepitus quickly increase, and dark purple taneous’ myonecrosis is most often caused by haematogenous spread patches develop and rapidly progress to extensive scrotal gangrene. of Clostridium septicum from a colonic lesion, which is commonly Downloaded from 183 With prompt and appropriate treatment (including surgery, anti- occult and frequently malignant. Diverticulitis, ischaemic bowel biotics and sometimes hyperbaric oxygen), the testicles can often be and neutropenic enterocolitis are other recognized underlying con- preserved and the scrotum will usually regenerate.154–160 ditions. A few patients with spontaneous gas gangrene have multiple discrete sites of infection related to bacteraemic seeding. The usual incubation period between injury and the onset of Synergic necrotizing cellulitis. Synergic necrotizing cellulitis is a http://jac.oxfordjournals.org/ 175 confusing misnomer for a distinctive clinical variant of necrotizing clostridial myonecrosis is 2–3 days, but may be as short as 6 h. fasciitis with prominent direct involvement of muscle.20 Most infec- Typically the onset is alarmingly sudden, with pain out of proportion tions involve the lower extremities. The lesion begins as a clustered to the inciting injury. Cutaneous signs soon become apparent and the area of small ulcers draining foul-smelling ‘dishwater’ pus. Circum- patient appears toxaemic. Brownish foul-smelling drainage, contain- scribed areas of gangrene develop around the draining areas, while ing numerous organisms but few leucocytes, exudes from the wound. the intervening skin remains uninvolved despite underlying necrosis. Gas bubbles may be visible in the discharge and deep crepitus is Local pain and tenderness are marked; crepitus, systemic toxicity and usually present, although generally not prominent. Tense blebs con- bacteraemia are common. taining serosanguineous fluid develop in skin overlying necrotic muscle. Radiographs of involved areas reveal characteristic gaseous at :: on August 8, 2012 dissection of muscle and fascial planes. At surgery, infected muscle Streptococcal gangrene (type II necrotizing fasciitis) may initially exhibit only pallor, oedema and loss of elasticity, but Streptococcal gangrene occurs after minor trauma or surgery, partic- rapidly becomes frankly necrotic. In the earliest stages, prior to its ularly in patients with diabetes or peripheral vascular disease.24,161–163 discolouration and dissolution, non-viable muscle may be identified Groups B, C and G streptococci cause infections indistinguishable by its lack of bleeding and failure to contract on stimulation. from the classic group A gangrene, and tend to affect diabetic Gram-stained smears of wound exudate or a bleb aspirate reveal patients.41,99 Necrotizing fasciitis caused by streptococci may be only a few neutrophils but many large Gram-positive bacilli with a prominent part of the ‘toxic strep syndrome’.162,164–166 Even when blunt ends. C. perfringens is the most common isolate, followed in S. aureus is isolated from necrotic tissue, it usually contributes little frequency by Clostridium novyi and C. septicum.184 E. coli, other to the pathogenesis. Streptococcal gangrene can be confined to the enteric Gram-negative bacilli and enterococci are sometimes dermis, but the fascia is often the major site of involvement and recovered, reflecting the contaminated nature of the initiating myonecrosis may be the dominant process.38,162,167 Startlingly fast lesion.152 C. perfringens may produce gas in anaerobic broth within progression of erythema with distinct borders, quickly followed by 6 h of inoculation, providing an early presumptive identification of gangrene, is a hallmark of this infection.168 Gram-stained smears of the infecting species. exudate usually reveal chains of Gram-positive cocci, sometimes The approach to suspected gas gangrene includes urgent surgical with interspersed clusters of larger Gram-positive cocci, and blood exploration to define the nature and extent of the process, to obtain cultures may yield streptococci. High-dose penicillin and/or clinda- specimens for stains and cultures, and to carry out appropriate mycin appear to be the treatment of choice.150,166,169–171 However, debridement. Prompt, extensive, and often repeated, surgery is the prompt and adequate debridement, often requiring repeated oper- principal treatment. All involved muscle, regardless of its gross ative procedures, is essential.153,172,173 appearance, should be resected so that the margins contain healthy bleeding tissue. Liberal fasciotomies to drain and decompress Myonecrosis swollen fascial compartments may be necessary; unfortunately, limb amputation is sometimes required. Antibiotic therapy has tradition- Bacterial myonecrosis can be caused by a variety of organisms, ally consisted of high-dose intravenous penicillin. Currently, clinda- some of which produce gas deep within the involved muscles.174 mycin (600 mg intravenously every 6–8 h) is often added to penicillin Myonecrosis often coexists with wound infection and necrotizing on the basis that the combination of penicillin with clindamycin has

ii42 Skin and skin-structure infections been shown to provide greater efficacy than either agent alone in a after admission.161 Many patients with cellulitis can be discharged in murine model of gas gangrene.175,185–190 less than a day by using outpatient antimicrobial therapy.205 Ancillary therapy for gas gangrene includes fluid and electrolyte replacement and often blood transfusions. The efficacy of hyperbaric Diagnostic tests oxygen therapy has not been conclusively established, but it may have an important role early in the treatment of seriously ill patients Basic haematological studies and serum chemistries are appropriate or in those with extensive involvement of the trunk in whom defini- for seriously ill patients. Leucocytosis (white blood cell count 3 tive surgical excision would be impossible or excessively > 15 400/mm ) or hyponatraemia (serum sodium < 135 meq/L) mutilating.176,191–201 The usefulness of intravenously administered increases the likelihood that necrotizing fasciitis is present in a 21 polyvalent antitoxin has not been demonstrated in this setting.175,202 patient with a severe soft-tissue infection. The utility of diagnostic aspiration of cellulitic skin is debated because of its low yield in 206,207 Non-clostridial myonecrosis identifying pathogens by stain or culture. When performed, aspirating near the advancing border of erythema or the point of Non-clostridial myonecrosis encompasses at least four relatively maximal inflammation is usually advocated to increase the yield.208 distinct entities that differ from gas gangrene in their pathogenesis, On occasion, a Gram-stained smear of an appropriate wound speci- clinical features and bacteriology: streptococcal myositis ± type II men may provide rapid guidance to the causative organisms.7 A posi- fasciitis (see earlier discussion under ‘Necrotizing fasciitis’); synergic tive Gram-stained smear can be highly predictive of culture results necrotizing ‘cellulitis’ with type I fasciitis and myonecrosis (see when adequate samples from the site of infection can be obtained and earlier discussion under ‘Necrotizing fasciitis’); A. hydrophila myone- properly processed. The presence of abundant neutrophils usually crosis; and superinfected dry gangrene. The last entity is a focal, usu- indicates the adequacy of the sample, but some histolytic bacteria, ally indolent and primarily ischaemic process in the small muscles of such as C. perfringens, destroy inflammatory cells at the site of infec- Downloaded from a distal lower extremity already gangrenous from arterial insuffic- tion. White cells also adhere to cotton-tipped swabs used to culture iency. Diabetic patients are prone to develop this complication, exudate, which dry quickly and absorb the specimen. For many com- which usually does not extend beyond the area of vascular gangrene plicated infections, a culture may provide useful information in to involve viable muscle. selecting appropriate therapy; tissue specimens or aspirates are gen- Rapidly progressive myonecrosis resembling clostridial gangrene erally preferred to wound swabs. Blood cultures may be informative http://jac.oxfordjournals.org/ but caused by Aeromonas species (usually A. hydrophila) may occur in those patients with systemic signs and symptoms, but their cost- after injuries sustained (or contaminated) in a freshwater environ- effectiveness is low in uncomplicated patients with routine cellu- 165,203,204 ment, or in conjunction with medicinal leech therapy. lites.209,210 In a study of adults with cellulitis, the probability of a con- 117 Lymphoma and leukaemia are predisposing factors in some cases. taminant (3.6%) exceeded that of a true pathogen (2.0%).209 Cellulitis often develops within 12–48 h, accompanied by excruciat- Other diagnostic tests can supply important information in diffi- ing pain, marked oedema and bullae, serosanguineous drainage and cult or confusing cases. Soft-tissue radiographs may demonstrate a systemic toxicity. Bacteraemia is frequently documented. Treatment foreign body or gas in deep tissues.211 Computerized tomography or requires prompt antimicrobial therapy and wide surgical debride- magnetic resonance imaging sometimes assist in defining the depth ment. Most isolates of Aeromonas are susceptible in vitro to genta- at :: on August 8, 2012 and extent of the process when the diagnosis of fasciitis or myo- micin, trimethoprim–sulfamethoxazole, ciprofloxacin and third- or necrosis is in doubt.175,211 Often, surgical exploration is the most fourth-generation cephalosporins. prudent diagnostic approach for seriously ill patients in whom the window of opportunity may be narrow.212 General observations on diagnosis and management Initial management Medical treatment The first and economically most important decision in treating skin Antibiotic therapy. The clinical setting, apparent severity of the infections concerns the need for hospitalization. Most pyodermas infection and the results of available laboratory tests should largely and uncomplicated soft-tissue infections do not require hospitaliza- guide the choice of an antibiotic regimen. Initial empirical therapy tion. Complicated infections often require admission, especially if of cellulitis should almost invariably cover aerobic Gram-positive muscle or fascial involvement is suspected, the process is rapidly cocci. The clinical course, epidemiological clues, local antibiotic progressing, signs of toxaemia are developing, the diagnosis or prog- susceptibility trends and/or microscopic examination of appropriate nosis is in doubt, exploratory surgery is contemplated or the patient specimens may dictate broader coverage. Definitive therapy should cannot adequately comply with outpatient treatment. At the time of ultimately be based on culture and susceptibility results as well as the admission, the physician should carefully delineate and record the clinical response to empirical therapy. apparent extent of infection and consider whether or not surgical consultation or further diagnostic measures (see section ‘Diagnostic Ancillary measures. Simple measures such as elevation or com- tests’) are indicated. Concurrently, any necessary supportive care pression of an oedematous213 or inflamed extremity and therapy with should be initiated and an empirical antimicrobial regimen selected. non-steroidal anti-inflammatory drugs may hasten symptomatic reso- Watchful waiting for 24–48 h on antibiotics to assess the clinical lution.214,215 Although non-steroidal drugs have been implicated in response to medical therapy may be appropriate in stable hospitalized accelerating the progression of streptococcal gangrene and other patients. Local signs of cellulitis may initially worsen after starting forms of necrotizing fasciitis,216–220 the weight of evidence suggests therapy, but then should gradually improve. Rapidly advancing ery- that the putative association simply reflects masking of serious signs thema or unrelenting toxaemia should precipitate more aggressive and symptoms by these anti-inflammatory agents,221–224 delaying accur- evaluation. These patients must be reassessed regularly and frequently ate diagnosis and appropriate management.167 Glucocorticoids may (preferably by the same clinician) during the first few hours and days accelerate healing and reduce long-term relapse rates in patients with

ii43 M. J. DiNubile and B. A. Lipsky erysipelas.225,226 Hyperbaric oxygen appears to benefit some patients 7. Bowler, P. G., Duerden, B. I. & Armstrong, D. G. (2001). Wound with necrotizing or anaerobic infections.17,158,160,179,194,195,227–231 How- microbiology and associated approaches to wound management. Clin- ever, use of hyperbaric oxygen should not delay the prompt institution ical Microbiology Reviews 14, 244–69. of necessary antibiotic therapy and surgical debridement. Intra- 8. Sharma, S. & Verma, K. K. (2001). Skin and soft tissue infection. Indian Journal of Pediatrics 68, Suppl. 3, S46–50. venous γ-globulin, antitoxins, granulocyte-stimulating factors and 9. Trent, J. T., Federman, D. & Kirsner, R. S. (2001). Common bac- 166,202,232–238 other growth factors may have roles in selected patients. terial skin infections. Ostomy/Wound Management 47, 30–4. Attention to fluid and electrolyte requirements as well as other meta- 10. Lipsky, B. A. (2002). Cellulitis, erysipelas, and necrotizing soft- bolic parameters is an important aspect of supportive care. Optimiz- tissue infections. Best Practice of Medicine. http://www.bestpracticeof- ing control of hyperglycaemia in diabetic patients appears to be medicine.com (date last accessed December 2003). important.239–242 Blood transfusions may be needed to replace peri- 11. Lodha, A., Wales, P. W., James, A. et al. (2003). Acute appendic- operative loss. Patients with extensive necrosis may benefit from itis with fulminant necrotizing fasciitis in a neonate. Journal of Pediatric management in a burn centre.243 The application of sterile maggots Surgery 38, E5–6. has been advocated for selected patients with massive swelling 12. Mahler, C. W., Boermeester, M. A. & Busch, O. R. (2003). Acute diverticulitis mimicking necrotizing fasciitis. Journal of the American following debridement of necrotizing fasciitis involving the head and College of Surgeons 197, 517. 244 neck, as well as for infected diabetic foot ulcers. Vacuum-assisted 13. Francque, S. M., Van Laer, C., Struyf, N. et al. (2001). Perforating wound closure devices have been advocated for some infected oesophageal carcinoma presenting as necrotizing fasciitis of the neck. wounds, but compelling data are scarce.7,245,246 Ultrasound and elec- European Journal of Gastroenterology and Hepatology 13, 1261–4. trical stimulation have been used to promote healing of chronic 14. Center for Drug Evaluation and Research (CDER). (1998). ischaemic wounds.247,248 Uncomplicated and complicated skin and skin structure infections— developing antimicrobial drugs for treatment. Guidance for Industry. http://www.fda.gov/cder/guidance/2566dft.pdf (date last accessed Downloaded from Surgical indications. Operative intervention is rarely needed for December 2003). uncomplicated pyodermas or cellulitis, but is critical to the diagnosis 15. Winthrop, K. L., Abrams, M., Yakrus, M. et al. (2002). An outbreak 153 and therapy of necrotizing fasciitis and myonecroses. Prompt sur- of mycobacterial furunculosis associated with footbaths at a nail salon. gical exploration with histopathological examination,213 extensive New England Journal of Medicine 346, 1366–71. debridement of devitalized tissue and decompression of ischaemic 16. Bangsberg, D. R., Rosen, J. I., Aragon, T. et al. (2002). Clostridial http://jac.oxfordjournals.org/ compartments is widely advocated for patients with suspected myonecrosis cluster among injection drug users: a molecular epidemio- deep tissue necrosis, especially if rapidly advancing or adjacent to the logy investigation. Archives of Internal Medicine 162, 517–22. trunk.172,173 The timing and extent of debridement are often the most 17. Centers for Disease Control. (2002). Update: cutaneous anthrax in hotly debated management decisions at the bedside in patients with a laboratory worker—Texas, 2002. Morbidity and Mortality Weekly Report 51, 482. diffuse necrotizing infections. As a general rule, it is usually safer to 18. Weil, H. P., Fischer-Brugge, U. & Koch, P. (2002). Potential err on the side of too early or too much surgery. Getting ahead of a hazard of wound licking. New England Journal of Medicine 346, 1336. rapidly advancing fasciitis or myonecrosis offers the only real chance 19. Sharkawy, A., Low, D. E., Saginur, R. et al. (2002). Severe group of cure in many cases. A streptococcal soft-tissue infections in Ontario: 1992–1996. Clinical Infectious Diseases 34, 454–60. at :: on August 8, 2012 Transparency declarations 20. Dellinger, E. P. (1981). Severe necrotizing soft-tissue infections. Multiple disease entities requiring a common approach. Journal of the B.A.L. has received fees for speaking at symposia organized on American Medical Association 246, 1717–21. 21. Wall, D. B., Klein, S. R., Black, S. et al. (2000). A simple model to behalf of Merck & Co., Inc., and Pharmacia, and has also received help distinguish necrotizing fasciitis from nonnecrotizing soft tissue infec- funds for research from Merck & Co., Inc., Pharmacia and Wyeth- tion. Journal of the American College of Surgeons 191, 227–31. Ayerst. B.A.L. is also a member of advisory boards for Merck & Co., 22. Landowski, M. A. (2002). Necrotizing fasciitis: a diagnostic and Inc., Pharmacia, Wyeth-Ayerst, Cubist and Vicuron. M.J.D. is an management challenge. Ostomy/Wound Management 48, 18–21. employee of Merck & Co., Inc., and potentially owns stock and/or 23. Trent, J. T. & Kirsner, R. S. (2002). Diagnosing necrotizing fasci- holds stock options in the Company. itis. Advances in Skin Wound Care 15, 135–8. 24. Smolyakov, R., Riesenberg, K., Schlaeffer, F. et al. (2002). Streptococcal septic arthritis and necrotizing fasciitis in an intravenous References drug user couple sharing needles. Israel Medical Association Journal 4, 302–3. 1. Lewis, R. T. (1988). Soft tissue infections. World Journal of Sur- 25. Adams, E. M., Gudmundsson, S., Yocum, D. E. et al. (1985). gery 22, 146–51. Streptococcal myositis. Archives of Internal Medicine 145, 1020–3. 2. Nichols, R. L. & Florman, S. (2001). Clinical presentations of soft- 26. Patel, V. J., Gardner, E. & Burton, C. S. (2002). Vibrio vulnificus Clinical Infectious Diseases tissue infections and surgical site infections. septicemia and leg ulcer. Journal of the American Academy of Dermat- 33, Suppl. 2, S84–93. ology 46, Suppl. 5, S144–5. 3. Rhody, C. (2000). Bacterial infections of the skin. Primary Care 27, 27. King, C. T. (1994). Sternal wound infection due to Corynebac- 459–73. terium xerosis. Clinical Infectious Diseases 19, 1171–2. 4. Goldstein, E. J. (1992). Bite wounds and infection. Clinical Infec- 28. Brook, I. (1989). Microbiology of postthoractomy sternal wound tious Diseases 14, 633–8. infection. Journal of Clinical Microbiology 27, 806–7. 5. Bisno, A. L. (1984). Cutaneous infections: microbiologic and epi- 29. Dancer, S. J., McNair, D., Finn, P. et al. (2002). Bacillus cereus demiologic considerations. American Journal of Medicine 76, 172–9. cellulitis from contaminated heroin. Journal of Medical Microbiology 51, 6. Centers for Disease Control. (2001). Incidence of soft tissue 278–81. infections: San Francisco General Hospital—1996–2000. Morbidity and 30. Meredith, F. T., Fowler, V. G., Gautier, M. et al. (1997). Bacillus Mortality Weekly Report 50, 381–4. cereus necrotizing cellulitis mimicking clostridial myonecrosis: case

ii44 Skin and skin-structure infections report and review of the literature. Scandinavian Journal of Infectious 50. Brewer, V. H., Hahn, K. A., Rohrbach, B. W. et al. (2000). Risk Diseases 29, 528–9. factor analysis for breast cellulitis complicating breast conservation ther- 31. Hook, E. W., III, Hooton, T. M., Horton, C. A. et al. (1986). Micro- apy. Clinical Infectious Diseases 31, 654–9. biologic evaluation of cutaneous cellulitis in adults. Archives of Internal 51. Dupuy, A., Benchikhi, H., Roujeau, J. C. et al. (1999). Risk factors Medicine 146, 295–7. for erysipelas of the leg (cellulitis): case–control study. British Medical 32. Doern, G. V., Jones, R. N., Pfaller, M. A. et al. (1999). Bacterial Journal 318, 1591–4. pathogens isolated from patients with skin and soft tissue infections: fre- 52. Baddour, L. M. & Bisno, A. L. (1984). Recurrent cellulitis after quency of occurrence and antimicrobial susceptibility patterns from the coronary bypass surgery. Association with superficial fungal infection in SENTRY Antimicrobial Surveillance Program (United States and Can- saphenous venectomy limbs. Journal of the American Medical Asso- ada, 1997). SENTRY Study Group (North America). Diagnostic Micro- ciation 251, 1049–52. biology and Infectious Disease 34, 65–72. 53. Boyce, J. M. (1996). Preventing staphylococcal infections by 33. Laube, S. & Farrell, A. M. (2002). Bacterial skin infections in the eradicating nasal carriage of Staphylococcus aureus: proceeding with elderly: diagnosis and treatment. Drugs and Aging 19, 331–42. caution. Infection Control and Hospital Epidemiology 17, 775–9. 34. Manfredi, R., Calza, L. & Chiodo, F. (2002). Epidemiology and 54. Perl, T. M., Cullen, J. J., Wenzel, R. P. et al. (2002). Intranasal microbiology of cellulitis and bacterial soft tissue infection during HIV dis- mupirocin to prevent postoperative Staphylococcus aureus infections. ease: a 10-year survey. Journal of Cutaneous Pathology 29, 168–72. New England Journal of Medicine 346, 1871–7. 35. Fluit, A. C., Schmitz, F. J. & Verhoef, J. (2001). Frequency of iso- 55. Cimochowski, G. E., Harostock, M. D., Brown, R. et al. (2001). lation of pathogens from bloodstream, nosocomial pneumonia, skin and Intranasal mupirocin reduces sternal wound infection after open heart soft tissue, and urinary tract infections occurring in European patients. surgery in diabetics and nondiabetics. Annals of Thoracic Surgery 71, European Journal of Clinical Microbiology and Infectious Diseases 20, 1572–8. 188–91. 56. Ruef, C., Fanconi, S. & Nadal, D. (1996). Sternal wound infection after heart operations in pediatric patients associated with nasal carriage

36. Chiller, K., Selkin, B. A. & Murakawa, G. J. (2001). Skin microflora Downloaded from and bacterial infections of the skin. Journal of Investigative Dermatology. of Staphylococcus aureus. Journal of Thoracic and Cardiovascular Sur- gery , 681–6. Symposium Proceedings 6, 170–4. 112 57. Klempner, M. S. & Styrt, B. (1988). Prevention of recurrent 37. Sigurdsson, A. F. & Gudmundsson, S. (1989). The etiology of staphylococcal skin infections with low-dose oral clindamycin therapy. bacterial cellulitis as determined by fine-needle aspiration. Scandinavian Journal of the American Medical Association 260, 2682–5. Journal of Infectious Diseases 21, 537–42. 58. Kremer, M., Zuckerman, R., Avraham, Z. et al. (1991). Long-term 38. Bisno, A. L. & Stevens, D. L. (1996). Streptococcal infections of http://jac.oxfordjournals.org/ antimicrobial therapy in the prevention of recurrent soft-tissue infections. skin and soft tissues. New England Journal of Medicine 334, 240–5. Journal of Infection 22, 37–40. 39. Lewthwaite, P., Parsons, H. K., Bates, C. J. et al. (2002). Group G 59. Wang, J. H., Liu, Y. C., Cheng, D.L. et al. (1997). Role of benza- streptococcal bacteraemia: an opportunistic infection associated with thine penicillin G in prophylaxis for recurrent streptococcal cellulitis of the immune senescence. Scandinavian Journal of Infectious Diseases 34, lower legs. Clinical Infectious Diseases 25, 685–9. 83–7. 60. Pauszek, M. E. (1991). Prophylaxis for recurrent cellulitis compli- 40. Baddour, L. M. & Bisno, A. L. (1985). Non-group A β-hemolytic cating venous and lymphatic insufficiency. Indiana Medicine 84, 252–3. streptococcal cellulitis. Association with venous and lymphatic compro- 61. Berlyne, G. M., Kwan, T., Li, J. et al. (1989). Oedema protein con- mise. American Journal of Medicine 79, 155–9. centrations for differentiation of cellulitis and deep vein thrombosis. 41. Riefler, J., III, Molavi, A., Schwartz, D. et al. (1988). Necrotizing Lancet ii, 728–9. at :: on August 8, 2012 fasciitis in adults due to group B streptococcus. Report of a case and 62. Houston, D. S. & Curran, J. (2001). Charcot foot. Orthopaedic review of the literature. Archives of Internal Medicine 148, 727–9. Nursing 20, 11–15. 42. Barzilai, A. & Choen, H. A. (1998). Isolation of group A strepto- 63. Phelps, R. G. & Shoji, T. (2001). Update on panniculitis. Mount cocci from children with perianal cellulitis and from their siblings. Pediatric Sinai Journal of Medicine 68, 262–7. Infectious Disease Journal 17, 358–60. 64. Stein, M. E., Drumea, K., Abu-Rasmi, R. et al. (1997). Taxol- 43. Bisno, A. L., Cockerill, F. R., III & Bermudez, C. T. (2000). The ini- induced cellulitis after extravasation: a rarely reported event. American tial outpatient–physician encounter in group A streptococcal necrotizing Journal of Clinical Oncology 20, 540. fasciitis. Clinical Infectious Diseases 31, 607–8. 65. Foo, K. F., Tao, M. & Tan, E. H. (2002). Gastric carcinoma pre- 44. Morgan, K. W. & Callan, J. P. (2001). Sweet’s syndrome in acute senting with cellulitis-like cutaneous metastasis. Singapore Medical Jour- myelogenous leukemia presenting as periorbital cellulitis with an infiltrate nal 43, 37–8. of leukemic cells. Journal of the American Academy of Dermatology 45, 66. Faber, J., Shroeder, L., Thill, M. P. et al. (2002). Clinical picture: 590–5. carcinoma erysipeloides of the neck. Lancet 359, 1025. 45. Turner, D. P., Nagra, R. S., Large, S. et al. (1995). Recurrent 67. Edwards, E. A. (1963). Recurrent febrile episodes and lymph- cellulitis following coronary bypass surgery. Journal of Hospital Infection odema. Journal of the American Medical Association 184, 858–62. 30, 78–80. 68. Bruce, A. J., Bennett, D. D., Lohse, C. M. et al. (2002). Lipo- 46. Hurwitz, R. M. & Tisserand, M. E. (1985). Streptococcal cellulitis dermatosclerosis: review of cases evaluated at Mayo Clinic. Journal of proved by skin biopsy in a coronary artery bypass graft patient. Archives the American Academy of Dermatology 46, 187–92. of Dermatology 121, 908–9. 69. Demitsu, T., Okada, O., Yoneda, K. et al. (1999). Lipodermato- 47. Woo, P. C., Lum, P. N., Wong, S. S. et al. (2000). Cellulitis compli- sclerosis—report of three cases and review of the literature. Dermatology cating lymphoedema. European Journal of Clinical Microbiology and 199, 271–3. Infectious Diseases 19, 294–7. 70. Stanley, T. V. & Jogose, M. (1991). Haemophilus influenzae type 48. Ellison, R. T., III & McGregor, J. A. (1987). Recurrent postcoital b cellulitis. New Zealand Medical Journal 104, 334–6. lower-extremity streptococcal erythroderma in women. Streptococcal- 71. Sankrithi, U. M. & LiPuma, J. J. (1991). Clinically inapparent sex syndrome. Journal of the American Medical Association 257, 3260– meningitis complicating periorbital cellulitis. Pediatric Emergency Care 7, 2. 28–9. 49. Mertz, K. R., Baddour, L. M., Bell, J. L. et al. (1998). Breast 72. Ambati, B. K., Ambati, J., Azar, N. et al. (2000). Periorbital and cellulitis following breast conservation therapy: a novel complication of orbital cellulitis before and after the advent of Haemophilus influenzae medical progress. Clinical Infectious Diseases 26, 481–6. type B vaccination. Ophthalmology 107, 1450–3.

ii45 M. J. DiNubile and B. A. Lipsky

73. Donahue, S. P. & Schwartz, G. (1998). Preseptal and orbital cellul- 96. Chen, Y., Satoh, T. & Tokunaga, O. (2002). Vibrio vulnificus infec- itis in childhood. A changing microbiologic spectrum. Ophthalmology tion in patients with liver disease: report of five autopsy cases. Virchows 105, 1902–5. Archiv 441, 88–92. 74. Adams, W. G., Deaver, K. A., Cochi, S. L. et al. (1993). Decline of 97. Lipsky, B. A. (2001). Diabetic foot infections: progress in a ped- childhood Haemophilus influenzae type b (Hib) disease in the Hib vac- estrian problem. Contemporary Surgery 57, Suppl., S7–19. cine era. Journal of the American Medical Association 269, 221–6. 98. Gerding, D. N. (1995). Foot infections in diabetic patients: the role 75. DiNubile, M. J., Albornoz, M. A., Stumacher, R. J. et al. (1991). of anaerobes. Clinical Infectious Diseases 20, Suppl. 2, S283–8. Pneumococcal soft-tissue infections: possible association with connec- 99. Reyzelman, A. M., Lipsky, B. A., Hadi, S. A. et al. (1999). The tive tissue diseases. Journal of Infectious Diseases 163, 897–900. increased prevalence of severe necrotizing infections caused by non- 76. Patel, M., Ahrens, J. C., Moyer, D. V. et al. (1994). Pneumococcal group A streptococci. Journal of the American Podiatric Medical Asso- soft-tissue infections: a problem deserving more recognition. Clinical ciation 89, 454–7. Infectious Diseases 19, 149–51. 100. Ovens, N. & Fairhurst, J. (2002). Management of a heavily exud- 77. Hillenbrand, F. K. M. (1953). Whale finger and seal finger: their ing, painful wound with necrotising subcutaneous infection. Journal of relation to erysipeloid. Lancet i, 680–1. Wound Care 11, 25–7. 78. Brooke, C. J. & Riley, T. V. (1999). Erysipelothrix rhusiopathiae: 101. Fourrier, F., Lestavel, P., Chopin, C. et al. (1990). Meningo- bacteriology, epidemiology and clinical manifestations of an occupational coccemia and purpura fulminans in adults: acute deficiencies of pathogen. Journal of Medical Microbiology 48, 789–99. C and S and early treatment with antithrombin III concentrates. Intensive 79. Hill, D. C. & Ghassemian, J. N. (1997). Epidemiologic investi- Care Medicine 16, 121–4. gations of bioterrorism-related anthrax, New Jersey, 2001. Emerging 102. Powars, D. R., Rogers, Z. R., Patch, M. J. et al. (1987). Purpura Infectious Diseases, 8, 1048–55. fulminans in meningococcemia: association with acquired deficiencies of 80. Swartz, M. N. (2002). Recognition and management of anthrax— proteins C and S. New England Journal of Medicine 317, 571–2. an update. New England Journal of Medicine 345, 1621–5. 103. Welchon, J. G., Armstrong, D. G. & Harkless, L. B. (1996). Pedal

81. Greene, C. M., Reefhuis, J., Tan, C. et al. (2002). Epidemiologic manifestations of meningococcal septicemia. Journal of the American Downloaded from investigations of bioterrorism-related anthrax, New Jersey, 2001. Emer- Podiatric Medical Association 86, 129–31. ging Infectious Diseases 8, 1048–55. 104. Genoff, M. C., Hoffer, M. M., Achauer, B. et al. (1992). Extremity 82. Brachman, P. S. (2002). Bioterrorism: an update with a focus on amputations in meningococcemia-induced purpura fulminans. Plastic anthrax. American Journal of Epidemiology 155, 981–7. and Reconstructive Surgery 89, 878–81. 83. Inglesby, T. V., O’Toole, T., Henderson, D. A. et al. (2002). 105. Jacobsen, S. T. & Crawford, A. H. (1984). Amputation following

Anthrax as a biological weapon, 2002: updated recommendations for meningococcemia. A sequela to purpura fulminans. Clinical Ortho- http://jac.oxfordjournals.org/ management. Journal of the American Medical Association 287, 2236– paedics and Related Research 185, 214–9. 52. 106. Gamper, G., Oschatz, E., Herkner, H. et al. (2001). Sepsis- 84. Claeys, L. G. & Matamoros, R. (2002). Anaerobic cellulitis as the associated purpura fulminans in adults. Wiener Klinische Wochenschrift result of Clostridium perfringens: a rare cause of vascular access graft 113, 107–12. infection. Journal of Vascular Surgery 35, 1287–8. 107. Powars, D., Larsen, R., Johnson, J. et al. (1993). Epidemic 85. Syrjala, H., Karvonen, J. & Salminen, A. (1984). Skin manifesta- meningococcemia and purpura fulminans with induced protein C defi- tions of : a study of 88 cases in northern Finland during 16 years ciency. Clinical Infectious Diseases 17, 254–61. (1967–1983). Acta Dermato-Venereologica 64, 513–6. 108. Sevinsky, L. D., Viecens, C., Ballesteros, D. O. et al. (1993).

86. Pelak, B. A., Bartizal, K., Woods, G. L. et al. (2002). Comparative Ecthyma gangrenosum: a cutaneous manifestation of Pseudomonas at :: on August 8, 2012 in vitro activities of ertapenem against aerobic and facultative bacterial aeruginosa sepsis. Journal of the American Academy of Dermatology pathogens from patients with complicated skin and skin structure infec- 29, 104–6. tions. Diagnostic Microbiology and Infectious Disease 43, 129–33. 109. Fergie, J. E., Patrick, C. C. & Lott, L. (1991). Pseudomonas 87. Arons, M. S., Fernando, L. & Polayes, I. M. (1982). Pasteurella aeruginosa cellulitis and ecthyma gangrenosum in immunocompromised multocida—the major cause of hand infections following domestic animal children. Pediatric Infectious Disease Journal 10, 496–500. bites. Journal of Hand Surgery 7, 47–52. 110. Del Pozo, J., Garcia-Silva, J., Almagro, M. et al. (1998). Ecthyma 88. Weber, D. J., Wolfson, J. S., Swartz, M. N. et al. (1984). Pasteur- gangrenosum-like eruption associated with Morganella morganii infec- ella multocida infections. Report of 34 cases and review of the literature. tion. British Journal of Dermatology 139, 520–1. Medicine (Baltimore) 63, 133–54. 111. Agger, W. A. & Mardan, A. (1995). 89. Hara, H., Ochiai, T., Morishima, T. et al. (2002). Pasteurella canis infections of intact skin. Clinical Infectious Diseases 20, 302–8. osteomyelitis and cutaneous abscess after a domestic dog bite. Journal 112. Keene, W. E., Markum, A. C. & Samadpour, M. (2004). Outbreak of the American Academy of Dermatology 46, Suppl. 5, S151–2. of Pseudomonas aeruginosa infections caused by commercial piercing 90. von Eiff, C., Becker, K., Machka, K. et al. (2001). Nasal carriage as of upper ear cartilage. Journal of the American Medical Association 291, a source of Staphylococcus aureus bacteremia. Study Group. New Eng- 981–5. land Journal of Medicine 344, 11–6. 113. Tweeten, S. S. M. & Rickman, L. S. (1998). Infectious complica- 91. Talan, D. A., Citron, D. M., Abrahamian, F. M. et al. (1999). Bac- tions of body piercing. Clinical Infectious Diseases 26, 735–40. teriologic analysis of infected dog and cat bites. New England Journal of 114. Centers for Disease Control and Prevention. (2000). Pseudono- Medicine 340, 85–92. nas dermatitis/folliculitis associated with pools and hot tubs—Colorado 92. Upton, A. & Taylor, S. (2002). Vibrio vulnificus necrotising fasciitis and Maine, 1999–2000. Morbidity and Mortality Weekly Report 49, and septicaemia. New Zealand Medical Journal 115, 108–9. 1087–91. 93. Joynt, G. M., Gomersall, C. D. & Lyon, D. J. (1999). Severe necro- 115. Inaba, A. S., Zukin, D. D. & Perro, M. (1992). An update on the tising fasciitis of the extremities caused by Vibrionaceae: experience of a evaluation and management of plantar puncture wounds and Pseudo- Hong Kong tertiary hospital. Hong Kong Medical Journal 5, 63–8. monas osteomyelitis. Pediatric Emergency Care 8, 38–44. 94. Bonner, J. R., Coker, A. S., Berryman, C. R. et al. (1983). Spec- 116. Fustes-Morales, A., Gutierrez-Castrellon, P., Duran-Mckinster, C. trum of Vibrio infections in a Gulf Coast community. Annals of Internal et al. (2002). Necrotizing fasciitis: report of 39 pediatric cases. Archives Medicine 99, 464–9. of Dermatology 138, 893–9. 95. Potasman, I., Paz, A. & Odeh, M. (2002). Infectious outbreaks 117. Johnston, D. L., Waldhausen, J. H. & Park, J. R. (2001). Deep soft associated with bivalve shellfish consumption: a worldwide perspective. tissue infections in the neutropenic pediatric oncology patient. Journal of Clinical Infectious Diseases 35, 921–8. Pediatric Hematology and Oncology 23, 443–7.

ii46 Skin and skin-structure infections

118. Summanen, P. H., Talan, D. A., Strong, C. et al. (1995). Bacteriol- Skin and Skin Structure Infection Group. Journal of Antimicrobial Chemo- ogy of skin and soft-tissue infections: comparison of infections in intra- therapy 44, 263–73. venous drug users and individuals with no history of intravenous drug 135. Stevens, D. L., Smith, L. G., Bruss, J. B. et al. (2000). Randomized use. Clinical Infectious Diseases 20, Suppl. 2, S279–82. comparison of linezolid (PNU-100766) versus oxacillin–dicloxacillin for 119. Eady, E. A. & Cove, J. H. (2003). Staphylococcal resistance revis- treatment of complicated skin and soft tissue infections. Antimicrobial ited: community-acquired methicillin resistant Staphylococcus aureus— Agents and Chemotherapy 44, 3408–13. an emerging problem for the management of skin and soft tissue infec- 136. Shah, P. M. & Isaacs, R. D. (2003). Ertapenem, the first of a new tions. Current Opinion in Infectious Diseases 16, 103–24. group of carbapenems. Journal of Antimicrobial Chemotherapy 52, 538– 120. Salgado, C. D., Farr, B. M. & Calfee, D. P. (2003). Community- 43. acquired methicillin-resistant Staphylococcus aureus: a meta-analysis of 137. Fass, R. J., Freimer, E. H. & McCloskey, R. V. (1985).Treatment prevalence and risk factors. Clinical Infectious Diseases 36, 131–9. of skin and soft tissue infections with imipenem/cilastatin. American Jour- 121. Vandenesch, F., Naimi, T., Enright, M. C. et al. (2003). Community- nal of Medicine 78, 110–2. acquired methicillin-resistant Staphylococcus aureus carrying the 138. Nichols, R. L., Smith, J. W., Geckler, R. W. et al. (1995). Mero- Panton–Valentine leukocidin genes: worldwide emergence. Emerging penem versus imipenem/cilastatin in the treatment of hospitalized Infectious Diseases 9, 978–84. patients with skin and soft tissue infections. Southern Medical Journal 88, 122. Brook, I. (1993). Antimicrobial therapy of skin and soft tissue infec- 397–404. tion in children. Journal of the American Podiatric Medical Association 139. Goldstein, E. J., Citron, D. M., Merriam, C. V. et al. (2001). Com- 83, 398–405. parative in vitro activity of ertapenem and 11 other antimicrobial agents 123. Tan, J. S., Wishnow, R. M., Talan, D. A. et al. (1993). Treatment of against aerobic and anaerobic pathogens isolated from skin and soft hospitalized patients with complicated skin and skin structure infections: tissue animal and human bite wound infections. Journal of Antimicrobial double-blind, randomized, multicenter study of piperacillin–tazobactam Chemotherapy 48, 641–51. versus ticarcillin–clavulanate. The Piperacillin/Tazobactam Skin and 140. Chow, A. T., Chen, A., Lattime, H. et al. (2002). Penetration of lev- Skin Structure Study Group. Antimicrobial Agents and Chemotherapy 37, ofloxacin into skin tissue after oral administration of multiple 750 mg Downloaded from 1580–6. once-daily doses. Journal of Clinical Pharmacy and Therapeutics 27, 124. Lipsky, B. A., Baker, P. D., Landon, G. C. et al. (1997). Antibiotic 143–50. therapy for diabetic foot infections: comparison of two parenteral-to-oral 141. Nicodemo, A. C., Robledo, J. A., Jasovich, A. et al. (1998). A multi- regimens. Clinical Infectious Diseases 24, 643–8. centre, double-blind, randomised study comparing the efficacy and 125. Graham, D. R., Lucasti, C., Malafaia, O. et al. (2002). Ertapenem safety of oral levofloxacin versus ciprofloxacin in the treatment of uncom- http://jac.oxfordjournals.org/ once daily versus piperacillin–tazobactam 4 times per day for treatment plicated skin and skin structure infections. International Journal of Clin- of complicated skin and skin-structure infections in adults: results of a ical Practice 52, 69–74. prospective, randomized, double-blind multicenter study. Clinical Infec- 142. Lipsky, B. A., Pecoraro, R. E., Larson, S. A. et al. (1990). Out- tious Diseases 34, 1460–8. patient management of uncomplicated lower-extremity infections in dia- 126. Lautre, G. & Baker, L. W. (1972). Cephalexin therapy of soft-tissue betic patients. Archives of Internal Medicine 150, 790–7. infection. South African Medical Journal 46, 837–9. 143. Legua, P., Lema, J., Moll, J. et al. (2002). Safety and local tolera- 127. Brown, G., Chamberlain, R., Goulding, J. et al. (1996). Ceftriaxone bility of intramuscularly administered ertapenem diluted in lidocaine: a versus cefazolin with probenecid for severe skin and soft tissue infec- prospective, randomized, double-blind study versus intramuscular ceftri- tions. Journal of Emergency Medicine 14, 547–51. axone. Clinical Therapeutics 24, 434–44. at :: on August 8, 2012 128. Grayson, M. L., McDonald, M., Gibson, K. et al. (2002). Once-daily 144. Nathwani, D. (2001). The management of skin and soft tissue intravenous cefazolin plus oral probenecid is equivalent to once-daily infections: outpatient parenteral antibiotic therapy in the United Kingdom. intravenous ceftriaxone plus oral placebo for the treatment of moderate- Chemotherapy 47, Suppl. 1, 17–23. to-severe cellulitis in adults. Clinical Infectious Diseases 34, 1440–8. 145. Martinez-Aguilar, G., Hammerman, W. A., Mason, E. O., Jr et al. 129. Siami, G., Christou, N., Eiseman, I. et al. (2001). Clinafloxacin (2003). Clindamycin treatment of invasive infections caused by community- versus piperacillin–tazobactam in treatment of patients with severe skin acquired, methicillin-resistant and methicillin-susceptible Staphylococcus and soft tissue infections. Antimicrobial Agents and Chemotherapy 45, aureus in children. Pediatric Infectious Disease Journal 22, 593–8. 525–31. 146. Stevens, D. L. (2003). Community-acquired Staphylococcus 130. Graham, D. R., Talan, D. A., Nichols, R. L. et al. (2002). Once- aureus infections: increasing virulence and emerging methicillin resist- daily, high-dose levofloxacin versus ticarcillin–clavulanate alone or fol- ance in the new millennium. Current Opinion in Infectious Diseases 16, lowed by amoxicillin–clavulanate for complicated skin and skin-structure 189–91. infections: a randomized, open-label trial. Clinical Infectious Diseases 147. Naimi, T. S., LeDell, K. H., Como-Sabetti, K. et al. (2003). Com- 35, 381–9. parison of community- and health care-associated methicillin-resistant 131. Gentry, L. O., Rodriguez-Gomez, G., Zeluff, B. J. et al. (1989). A Staphylococcus aureus infection. Journal of the American Medical Asso- comparative evaluation of oral ofloxacin versus intravenous cefotaxime ciation 290, 2976–84. therapy for serious skin and skin structure infections. American Journal 148. Ghanem, A. N., Halim, I. A., Embil, J. M. et al. (2002). Necrotizing of Medicine 87, 57S–60S. fasciitis. Saudi Medical Journal 23, 608–10. 132. Gentry, L. O., Ramirez-Ronda, C. H., Rodriguez-Noriega, E. et al. 149. Childers, B. J., Potyondy, L. D., Nachreiner, R. et al. (2002). (1989). Oral ciprofloxacin vs parenteral cefotaxime in the treatment of Necrotizing fasciitis: a fourteen-year retrospective study of 163 consec- difficult skin and skin structure infections. A multicenter trial. Archives of utive patients. American Surgery 68, 109–16. Internal Medicine 149, 2579–83. 150. Seal, D. V. (2001). Necrotizing fasciitis. Current Opinion in Infec- 133. Vinken, A., Li, Z., Balan, D. et al. (2001). Economic evaluation of tious Diseases 14, 127–32. linezolid, flucloxacillin and vancomycin in the empirical treatment of 151. Hoeffel, J. C. & Hoeffel, F. (2002). Necrotizing fasciitis and purpura cellulitis in UK hospitals: a decision analytical model. Journal of Hospital fulminans. Plastic and Reconstructive Surgery 109, 2165. Infection 49, Suppl. A, S13–24. 152. Brook, I. & Frazier, E. H. (1998). Aerobic and anaerobic micro- 134. Nichols, R. L., Graham, D. R., Barriere, S. L. et al. (1999). Treat- biology of infection after trauma. American Journal of Emergency Medi- ment of hospitalized patients with complicated Gram-positive skin and cine 16, 585–91. skin structure infections: two randomized, multicentre studies of quinu- 153. Brandt, M. M., Corpron, C. A. & Wahl, W. L. (2000). Necrotizing pristin/dalfopristin versus cefazolin, oxacillin or vancomycin. Synercid soft tissue infections: a surgical disease. American Surgery 66, 967–70.

ii47 M. J. DiNubile and B. A. Lipsky

154. Villanueva-Saenz, E., Martinez Hernandez-Magro, P., Valdes, O. 177. Meltzer, R. M., Engel, E. D. & Turf, R. (1983). Postoperative gas M. et al. (2002). Experience in management of Fournier’s gangrene. gangrene. Journal of Foot Surgery 22, 126–33. Techniques in Coloproctology 6, 5–13. 178. Van Hook, R. & Vandevelde, A. G. (1975). Gas gangrene after 155. Hodgson, D. J., Peters, J. L. & Shah, P. J. (2002). Perineal necro- intramuscular injection of epinephrine: report of fatal case. Annals of tizing fasciitis with dilatation of Cowper’s gland. Journal of the Royal Internal Medicine 83, 669–70. Society of Medicine 95, 136–7. 179. Kalani, M., Jorneskog, G., Naderi, N. et al. (2002). Hyperbaric 156. Hollabaugh, R. S., Jr, Dmochowski, R. R., Hickerson, W. L. et al. oxygen (HBO) therapy in treatment of diabetic foot ulcers. Long-term (1998). Fournier’s gangrene: therapeutic impact of hyperbaric oxygen. follow-up. Journal of Diabetes Complications 16, 153–8. Plastic and Reconstructive Surgery 101, 94–100. 180. Parish, L. C., Routh, H. B., Miskin, B. et al. (2000). Moxifloxacin 157. Korhonen, K., Hirn, M. & Niinikoski, J. (1998). Hyperbaric oxygen versus cephalexin in the treatment of uncomplicated skin infections. in the treatment of Fournier’s gangrene. European Journal of Surgery International Journal of Clinical Practice 54, 497–503. 164, 251–5. 181. Gorbach, S. L. & Thadepalli, H. (1975). Isolation of Clostridium in 158. Lucca, M., Unger, H. D. & Devenny, A. M. (1990). Treatment of human infections: evaluation of 114 cases. Journal of Infectious Dis- Fournier’s gangrene with adjunctive hyperbaric oxygen therapy. Ameri- eases 131, Suppl., S81–5. can Journal of Emergency Medicine 8, 385–7. 182. Ramsay, A. M. (1949). The significance of Clostridium welchii in 159. Riegels-Nielsen, P., Hesselfeldt-Nielsen, J., Bang-Jensen, E. et al. the cervical swab and blood-stream in postpartum and postabortium (1984). Fournier’s gangrene: 5 patients treated with hyperbaric oxygen. sepsis. Journal of Obstetrics and Gynaecology 56, 247–8. Journal of Urology 132, 918–20. 183. Jacob, Z. C., Dedekian, M. & Seoudi, H. (2002). Nontraumatic 160. Ziser, A., Girsh, Z., Gozal, D. et al. (1985). Hyperbaric oxygen clostridial myonecrosis: an indication for colonoscopy? American Sur- therapy for Fournier’s gangrene. Critical Care Medicine 13, 773–4. gery 68, 463–5. 161. Yoder, E. L., Mendez, J. & Khatib, R. (1987). Spontaneous gan- 184. Noone, M., Tabaqchali, M. & Spillane, J. B. (2002). Clostridium grenous myositis induced by Streptococcus pyogenes: case report and

novyi causing necrotising fasciitis in an injecting drug user. Journal of Downloaded from review of the literature. Reviews of Infectious Diseases 9, 382–5. Clinical Pathology 55, 141–2. 162. Ben Abraham, R., Keller, N., Vered, R. et al. (2002). Invasive 185. Muhvich, K. H., Anderson, L. H. & Mehm, W. J. (1994). Evaluation group A streptococcal infections in a large tertiary center: epidemiology, of antimicrobials combined with hyperbaric oxygen in a mouse model of characteristics and outcome. Infection 30, 81–5. clostridial myonecrosis. Journal of Trauma 36, 7–10. 163. Francis, B. A., Mui, A. L. & Calonje, D. (2001). Orbital strepto- 186. Bodey, G. P., Rodriguez, S., Fainstein, V. et al. (1991). Clostridial coccal gangrene and AIDS. Orbit 20, 243–8. bacteremia in cancer patients. A 12-year experience. Cancer 67, 1928– http://jac.oxfordjournals.org/ Lang, M. E., Vaudry, W. & Robinson, J. L. (2003). Case report and 164. 42. literature review of late-onset group B streptococcal disease manifesting 187. Traub, W. H. (1988). Chemotherapy of experimental (murine) as necrotizing fasciitis in preterm infants: is this a new syndrome? Clin- Clostridium perfringens type A gas gangrene. Chemotherapy 34, 472–7. ical Infectious Diseases 37, e132–5. 188. Stevens, D. L., Laine, B. M. & Mitten, J. E. (1987). Comparison of 165. Barnham, M. R., Weightman, N. C., Anderson, A. W. et al. (2002). single and combination antimicrobial agents for prevention of experi- Streptococcal toxic shock syndrome: a description of 14 cases from mental gas gangrene caused by Clostridium perfringens. Antimicrobial North Yorkshire, UK. Clinical Microbiology and Infection 8, 174–81. Agents and Chemotherapy 31, 312–6. 166. Shaaban, H., Bayat, A., Davenport, P. et al. (2002). Necrotising 189. Stevens, D. L., Maier, K. A. & Mitten, J. E. (1987). Effect of fasciitis in an infant with congenital insensitivity to pain syndrome. British antibiotics on toxin production and viability of Clostridium perfringens. at :: on August 8, 2012 Journal of Plastic Surgery 55, 160–3. Antimicrobial Agents and Chemotherapy 31, 213–8. 167. Diazgranados, C. & Bisno, A. (2001). Clues to early diagnosis of group A streptococcal necrotizing fasciitis. Infections in Medicine 18, 190. Stevens, D. L., Maier, K. A., Laine, B. M. et al. (1987). Comparison 198–206. of clindamycin, rifampin, tetracycline, metronidazole, and penicillin for efficacy in prevention of experimental gas gangrene due to Clostridium 168. Chelsom, J., Halstensen, A., Haga, T. et al. (1994). Necrotising perfringens. Journal of Infectious Diseases , 220–8. fasciitis due to group A streptococci in western Norway: incidence and 155 clinical features. Lancet 344, 1111–5. 191. Korhonen, K., Klossner, J., Hirn, M. et al. (1999). Management of 169. Stevens, D. L., Gibbons, A. E., Bergstrom, R. et al. (1988). The clostridial gas gangrene and the role of hyperbaric oxygen. Annales Eagle effect revisited: efficacy of clindamycin, erythromycin, and penicil- chirurgiae et gynaecologiae 88, 139–42. lin in the treatment of streptococcal myositis. Journal of Infectious Dis- 192. Stephens, M. B. (1996). Gas gangrene: potential for hyperbaric eases 158, 23–8. oxygen therapy. Postgraduate Medicine 99, 217–24. 170. Stevens, D. L. (2000). Streptococcal toxic shock syndrome asso- 193. Hirn, M., Niinikoski, J. & Lehtonen, O. P. (1992). Effect of hyper- ciated with necrotizing fasciitis. Annual Review of Medicine 51, 271–88. baric oxygen and surgery on experimental gas gangrene. European Sur- 171. Stevens, D. L., Madaras-Kelly, K. J. & Richards, D. M. (1998). In gery Research 24, 356–62. vitro antimicrobial effects of various combinations of penicillin and clinda- 194. Trivedi, D. R. & Raut, V. V. (1990). Role of hyperbaric oxygen mycin against four strains of Streptococcus pyogenes. Antimicrobial therapy in the rapid control of gas gangrene infection and its toxaemia. Agents and Chemotherapy 42, 1266–8. Journal of Postgraduate Medicine 36, 13–15. 172. Burge, T. S. (1995). Necrotizing fasciitis—the hazards of delay. 195. Slack, W. K. (1976). Hyperbaric oxygen therapy in anaerobic Journal of the Royal Society of Medicine 88, 342P–343P. infections: gas gangrene. Proceedings of the Royal Society of Medicine 173. Bilton, B. D., Zibari, G. B., McMillan, R. W. et al. (1998). Aggres- 69, 326–7. sive surgical management of necrotizing fasciitis serves to decrease 196. Larcan, A., Laprevote-Heully, M. C., Lambert, H. et al. (1974). Gas mortality: a retrospective study. American Surgery 64, 397–400. gangrene—on the basis of 19 observations. Advantages of combining 174. Qvist, G. (1941). Anaerobic cellulitis and gas gangrene. British surgery with hyperbaric oxygen therapy. Materia Medica Polona 6, 116– Medical Journal 2, 217. 9. 175. Cline, K. A. & Turnbull, T. L. (1985). Clostridial myonecrosis. 197. Schweigel, J. F. & Shim, S. S. (1973). A comparison of the treat- Annals of Emergency Medicine 14, 459–66. ment of gas gangrene with and without hyperbaric oxygen. Surgery, 176. Johnson, J. T., Gillespie, T. E., Cole, J. R. et al. (1969). Hyperbaric Gynecology and Obstetrics 136, 969–70. oxygen therapy for gas gangrene in war wounds. American Journal of 198. Maudsley, R. H. & Arden, G. P. (1972). Gas gangrene and hyper- Surgery 118, 839–43. baric oxygen. British Medical Journal 4, 362.

ii48 Skin and skin-structure infections

199. Roding, B., Groeneveld, P. H. & Boerema, I. (1972). Ten years of 221. Barnham, M. (1997). Nonsteroidal antiinflammatory drugs: con- experience in the treatment of gas gangrene with hyperbaric oxygen. current or causative drugs in serious infection? Clinical Infectious Dis- Surgery, Gynecology and Obstetrics 134, 579–85. eases 25, 1272–3. 200. Slack, W. K., Hanson, G. C. & Chew, H. E. (1969). Hyperbaric 222. Barnham, M. & Anderson, A. W. (1997). Non-steroidal anti- oxygen in the treatment of gas gangrene and clostridial infection. A report inflammatory drugs (NSAIDs). A predisposing factor for streptococcal of 40 patients treated in a single-person hyperbaric oxygen chamber. bacteraemia? Advances in Experimental Medicine and Biology 418, British Journal of Surgery 56, 505–10. 145–7. 201. Colwill, M. R. & Maudsley, R. H. (1968). The management of gas 223. Stevens, D. L. (1995). Could nonsteroidal antiinflammatory drugs gangrene with hyperbaric oxygen therapy. Journal of Bone and Joint Sur- (NSAIDs) enhance the progression of bacterial infections to toxic shock gery. British volume 50, 732–42. syndrome? Clinical Infectious Diseases 21, 977–80. 202. Hamsa, W. R. & Burney, D. W., Jr (1966). Gas gangrene infection: 224. Aronoff, D. M. & Bloch, K. C. (2003). Assessing the relationship combined treatment including 3,432,000 international units of polyvalent between the use of nonsteroidal antiinflammatory drugs and necrotizing gas gangrene antitoxin. Nebraska State Medical Journal 51, 85–9. fasciitis caused by Group A Streptococcus. Medicine 82, 225–35. 203. Gold, W. L. & Salit, I. E. (1993). Aeromonas hydrophila infections 225. Bergkvist, P. I. & Sjobeck, K. (1997). Antibiotic and prednisolone of skin and soft tissue: report of 11 cases and review. Clinical Infectious therapy of erysipelas: a randomized, double blind, placebo-controlled Diseases 16, 69–74. study. Scandinavian Journal of Infectious Diseases 29, 377–82. 204. Mackay, D. R., Manders, E. K., Saggers, G. C. et al. (1999). 226. Bergkvist, P. I. & Sjobeck, K. (1998). Relapse of erysipelas follow- Aeromonas species isolated from medicinal leeches. Annals of Plastic ing treatment with prednisolone or placebo in addition to antibiotics: a Surgery 42, 275–9. 1-year follow-up. Scandinavian Journal of Infectious Diseases 30, 206–7. 205. Eron, L. J. & Passos, S. (2001). Early discharge of infected 227. Baroni, G., Porro, T., Faglia, E. et al. (1987). Hyperbaric oxygen in patients through appropriate antibiotic use. Archives of Internal Medicine diabetic gangrene treatment. Diabetes Care 10, 81–6. 161, 61–5. 228. Eltorai, I. M., Hart, G. B., Strauss, M. B. et al. (1986). The role of

206. Kielhofner, M. A., Brown, B & Dall, L. (1988). Influence of under- hyperbaric oxygen in the management of Fournier’s gangrene. Inter- Downloaded from lying disease process on the utility of cellulitis needle aspirates. Archives national Surgery 71, 53–8. of Internal Medicine 148, 2451–2. 229. McDonald, D. F., Hulet, W. H. & Cowan, J. W. (1975). Scrotal 207. Sachs, M. K. (1990). The optimum use of needle aspiration in the gangrene treated with oxygen under high pressure. Journal of Urology bacteriologic diagnosis of cellulitis in adults. Archives of Internal Medi- 113, 364–6. cine 150, 1907–12. [Erratum, (1991). Archives of Internal Medicine 151, 230. Yuen, J. C. & Feng, Z. (2002). Salvage of limb and function in

244] necrotizing fasciitis of the hand: role of hyperbaric oxygen treatment and http://jac.oxfordjournals.org/ 208. Howe, P. M., Eduardo Fajardo, J. & Orcutt, M. A. (1987). Etiologic free muscle flap coverage. Southern Medical Journal 95, 255–7. diagnosis of cellulitis: comparison of aspirates obtained from the leading 231. Ulkur, E., Yuksel, F., Acikel, C. et al. (2002). Effect of hyperbaric edge and the point of maximal inflammation. Pediatric Infectious Disease oxygen on pedicle flaps with compromised circulation. Microsurgery 22, Journal 6, 685–6. 16–20. 209. Perl, B., Gottehrer, N. P., Raveh, D. et al. (1999). Cost-effective- 232. Owen-Smith, M. S. (1968). Antibiotics and antitoxin therapy in the ness of blood cultures for adult patients with cellulitis. Clinical Infectious prophylaxis of experimental gas gangrene. British Journal of Surgery 55, Diseases 29, 1483–8. 43–5. 210. Sadow, K. B. & Chamberlain, J. M. (1998). Blood cultures in the 233. Shemanova, G. F., Vlasova, E. V. & Postnikova, T. M. (1988). On

evaluation of children with cellulitis. Pediatrics 101, E4. the in vitro neutralization test of Clostridium perfringens type A toxin. at :: on August 8, 2012 211. Struk, D. W., Munk, P. L., Lee, M. J. et al. (2001). Imaging of soft Journal of Hygiene, Epidemiology, Microbiology, and Immunology 32, tissue infections. Radiologic Clinics of North America 39, 277–303. 219–26. 212. Stamenkovic, I. & Lew, P. D. (1984). Early recognition of poten- 234. Jude, E. B., Blakytny, R., Bulmer, J. et al. (2002). Transforming tially fatal necrotizing fasciitis. The use of frozen-section biopsy. New -β 1, 2, 3 and receptor type I and II in diabetic foot ulcers. England Journal of Medicine 310, 1689–93. Diabetic Medicine 19, 440–7. 213. Armstrong, D. G. & Nguyen, H. C. (2000). Improvement in healing 235. Shackelford, D. P., Fackler, E., Hoffman, M. K. et al. (2002). Use with aggressive edema reduction after debridement of foot infection in of topical recombinant human platelet-derived growth factor BB in persons with diabetes. Archives of Surgery 135, 1405–9. abdominal wound separation. American Journal of Obstetrics and Gyne- 214. Edmonds, M., Wilson, M. & Foster, A. (1999). Diabetic foot ulcers. cology 186, 701–4. Nursing Standard 14, 39–45. 236. Fu, X., Shen, Z., Guo, Z. et al. (2002). Healing of chronic cutane- 215. Cox, N. H. (2002). Management of lower leg cellulitis. Clinical ous wounds by topical treatment with basic fibroblast growth factor. Medicine 2, 23–7. Chinese Medical Journal (Engl) 115, 331–5. 216. Brun-Buisson, C. J., Saada, M., Trunet, P. et al. (1985). Haemo- 237. Mann, A., Breuhahn, K., Schirmacher, P. et al. (2001). Keratinoc- lytic streptococcal gangrene and non-steroidal antiinflammatory drugs. yte-derived granulocyte-macrophage colony stimulating factor acceler- British Medical Journal (Clinical Research Edition) 290, 1786. ates : stimulation of keratinocyte proliferation, granulation 217. Choo, P. W., Donahue, J. G. & Platt, R. (1997). Ibuprofen and skin tissue formation, and vascularization. Journal of Investigative Dermatol- and soft tissue superinfections in children with varicella. Annals of Epi- ogy 117, 1382–90. demiology 7, 440–5. 238. Rumalla, V. K. & Borah, G. L. (2001). Cytokines, growth factors, 218. Forbes, N. & Rankin, A. P. (2001). Necrotizing fasciitis and non- and plastic surgery. Plastic and Reconstructive Surgery 108, 719–33. steroidal antiinflammatory drugs: a case series and review of the litera- 239. Golden, S. H., Peart-Vigilance, C., Kao, W. H. et al. (1999). Peri- ture. New Zealand Medical Journal 114, 3–6. operative glycemic control and the risk of infectious complications in a 219. Guibal, F., Muffat-Joly, M., Terris, B. et al. (1998). Effects of cohort of adults with diabetes. Diabetes Care 22, 1408–14. diclofenac on experimental streptococcal necrotizing fasciitis (NF) in 240. Pomposelli, J. J., Baxter, J. K., III, Babineau, T. J. et al. (1998). rabbit. Archives in Dermatological Research 290, 628–33. Early postoperative glucose control predicts nosocomial infection rate in 220. Kaul, R., McGeer, A., Low, D. E. et al. (1997). Population-based diabetic patients. Journal of Parenteral and Enteral Nutrition 22, 77–81. surveillance for group A streptococcal necrotizing fasciitis: clinical 241. Van den Berge, G. (2002). Intensive insulin therapy in critically ill features, prognostic indicators, and microbiologic analysis of seventy- patients. New England Journal of Medicine 345, 1359–67. seven cases. Ontario Group A Streptococcal Study. American Journal of 242. Furnary, A. P., Zerr, K. J., Grunkemeier, G. L. et al. (1999). Medicine 103, 18–24. Continuous intravenous insulin infusion reduces the incidence of deep

ii49 M. J. DiNubile and B. A. Lipsky sternal wound infection in diabetic patients after cardiac surgical pro- 246. Ford, C. N., Reinhard, E. R., Yeh, D. et al. (2002). Interim analysis cedures. Annals of Thoracic Surgery 67, 352–60. of a prospective, randomized trial of vacuum-assisted closure versus the 243. Faucher, L. D., Morris, S. E., Edelman, L. S. et al. (2001). Burn healthpoint system in the management of pressure ulcers. Annals of center management of necrotizing soft-tissue surgical infections in Plastic Surgery 49, 55–61. unburned patients. American Journal of Surgery 182, 563–9. 247 Gardner, S. E. , Frantz, R. A. & Schmidt, F. L. (1999). Effect of elec- 244. Dunn, C., Raghavan, U. & Pfleiderer, A. G. (2002). The use of trical stimulation on chronic wound healing: a meta-analysis. Wound maggots in head and neck necrotizing fasciitis. Journal of Laryngology Repair Regeneration 7, 495–503. and Otology 116, 70–2. 248 Peters, E. J., Lavery, L. A., Armstrong, D. J. et al. (2001). Electric 245. Armstrong, D. G., Lavery, L. A., Abu-Rumman, P. et al. (2002). stimulation as an adjunct to heal diabetic foot ulcers: a randomized clin- Outcomes of subatmospheric pressure dressing therapy on wounds of ical trial. Archives Physical and Medical Rehabilitation 82, 721–5. the diabetic foot. Ostomy/Wound Management 48, 64–8. Downloaded from http://jac.oxfordjournals.org/ at :: on August 8, 2012

ii50 IDSA GUIDELINES

Practice Guidelines for the Diagnosis and Management of Skin and Soft-Tissue Infections

Dennis L. Stevens,1,3 Alan L. Bisno,5 Henry F. Chambers,6,7 E. Dale Everett,13 Patchen Dellinger,2 Ellie J. C. Goldstein,8,9 Sherwood L. Gorbach,14 Jan V. Hirschmann,3,4 Edward L. Kaplan,15,16 Jose G. Montoya,10,11,12 and James C. Wade17 1Infectious Diseases Section, Veterans Affairs Medical Center, Boise, Idaho; 2Department of Surgery, 3University of Washington School of Medicine, and 4Seattle Veterans Affairs Medical Center, Seattle, Washington; 5University of Miami Miller School of Medicine, Miami, Florida; 6Infectious Diseases, San Francisco General Hospital, and 7University of California–San Francisco, San Francisco, 8R. M. Alden Research Laboratory, Santa Monica, 9University of California, Los Angeles School of Medicine, Los Angeles, and 10Department of Medicine and 11Division of Infectious Diseases and Geographic Medicine, Stanford University School of Medicine, and 12Research Institute, Palo Alto Medical Foundation, Palo Alto, California; 13University of Missouri Health Science Center, University of Missouri, Columbia; 14Tufts University School of Medicine, Boston, Massachusetts; 15University of Minnesota Medical School and 16Division of Epidemiology, University of Minnesota School of Public Health, Minneapolis, Minnesota; and 17Division of Neoplastic Diseases and Related Disorders, Medical College of Wisconsin, Milwaukee, Wisconsin

EXECUTIVE SUMMARY ters: results of blood culture and drug susceptibility tests, complete blood cell count with differential, and Soft-tissue infections are common, generally of mild to creatinine, bicarbonate, creatine phosphokinase, and C- modest severity, and are easily treated with a variety of reactive protein levels. In patients with hypotension agents. An etiologic diagnosis of simple cellulitis is fre- and/or an elevated creatinine level, low serum bicar- quently difficult and generally unnecessary for patients bonate level, elevated creatine phosphokinase level (2– with mild signs and symptoms of illness. Clinical as- 3 times the upper limit of normal), marked left shift, sessment of the severity of infection is crucial, and sev- or a C-reactive protein level 113 mg/L, hospitalization eral classification schemes and algorithms have been should be considered and a definitive etiologic diag- proposed to guide the clinician [1]. However, most nosis pursued aggressively by means of procedures such clinical assessments have been developed from either as Gram stain and culture of needle aspiration or punch retrospective studies or from an author’s own “clinical biopsy specimens, as well as requests for a surgical con- experience,” illustrating the need for prospective studies sultation for inspection, exploration, and/or drainage. with defined measurements of severity coupled to man- Other clues to potentially severe deep soft-tissue infec- agement issues and outcomes. tion include the following: (1) pain disproportionate Until then, it is the recommendation of this com- to the physical findings, (2) violaceous bullae, (3) cu- mittee that patients with soft-tissue infection accom- taneous hemorrhage, (4) skin sloughing, (5) skin an- panied by signs and symptoms of systemic toxicity (e.g., esthesia, (6) rapid progression, and (7) gas in the tissue. fever or hypothermia, tachycardia [heart rate, 1100 beats/min], and hypotension [systolic blood pressure, Unfortunately, these signs and symptoms often appear !90 mm Hg or 20 mm Hg below baseline]) have blood later in the course of necrotizing infections. In these drawn to determine the following laboratory parame- cases, emergent surgical evaluation is of paramount im- portance for both diagnostic and therapeutic reasons. Emerging antibiotic resistance among Staphylococcus

Received 13 July 2005; accepted 14 July 2005; electronically published 14 aureus (methicillin resistance) and Streptococcus pyoge- October 2005. nes (erythromycin resistance) are problematic, because These guidelines were developed and issued on behalf of the Infectious both of these organisms are common causes of a variety Diseases Society of America. Reprints or correspondence: Dr. Dennis L. Stevens, Infectious Disease Section, of skin and soft-tissue infections and because empirical VAMC, 500 West Fort St. (Bldg. 45), Boise, ID 83702 ([email protected]). choices of antimicrobials must include agents with ac- Clinical Infectious Diseases 2005;41:1373–406 2005 by the Infectious Diseases Society of America. All rights reserved. tivity against resistant strains. Minor skin and soft-tis- 1058-4838/2005/4110-0001$15.00 sue infections may be empirically treated with semi-

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1373 synthetic penicillin, first-generation or second-generation oral decision of how to treat impetigo depends on the number of cephalosporins, macrolides, or clindamycin (A-I); however, lesions, their location (face, eyelid, or mouth), and the need 50% of methicillin-resistant S. aureus (MRSA) strains have in- to limit spread of infection to others. The best topical agent is ducible or constitutive clindamycin resistance [2] (table 1). mupirocin (A-I), although resistance has been described [5]; Most community-acquired MRSA strains remain susceptible to other agents, such as bacitracin and neomycin, are considerably trimethoprim-sulfamethoxazole and tetracycline, though treat- less effective treatments. Patients who have numerous lesions ment failure rates of 21% have been reported in some series or who are not responding to topical agents should receive oral with doxycycline or minocycline [3]. Therefore, if patients are antimicrobials effective against both S. aureus and S. pyogenes sent home receiving these regimens, it is prudent to reevaluate (A-I) (table 2). Although rare in developed countries (!1 case/ them in 24–48 h to verify a clinical response. Progression de- 1,000,000 population per year), glomerulonephritis following spite receipt of antibiotics could be due to infection with re- streptococcal infection may be a complication of impetigo sistant microbes or because a deeper, more serious infection caused by certain strains of S. pyogenes, but no data demonstrate exists than was previously realized. that treatment of impetigo prevents this sequela. Patients who present to the hospital with severe infection or Classically, erysipelas, is a fiery red, tender, painful plaque whose infection is progressing despite empirical antibiotic ther- with well-demarcated edges and is commonly caused by strep- apy should be treated more aggressively, and the treatment tococcal species, usually S. pyogenes. strategy should be based upon results of appropriate Gram Cellulitis may be caused by numerous organisms that are stain, culture, and drug susceptibility analysis. In the case of S. indigenous to the skin or to particular environmental niches. aureus, the clinician should assume that the organism is resis- Cellulitis associated with furuncles, , or abscesses is tant, because of the high prevalence of community-associated usually caused by S. aureus. In contrast, cellulitis that is diffuse MRSA strains, and agents effective against MRSA (i.e., van- or unassociated with a defined portal is most commonly caused comycin, linezolid, or daptomycin) should be used (A-I). Step- by streptococcal species. Important clinical clues to other causes down to treatment with other agents, such as tetracycline or include physical activities, trauma, water contact, and animal, trimethoprim-sulfamethoxazole, for MRSA infection may be insect, or human bites. In these circumstances appropriate cul- possible, based on results of susceptibility tests and after an ture material should be obtained, as they should be in patients initial clinical response. In the United States, not all laboratories who do not respond to initial empirical therapy directed against perform susceptibility testing on S. pyogenes. However, the Cen- S. aureus and S. pyogenes and in immunocompromised hosts. ters for Disease Control and Prevention has provided national Unfortunately, aspiration of skin is not helpful in 75%–80% surveillance data that suggest a gradual trend of increasing mac- of cases of cellulitis, and results of blood cultures are rarely rolide resistance of S. pyogenes from 4%–5% in 1996–1998 to positive (!5% of cases). 8%–9% in 1999–2001 [4]. Of interest, 99.5% of strains remain Penicillin, given either parenterally or orally depending on susceptible to clindamycin, and 100% are susceptible to clinical severity, is the treatment of choice for erysipelas (A-I). penicillin. For cellulitis, a penicillinase-resistant semisynthetic penicillin Impetigo, erysipelas, and cellulitis. Impetigo may be or a first-generation cephalosporin should be selected (A-I), caused by infection with S. aureus and/or S. pyogenes. The unless streptococci or staphylococci resistant to these agents

Table 1. Infectious Diseases Society of America–US Public Health Service Grading System for ranking recommendations in clinical guidelines.

Category, grade Definition Strength of recommendation A Good evidence to support a recommendation for use; should always be offered B Moderate evidence to support a recommendation for use; should generally be offered C Poor evidence to support a recommendation; optional D Moderate evidence to support a recommendation against use; should generally not be offered E Good evidence to support a recommendation against use; should never be offered Quality of evidence I Evidence from у1 properly randomized, controlled trial II Evidence from у1 well-designed clinical trial, without randomization; from cohort or case-con- trolled analytic studies (preferably from 11 center); from multiple timeseries; or from dra- matic results from uncontrolled experiments III Evidence from opinions of respected authorities, based on clinical experience, descriptive studies, or reports of expert committees

1374 • CID 2005:41 (15 November) • Stevens et al. are common in the community. For penicillin-allergic patients, the bite occurred. Patients not allergic to penicillin should re- choices include clindamycin or vancomycin. ceive treatment with oral amoxicillin-clavulanate or with in- Lack of clinical response could be due to unusual organisms, travenous ampicillin-sulbactam or ertapenem (B-II), because resistant strains of staphylococcus or streptococcus, or deeper agents such as dicloxacillin, cephalexin, erythromycin, and clin- processes, such as necrotizing fasciitis or myonecrosis. In pa- damycin have poor activity against Pasteurella multocida.Al- tients who become increasingly ill or experience increasing tox- though cefuroxime, cefotaxime, and ceftriaxone are effective icity, necrotizing fasciitis, myonecrosis, or toxic shock syn- against P. multocida, they do not have good anaerobic spectra. drome should be considered, an aggressive evaluation initiated, Thus, cefoxitin or carbapenem antibiotics could be used par- and antibiotic treatment modified, on the basis of Gram stain enterally in patients with mild penicillin . Patients with results, culture results, and antimicrobial susceptibilities of or- previous severe reactions can receive oral or intravenous doxy- ganisms obtained from surgical specimens. cycline, trimethoprim-sulfamethoxazole, or a fluoroquinolone Necrotizing infections. Necrotizing fasciitis may be mon- plus clindamycin. omicrobial and caused by S. pyogenes, Vibrio vulnificus, or Aero- Human bites may occur from accidental injuries, purposeful monas hydrophila. Recently, necrotizing fasciitis was described biting, or closed fist injuries. The bacteriologic characteristics in a patient with MRSA infection [7]. Polymicrobial necrotizing of these wounds are complex but include infection with aerobic fasciitis may occur following surgery or in patients with pe- bacteria, such as streptococci, S. aureus, and Eikenella corrodens, ripheral vascular disease, diabetes mellitus, decubitus ulcers, as well as with multiple anaerobic organisms, including Fuso- and spontaneous mucosal tears of the gastrointestinal or gas- bacterium, Peptostreptococcus, Prevotella, and Porphyromonas trourinary tract (i.e., ). As with clostridial species. E. corrodens is resistant to first-generation cephalospo- myonecrosis, gas in the deep tissues is frequently found in these rins, macrolides, clindamycin, and aminoglycosides. Thus, in- mixed infections. travenous treatment with ampicillin-sulbactam or cefoxitin is Gas gangrene is a rapidly progressive infection caused by the best choice (B-III). Clostridium perfringens, Clostridium septicum, Clostridium his- Infections associated with animal contact. Infections as- tolyticum, or Clostridium novyi. Severe penetrating trauma or sociated with animal contact, although uncommon, are fre- crush injuries associated with interruption of the blood supply quently severe, sometimes lethal, and diagnosticallychallenging. are the usual predisposing factors. C. perfringens and C. novyi The potential use of Bacillus anthracis, Francisella tularensis, infections have recently been described among heroin abusers and Yersinia pestis for bioterrorism has generated great interest following intracutaneous injection of black tar heroin. C. sep- in rapid diagnostic techniques, because early recognition and ticum, a more aerotolerant Clostridium species, may cause treatment are essential. Doxycycline or ciprofloxacin therapy is spontaneous gas gangrene in patients with colonic lesions (such recommended in standard doses for nonpregnant adults and as those due to diverticular disease), adenocarcinoma, or children 18 years of age, pending identification of the offending neutropenia. agent (B-III). Necrotizing fasciitis and gas gangrene may cause necrosis of Adults and children who receive a diagnosis of tularemia skin, subcutaneous tissue, and muscle. Cutaneous findings of should receive an aminoglycoside, preferably streptomycin or purple bullae, sloughing of skin, marked edema, and systemic gentamicin, for 7–10 days. In mild cases, doxycycline or tet- toxicity mandate prompt surgical intervention. For severe racycline for 14 days is recommended (B-III) (comments re- group A streptococcal and clostridial necrotizing infections, garding treatment of children !8 years of age are specified in parenteral clindamycin and penicillin treatment is recom- table 3). Patients with bubonic should receive strepto- mended (A-II). A variety of antimicrobials directed against aer- mycin, tetracycline, or chloramphenicol for 10–14 days and obic gram-positive and gram-negative bacteria, as well as should be placed in isolation for 48 h after initiation of treat- against anaerobes, may be used in mixed necrotizing infections ment, because some patients may develop secondary pneu- (B-II). monic plague (B-III). Infections following animal or human bites. Animal bites Data regarding antibiotic efficacy for treatment of cat-scratch account for 1% of all emergency department visits, and dog disease are inconclusive, although 1 small study demonstrated bites are responsible for 80% of such cases. Although Pasteurella more-rapid lymph node regression in patients receiving azith- species are the most common isolates, cat and dog bites contain romycin, compared with patients receiving no treatment. Cu- an average of 5 different aerobic and anaerobic bacteria per taneous has not been systematicallystud- wound, often including S. aureus, Bacteroides tectum, and Fu- ied, but treatment with erythromycin or doxycycline in sobacterium, Capnocytophaga, and Porphyromonas species. The standard doses for 4 weeks has been effective in very small decision to administer oral or parenteral antibiotics depends series (B-III). on the depth and severity of the wound and on the time since On the basis of very incomplete data, erysipeloid is best

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1375 Table 2. Antimicrobial therapy for impetigo and for skin and soft-tissue infections.

Dosage Antibiotic therapy, by disease Adults Childrena Comment Impetigob Dicloxacillin 250 mg 4 times per day po 12 mg/kg/day in 4 divided doses po … Cephalexin 250 mg 4 times per day po 25 mg/kg/day in 4 divided doses po … Erythromycin 250 mg 4 times per day poc 40 mg/kg/day in 4 divided doses po Some strains of Staphylococcus aureus and Streptococcus pyogenes may be resistant Clindamycin 300–400 mg 3 times per day po 10–20 mg/kg/day in 3 divided doses po … Amoxicillin/clavulanate 875/125 mg twice per day po 25 mg/kg/day of the amoxicillin compo- … nent in 2 divided doses po Mupirocin ointment Apply to lesions 3 times per day Apply to lesions 3 times per day For patients with a limited number of lesions MSSA SSTI Nafcillin or oxacillin 1–2 g every 4 h iv 100–150 mg/kg/day in 4 divided doses Parental drug of choice; inactive against MRSA Cefazolin 1 g every 8 h iv 50 mg/kg/day in 3 divided doses For penicillin-allergic patients, except those with immediate hypersensitivity reactions Clindamycin 600 mg/kg every 8 h iv or 300–450 25–40 mg/kg/day in 3 divided doses iv Bacteriostatic; potential of cross-resis- mg 3 times per day po or 10–20 mg/kg/day in 3 divided tance and emergence of resistance in doses po erythromycin-resistant strains; inducible resistance in MRSA Dicloxacillin 500 mg 4 times per day po 25 mg/kg/day in 4 divided doses po Oral agent of choice for methicillin-sus- ceptible strains Cephalexin 500 mg 4 times per day po 25 mg/kg/day in 4 divided doses po For penicillin-allergic patients, except those with immediate hypersensitivity reactions Doxycycline, minocycline 100 mg twice per day po Not recommended for persons aged Bacteriostatic; limited recent clinical !8 yearsd experience TMP-SMZ 1 or 2 double-strength tablets twice 8–12 mg/kg (based on the trimethoprim Bactericidal; efficacy poorly documented per day po component) in either 4 divided doses iv or 2 divided doses po MRSA SSTI Vancomycin 30 mg/kg/day in 2 divided doses iv 40 mg/kg/day in 4 divided doses iv For penicillin-allergic patients; parenteral drug of choice for treatment of infec- tions caused by MRSA Linezolid 600 mg every 12 h iv or 600 mg 10 mg/kg every 12 h iv or po Bacteriostatic; limited clinical experience; twice per day po no cross-resistance with other antibi- otic classes; expensive; may eventually replace other second-line agents as a preferred agent for oral therapy of MRSA infections Clindamycin 600 mg/kg every 8 h iv or 300–450 25–40 mg/kg/day in 3 divided doses iv Bacteriostatic; potential of cross-resis- mg 3 times per day po or 10–20 mg/kg/day in 3 divided tance and emergence of resistance in doses po erythromycin-resistant strains; inducible resistance in MRSA Daptomycin 4 mg/kg every 24 h iv Not applicable Bactericidal; possible myopathy Doxycycline, minocycline 100 mg twice per day po Not recommended for persons aged Bacteriostatic, limited recent clinical !8 yearsd experience TMP-SMZ 1 or 2 double-strength tablets twice 8–12 mg/kg/day (based on the trimetho- Bactericidal; limited published efficacy per day po prim component) in either 4 divided data doses iv or 2 divided doses po

NOTE. MRSA, methicillin-resistant S. aureus; MSSA, methicillin-susceptible S. aureus; SSTI, skin and soft-tissue infection; TMP-SMZ, trimethoprim-sulfa- methoxazole. iv, intravenously; po, orally. a Doses listed are not appropriate for neonates. Refer to the report by the Committee on Infectious Diseases, American Academy of Pediatrics [6] for neonatal doses. b Infection due to Staphylococcus and Streptococcus species. Duration of therapy is ∼7 days, depending on the clinical response. c Adult dosage of erythromycin ethylsuccinate is 400 mg 4 times per day po. d See [6] for alternatives in children. Table 3. Antibiotic therapy for community-acquired and bioterrorism-related cutaneous anthrax.

Dosage Antibiotic therapy, by route of anthrax acquisition Adults Childrena Community acquired Penicillin V 200–500 mg po 4 times daily in divided doses 25–50 mg/kg/day in divided doses 2 or 4 times per day Penicillin G 8–12 MU/day iv in divided doses every 4-6 h 100,000–150,000 U/kg/day iv in divided doses every 4-6 h Amoxicillin 500 mg po every 8 h Persons who weigh р20 kg: 500 mg po every 8 h; persons who weigh !20 kg: 40 mg/kg po in divided doses every 8 h Erythromycin 250 mg po every 6 h 40 mg/kg/day in divided doses every 6 h Erythromycin lactobionate 15–20 mg/kg (4 g maximum) iv in divided 20–40 mg/kg/day iv in divided doses every 6 h doses every 6 h Tetracycline 250–500 mg po or iv every 6 h … Doxycyclineb 100 mg twice per day po or iv … Ciprofloxacinb 500 mg twice per day or 400 mg iv every 12 h … Bioterrorism or suspected bioterrorism Doxycyclineb 100 mg twice per day po or iv Persons who weigh р45 kg: 2.2 mg/kg every 12 h; persons who weigh 145 kg: 100 mg twice per day po or iv Ciprofloxacinb 500 mg twice per day 10–15 mg/kg every 12 h po or iv (not to exceed 1 g in 24 h)

NOTE. As a rule, the use of fluoroquinolones is contraindicated by the US Food and Drug Administration for children and adolescents !18 years of age. It should also be noted that tetracyclines are rarely used in children !8 years of age. Alternatives should be strongly considered for these 2 antibiotics [6]. iv, intravenously; po, orally. a Dosages listed for children are not appropriate for neonates. Refer to the report by the Committee on Infectious Diseases, American Academy of Pediatrics [6] for neonatal dosing regimens. b Doxycycline, tetracycline, and ciprofloxacin are not generally recommended during pregnancy or for children !8 years of age, except in exceptional circumstances. treated with oral penicillin or amoxicillin for 10 days (B-III). occur as part of a broader systemic infection; and (3) the degree E. rhusiopathiae is resistant in vitro to vancomycin, teicoplanin, and type of immune deficiency attenuate the clinical findings. and daptomycin (E-III). The importance of establishing a diagnosis and performing Surgical site infections. Surgical soft-tissue infections in- susceptibility testing is crucial, because many infections are clude those occurring postoperatively and those severe hospital acquired, and mounting resistance among both gram- enough to require surgical intervention for diagnosis and positive and gram-negative bacteria make dogmatic empirical treatment. The algorithm presented clearly indicates that sur- treatment regimens difficult, if not dangerous. In addition, fun- gical site infection rarely occurs during the first 48 h after gal infections may present with cutaneous findings. surgery, and fever during that period usually arises from non- Immunocompromised patients who are very ill or experi- infectious or unknown causes. In contrast, after 48 h, surgical encing toxicity typically require very broad-spectrum empirical site infection is a more common source of fever, and careful agents that include specific coverage for resistant gram-positive inspection of the wound is indicated. For patients with a bacteria, such as MRSA (e.g., vancomycin, linezolid, dapto- temperature !38.5C and without tachycardia, observation, mycin, or quinupristin/dalfopristin). Coverage for gram-neg- dressing changes, or opening the incision site suffices. Patients ative bacteria may include monotherapy with a cephalosporin with a temperature 138.5C or a heart rate 1110 beats/min possessing activity against Pseudomonas species, with carba- generally require antibiotics as well as opening of the suture penems, or with a combination of either a fluoroquinolone or line. Infections developing after surgical procedures involving an aminoglycoside plus either an extended-spectrum penicillin nonsterile tissue, such as colonic, vaginal, biliary or respira- or cephalosporin. tory mucosa, may be caused by a combination of aerobic and Infections in patients with cell-mediated immunodeficiency anaerobic bacteria. These infections can rapidly progress and (such as that due to Hodgkin disease, lymphoma, HIV infec- involve deeper structures than just the skin, such as fascia, tion, bone marrow transplantation, and receipt of long-term fat, or muscle (see table 4). high-dose immunosuppressive therapy) can be caused by either Infections in the immunocompromised host. Skin and soft common or unusual bacteria, viruses, protozoa, helminths, or tissues are common sites of infection in compromised hosts fungi. Although infection may begin in the skin, cutaneous and usually pose major diagnostic challenges for the following lesions can also be the result of hematogenous seeding. A well- 3 reasons: (1) infections are caused by diverse organisms, in- planned strategy for prompt diagnosis, including biopsy and cluding organisms not ordinarily considered to be pathogens aggressive treatment protocols, is essential. Diagnostic strategies in otherwise healthy hosts; (2) infection of the soft tissues may require laboratory support capable of rapid processing and early

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1377 detection of bacteria (including Mycobacteria and Nocardia spe- Table 4. Antibiotic choices for incisional surgical site infec- cies), viruses, and fungi. The algorithm presented provides an tions (SSIs). approach to diagnosis and treatment. The empirical antibiotic Antibiotic therapy for SSIs, by site of operation guidelines are based on results of clinical trials, national sur- veillance antibiograms, and consensus meetings. Because an- Intestinal or genital tract timicrobial susceptibilities vary considerably across the nation, Single agents Cefoxitin clinicians must base empirical treatment on the antibiograms Ceftizoxime in their own location. Ampicillin/sulbactam Microbiologic cultures are important in establishing a spe- Ticarcillin/clavulanate cific diagnosis, and testing the drug susceptibility of organisms Piperacillin/tazobactam is critical for optimal antimicrobial treatment. This guideline Imipenem/cilastatin offers recommendations for empirical treatment of specific Meropenem community-acquired and hospital-acquired infections. None- Ertapenem theless, therapy may fail for several reasons: (1) the initial di- Combination agents agnosis and/or treatment chosen is incorrect, (2) the etiologic Facultative and aerobic activity agent from a given locale is resistant to antibiotics, (3) anti- Fluoroquinolone microbial resistance develops during treatment, and (4) the Third-generation cephalosporin a infection is deeper and more complex than originally estimated. Aztreonam Aminoglycoside Anaerobic activity INTRODUCTION Clindamycin a This practice guideline provides recommendations for diag- Metronidazole nosis and management of skin and soft-tissue infections in Chloramphenicol Penicillin agent plus -lactamase inhibitor otherwise healthy hosts and compromised hosts of all age b Nonintestinal groups. These infections have diverse etiologies that depend, Trunk and extremities away from axilla or perineum in part, on the epidemiological setting. Thus, obtaining a careful Oxacillin history, including information about the patient’s immune First-generation cephalosporin status, the geographical locale, travel history, recent trauma or Axillary or perineum surgery, previous antimicrobial therapy, lifestyle, hobbies, and Cefoxitin animal exposure or bites is key to developing an adequate dif- Ampicillin/sulbactam ferential diagnosis and an appropriate index of suspicion for Other single agents as described above for intestinal and specific etiological agents. Recognizing the physical examina- genital operations tion findings and understanding the anatomical relationships a Do not combine aztreonam with metronidazole, because this combination of skin and soft tissue are also crucial for establishing the correct has no activity against gram-positive cocci. diagnosis. In some cases, this information is insufficient, and biopsy or aspiration of tissue may be necessary. In addition, Public Health Service grading system for ranking recommen- radiographic procedures may be useful to determine the level dations in clinical guidelines (table 1). of infection and the presence of gas or abscess. Finally, surgical IMPETIGO exploration or debridement is an important diagnostic, as well as therapeutic, procedure in immunocompromised hosts or in Impetigo, a skin infection that is common throughout the patients with necrotizing infections or myonecrosis. world, consists of discrete purulent lesions that are nearly al- Three contemporary problems confounding the clinical eval- ways caused by b-hemolytic streptococci and/or S. aureus.Im- uation of patients with skin and soft-tissue infection are di- petigo occurs most frequently among economically disadvan- agnosis, severity of infection, and pathogen-specific antibiotic taged children in tropical or subtropical regions, but it is also resistance patterns. Dozens of microbes may cause soft-tissue prevalent in northern climates during the summer months [8]. infections, and although specific bacteria may cause a particular Its peak incidence is among children aged 2–5 years, although type of infection, considerable overlaps in clinical presentations older children and adults may also be afflicted [9, 10]. There exist. Clues to the diagnosis or algorithmic approaches to di- is no sex predilection, and all races are susceptible. agnosis are covered in detail in the text to follow. Specific Prospective studies of streptococcal impetigo have demon- recommendations for therapy are given, each with a rating that strated that the responsible microorganisms initially colonize indicates the strength of and evidence for recommendations, the unbroken skin [8], an observation that probably explains expressed using the Infectious Diseases Society of America–US the influence of personal hygiene on disease incidence. Skin

1378 • CID 2005:41 (15 November) • Stevens et al. colonization with a given streptococcal strain precedes the de- cently, S. pyogenes, are prevalent. Topical therapy with mupi- velopment of impetiginous lesions by a mean duration of 10 rocin is equivalent to oral systemic antimicrobials [21, 22] (A- days. Inoculation of surface organisms into the skin by abra- I) and may be used when lesions are limited in number. It is sions, minor trauma, or insect bites then ensues. During the expensive, however, and some strains of staphylococci are re- course of 2– 3 weeks, streptococcal strains may be transferred sistant [5]. Suppurative complications of streptococcal impetigo from the skin and/or impetigo lesions to the upper respiratory are uncommon, and for as yet unexplained reasons, rheumatic tract. In contrast, in patients with staphylococcal impetigo, the fever has never occurred after streptococcal impetigo. On the pathogens are usually present in the nose before causing cu- other hand, cutaneous infections with nephritogenic strains of taneous disease. group A streptococci are the major antecedent of poststrep- Impetigo usually occurs on exposed areas of the body, most tococcal glomerulonephritis in many areas of the world. No frequently the face and extremities. The lesions remain well- conclusive data indicate that treatment of streptococcal pyo- localized but are frequently multiple and may be either bullous derma prevents nephritis [23], but such therapy is important or nonbullous in appearance. Bullous lesions appear initially as an epidemiologic measure in eradicating nephritogenic as superficial vesicles that rapidly enlarge to form flaccid bullae strains from the community. filled with clear yellow fluid, which later becomes darker, more turbid, and sometimes purulent. The bullae may rupture, often ABSCESSES, CELLULITIS, AND ERYSIPELAS leaving a thin brown crust resembling lacquer [11]. The lesions Cutaneous abscesses. Cutaneous abscesses are collections of of nonbullous impetigo begin as papules that rapidly evolve pus within the dermis and deeper skin tissues. They are into vesicles surrounded by an area of erythema and then be- usually painful, tender, and fluctuant red nodules, often sur- come pustules that gradually enlarge and break down over a mounted by a pustule and surrounded by a rim of erythem- period of 4–6 days to form characteristic thick crusts. The atous swelling. Cutaneous abscesses are typically poly- lesions heal slowly and leave depigmented areas. A deeply ul- microbial, containing bacteria that constitute the normal cerated form of impetigo is known as ecthyma. Although re- regional skin flora, often combined with organisms from ad- gional lymphadenitis may occur, systemic symptoms are usually jacent mucous membranes [24–30]. S. aureus is present, usu- absent. ally as a single pathogen, in only ∼25% of cutaneous abscesses is caused by strains of S. aureus that pro- overall. Epidermoid cysts, often erroneously labeled “seba- duce a toxin causing cleavage in the superficial skin layer. In ceous cysts,” ordinarily contain skin flora in the cheesy ke- the past, nonbullous lesions were usually caused by streptococci. ratinous material, even when uninflamed. Cultures of in- Now, most cases are caused by staphylococci alone or in com- flamed cysts also yield the same organisms, suggesting that bination with streptococci [12, 13]. Streptococci isolated from the inflammation and purulence occur as a reaction to rupture lesions are primarily group A organisms, but occasionally, other of the cyst wall and extrusion of its contents into the dermis, serogroups (such as C and G) are responsible. rather than as an infectious complication [31]. Assays of streptococcal antibodies are of no value in the Effective treatment of abscesses and inflamed epidermoid diagnosis and treatment of impetigo, but they provide helpful cysts entails incision, thorough evacuation of the pus, and prob- supporting evidence of recent streptococcal infection in patients ing the cavity to break up loculations (A-I). Simply covering suspected of having poststreptococcal glomerulonephritis. The the surgical site with a dry dressing is usually the easiest and anti–streptolysin O response is weak in patients with strepto- most effective treatment of the wound [32, 33], although some coccal impetigo [14, 15], presumably because skin lipids sup- clinicians pack it with gauze or suture it closed. Gram stain, press streptolysin O response [16], but anti–DNAse B levels are culture, and systemic antibiotics are rarely necessary (E-III). consistently elevated [14, 15]. Unusual exceptions include the presence of multiple lesions, In the past, therapy directed primarily at group A strepto- cutaneous gangrene, severely impaired host defenses, extensive cocci (e.g., penicillin) was successful, both in healing the lesions surrounding cellulitis, or severe systemic manifestations of in- and decreasing recurrences of nonbullous impetigo for at least fection, such as high fever. several weeks [17, 18]. Because S. aureus currently accounts for Furuncles and carbuncles. Furuncles (or “”) are in- most cases of bullous impetigo, as well as for a substantial fections of the hair follicle, usually caused by S. aureus, in which portion of nonbullous infections [13, 19, 20], penicillinase- suppuration extends through the dermis into the subcutaneous resistant penicillins or first- generation cephalosporins are pre- tissue, where a small abscess forms. They differ, therefore, from ferred (A-I), although impetigo caused by MRSA is increasing folliculitis, in which inflammation is more superficial and pus in frequency [13] (table 2). Erythromycin has been a mainstay is present in the epidermis. Furuncles can occur anywhere on of pyoderma therapy, but its utility may be lessened in areas hairy skin. Each lesion consists of an inflammatory nodule and where erythromycin-resistant strains of S. aureus, or more re- an overlying pustule through which hair emerges. When in-

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1379 fection extends to involve several adjacent follicles, producing practice, however, distinguishing between cellulitis and erysip- a coalescent inflammatory mass with pus draining from mul- elas clinically may be difficult, and some physicians, especially tiple follicular orifices, the lesion is called a carbuncle. Car- in northern Europe, use the term “erysipelas” to describe both buncles tend to develop on the back of the neck and are es- infections. pecially likely to occur in diabetic persons. Erysipelas is distinguished clinically from other forms of cu- For small furuncles, moist heat, which seems to promote taneous infection by the following 2 features: the lesions are drainage, is satisfactory. Larger furuncles and all carbuncles raised above the level of the surrounding skin, and there is a require incision and drainage. Systemic antibiotics are usually clear line of demarcation between involved and uninvolved unnecessary, unless extensive surrounding cellulitis or fever oc- tissue [41]. This disorder is more common among infants, curs (E-III). Outbreaks of furunculosis caused by MSSA, as well young children, and older adults. It is almost always caused by as by MRSA, may occur in families and other settings involving b-hemolytic streptococci (usually group A), but similar lesions close personal contact (e.g., prisons), especially when skin in- can be caused by streptococci from serogroups C or G. Rarely, jury is common, such as sports teams or outdoor recreation group B streptococci or S. aureus may be involved. In older groups [34–36]. Inadequate personal hygiene and exposure to reports, erysipelas characteristically involved the butterfly area others with furuncles are important predisposing factors in of the face, but at present, the lower extremities are more fre- these settings. In some cases, fomites may harbor the organism quently affected [42, 43]. and facilitate transmission of the infection. Depending on the With early diagnosis and proper treatment, the prognosis is individual circumstances, control of outbreaks may require excellent. Rarely, however, the infection may extend to deeper bathing with antibacterial soaps, such as chlorhexidine; thor- levels of the skin and soft tissues. Penicillin, given either par- ough laundering of clothing, towels, and bed wear; separate enterally or orally depending on clinical severity, is the treat- use of towels and washcloths; and attempted eradication of ment of choice (A-III). If is suspected, staphylococcal carriage among colonized persons [36] (B-III). a penicillinase-resistant semisynthetic penicillin or a first-gen- Some individuals have repeated attacks of furunculosis. A eration cephalosporin should be selected [44] (A-III). In a ran- few of these persons, particularly children, have abnormal sys- domized, prospective multicenter trial [45], the efficacy of rox- temic host responses, but for most, the only identifiable pre- ithromycin, a macrolide antimicrobial, was equivalent to that disposing factor is the presence of S. aureus in the anterior for penicillin. Macrolide resistance among group A strepto- nares or, occasionally, elsewhere, such as the perineum [37]. cocci, however, is increasing in the United States [46, 47]. The prevalence of nasal staphylococcal colonization in the gen- Cellulitis is an acute spreading infection of the skin, extend- eral population is 20%–40%, but why some carriers develop ing more deeply than erysipelas to involve the subcutaneous recurrent skin infections and others do not is usually unclear. tissues. It therefore lacks the distinctive anatomical features The major method of controlling recurrent furunculosis is described above for erysipelas. Although most cellulitis is the use of antibacterial agents to eradicate staphylococcal car- caused by b-hemolytic streptococci, a number of other micro- riage. For persons with nasal colonization, one approach is the organisms may give rise to this disorder (see below). application of mupirocin ointment twice daily in the anterior Both erysipelas and cellulitis are manifested clinically by rap- nares for the first 5 days each month [38] (A-I). This regimen idly spreading areas of edema, redness, and heat, sometimes reduces recurrences by ∼50%. Few systemic antibiotics attain accompanied by and inflammation of the regional adequate levels in the nasal secretions to achieve protracted lymph nodes. The skin surface may resemble an orange peel elimination of staphylococci [39]. Clindamycin is an exception, (i.e., peau d’orange) because superficial cutaneous edema sur- and probably the best program for recurrent furunculosis rounds the hair follicles, which causes dimpling in the skin caused by susceptible S. aureus is a single oral daily dose of because they remain tethered to the underlying dermis. Vesicles, 150 mg of this agent for 3 months, which decreases subsequent bullae, and cutaneous hemorrhage in the form of petechiae or infections by ∼80% [40] (A-I). ecchymoses may develop on the inflamed skin. Systemic man- Cellulitis and erysipelas. These terms refer to diffuse, ifestations are usually mild, but fever, tachycardia, confusion, spreading skin infections, excluding infections associated with hypotension, and leukocytosis are sometimes present and may underlying suppurative foci, such as cutaneous abscesses, nec- even occur hours before the skin abnormalities appear. Vesicles rotizing fasciitis, septic arthritis, and osteomyelitis. Unfortu- and bullae filled with clear fluid are common. Petechiae and nately, physicians use the words “cellulitis” and “erysipelas” ecchymoses may develop in inflamed skin; if these are wide- inconsistently. For some, the distinction between the 2 terms spread and associated with systemic toxicity, a deeper infection relates to the depth of inflammation: erysipelas affects the upper such as necrotizing fasciitis should be considered. dermis, including the superficial lymphatics, whereas cellulitis These infections arise when organisms enter through involves the deeper dermis, as well as subcutaneous fat. In breaches in the skin. Predisposing factors for these infections

1380 • CID 2005:41 (15 November) • Stevens et al. include conditions that make the skin more fragile or local host infection in persons employed in aquaculture or meatpacking, defenses less effective, such as obesity, previous cutaneous dam- respectively. Periorbital cellulitis due to Haemophilus influenzae age, and edema from venous insufficiency or lymphatic ob- can occur in children. Diagnostic and therapeutic consider- struction or other causes [48]. The origin of the disrupted ations of this infection have been reported by the Committee cutaneous barrier may be trauma, preexisting skin infections on Infectious Diseases, American Academy of Pediatrics [6]. such as impetigo or ecthyma, ulceration, fissured toe webs from In neutropenic hosts, infection may be due to Pseudomonas maceration or fungal infection, and inflammatory dermatoses, aeruginosa or other gram-negative bacilli, and in patients in- such as eczema. Often, however, the breaks in the skin are small fected with HIV, the responsible organism may be Helicobacter and clinically inapparent. These infections can occur at any cinaedi [71]. Occasionally, Cryptococcus neoformans causes cel- location but are most common on the lower legs. lulitis in patients with deficient cell-mediated immunity. Surgical procedures that increase the risk for cellulitis, pre- Because of their very low yield, blood cultures are not fruitful sumably due to disruption of lymphatic drainage, include sa- for the typical case of erysipelas or cellulitis, unless it is par- phenous venectomy [49, 50], axillary node dissection for breast ticularly severe [55]. Needle aspirations and skin biopsies are cancer [51, 52], and operations for gynecologic malignancies also unnecessary in typical cases, which should respond to an- that involve lymph node dissection, especially when followed tibiotic therapy directed against streptococci and staphylococci. by radiation therapy, such as radical vulvectomy and radical These procedures may be more rewarding [56] for patients with hysterectomy [53, 54]. diabetes mellitus, malignancy, and unusual predisposing fac- Blood culture results are positive in р5% of cases [55]. Re- tors, such as immersion injury, animal bites, neutropenia, and sults of culture of needle aspirations of the inflamed skin are immunodeficiency. bewilderingly variable, varying from р5% to ∼40% in reported Diseases sometimes confused with cellulitis include acute series [56–63], and probably depending on the patient popu- dermatitis, such as that due to contact with an allergen; gout, lation, the definition of cellulitis, the inclusion or exclusion of with marked cutaneous inflammation extending beyond the cases with associated abscesses, and the determination of joint involved; and herpes zoster. Acute lipodermatosclerosis, whether isolates are pathogens or contaminants. Culture of a panniculitis that occurs predominantly in obese women with punch biopsy specimens yields an organism in 20%–30% of lower extremity venous insufficiency, causes painful, erythem- cases [57, 64], but the concentration of bacteria is usually quite atous, tender, warm, indurated, and sometimes scaly areas in low [64]. Culture of these specimens, as well as other available the medial leg that resemble cellulitis [72]. evidence, including serologic studies [42, 59, 65] and techniques Therapy for the typical case of erysipelas or cellulitis should employing immunofluorescent antibodies to detect antigens in include an antibiotic active against streptococci. Many clini- skin biopsy specimens [66, 67], indicate that most of the in- cians choose an agent that is also effective against S. aureus, fections arise from streptococci, often group A, but also from although this organism rarely causes cellulitis unless associated other groups, such as B, C, or G. The source of the pathogens with an underlying abscess or penetrating trauma. A large per- is frequently unclear, but in many infections of the lower ex- centage of patients can receive oral medications from the start tremities, the responsible streptococci are present in the mac- [73]. Suitable agents include dicloxacillin, cephalexin, clinda- erated or fissured interdigital toe spaces [68, 69], emphasizing mycin, or erythromycin, unless streptococci or staphylococci the importance of detecting and treating tinea pedis and other resistant to these agents are common in the community (A-I). causes of toe web abnormalities in these patients. Occasionally, Macrolide resistance among group A streptococci has in- the reservoir of streptococci is the anal canal [70] or the vagina, creased regionally in the United States. For parenteral therapy, especially for group B streptococci causing cellulitis in patients which is indicated for severely ill patients or for those unable with previous gynecologic cancer treated with surgery and ra- to tolerate oral medications, reasonable choices include a pen- diation therapy. S. aureus less frequently causes cellulitis, often icillinase-resistant penicillin such as nafcillin, a first-generation associated with previous penetrating trauma, including injec- cephalosporin such as cefazolin, or, for patients with life-threat- tion sites of illicit drug use. ening penicillin allergies, clindamycin or vancomycin (A-I). In Many other infectious agents can produce cellulitis, but usu- cases of uncomplicated cellulitis, 5 days of antibiotic treatment ally only in special circumstances. With cat or dog bites, for is as effective as a 10-day course [74]. example, the organism responsible is typically Pasteurella spe- Antibiotic treatment alone is effective in most patients with cies, especially P. multocida, or Capnocytophaga canimorsus. A. cellulitis. However, patients who are slow to respond may have hydrophila may cause cellulitis following immersion in fresh a deeper infection or underlying conditions, such as diabetes, water, whereas infection after saltwater exposure can arise from chronic venous insufficiency, or lymphedema. In some patients, Vibrio species, particularly V. vulnificus in warm climates. In cutaneous inflammation sometimes worsens after initiating rare cases, Streptococcus iniae or E. rhusiopathiae may cause therapy, probably because the sudden destruction of pathogens

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1381 releases potent that increase local inflammation. In a soft-tissue infections caused by community-acquired MRSA. single randomized, double-blind, placebo-controlled trial, sys- Traditionally regarded as a nosocomial pathogen, MRSA isolates temic corticosteroids attenuated this reaction and hastened res- causing community-onset disease differ from their hospital olution [75]. Specifically, 108 patients with a diagnosis of un- counterparts in several ways [82–84]. Community strains cause complicated erysipelas were randomized to receive antibiotics infections in patients lacking typical risk factors, such as hos- (90% received benzyl penicillin) plus either an 8-day tapering pital admission or residence in a long-term care facility; they oral course of corticosteroid therapy beginning with 30 mg of are often susceptible to non–b-lactam antibiotics, including prednisolone or a placebo. Subjects !18 years of age, diabetic doxycycline, clindamycin, trimethoprim-sulfamethoxazole, patients, and pregnant women were excluded. One-third of fluoroquinolones, or rifampin; genotypically, they appear not enrolled subjects had a previous episode of erysipelas at the to be related to local hospital strains and to contain type IV current site of infection. Median healing time, median treat- SCCmec cassette not typical of hospital isolates [85, 86]. Finally, ment time with intravenous antibiotics, and median duration community isolates have frequently contained genes for Pan- of hospital stay were all shortened by 1 day in the prednisolone- ton-Valentine leukocidin [87], which has been associated with treated group [75]. Long-term follow-up of these patients mild to severe skin and soft-tissue infections [7]. Outbreaks showed no difference in relapse or recurrence [76]. Further caused by community-acquired MRSA isolates have occurred studies are warranted, but in the meantime, clinicians may wish among prison and jail inmates, injection drug users, Native to consider systemic corticosteroids as an optional adjunct for American populations, gay men, participants in contact sports, treatment of uncomplicated cellulitis and erysipelas in selected and children [88, 89]. Thus, recurrent or persistent furuncles adult patients. and impetigo, particularly in these high-risk groups, that do Elevation of the affected area, an important and often ne- not respond to oral b-lactam antibiotic therapy are increasingly glected aspect of treatment, quickens improvement by pro- likely to be caused by MRSA. Such lesions should be cultured moting gravity drainage of the edema and inflammatory sub- and antibiotic susceptibilities determined. Fluctuant lesions stances. Patients should also receive appropriate therapy for should be drained. An oral agent to which the isolate is sus- any underlying condition that may have predisposed to the ceptible should be used as initial therapy (table 2). Most com- infection, such as tinea pedis, venous eczema (“stasis derma- munity-acquired strains are susceptible to doxycycline or min- р titis”), or trauma. ocycline, but these should be avoided in children 8 years old Each attack of cellulitis causes lymphatic inflammation and and during pregnancy. Clindamycin has excellent antistaphy- possibly some permanent damage. Severe or repeated episodes lococcal activity, but there is the potential for emergence of resistance with high-inoculum infections caused by strains in- of cellulitis may lead to lymphedema, sometimes substantial ducibly resistant to erythromycin. Linezolid, daptomycin, and enough to cause elephantiasis. Measures to reduce recurrences vancomycin have excellent efficacy in skin and soft-tissue in- of cellulitis include treating interdigital maceration, keeping fections in general and against those due to MRSA specifically the skin well hydrated with emollients to avoid dryness and [90, 91] (A-I). However, these agents should be reserved for cracking, and reducing any underlying edema by such meth- patients who have severe infections requiring hospitalization or ods as elevation of the extremity, compressive stockings or who have not responded to attempts to eradicate the infection. pneumatic pressure pumps, and, if appropriate, diuretic ther- Trimethoprim-sulfamethoxazole has been used to treat serious apy. If frequent infections occur despite such measures, pro- staphylococcal infections, including those due to MRSA. In one phylactic antibiotics appear reasonable; however, published double-blind, randomized trial in which 47% of the isolates results demonstrating efficacy have been mixed [77–80]. Be- were MRSA, cures were documented in 37 of the 43 patients cause streptococci cause most recurrent cellulitis, options in- receiving trimethoprim-sulfamethoxazole, compared with 57 of clude monthly intramuscular benzathine penicillin injections 58 patients in the vancomycin group; trimethoprim-sulfa- of 1.2 MU in adults or oral therapy with twice-daily doses of methoxazole failures occurred mostly in patients with MSSA either 250 mg of erythromycin or 1 g of penicillin V (B-II). infections [92]. If a fluoroquinolone is chosen, one with en- An alternative, but untested, option for reliable patients with hanced activity against gram-positive bacteria should be used recurrent cellulitis is to try to shorten each episode by pro- (e.g., levofloxacin, gatifloxacin, or moxifloxacin), but still there viding oral antibiotics for them to initiate therapy as soon as is the possibility of emergence of resistance. symptoms of infection begins. One trial of oral selenium dem- onstrated a reduced recurrence rate of erysipelas in secondary NECROTIZING SKIN AND SOFT-TISSUE lymphedema by 80% [81]. This report requires independent INFECTIONS confirmation. Soft-tissue infections and the evaluation of MRSA infection. Necrotizing skin and soft-tissue infections differ from the An emerging problem is the increasing prevalence of skin and milder, superficial infections by clinical presentation, coexisting

1382 • CID 2005:41 (15 November) • Stevens et al. systemic manifestations, and treatment strategies [93, 94]. They presentation is that of cellulitis, which can advance rapidly or are often deep and devastating. They are deep because they slowly. As it progresses, there is systemic toxicity with high may involve the fascial and/or muscle compartments; they are temperatures. The patient may be disoriented and lethargic. devastating because they cause major destruction of tissue and The local site shows the following features: cellulitis (90% of can lead to a fatal outcome. These conditions are usually “sec- cases), edema (80%), skin discoloration or gangrene (70%), ondary” infections, in that they develop from an initial break and anesthesia of involved skin (frequent, but the true incidence in the skin related to trauma or surgery. They can be mon- is unknown). omicrobial (usually involving streptococci or, rarely, staphy- A distinguishing clinical feature is the wooden-hard feel of lococci) or polymicrobial (involving a mixed aerobe-anaerobe the subcutaneous tissues. In cellulitis or erysipelas the subcu- bacterial flora). Although many specific variations of necrotiz- taneous tissues can be palpated and are yielding. But in fasciitis, ing soft-tissue infections have been described on the basis of the underlying tissues are firm, and the fascial planes and mus- etiology, microbiology, and specific anatomic location of the cle groups cannot be discerned by palpation. It is often possible infection, the initial approach to the diagnosis, antimicrobial to observe a broad erythematous tract in the skin along the treatment, and decision to use operative management are sim- route of the infection as it advances cephalad in an extremity. ilar for all forms and are more important than determining the If there is an open wound, probing the edges with a blunt specific variant. instrument permits ready dissection of the superficial fascial In the initial phases, distinguishing between a cellulitis that planes well beyond the wound margins. should respond to antimicrobial treatment alone and a nec- Bacteriologic characteristics. In the monomicrobial form, rotizing infection that requires operative intervention may be the pathogens are S. pyogenes, S. aureus, V. vulnificus, A. hy- difficult. Several clinical features suggest the presence of a nec- drophila, and anaerobic streptococci (i.e., Peptostreptococcus rotizing infection of the skin and its deeper structures: (1) species). Staphylococci and hemolytic streptococci can occur severe, constant pain; (2) bullae, related to occlusion of deep simultaneously. Most infections are community acquired and blood vessels that traverse the fascia or muscle compartments; present in the limbs, with approximately two-thirds of cases in (3) skin necrosis or ecchymosis (bruising) that precedes skin the lower extremities. There is often an underlying cause, such necrosis; (4) gas in the soft tissues, detected by palpation or as diabetes, arteriosclerotic vascular disease, or venous insuf- imaging; (5) edema that extends beyond the margin of ery- ficiency with edema. Sometimes, a chronic vascular ulcer thema; (6) cutaneous anesthesia; (7) systemic toxicity, mani- changes into a more acute process. Cases of necrotizing fasciitis fested by fever, leukocytosis, delirium, and renal failure; and that arise after varicella or trivial injuries, such as minor (8) rapid spread, especially during antibiotic therapy. Bullae alone are not diagnostic of deep infections, because they also scratches and insect bites, are almost always due to S. pyogenes. occur with erysipelas, cellulitis, scalded skin syndrome, dissem- The mortality in this group is high, approaching 50%–70% in inated intravascular coagulation, purpura fulminans, some tox- patients with hypotension and organ failure [97, 98]. ins (e.g., those associated with bite from a brown-recluse spi- In the polymicrobial form, up to 15 different anaerobic and der), and primary dermatologic conditions. aerobic organisms can be cultured from the involved fascial plane, with an average of 5 pathogens in each wound. Most of Necrotizing Fasciitis the organisms originate from the bowel flora (e.g., coliforms Necrotizing fasciitis is a relatively rare subcutaneous infection and anaerobic bacteria). that tracks along fascial planes and extends well beyond the The polymicrobial necrotizing infection is associated with 4 superficial signs of infection, such as erythema and other skin clinical settings: (1) surgical procedures involving the bowel or changes [95, 96]. The term fasciitis sometimes leads to the penetrating abdominal trauma, (2) decubitus ulcer or a perianal mistaken impression that the muscular fascia or aponeurosis abscess, (3) at the site of injection in injection drug users, and is involved. The fascia most commonly referred to is the su- (4) spread from a Bartholin abscess or a minor vulvovaginal perficial fascia, which is comprised of all of the tissue between infection. Although mixed infections are usually noted in this the skin and underlying muscles (i.e., subcutaneous tissue). latter setting, some cases are caused by a single pathogen, par- Clinical features. Extension from a skin lesion is seen in ticularly anaerobic Streptococcus species. 80% of cases. The initial lesion, such as a minor abrasion, insect Diagnosis. It may not be possible to diagnose fasciitis upon bite, injection site (in the case of drug addicts), or , often first seeing the patient. Overlying cellulitis is a frequent accom- is trivial. Rare cases have arisen in Bartholin gland abscess or paniment. That the process involves the deeper tissue planes perianal abscess, from which the infection spreads via fascial is suggested by the following features: (1) failure to respond to planes of the perineum, thigh, groin, and abdomen. The re- initial antibiotic therapy; (2) the hard, wooden feel of the sub- maining 20% of patients have no visible skin lesion. The initial cutaneous tissue, extending beyond the area of apparent skin

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1383 involvement; (3) systemic toxicity, often with altered mental toxicity, fever, hypotension, or advancement of the skin and status; (4) bullous lesions; and (5) skin necrosis or ecchymoses. soft-tissue infection during antibiotic therapy is an indication CT scan or MRI may show edema extending along the fascial for surgical intervention. Third, when the local wound shows plane. In practice, clinical judgment is the most important any skin necrosis with easy dissection along the fascia by a element in diagnosis. Data regarding the sensitivity and spec- blunt instrument, more complete incision and drainage are ificity of CT or MRI are unavailable, and requesting such studies required. Fourth, any soft-tissue infection accompanied by gas may delay definitive diagnosis and treatment. The most im- in the affected tissue suggests necrotic tissue and requires op- portant diagnostic feature of necrotizing fasciitis is the ap- erative drainage and/or debridement. pearance of the subcutaneous tissues or fascial planes at op- Most patients with necrotizing fasciitis should return to the eration. Upon direct inspection, the fascia is swollen and dull operating room 24–36 h after the first debridement and daily gray in appearance, with stringy areas of necrosis. A thin, thereafter until the surgical team finds no further need for brownish exudate emerges from the wound. Even during deep debridement. Although discrete pus is usually absent, these dissection, there is typically no true pus. Extensive undermining wounds can discharge copious amounts of tissue fluid; ag- of surrounding tissues is present, and the tissue planes can be gressive administration of fluid is a necessary adjunct. dissected with a gloved finger or a blunt instrument. A Gram Antimicrobial therapy must be directed at the pathogens and stain of the exudate demonstrates the presence of the pathogens used in appropriate doses (table 5) until repeated operative and provides an early clue to therapy. Gram-positive cocci in procedures are no longer needed, the patient has demonstrated chains suggest Streptococcus organisms (either group A or an- obvious clinical improvement, and fever has been absent for aerobic). Large gram-positive cocci in clumps suggest S. aureus, 48–72 h. Treatment of polymicrobial necrotizing fasciitis must but this is an unusual primary organism in these spreading include agents effective against both aerobes and anaerobes infections. Samples for culture are best obtained from the deep (table 5). In general, ampicillin is useful for coverage of sus- tissues. If the infection originated from a contaminated skin ceptible enteric aerobic organisms, such as E. coli, as well as wound, such as a vascular ulcer, the bacteriologic characteristics for gram-positive organisms, such as Peptostreptococcus species, of the superficial wound are not necessarily indicative of deep- group B, C, or G streptococci, and some anaerobes (A-III). tissue infection. Direct needle aspiration of the advancing edge Clindamycin is useful for coverage of anaerobes and aerobic as a means of obtaining material for culture can be helpful if gram-positive cocci, including most S. aureus serogroups. Me- fluid is obtained. A definitive bacteriologic diagnosis is best tronidazole has the greatest anaerobic spectrum against the established by culture of tissue specimens obtained during op- enteric gram-negative anaerobes, but it is less effective against eration or by positive blood culture results. In doubtful cases, the gram-positive anaerobic cocci. Gentamicin or a fluorinated the surgical procedure may provide both diagnosis and treat- quinolone, ticarcillin-clavulanate, or piperacillin-sulbactam is ment. If necrotizing infection is suspected but not confirmed, useful for coverage against resistant gram-negative rods. Thus, a small, exploratory incision should be made in the area of the best choice of antibiotics for community-acquired mixed maximum suspicion. If a necrotizing infection is present, it will infections is a combination of ampicillin-sulbactam plus clin- be obvious from the findings described above. If there is no damycin plus ciprofloxacin (A-III). necrosis on exploratory incision, the procedure can be termi- Necrotizing fasciitis and/or streptococcal toxic shock syn- nated with very little risk or morbidity to the patient. Some drome caused by group A streptococci should be treated with have suggested biopsy for frozen section analysis to make the clindamycin and penicillin (A-II). The rationale for clinda- diagnosis. However, if enough suspicion exists to do a biopsy, mycin is based on in vitro studies demonstrating both toxin the diagnosis is usually evident to gross inspection without suppression and modulation of cytokine (i.e., TNF) production, histological slides. on animal studies demonstrating superior efficacy versus that Treatment. Surgical intervention is the major therapeutic of penicillin, and on 2 observational studies demonstrating modality in cases of necrotizing fasciitis (A-III). Many cases of greater efficacy for clindamycin than for b-lactam antibiotics necrotizing fasciitis, however, probably begin as cellulitis, and [99, 100]. Penicillin should be added because of the increasing if necrotizing fasciitis is recognized early and treated aggres- resistance of group A streptococci to macrolides, although in sively, some patients may avoid potentially mutilating surgical the United States, only 0.5% of macrolide-resistant group A procedures. The decision to undertake aggressive surgery streptococci are also clindamycin resistant. should be based on several considerations. First, no response A recommendation to use intravenous g-globulin (IVIG) to to antibiotics after a reasonable trial is the most common index. treat streptococcal toxic shock syndrome cannot be made with A response to antibiotics should be judged by reduction in certainty (B-II). Although there is ample evidence for the role fever and toxicity and lack of advancement. Second, profound of extracellular streptococcal toxins in shock, organ failure, and

1384 • CID 2005:41 (15 November) • Stevens et al. Table 5. Treatment of necrotizing infections of the skin, fascia, and muscle.

First-line Antimicrobial agent(s) antimicrobial agent, for patients with severe by infection type Adult dosage penicillin hypersensitivity Mixed infection Ampicillin-sulbactam 1.5–3.0 g every 6–8 h iv Clindamycin or metronidazolea with an amino- or glycoside or fluoroquinolone piperacillin-tazobactam 3.37 g every 6–8 h iv plus clindamycin 600–900 mg/kg every 8 h iv plus ciprofloxacin 400 mg every 12 h iv Imipenem/cilastatin 1 g every 6–8 h iv … Meropenem 1 g every 8 h iv … Ertapenem 1 g every day iv … Cefotaxime 2 g every 6 h iv … plus metronidazole 500 mg every 6 h iv or clindamycin 600–900 mg/kg every 8 h iv Streptococcus infection Penicillin 2–4 MU every 4–6 h iv (adults) Vancomycin, linezolid, quinupristin/dalfopristin, plus or daptomycin clindamycin 600–900 mg/kg every 8 h iv S. aureus infection Nafcillin 1–2 g every 4 h iv Vancomycin, linezolid, quinupristin/dalfopristin, daptomycin Oxacillin 1–2 g every 4 h iv … Cefazolin 1 g every 8 h iv … Vancomycin (for resistant strains) 30 mg/kg/day in 2 divided doses iv … Clindamycin 600–900 mg/kg every 8 h iv Bacteriostatic; potential of cross-resistance and emergence of resistance in erythromy- cin-resistant strains; inducible resistance in methicillin-resistant S. aureus Clostridium infection Clindamycin 600–900 mg/kg every 8 h iv … Penicillin 2–4 MU every 4–6 h iv …

a If Staphylococcus infection is present or suspected, add an appropriate agent. iv, intravenously. tissue destruction, different batches of IVIG contain variable Anaerobic Streptococcal Myositis quantities of neutralizing antibodies to some of these toxins, Anaerobic streptococci cause a more indolent infection than and definitive clinical data are lacking [101]. One observational other streptococci. Unlike other necrotizing infections, infec- study demonstrated better outcomes in patients receiving IVIG, tion of the muscle and fascial planes by anaerobic streptococci but these patients were more likely to have had surgery and to usually is associated with trauma or a surgical procedure. have received clindamycin than were historical control subjects Incision and drainage are critical. Necrotic tissue and debris [102]. A second study, which was a double-blind, placebo- are resected but the inflamed, viable muscle should not be controlled trial from northern Europe, showed no statistically removed, because it can heal and regain function. The incision significant improvement in survival, and, specific to this sec- should be packed with moist dressings. Antibiotic treatment is tion, no reduction in the time to no further progression of highly effective. These organisms are all susceptible to penicillin necrotizing fasciitis (69 h for the IVIG group, compared with or ampicillin, which should be administered in high doses. 36 h for the placebo group) [103]. Results of these studies provide some promise. However, this committee believes that additional studies of the efficacy of IVIG are necessary before a recommendation can be made regarding use of IVIG for Pyomyositis, which is caused mainly by S. aureus, is the pres- treatment of streptococcal toxic shock syndrome. ence of pus within individual muscle groups. Occasionally, S.

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1385 pneumoniae or a gram-negative enteric bacillus is responsible. usually in mixed culture, but occasionally, S. aureus is the only Blood culture results are positive in 5%–30% of cases. Because pathogen. Pseudomonas is another common organism in the of its geographical distribution, this condition is often called mixed culture. As with other necrotizing infections, prompt “tropical pyomyositis,” but cases are increasingly recognized in and aggressive surgical exploration and appropriate debride- temperate climates, especially in patients with HIV infection ment is necessary to remove all necrotic tissue, sparing the or diabetes [104]. Presenting findings are localized pain in a deeper structures when possible (A-III). single muscle group, muscle spasm, and fever. The disease oc- curs most often in an extremity, but any muscle group can be Clostridial Myonecrosis involved, including the psoas or trunk muscles. Initially, it may Clostridial gas gangrene (i.e., myonecrosis) is most commonly not be possible to palpate a discrete abscess because the infec- caused by C. perfringens, C. novyi, C. histolyticum, and C. sep- tion is localized deep within the muscle, but the area has a ticum. C. perfringens is the most frequent cause of trauma- firm, wooden feel associated with pain and tenderness. In the associated gas gangrene. Increasingly severe pain beginning at early stages, ultrasonography or CT scan may be performed to the injury site р24 h after infection is the first reliable symptom. differentiate this entity from a deep venous thrombosis. In more Skin may initially be pale, but it quickly changes to bronze and advanced cases, a bulging abscess is usually clinically apparent. then to a purplish red. The infected region becomes tense and Appropriate antibiotics plus extensive surgical incision and tender, and bullae filled with reddish-blue fluid appear. Gas in drainage are required for appropriate management. the tissue, detected as crepitus or on the basis of imaging stud- ies, is universally present by this late stage. Signs of systemic Synergistic Necrotizing Cellulitis toxicity, including tachycardia, fever, and diaphoresis, develop This is simply a necrotizing soft-tissue infection that involves rapidly, followed by shock and multiple organ failure. muscle groups in addition to superficial tissues and fascia. The In contrast to traumatic gas gangrene, spontaneous gangrene level of involvement depends on the depth and the tissue planes is principally associated with the more aerotolerant C. septicum affected by the original operation or pathological process that and occurs predominantly in patients with neutropenia and precedes the infection. Major predisposing causes are perirectal gastrointestinal malignancy. It develops in normal skin in the and ischiorectal abscesses. Recognition and treatment are sim- absence of trauma as a result of hematogenous spread from a ilar to necrotizing fasciitis, but operative exploration reveals its colonic lesion, usually cancer. A rather innocuous early lesion deeper location. may evolve to all of the above signs over the course of 24 h. Frequently, the diagnosis is unsuspected until gas is detected Fournier Gangrene in tissue or systemic signs of toxicity appear. Early surgical This variant of necrotizing soft-tissue infection involves the inspection and debridement are necessary, and Gram stain of removed tissue shows large, spore-forming gram-positive scrotum and penis or vulva and can have an insidious or ex- bacilli. plosive onset [105, 106]. The mean age of onset is 50 years. Both traumatic and spontaneous clostridial gas gangrene are Most patients have significant underlying disease, particularly fulminant infections requiring meticulous intensive care, sup- diabetes, but 20% will have no discernible cause. Most patients portive measures, aggressive surgical debridement, and appro- initially have a perianal or retroperitoneal infection that has priate antibiotics. The role of hyperbaric oxygen treatment re- spread along fascial planes to the genitalia; a urinary tract in- mains unclear. Altemeier and Fullen [107] reported a significant fection, most commonly secondary to a urethral stricture, that reduction in mortality among patients with gas gangrene using involves the periurethral glands and extends into the penis and penicillin and tetracycline plus aggressive surgery in the absence scrotum; or previous trauma to the genital area, providing ac- of hyperbaric oxygen. Treatment of experimental gas gangrene cess of organisms to the subcutaneous tissues. has demonstrated that tetracycline, clindamycin, and chloram- The infection can begin insidiously with a discrete area of phenicol were more effective than penicillin [108, 109] or hy- necrosis in the perineum that progresses rapidly over 1–2 days perbaric oxygen treatment [110]. Because 5% of strains of C. with advancing skin necrosis. At the outset, it tends to cause perfringens are clindamycin resistant, the recommended anti- superficial gangrene, limited to skin and subcutaneous tissue, biotic treatment is penicillin plus clindamycin (B-III). and extending to the base of the scrotum. The testes, glans penis, and spermatic cord usually are spared, because they have ANIMAL BITES a separate blood supply. The infection may extend to the per- ineum and the anterior abdominal wall through the fascial One-half of all Americans are bitten during their lifetime, usu- planes. ally by a dog. Fortunately, 80% of the wounds are minor, but Most cases are caused by mixed aerobic and anaerobic flora. the remaining 20% that require medical care will account for Staphylococci and Pseudomonas species are frequently present, 1% of all emergency department visits and for 10,000 inpatient

1386 • CID 2005:41 (15 November) • Stevens et al. admissions yearly. Most bites are due to dogs or cats, but bites such as cefuroxime, ceftriaxone, and cefotaxime, may be used from exotic pets and from feral animals also occur. The pre- but may require the addition of an antianaerobic agent. dominant pathogens in these wounds are the normal oral flora Penicillin-allergic pregnant women constitute a special pop- of the biting animal, along with human skin organisms and ulation, because tetracyclines, sulfa compounds (during late occasional secondary invaders (e.g., S. aureus and S. pyogenes) pregnancy), and metronidazole are contraindicated. Similarly, [111, 112]. There are no published large case series on the the selection of an antimicrobial for penicillin-allergic children therapy of bite wounds, but there are many smaller series and is problematic when tetracyclines and fluoroquinolones are anecdotal reports especially focusing on complications. contraindicated. In these situations, macrolides (e.g., azithro- Bacteriologic characteristics. Patients who present !8haf- mycin 250–500 mg every day or telithromycin 400 mg, 2 tablets ter injury seek either wound care or tetanus toxoid, and some by mouth every day) are occasionally used. However, these are concerned about . Patients who seek medical care patients should be observed closely and the potential increased after 8–12 h of injury typically have established infection. The risk of failure noted. wounds may be nonpurulent (30% of dog bites and 42% of The duration of therapy varies by the severity of the injury/ cat bites), purulent (58% of dog bites and 39% of cat bites), infection. Cellulitis and abscess often respond to 5–10 days of or abscesses (12% of dog bites and 19% of cat bites). The therapy. The therapy for early presenting, noninfected wounds average wound yields 5 types of bacterial isolates (range, 0–16 remains controversial. Wounds that are moderate to severe, types of bacterial isolates), with ∼60% yielding mixed aerobic have associated crush injury, have associated edema (either pre- and anaerobic bacteria. Pasteurella species are isolated from existing or subsequent), that are on the hands or in proximity 50% of dog bite wounds and 75% of cat bite wounds. Staph- to a bone or a joint, or that are in compromised hosts should ylococci and streptococci are found in ∼40% of bites from both receive 3–5 days of “prophylactic” antimicrobial therapy. These types of animals. Capnocytophaga canimorsus (formerly known wounds are often colonized with potential pathogens (85% of as DF-2), a fastidious gram-negative rod, can cause bacteremia cases), and it is difficult to determine whether the wound will and fatal sepsis after animal bites, especially in patients with become infected. asplenia or underlying hepatic disease. Facultative gram-neg- Complications. Infectious complications of bite wounds ative rods are uncommon. Bacteroides species, fusobacteria, include septic arthritis, osteomyelitis, subcutaneous abscess for- Porphyromonas species, Prevotella heparinolytica, proprionibac- mation, tendonitis, and, rarely, bacteremia. Pain dispropor- teria, and peptostreptococci are common anaerobes isolated tionate to the severity of injury but located near a bone or joint from both dog bite wounds and cat bite wounds [113]. should suggest periosteal penetration. Hand wounds are often Antimicrobial therapy. Empirical treatment of dog and cat more serious than wounds to fleshy parts of the body. These bites is similar (table 6). Although cat bite wounds have little wound complications will necessitate prolonged therapy, such crush injury and less wound trauma than do dog bites, they as 4–6-week courses for osteomyelitis and 3–4 -week courses are often more severe and have a higher proportion of oste- for synovitis. Noninfectious complications include nerve or omyelitis and septic arthritis. Cat bites have a greater prevalence tendon injury or severance, compartment syndromes, postin- of anaerobes (65% vs. 50%) and P. multocida (75% vs. 50%) fectious and traumatic arthritis, fracture, and bleeding. than do dog bites. For oral, outpatient therapy, amoxicillin- Adjunctive therapeutic measures are often as important as clavulanate has been studied in a small series [114] and is antimicrobial therapy. Wounds should be cleansed with sterile recommended (B-II). Alternative oral agents include doxycy- normal saline (no need for iodine- or antibiotic-containing cline, as well as penicillin VK plus dicloxacillin. Other options, solutions) and superficial debris removed. Deeper debridement including fluoroquinolones (ciprofloxacin, levofloxacin, mox- is usually unnecessary, but, if performed, should be done very ifloxacin, and gatifloxacin), trimethoprim-sulfamethoxazole, cautiously to avoid enlarging the wound and impairing skin and cefuroxime, may require an additional agent active against closure. Infected wounds should not be closed. Suturing anaerobes, such as metronidazole or clindamycin. First-gen- wounds early (!8 h after injury) is controversial, and there are eration cephalosporins, such as cephalexin, penicillinase-resis- no studies to delineate guidelines; however, approximation of tant penicillins (e.g., dicloxacillin), macrolides (e.g., erythro- the margins by Steri-Strips (3M Health Care) and subsequent mycin), and clindamycin, all have poor in vitro activity against closure by either delayed primary or secondary intent seem P. multocida and should be avoided (D-III). prudent. Wounds on the face seem to be an exception and can Intravenous options include the b-lactam/b-lactamase com- be closed primarily if seen by a plastic surgeon, provided there binations (such as ampicillin sulbactam), piperacillin/tazobac- has been meticulous wound care, copious irrigation, and ad- tam, second-generation cephalosporins (such as cefoxitin), and ministration of prophylactic antibiotics. During the first few carbapenems (such as ertapenem, imipenem, and meropenem) days after injury, elevation of the injured body part, especially (B-II). Second-generation and third-generation cephalosporins, if swollen, accelerates healing. This should be accomplished

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1387 using a passive method (a sling for outpatients or a tubular wounds, although often quite small, may extend deeply into stockinet and an intravenous pole for inpatients). the hand tissues, and relaxation of the fist may carry organisms Outpatients should be followed up within 24 h either by into the deep compartments and potential spaces of the hand. phone or during an office visit. If infection progresses despite Exploration under tourniquet control may be necessary. good antimicrobial and ancillary therapy, hospitalization Clenched-fist injuries often require hospitalization and intra- should be considered. On occasion, a single initial dose of a venous antimicrobial therapy with agents such as cefoxitin (1 parenteral antimicrobial may be administered before starting g intravenously every 6–8 h), ampicillin-sulbactam (1.5–3 g oral therapy. Clinicians should insure that tetanus prophylaxis intravenously every 6 h), ertapenem (1 g intravenously every status is current. If it is outdated or if the status is unknown, 24 h), or some combination that covers S. aureus, Haemophilus then a dose of tetanus toxoid (0.5 mL intramuscularly) should species, E. corrodens, and b-lactamase–producing anaerobes (B- be administered. Rabies prophylaxis should be considered for III). E. corrodens is usually resistant to first-generation cepha- all feral and wild animal bites and in geographic areas where losporins (e.g., cefazolin and cephalexin), macrolides (e.g., there is a high prevalence of rabies. The local department of erythromycin), clindamycin, and aminoglycosides, and these health should be consulted about the risks and benefits of rabies agents should be avoided as monotherapy. In the type 1 b- prophylaxis (administration on day 0 of rabies immunoglob- lactam–allergic patient, fluoroquinolones (e.g., moxifloxacin ulin, followed by rabies human diploid cell vaccination at a and gatifloxacin) plus clindamycin, or trimethoprim-sulfa- different site). Only anecdotal literature exists regarding the methoxazole plus metronidazole may be useful. Ancillary mea- bacteriologic characteristics and therapy of exotic or wild an- sures include administration of tetanus toxoid as indicated. The imal bites, but the same general principles should apply. duration of therapy is typically 4 weeks for septic arthritis and 6 weeks for osteomyelitis. HUMAN BITES Complications. Complications are frequent and include tendon and nerve damage, fractures, septic arthritis, and os- Human bite wounds often result from aggressive behavior and teomyelitis. Splinting of the hand in a position of function is are frequently more serious than animal bites. Wounds may be often required, as is subsequent physical therapy. Residual joint either occlusive injuries, in which the teeth actually bite the stiffness is common after clenched fist injury and may affect body part, or clenched-fist injuries, which occur when the fist function. of one person strikes the teeth of another. Between 10% and 20% of occlusive wounds occur during sexual interactions. Bite SOFT-TISSUE INFECTIONS FOLLOWING wounds in children may be associated with sports-related ac- ANIMAL CONTACT tivity (look for imbedded teeth) but should also alert the cli- nician to possible child abuse. Anthrax. One of several clinical manifestations of anthrax is Bacteriologic characteristics. The bacteriologic character- a cutaneous lesion. After an incubation period of 1–12 days, istics of these wounds reflect the normal oral flora of the biter, pruritus begins at the entry site, followed by a papule, devel- with streptococci (especially viridans streptococci) in 80% of opment of vesicles on top of the papule, and, finally, a painless wounds, as well as staphylococci, Haemophilus species, and ulcer with a black scab. This eschar generally separates and Eikenella corrodens as prominent aerobic pathogens [112, 115]. sloughs after 12–14 days. Swelling surrounding the lesion can Other gram-negative rods are infrequent. Anaerobes, including be minor or severe (i.e., malignant edema). Mild-to-moderate Fusobacterium nucleatum and other Fusobacterium species, pep- fever, headaches, and malaise often accompany the illness. Re- tostreptococci, Prevotella species, and Porphyromonas species, gional is common, but pus in the lesion is are present in 160% of cases, but usually in mixed culture. absent unless a secondary infection occurs. WBC counts are Bacteroides fragilis is rarely present. Many of the anaerobes generally normal, but mild leukocytosis can occur. Blood cul- produce b-lactamases, making them resistant to penicillin and ture results are almost always negative. Cultures of untreated first-generation cephalosporins. Human bites also have the po- lesions, depending on the stage of evolution, have positive re- tential to transmit various viral diseases, such as herpes, hep- sults 180% of the time. Methods of specimen collection for atitis B and C, and HIV infection [116–120]. culture depend on the type of lesion. With vesicles, the blister Therapy. Evaluation and treatment should follow the gen- should be unroofed and 2 dry swabs soaked in the fluid. At a eral principles outlined for animal bites, with irrigation and later stage, 2 moist swabs should be rotated in the ulcer base topical wound cleansing, except that prophylactic antimicro- or beneath the eschar’s edge. Patients who have previously bials should be given as early as possible to all patients regardless received antimicrobials or who have negative results of tests of the appearance of the wound (table 6). An expert in hand but still have suspected cutaneous anthrax should have a punch care should evaluate clenched-fist injuries for penetration into biopsy specimen obtained that can be submitted for special the synovium, joint capsule, and the bone (B-III). These studies, such as immunohistochemical staining and/or PCR.

1388 • CID 2005:41 (15 November) • Stevens et al. When obtaining specimens, lesions should not be squeezed to though mainly a research tool, is also a diagnostic option. Rou- produce material for culture. Additional diagnostic methods tine histologic examination of a node, coupled with the clinical include serologic and skin tests. findings, may strongly suggest the diagnosis. Histologic ex- No randomized, controlled trials of therapy of cutaneous amination in conjunction with a Wharthin-Starry silver stain anthrax exist. Most published data indicate that penicillin is is helpful but does not differentiate the species of Bartonella. effective therapy (B-III) (table 3) and will “sterilize” most le- Aspiration of fluctuant nodes may exclude other causes of pu- sions between a few hours to 3 days but does not accelerate rulent lymphadenopathy and sometimes is appropriate to re- healing. Its value seems to be primarily in reducing mortality lieve pain. from as high as 20% to 0%. On the basis of even less evidence, Treatment of cat-scratch disease with antimicrobial agents tetracyclines, chloramphenicol, and erythromycin also appear has had variable, but rarely dramatic, results. A single, double- to be effective. blind, placebo-controlled study involved 29 patients, 14 of Suggested antimicrobials and dosages derive from 3 whom received azithromycin [124]. The lymph node size had publications (table 3) [121–123]. The optimal duration of treat- regressed 30 days after treatment more often in the azithro- ment is uncertain, but 5–9 days appears to be adequate. Sixty mycin-treated patients (P p .02 ). If antimicrobial therapy is days of treatment is recommended when infection is associated used, patients weighing 145.5 kg (1100 lbs) should receive with bioterrorism, because concomitant inhalation may have 500 mg of azithromycin orally on day 1, followed by 250 mg occurred. Until results of susceptibility tests are available, cip- once daily for 4 additional days (A-I). Those weighing less rofloxacin is rational empirical therapy (B-III), especially with than the weight listed above should receive 10 mg/kg orally the possibility of genetically altered B. anthracis. Other fluor- on day 1, followed by 5 mg/kg on days 2–5 [124]. Cutaneous oquinolones, such as levofloxacin, gatifloxacin, or moxifloxa- bacillary angiomatosis therapy has not been systematically cin, are also likely to be effective. Initiation of intravenous examined. On the basis of results of case reports and small versus oral therapy depends on the severity of the illness, par- series, either erythromycin (500 mg 4 times per day) or doxy- ticularly the degree of edema. cycline (100 mg twice per day) appear to be effective (B-III). Some have suggested systemic corticosteroid therapy for pa- The duration of initial therapy, although not standardized, tients who develop malignant edema, especially of the head and should be at least 4 weeks. With relapses, retreatment with neck, but studies supporting this recommendation are lacking. prolonged therapy (lasting several months) should be enter- Airway compromise requiring intubation or trachostomy may tained until immunocompetence returns. Other antimicro- occur with malignant edema. bials with some efficacy are rifampin, trimethoprim-sulfa- Cat-scratch disease and bacillary angiomatosis. Bartonella methoxazole, and ciprofloxacin [125]. henselae causes most cases of cat-scratch disease in immuno- Erysipeloid. Erysipeloid is a cutaneous infection caused by competent hosts. Bacillary angiomatosis, seen in immunocom- the thin, pleomorphic, non–spore-forming gram-positive rod promised patients, especially with AIDS, can occur from either E. rhusiopathiae. It is a seen in persons who handle B. henselae or Bartonella quintana. In classic cat-scratch disease, fish, marine animals, swine, or poultry. Between 1 and 7 days a papule or pustule develops 3–30 days after a scratch or a bite. after exposure, a red maculopapular lesion develops, usually on Regional adenopathy occurs ∼3 weeks after inoculation in the fingers or hands. Erythema spreads centrifugally with cen- nodes that drain the infected area. Extranodal disease (such as tral clearing. A blue ring with a peripheral red halo may appear, that found in the CNS, liver, spleen, bone, and lung) develops giving the lesion a target appearance. Regional lymphangitis in р2% of cases. In ∼10% of cases, the nodes suppurate. The and/or lymphadenopathy occurs in about one-third of cases. disease course varies, but lymphadenopathy generally resolves A severe, generalized cutaneous infection also occurs. However, within 1–6 months. systemic symptoms and leukocytosis are unusual. Culture of a Cutaneous bacillary angiomatosis has 2 clinical appearances. lesion aspirate and/or biopsy specimen establishes the diagnosis, The dermal form is a red papule that varies in size from 1 but the results of blood cultures are rarely positive. Untreated millimeter to several centimeters, and the number of lesions erysipeloid resolves during a period of 3–4 weeks, but treatment may vary from 1 to 11000. The second form is a painful sub- probably hastens healing and perhaps reduces systemic com- cutaneous nodule with overlying skin having a normal or dusky plications. Most of the literature concerning therapy relates to hue. endocarditis, in which high-dose penicillin is generally used. Definitive confirmation of Bartonella infections may be dif- On the basis of in vitro susceptibilities and anecdotal state- ficult, because these fastidious organisms infrequently grow ments, penicillin is appropriate (B-III), although the optimum from pus or nodal tissue. Serologic testing supports the diag- duration of therapy is unknown. For cutaneous infection, pen- nosis. However, cross-reactivity occurs between B. henselae and icillin (500 mg orally 4 times per day) or amoxicillin (500 mg B. quintana, as well as with a few other organisms. PCR, al- 3 times per day) for 7–10 days seems to be rational. For patients

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1389 Table 6. Recommended therapy for infections following animal or human bites.

Route of drug administration Antimicrobial agent, by type of bite Oral Intravenous Comment Animal bite Amoxicillin/clavulanate 500/875 mg twice per daya … Some gram-negative rods are resistant; misses MRSA Ampicillin-sulbactam … 1.5–3.0 g every 6–8 h Some gram-negative rods are resistant; misses MRSA Piperacillin/tazobactam … 3.37 g every 6–8 h Carbapenem Misses MRSA Ertapenem … 1 g every day Imipenem … 1 g every 6–8 h Meropenem … 1 g every 8 h Doxycycline 100 mg twice per day … Excellent activity against Pas- teurella multocida; some strepto- cocci are resistant Penicillin 500 mg 4 times per day … plus dicloxacillin 500 mg 4 times per day TMP-SMZ 160–800 mg twice per day … Good activity against aerobes; poor activity against anaerobes Metronidazole 250–500 mg 4 times per day … Good activity against anaerobes; no activity against aerobes Clindamycin 300 mg 3 times per day … Good activity against staphylococci, streptococci and anaerobes; misses P. multocida First-generation cephalosporin Good activity against staphylococci and streptococci; misses P. multo- cida and anaerobes Cephalexin 500 mg 3 times per day … Cefazolin … 1 g every 8 h Second-generation cephalosporin Good activity against P. multocida; misses anaerobes Cefuroxime 500 mg twice per day 1 g every day Cefoxitin … 1 g every 6–8 h Third-generation cephalosporin Ceftriaxone … 1 g every 12 h Cefotaxime … 2 g every 6 h Fluoroquinolones Good activity against P. multocida; misses MRSA and some anaerobes Ciprofloxacin 500–750 mg twice per day 400 mg every 12 h Gatifloxacin 400 mg every day … Moxifloxacin 400 mg every day 400 mg every day Human bite Amoxicillin/clavulanate 500 mg every 8 ha … Some gram-negative rods are resistant; misses MRSA Ampicillin/sulbactam … 1.5– 3.0 g every 6 h Some gram-negative rods are resistant; misses MRSA Carbapenem Misses MRSA Ertapenem … 1 g every day Imipenem … 1 g every day Meropenem … 1 g every day Doxycycline 100 mg twice per day … Good activity against Eikenella spe- cies, staphylococci, and anaerobes; some streptococci are resistant (continued) Table 6. (Continued.)

Route of drug administration Antimicrobial agent, by type of bite Oral Intravenous Comment TMP-SMZ 160–800 mg twice per day … Good activity against aerobes; poor activity against anaerobes Metronidazole 250–500 mg 4 times per day … Good activity against anaerobes; poor activity against aerobes Clindamycin 300 mg 3 times per day … Good activity against staphylococci, streptococci, and anaerobes; misses Eikenella corrodens Cephalosporin Good activity against staphylococci and streptococci; misses E. corro- dens and gram-negative anaerobes Cephalexin 500 mg 4 times per day … Cefazolin … 1 g every 8 h Fluoroquinolone Good activity against E. corrodens; misses MRSA and some anaerobes Ciprofloxacin 500–750 mg twice per day 400 mg every 12 h Gatifloxacin 400 mg every day 400 mg every day Moxifloxacin 400 mg every day 400 mg every day

NOTE. As a rule, the use of fluoroquinolones is contraindicated by the US Food and Drug Administration for children and adolescents !18 years of age. It should also be noted that tetracyclines are rarely used in children younger than 8 years of age. Alternatives should be strongly considered for these two antibiotics [6]. MRSA, methicillin-resistant Staphylococcus aureus. TMP-SMZ, trimethoprim-sulfamethoxazole. a Should be given with food. who are intolerant of penicillins, treatment with cephalospo- bonic plague may develop septicemia and secondary plague rins, clindamycin, or fluoroquinolones should be effective. E. pneumonia, the latter permitting person-to-person transmis- rhusiopathiae is resistant to vancomycin, teicoplanin, and dap- sion. Diagnosis can be made by blood cultures and by aspirating tomycin [125, 126]. lymph nodes for staining and culture. PCR and other more . Glanders, caused by the aerobic gram-negative sophisticated tests are generally available only at reference lab- rod Burkholderia mallei, is mainly a disease of solipeds (e.g., oratories. Results of serologic tests may provide retrospective horses and mules). Humans become accidental hosts either by confirmation. inhalation or skin contact. Although other organs may be in- No controlled comparative trials of therapy for plague exist. volved, pustular skin lesions and lymphadenopathy with sup- Streptomycin has been the drug of choice (B-III), although purative nodes can be a prominent feature. Almost all glanders tetracycline and chloramphenicol are also considered to be ap- infections preceded the antibiotic era. Results of in vitro sus- propriate therapy (table 7). Although there have been no recent ceptibility tests suggest that ceftazidime, gentamicin, imipenem, reports of treatment of any sizable numbers of cases of plague, doxycycline, and ciprofloxacin should be effective. A recent studies from the Vietnam War period showed that most patients laboratory-acquired case was successfully treated with imipe- actually received streptomycin plus either tetracycline or chlor- nem and doxycycline for 2 weeks, followed by azithromycin amphenicol. Some patients have been successfully treated with and doxycycline for an additional 6 months [127]. kanamycin. Gentamicin has been suggested as a substitute for Bubonic plague. Plague results from infection with Y. pes- streptomycin, but its use in humans has been limited. On the tis, a facultative, anaerobic gram-negative coccobacillus. It pri- basis of in vitro susceptibilities and murine models, fluoro- marily affects rodents, being maintained in nature by several quinolones are another option. A multidrug-resistant strain of species of fleas that feed on them. Three plague syndromes Y. pestis has been isolated in Madagascar, and it is suspected occur in humans: septicemic, pneumonic, and bubonic. Bu- that an antimicrobial-resistant strain of the plague bacillus has bonic plague, the most common and classic form, develops been developed for biologic warfare. Unless introduced into when humans are bitten by infected fleas or have a breach in the rodent population, however, Y. pestis as a biowarfare agent the skin when handling infected animals. Domestic cat scratches is much more likely to be used as an aerosol, thus producing or bites may also transmit bubonic plague. Patients usually pneumonic plague rather than bubonic plague. Ciprofloxacin develop fever, headache, chills, and tender regional lymphade- has been suggested as a drug for both treatment and prevention nopathy 2–6 days after contact with the organism. A skin lesion of plague due to biowarfare agents, despite a lack of docu- at the portal of entry is sometimes present. Patients with bu- mented efficacy in humans. The optimal duration for treating

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1391 Table 7. Therapy for bubonic plague.

Dosage Adults (including Drug pregnant women) Childrena Streptomycinb 1 g im twice per day 30 mg/kg im daily in 2 divided doses Gentamicinb 2 mg/kg loading dose, followed by 1.7 mg/kg/day in 3 2 mg/kg every 8 h iv divided doses iv Tetracyclinec 500 mg po every 6 h … Chloramphenicol 25 mg/kg iv every 6 h (not to exceed 6 g total dose daily) 25 mg/kg iv every 6 h (not to exceed 6 g total dose daily) Doxycyclinec 100 mg iv or po twice daily Persons who weigh 145 kg: 100 mg iv or po twice daily; persons who weigh р45 kg: 2.2 mg/kg iv or po twice daily Ciprofloxacinc 500 mg po twice daily or 400 mg iv twice daily 20 mg/kg po twice daily or 15 mg/kg iv twice daily

NOTE. Agents of bioterrorism may be genetically altered for antimicrobial resistance. im, intramuscularly; iv, intravenously; po, orally. a Not appropriate for neonates. b Aminoglycoside dosages need adjustment according to renal function. c Doxycycline, tetracycline, and ciprofloxacin should be used only under exceptional circumstances in children !8 years of age or during pregnancy. bubonic plague is unknown, but 10–14 days is probably ade- choice for tularemia for several decades (B-III). A 1994 review quate. In view of the forgoing, the recommendations in reviews found 294 cases treated with streptomycin but only 20, 43, and by Perry and Fetherston [128] and by Inglesby et al. [129] seem 36 patients treated with tetracycline, chloramphenicol, and gen- to be rational (table 7). Patients with bubonic plague should tamicin, respectively [130]. Since then, a few patients have been be placed in respiratory isolation until completion of 48 h of received fluoroquinolones. Francisella species are resistant to effective drug therapy, because some develop secondary pneu- most b-lactam antibiotics. Even with favorable in vitro sus- monic plague. ceptibilities, failure rates with ceftriaxone have been high. One Tularemia—ulceroglandular or glandular. F. tularensis, al- patient has responded to imipenem, and 2 patients have re- though hardy and persistent in nature, is a fastidious, aerobic, sponded to erythromycin. When static drugs such as tetra- gram-negative coccobacillus. Illness can often be categorized cyclines or chloramphenicol are used, relapses may be more into several fairly distinct syndromes—ulceroglandular, glan- common, but often the patients have received brief therapy dular, typhoidal, pneumonic, oculoglandular, or oropharyngeal. (duration, !7 to 10 days). The glandular varieties are generally acquired by handling in- Acutely ill adults or children should receive an aminogly- fected animals, by tick bites, and sometimes by animal bites, coside, preferably streptomycin or possibly gentamicin. For especially from cats. Biting flies occasionally transmit the illness adults, the regimen for streptomycin is 30 mg/kg per day in 2 in the United States, whereas mosquitoes are common vectors divided doses (!2 g daily) or gentamicin 3–5 mg/kg per day in Europe. After an incubation period of 3–10 days, the patient in 3 divided doses. For children, streptomycin should be ad- typically develops a skin lesion (an ulcer or an eschar) at the ministered at 30 mg/kg per day in 2 divided doses and gen- entry site of the organism, along with tender regional adeno- tamicin at 6 mg/kg per day in 3 divided doses [130]. Treatment pathy in the lymph nodes—thus the term “ulceroglandular.” duration of 7–10 days is appropriate, with dosages of amino- In some patients, the skin lesion is inconspicuous or healed by glycosides adjusted according to renal function. Although no the time that they seek medical care, resulting in “glandular” tularemia. The illness is often associated with substantial fever, data exist, treatment with a parenteral agent until the acute chills, headache, and malaise. illness is controlled, followed by an oral agent, seems to be Confirmation of the diagnosis is usually accomplished by rational. means of serologic testing. Results of routine cultures are often In mild-to-moderate disease, oral tetracycline (500 mg 4 negative unless cysteine-supplemented media are used. Unsus- times per day) or doxycycline (100 mg twice per day) is ap- pected growth of Francisella species can cause laboratory-ac- propriate. Chloramphenicol (2–3 g daily in 4 divided doses) quired disease. PCR shows considerable promise for diagnosis. has been used in adults. Oral chloramphenicol is no longer No prospective controlled or randomized trials of therapy distributed in the United States, and the rare, but serious ad- for tularemia have been performed, nor has the optimal du- verse effect—bone marrow aplasia—makes it an undesirable ration of treatment been established, but many patients will agent. A few cases have been treated with fluoroquinolones, require initiation of treatment before confirmation of the di- with mixed results [131–133]. Oral levofloxacin (500 mg daily) agnosis. Streptomycin has been considered to be the drug of or ciprofloxacin (750 mg twice per day) in adults may be rea-

1392 • CID 2005:41 (15 November) • Stevens et al. sonable for mild to moderate illness. With oral regimens, pa- erythematous changes can occur around or near a surgical tients should receive at least 14 days of therapy. incision during the first week without swelling or wound drain- age. Most resolve without any treatment, including antibiotics. SURGICAL SITE INFECTIONS (SSIs) The cause is unknown but may relate to tape sensitivity or to other local tissue insult not involving bacteria. Numerous ex- Infections of surgical wounds are the most common adverse perimental studies and clinical trials examining the prevention events affecting hospitalized patients who have undergone of SSIs demonstrate that antibiotic therapy that is begun im- surgery [134]. Data from the National Nosocomial Infection mediately after surgery or that is continued for long periods Surveillance System show an average SSI incidence of 2.6%, after the procedure does not prevent or cure this inflammation accounting for 38% of nosocomial infections in surgical pa- tients [135]. The frequency of SSI is clearly related to the or infection [138–143]. Therefore, the suspicion of possible SSI category of operation, with clean and low-risk operations (as does not justify use of antibiotics without a definitive diagnosis defined by the National Nosocomial Infection Surveillance and the initiation of other therapeutic measures, such as open- System classification) having the lowest rate of infection and ing the wound (B-III) (figure 1). contaminated and high-risk operations having greater infec- Most SSIs have no clinical manifestations for at least 5 days tion rates [136]. Very few sources of objective evidence com- after the operation, and many may not become apparent for pare treatments for SSI. up to 2 weeks. Later infections are less likely, but surveillance SSIs are divided into the categories of superficial incisional standards mandate a follow-up duration of 30 days. Rarely does SSI, deep incisional SSI, and organ/space SSI [135]. Superficial any bacterial pathogen cause fever and clinical evidence of soft- incisional SSIs involve only the subcutaneous space, between tissue infection within the first 48 h after an operation or injury. the skin and underlying muscular fascia, occur within 30 days Infections that do occur in this time frame are almost always of the index operation, and are documented with at least 1 of due to S. pyogenes or Clostridium species. Accordingly, fever or the following findings: (1) purulent incisional drainage; (2) systemic signs during the first several days after surgery should positive results of culture of aseptically obtained fluid or tissue be followed by direct examination of the wound to rule out from the superficial wound; (3) local signs and symptoms of signs suggestive of streptococcal or clostridial infection but pain or tenderness, swelling, and erythema, with the incision should not otherwise cause further manipulation of the wound. opened by the surgeon (unless culture results are negative); or Patients with an early infection due to streptococci or clostridia (4) diagnosis of SSI by the attending surgeon or physician. have wound drainage with the responsible organisms present A deep incisional infection involves the deep layers of soft on Gram stain. WBCs may not be evident in most clostridial tissue (e.g., fascia and muscle) in the incision and occurs within and some early streptococcal infections. Another rare cause of 30 days after the operation or within 1 year after the operation early fever and systemic signs after operation is toxic shock if a prosthesis was inserted and has the same findings as de- syndrome due to staphylococcal wound infection [144, 145]. scribed for a superficial incisional SSI. In these cases, the wound is often deceptively benign in ap- An organ/space SSI has the same time constraints and evi- pearance. Erythroderma occurs early but not immediately, and dence for infection as a deep incisional SSI and involves any desquamation occurs late. Fever, hypotension, abnormal he- part of the anatomy (organs or spaces) other than the incision patic and renal blood findings, and diarrhea may be early find- opened during the operation [135]. Superficial and deep in- ings. Treatment is to open the incision, obtain and culture a cisional SSIs are skin and soft-tissue infections and will be wound specimen, and begin antistaphylococcal treatment. discussed in this guideline. Organ/space SSIs are usually dealt The primary, and most important, therapy for SSI is to open with separately as infections related to the relevant organ and space. Any deep SSI that does not resolve in the expected man- the incision, evacuate the infected material, and continue dress- ner after treatment should be investigated as a possible super- ing changes until the wound heals by secondary intention. ficial manifestation of a deeper organ/space infection. Although patients commonly receive antibiotics when SSI is In diagnosing SSIs, the physical appearance of the incision first diagnosed, there is little or no evidence supporting this probably provides the most reliable information. Local signs of practice. Studies of subcutaneous abscesses found no benefit pain, swelling, erythema, and purulent drainage are usually for antibiotic therapy when combined with drainage [24, 33]. present. In morbidly obese patients or in patients with deep, The single published trial of antibiotic administration for SSIs multilayer wounds (such as wounds following thoracotomy), found no clinical benefit associated with this treatment [146]. the external signs of SSIs may be very late but always appear. Most textbooks of surgery, infectious diseases, or even surgical Although many patients with SSIs will have fever, it usually infectious diseases extensively discuss the epidemiologic char- does not occur immediately after operation, and in fact, most acteristics, prevention, and surveillance of SSIs but not their postoperative fevers are not associated with SSI [137]. Flat, treatment [147–153]. Two articles contain simple, unrefer-

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1393 Figure 1. Algorithm for the management and treatment of surgical site infections. *For patients with type 1 (anaphylaxis or hives) to b- lactam antibiotics. Where the rate of infection with methicillin-resistant Staphylococcus aureus infection is high, consider vancomycin, daptomycin, or linezolid, pending results of culture and susceptibility tests. Adapted and modified with permission from [154]. GI, gastrointestinal. enced, recommendations to open an infected wound without mon organisms. Because incisions in the axilla have a significant using antibiotics [154, 155]. recovery of gram-negative organisms and incisions in the per- A common practice, endorsed by expert opinion, is to open ineum have a higher incidence of gram-negative organisms and all infected wounds (B-III). If there is minimal surrounding anaerobes [24, 26, 157], antibiotic choices should be made evidence of invasive infection (!5 cm of erythema and indu- accordingly (table 4). Figure 1 presents a schematic algorithm ration), and if the patient has minimal systemic signs of in- to approach patients with suspected SSI [154] and includes fection (a temperature of !38.5C and a pulse rate of !100 specific antibiotic recommendations [158]. beats/min), antibiotics are unnecessary. Because incision and drainage of superficial abscesses rarely causes bacteremia [156], INFECTIONS IN THE IMMUNE COMPROMISED antibiotics are not needed. For patients with a temperature of HOST 138.5C or a pulse rate of 1100 beats/min, a short course of antibiotics, usually for a duration of 24–48 h, may be indicated. Immunocompromised patients, by definition, are at increased The antibiotic choice is usually empirical but can be supported risk of infection and have a decreased ability to control local by findings of Gram stain and results of culture of the wound infection [159–161]. Skin and soft-tissue infections are com- contents. SSIs that occur after an operation on the intestinal mon, and because they are caused by a wide range of pathogens tract or female genitalia have a high probability of having a and are often part of a widely disseminated infection, they mixed gram-positive and gram-negative flora with both fac- frequently pose a difficult clinical problem [162, 163]. Infection ultative and anaerobic organisms. If such an infection is being prevention in immunocompromised patients is important and treated with empirical antibiotics, any antibiotic considered to demands careful attention to measures that protect the skin be appropriate for treatment of intra-abdominal infection is from unnecessary trauma, maceration, or alterations in the reasonable (table 4). If the operation was a clean procedure normal microbial flora. When infections do develop, it is critical that did not enter the intestinal or genital tracts, S. aureus to establish a specific etiologic diagnosis, because many are (including MRSA) and streptococcal species are the most com- nosocomial and are caused by pathogens with increased anti-

1394 • CID 2005:41 (15 November) • Stevens et al. microbial resistance. Skin lesions, no matter how small or in- Initial Infection in Neutropenic Patients nocuous in appearance, should be carefully evaluated, and the Historically, the primary gram-negative pathogens have been clinician must remember that their gross appearance is fre- E. coli, Klebsiella species, and P. aeruginosa, but there is wide quently altered by the decreased inflammatory response. Thus, variability in the pathogens isolated in different treatment cen- the initial clinical impressions must be supplemented with a ters [159, 160, 164, 165]. The relative incidence of gram-neg- systematic approach for diagnosis and treatment [164, 165]. ative bacilli as causes of initial infections has decreased signif- After considering the important patient-specific factors con- icantly during the past 2 decades, but they remain important cerning the patient’s immune compromised status (e.g., neu- pathogens for patients with profound neutropenia (!100 poly- tropenia or neutrophil defects, cellular immune defect, and morphonuclear leukocytes/mL) with a prolonged duration (7– iatrogenic procedures), the gross morphologic characteristics 10 days) or for patients who have not received antibacterial of the skin lesion(s) should be characterized, the extent of the prophylaxis during their period of neutropenia [168]. Der- infection determined (e.g., localized vs. disseminated), and ap- matologic manifestations of gram-negative skin and soft-tissue propriate diagnostic tests undertaken to identify the infecting infections include erythematous maculopapular lesions, focal pathogen. Finally, antimicrobial therapy should be initiated, on or progressive cellulitis, cutaneous nodules [167], and ecthyma the basis of the important clinical parameters identified and gangrenosum. Ecthyma gangrenosum begins as painless, ery- the most likely offending pathogens [164, 165]. Although blood thematous, macules that rapidly become painful and necrotic cultures or tests for detection of antigen in blood or vesicular during a 12–24-h period. They may be discrete or multiple; are fluid may be helpful, the most specific method is aspiration or found preferentially in the groin, axilla, or trunk; and can in- 1 ! biopsy of the lesion to obtain material for histological and crease in size from 1 cm to 10 cm in 24 h. Ecthyma gan- microbiological evaluation. Analysis of lesion biopsy specimens grenosum is a cutaneous vasculitis caused by bacterial invasion of the media and adventitia of the vessel wall. Progression of yields positive results for only 20% of otherwise healthy patients the lesion leads to dermal necrosis, and bacteria are often visible with focal skin lesions [57]. Similar prospective studies in- during microscopic analysis of biopsy specimens. Ecthyma gan- volving immunocompromised patients have not been per- grenosum has classically been reported to occur with P. aeru- formed. Consequently, most clinicians who treat immunocom- ginosa infections, but similar lesions can occur with dissemi- promised patients combine blood cultures, tests for antigen nated infections caused by other Pseudomonas species, detection, and radiographic imaging with analysis of a biopsy Aeromonas species, Serratia species, S. aureus, Stenotrophomonas specimen obtained from the abnormal skin lesion to optimize maltophilia, Candida species, and fungi, including Aspergillus, recovery of the offending pathogen and to direct pathogen- Mucor, and Fusarium species [166]. specific antimicrobial therapy and local surgical management. The increased use of antimicrobial prophylaxis with fluor- oquinolones or trimethoprim-sulfamethoxazole and the fre- Predisposition to Infection: Neutropenia quent reliance on indwelling vascular access devices have re- Patients with neutropenia are predisposed to infection because sulted in gram-positive organisms being the most frequently of insufficient circulating neutrophils, lack of adequate myeloid isolated pathogens in initial infections [169]. These organisms, marrow reserve, or congenital or acquired defects in neutrophil in order of decreasing prevalence, include coagulase-negative function [159–163, 165]. Neutropenia is frequently associated staphylococci, viridans streptococci, enterococci, S. aureus, Co- with mucosal or integumentary barrier disruption, and the in- rynebacterium species, Clostridium species, and Bacillus species digenous colonizing florae are responsible for most infections. and often represent part of the patient’s normal skin flora. Soft- More than 20% of patients with chemotherapy-induced neu- tissue infections due to these pathogens usually begin as a focal tropenia develop skin and soft-tissue infections, many of which area of erythematous cutaneous tenderness, a macular or ma- are due to hematogenous dissemination from other sites, such culopapular eruption, or as cellulitis. The most frequent in- as the sinuses, lungs, and the alimentary tract [162, 163, 166]. fection sites are the groin, axilla, areas of cutaneous disruption Important pathogens for neutropenic patients can be separated (e.g., vascular catheter or bone marrow aspiration sites), or into organisms most likely to cause an “initial infection” (char- other portions of skin that are moist and frequently abraded. acterized by !7 days of fever and neutropenia) and those more Hematogenous dissemination of these gram-positive organisms likely to cause a “subsequent infection” (with an onset after 7 to the skin and soft tissue is uncommon except for S. aureus days of neutropenia) [159, 167]. Pathogens causing initial in- and some Clostridium species. A toxic shock–like syndrome has fections are usually bacteria, including both gram-negative and been described with blood stream infections caused by toxin- gram-positive organisms. Pathogens causing subsequent infec- producing viridans streptococci, and diffuse erythroderma can tions are usually antibiotic-resistant bacteria, yeast, or fungi be part of the early clinical presentation [170]. (table 8). The foundation of the initial treatment of patients with neu-

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1395 Table 8. Skin and soft-tissue infections in the immune compromised host: treatment and management.

Frequency or reason for Predisposing factor, pathogen Type of therapy Duration of therapy surgery Adjunct Neutropenia Initial infection Bacteria Gram negative Monotherapy or antibiotic 7–14 days Rare G-CSF/GM-CSF; combination granulocyte therapya Gram positive Pathogen specific 7–10 days Rare No Subsequent infection Antibiotic-resistant Pathogen specific 7–14 days Rare G-CSF/GM-CSF;b bacteria granulocyte therapya Fungi Amphotericin B, voriconazole, or Clinical and radiologic For localized Catheter removal; caspofungin resolution infection G-CSF/GM-CSF;b granulocyte therapya Cellular immune deficiency Bacteria Nocardia species Trimethoprim-sulfamethoxazole or 3–12 months Rare No sulfadiazine Atypical mycobacteria Antibiotic combination (including 3–6 weeks Yes No a macrolide) Fungi Cryptococcus species Amphotericin B plus 5-fluorocyto- 8–12 weeks No No sine or fluconazole Histoplasma species Amphotericin B or itraconazole Viruses Varicella-zoster virus Acyclovir 7–10 days No No famciclovir valacyclovir Herpes simplex virus Acyclovir 7 days No No famciclovir valacyclovir Cytomegalovirus Ganciclovir 21 days No No

NOTE. G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte-monocyte colony-stimulating factor. a Use if gram-negative bacillary infection is unresponsive to appropriate antimicrobial therapy or if the patient has invasive fungal infection. b Progressive infection, pneumonia, and invasive fungal infection. tropenia is the administration of empirical, broad-spectrum regimen. This strategy, however, has no impact on the survival antibiotics at the first clinical signs or symptoms of infection, of adult patients with neutropenia-associated bloodstream in- including fever [159–161, 164, 165]. Antibiotic selection should fections due to gram-positive organisms [171], and because of follow the clinical care guidelines developed by the Infectious the increasing prevalence of vancomycin-resistant organisms, Diseases Society of America and the National Comprehensive current guidelines restrict the empirical use of this agent [164, Cancer Network [164, 165]. Excellent results have been re- 165]. Thus, if empirical vancomycin is administered, it should ported for gram-negative infections using broad-spectrum be discontinued if culture results remain negative after 72–96 monotherapy with carbapenems, cephalosporins that possess h [164, 165]. Decisions regarding initial empirical antibiotic antipseudomonal activity, or piperacillin/tazobactam [164]. regimens and the subsequent antimicrobial adjustments, how- Antibiotic combinations using an aminoglycoside plus an an- ever, must consider adequate antimicrobial coverage against the tipseudomonal-penicillin or a extended-spectrum cephalospo- more virulent gram-positive organisms (S. aureus, viridans rin, or the combination of an extended-spectrum penicillin and streptococci, or antibiotic-resistant pathogens, such as MRSA, ciprofloxacin, are also frequently recommended [164, 165]. vancomycin-resistant enterococci, or penicillin-resistant S. Treatment of neutropenia-associated infections due to gram- pneumoniae.) [170, 172–175]. Linezolid or daptomycin may be positive organisms is now dictated by the increasing resistance acceptable alternatives to vancomycin. Linezolid is the drug of of these pathogens, leading many clinicians to consider the choice for infections caused by vancomycin-resistant entero- empirical use of vancomycin as part of the initial antibiotic cocci, but potential hematologic toxicity and cost should limit

1396 • CID 2005:41 (15 November) • Stevens et al. its use to individuals with pathogen-directed needs [176]. Al- routine use of azole antifungal prophylaxis was reported to be though linezolid and daptomycin have US Food and Drug as high as 12% for patients with profound and prolonged neu- Administration approval for skin and soft-tissue infections, no tropenia or recipients of blood or bone marrow transplants prospective, randomized studies involving compromised pa- [183]. Candida albicans (62% of candidiasis cases) and Candida tients have been performed. tropicalis (21% of candidiasis cases) were most frequently iso- Surgical intervention is rarely appropriate early during neu- lated. Between 6% and 13% of patients with invasive candidiasis tropenia-associated infection but may be necessary to drain a develop single or multiple nodular skin lesions [166, 183]. Such soft-tissue abscess after marrow recovery or for treatment of a lesions are discrete, pink-to-red subcutaneous papules or nod- progressive polymicrobial fasciitis. Most such infections do not ules and are most commonly found on the trunk and extrem- require adjunct colony-stimulating factor therapy or granulo- ities. The nodules are usually smaller (diameter, 0.5–0.8 cm) cyte transfusions, but these therapies are often considered when than ecthyma gangrenosum lesions, are initially nontender, and infection progresses despite appropriate antimicrobial treat- may evolve to develop central pallor; the nodules may become ment [159–161, 176, 177]. hemorrhagic in thrombocytopenic patients [160, 166]. Myositis can develop as a consequence of hematogenous infection and Subsequent Infection in Neutropenic Patients is most common with C. tropicalis infections [185, 186]. In Subsequent infections are the major cause of infection-asso- these cases, pain is often the chief initial complaint. Muscle ciated morbidity and mortality for patients with prolonged and soft-tissue abscess formation is uncommon, but when re- (duration, 7–10 days) and profound (!100 polymorphonuclear ported, it has usually followed bone marrow recovery. leukocytes/mL) neutropenia [159–161]. Of such patients, 25%– Trichosporon beigelii. T. beigelii is an uncommon, but 50% develop a second or subsequent episode of fever and/or frequently fatal disseminated fungal infection that often in- infection [167]. Although the skin and soft tissues are less fre- volves the skin [187]. Dermatologic manifestations vary from quently infected (10%–15% of cases), they may represent an multiple erythematous macules to maculopapular lesions, and early site of infection dissemination. Among subsequent infec- analysis of tissue biopsy specimens reveals a mixture of true tions, 10%–15% are caused by antibiotic-resistant gram-neg- hyphae, pseudohyphae, budding yeast, and arthroconidia that ative bacilli, 30%–40% are caused by antibiotic-resistant gram- can be easily mistaken for Candida species [166]. positive organisms (coagulase-negative staphylococci and Aspergillus species. Infections due to Aspergillus species vancomycin-resistant enterococci, most commonly), and 150% occur in 2%–10% of patients with profound and prolonged are caused by fungi [167]. Despite the incidence of subsequent neutropenia, and they may be increasing in frequency [188, infections caused by antibiotic-resistant gram-positive patho- 189]. Mortality remains high for all of these infections [188, gens, the empirical administration of vancomycin is unjustified 190]. Aspergillus fumigatus is the most frequently isolated spe- for patients with neutropenia and persistent fever (!96 h after cies (50% of cases), followed by Aspergillus flavus, Aspergillus initiation of empirical antibiotic therapy) who are clinically niger, and Aspergillus terreus [188]. Isolation of Aspergillus spe- stable and have no identified site of infection [178]. Empirical cies from blood cultures is infrequent, but dissemination to the antifungal therapy for patients with neutropenia and persistent brain, gastrointestinal tract, and other visceral organs is com- fever remains a common clinical practice, as revealed by 2 monly revealed during autopsy [191]. Cutaneous infections are clinical studies using amphotericin B that were conducted in unusual, but they may occur secondary to hematogenous dis- the 1980s [179, 180]. Recently, 2 randomized, international, semination or locally at sites of intravenous catheter insertion multicenter trials found that caspofungin [181] and voricon- or at nail bed and cuticle junctions on fingers and toes [192, azole [182] were each suitable alternatives to amphotericin B 193]. Because Aspergillus organisms have a propensity for an- in this patient population. Thus, profoundly neutropenic pa- gioinvasion, they produces painful skin nodules that may rap- tients with persistent fever who are systemically ill despite em- idly become necrotic and resemble pyoderma gangrenosum pirical antibiotic therapy may benefit from empirical antifungal lesions [166]. treatment (B-I). Rhizopus and Mucor species. Cutaneous infections due to Candida species. The frequency and occurrence of can- organisms from the Rhizopus and Mucor genera are uncom- didiasis has been well described [183, 184]. More than 80% of mon, but similar to infections due to Aspergillus species, epi- high-risk patients who develop neutropenia are colonized with dermal and dermal necrosis may develop, because of the ten- Candida species, and superficial mucosal and cutaneous infec- dency of these organism to invade blood vessels. Skin lesions tions are common. These noninvasive infections can be effec- are usually erythematous, nodular, and tender. Local Mucor tively treated with improved skin care and a topical antifungal infections have occurred as a consequence of contaminated agent or with a short course systemic azole antibiotic (e.g., bandages or other skin trauma, but patients with pulmonary fluconazole). The incidence of invasive candidiasis before the Mucor infection may also develop secondary cutaneous in-

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1397 volvement from presumed hematogenous dissemination [194, arise from local skin inoculation, whereas others result from 195]. Disseminated infections are almost never associated with hematogenous dissemination. positive blood culture results, but even without a documented Bacteria. Nontuberculous mycobacteria are ubiquitous, fungal bloodstream infection, the mortality rate for these in- and most cutaneous mycobacteria infections occur after pri- fections remains very high [196]. mary inoculation at sites of skin disruption or trauma, but Fusarium species. Fusarium species are now more fre- hematogenous dissemination does occur [210–215]. Dissemi- quently identified as the infecting pathogens in patients with nated infection with Mycobacterium avium complex occurs prolonged and profound neutropenia [196–198]. Patients com- preferentially among patients with HIV disease, whereas blood- monly have myalgias and persistent fever despite antimicrobial stream infections with Mycobacterium fortuitum, Mycobacte- therapy. Skin lesions occur in 60%–80% of these infections and rium chelonae, Mycobacterium abscessus, Mycobacterium ulcer- begin as multiple erythematous macules with central pallor that ans, or Mycobacterium mucogenicum are more frequent among quickly evolve to papules and necrotic nodules. Lesions localize compromised hosts with indwelling vascular-access devices preferentially to the extremities but also occur on the face and [216]. Sporadic cases in compromised hosts are also reported trunk. Recovery of Fusarium species from blood cultures is with Mycobacterium kansasii, Mycobacterium haemophilum, and common (40%–50% of cases) [191]. Mortality from this in- Mycobacterium marinum. Dermatologic manifestations include fection remains high among patients with persistent immu- a poorly resolving cellulitis, painless 1–2-cm nodules, necrotic nodeficiency, although the new azole antifungal agents appear ulcers, and subcutaneous abscesses. to be promising [199]. Treatment of nontuberculous mycobacterial infections of the The clinician must remember that yeast and fungal infections skin and soft tissues requires prolonged combination therapy remain the primary cause of infection-associated death among (duration, 6–12 weeks) that should include a macrolide anti- patients with neutropenia or patients who undergo blood or biotic (e.g., clarithromycin). Surgical debridement is appro- bone marrow transplantation [200, 201]. Diagnosis of these priate and often necessary to remove devitalized tissue and to infections remains difficult, and recovery of fungi from an as- promote skin and soft-tissue healing [216]. piration or biopsy of skin or soft tissue almost always warrants Cutaneous Nocardia infections usually represent metastatic aggressive therapy. Amphotericin B and lipid formulations of foci of infection from a primary pulmonary source. Nocardia amphotericin B have been the gold standard of treatment, but asteroides, Nocardia farcinica, and Nocardia brasiliensis have newer antifungal agents, such as voriconazole and caspofungin, been associated with cutaneous disease [217, 218]. The der- appear to be at least as effective against Aspergillus species, matologic manifestations are usually limited to subcutaneous Fusarium species, and non-albicans species of Candida [164, nodules or abscess and panniculitis. Soft-tissue abscesses are 165, 184, 202–204]. All of the new antifungal agents have less frequently painless and are cold to the touch. The incidence of serious acute toxicity and less nephrotoxicity but are also more local and disseminated Nocardia infections has decreased with expensive than conventional amphotericin B [203–208]. The the routine use of trimethoprim-sulfamethoxazole prophylaxis importance of treatment with adjunct growth factor or gran- for patients who experience prolonged periods of cellular im- ulocyte transfusion is unsubstantiated, but they are frequently mune deficiency. considered for patients who remain profoundly neutropenic Trimethoprim-sulfamethoxazole remains the treatment of and unresponsive to antimicrobial therapy [177]. The routine choice [218], but other sulfa antibiotics (e.g., sulfadiazine and use of azole prophylaxis in high-risk patients has dramatically sulfasoxazole) or imipenem are effective. Prolonged therapy is decreased the incidence of invasive C. albicans infections but important, and the duration of treatment (6–24 months) has increased the incidence of infections due to azole-resistant should take into account the presence of disseminated disease yeast, including C. glabrata or C. krusei [209]. and the extent of the patient’s underlying immune suppression. Surgical debridement is recommended for necrotic nodules or Predisposition to Infections: Cellular Immune Deficiency large subcutaneous abscesses. Patients with Hodgkin lymphoma or non-Hodgkin lymphoma; Fungi. Cryptococcal infections originate in the lungs, often recipients of blood, marrow, or solid organ transplants; and with early hematogenous dissemination to the meninges and patients being treated with corticosteroids and other immune skin or soft tissues [219], but primary cutaneous cryptococcus suppressants are predisposed to infection because of abnor- also occurs [220]. Single or multiple painless skin lesions in- malities of their cellular (lymphocyte-mediated) immune func- volving the face and scalp develop in 5%–10% of clinically tion. These patients are at increased risk for infections, and the infected patients, and in some patients, these lesions may pre- infections are caused by a select group of bacteria, fungi, viruses, cede documented cryptococcal meningitis by several weeks. Cu- protozoa, and helminthes, but only a few of these cause skin taneous cryptococcal infections may appear as papules (often and soft-tissue infections (table 8). Some of these infections similar to moluscum contagiosum lesions), nodules, or pustules

1398 • CID 2005:41 (15 November) • Stevens et al. or as chronic draining necrotic ulcers [220]. Cryptococcal cel- for at least 4–6 days, with the entire disease duration being !2 lulitis has occurred in recipients of blood, bone marrow, or weeks. In immune suppressed hosts, lesions may continue to solid organ transplants [221], although the incidence has dra- develop over a longer period (7–14 days) and generally heal matically decreased with the prophylactic use of the newer azole more slowly unless effective antiviral therapy is administered agents, particularly fluconazole. Fluconazole is often used as [164, 165, 230, 231]. Without adequate treatment, some im- initial treatment, for patients with more mild infections, or to mune suppressed patients develop chronic ulcerations with per- complete treatment after the patient has shown clinical and sistent viral replication complicated by secondary bacterial and microbiologic improvement with amphotericin B and 5-flu- fungal superinfection. Disseminated VZV lesions characteris- cytosine induction therapy [222, 223]. Surgical debridement tically begin on the face and trunk and then evolve peripherally. and/or drainage are not helpful in the management of skin or Cutaneous VZV, unlike smallpox, usually show lesions simul- soft-tissue cryptococcal infections [223]. taneously in the varied stages of infection progression. Preven- Cutaneous manifestations of acute progressive disseminated tion of viral reactivation with oral acyclovir, famciclovir, or histoplasmosis are rare [224] and usually occur in patients with valacyclovir is an important component of the treatment of severe cellular immune deficiency, where they appear as non- cutaneous VZV infection [164, 165]. Such therapy is usually specific maculopapular eruptions that may become hemor- administered to high-risk patients during the period of max- rhagic. Oral ulcers sometimes present, particularly in the sub- imum immunosuppression. Recipients of an allogenic blood acute, disseminated form of the disease. Histopathologic and bone marrow transplant routinely take acyclovir (800 mg analysis of these skin lesions reveals necrosis surrounding the twice per day) or valacyclovir (500 mg twice per day) during superficial dermal vessels, and with special stains, both intra- the first year after transplantation [165]. High-dose intravenous cellular and extracellular yeast may be seen. Prompt adminis- acyclovir remains the treatment of choice for VZV infections tration of amphotericin B therapy is the recommended treat- in compromised hosts [228] (B-III). Oral acyclovir, famciclovir, ment for patients with cellular immune deficiency and acute, and valacyclovir are beneficial for VZV infections in otherwise life-threatening, progressive disseminated histoplasmosis [225]. healthy hosts, but oral therapy should probably be reserved for Patients often show a rapid clinical improvement within 1–2 mild cases of VZV disease in patients with transient immune weeks, and itraconazole can then replace amphotericin B to suppression or as treatment to complete therapy once the pa- complete at least 6–12 months of treatment. Patients with ill- tient has shown a clinical response to intravenous acyclovir nesses that result in profound and prolonged immune sup- [165, 230, 231]. pression should receive long-term suppressive therapy with itra- Herpes simplex virus (HSV) has a worldwide distribution, conazole after the initial treatment course is complete. and 190% of adults have to HSV-1 by the fifth decade Viruses. Varicella zoster virus (VZV) is one of the 2 most of life [231, 232]. Antibodies against HSV-2 appear in puberty frequent herpesviruses to cause cutaneous infection in im- and correlate with sexual activity. The seroprevalence of HSV- munosuppressed patients [226–228]. Patients without a pre- 2 antibody among patients in the United States is now 20%– ceding history of varicella are at significant risk of developing 25% [232, 233]. HSV infections in compromised hosts are the disease if exposed, but herpes zoster with or without dis- almost exclusively due to viral reactivation [232]. Orofacial and semination is a more frequent clinical concern [227, 228]. Be- genital sites are the most common cutaneous locations, but tween 65% and 70% of adult patients are seropositive for VZV, autoinoculation can occur in almost any area. Infections of the and this identifies those patients at risk for future reactivation fingernail bed and cuticle (herpetic ) occur because of infection. Herpes zoster occurs most frequently during the first inoculation of HSV at sites of epidermal surface breakdown. year after treatment, or after receipt of a blood, bone marrow, Cutaneous lesions are often preceded by localized pain or a or a solid organ transplant [226, 229]. Depending on the in- tingling sensation. Early skin lesions are usually focal, erythem- tensity of treatment or type of transplantation, 25%–45% of atous, and maculopapular. These evolve to form thin-walled such patients develop dermatomal zoster, with a 10%–20% risk vesicles and then pustules before becoming small ulcers. Lesions of developing dissemination without prompt and effective an- frequently coalesce, and chronic, poorly healing ulcers are char- tiviral therapy. A few patients present initially with disseminated acteristic of HSV infections among immunocompromised cutaneous infection that mimics varicella. Herpes zoster (also hosts. These ulcerative lesions rarely include a vesicular com- known as “shingles”) causes a unilateral, vesicular eruption with ponent, thus making the clinical diagnosis of a chronic HSV dermatomal pain that often precedes the skin findings by 24– infection more difficult. Bloodborne HSV dissemination, man- 72 h (and sometimes longer). Early lesions are erythematous ifested by multiple vesicles over a widespread area of the trunk macules that rapidly evolve to papules and then to vesicles. or extremities, is uncommon, but when seen among compro- The vesicles frequently coalesce, form bullae, and scab before mised hosts, it is usually secondary HSV-2 infection. Acyclovir healing. Lesions in otherwise healthy hosts continue to erupt is the treatment of choice for HSV infections, although fam-

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1399 ciclovir and valacyclovir are also highly effective [164, 165]. positive when the catheter infection is limited to the entry site The development of acyclovir-resistant HSV isolates is well de- [164, 165, 236]. The skin manifestations of a tunnel infection scribed and occurs more frequently among immune compro- include a painful cellulitis that may progress to necrosis or mised patients [234]. Suppression of HSV reactivation or con- ulceration. Many early port-pocket infections are painless, hin- tinued treatment until the ulcerated skin or mucosal lesions dering the clinician’s ability to recognize the catheter as the site have totally healed may decrease the incidence of infections of infection. Gram-positive organisms cause two-thirds of the caused by acyclovir-resistant HSV strains. The treatment of vascular device infections. Whereas coagulase-negative staph- acyclovir-resistant HSV isolates is a prolonged course of intra- ylococci are the most frequent pathogens, gram-negative bacilli, venous foscarnet [234]. Surgery should be avoided in patients fungi, and atypical mycobacteria are other causes [165, 236]. with HSV infections, unless a documented bacterial or fungal The prevalence of infection due to gram-positive pathogens abscess is identified. justifies recommending the use of empirical intravenous van- Cutaneous cytomegalovirus infections have a highly variable comycin for treatment of clinically serious catheter-associated appearance, including cutaneous nodules, ulcers, indurated infections [164, 165]. Most entry-site infections can be treated plaques, maculopapular eruptions, and hemorrhagic vesicles. effectively with appropriate antimicrobial therapy without cath- The true prevalence of these cutaneous infection is uncertain, eter removal [164, 165, 236]. Tunnel or port-pocket infections because many have a bland appearance, biopsies are only rarely require catheter removal and culture, with modification of the performed, and infection sites usually do not contain cells that empirical antimicrobial therapy on the basis of culture and demonstrate cytomegalovirus inclusions [235]. Prolonged gan- susceptibility test results [165, 236]. Catheter-site infections ciclovir therapy is the treatment of choice [165]. caused by fungi or nontuberculosis mycobacteria routinely re- Parasites. Rarely, the skin and soft-tissue structures of im- quire catheter removal and debridement of devitalized soft tis- munosuppressed patients can also be affected by parasites, in- sues [211]. A recent report documented a 100% cure of tunnel cluding but not limited to Strongyloides stercoralis, free-living infections caused by nontuberculous mycobacteria with com- ameba (Acanthamoeba species and Balamuthia species), and bination antimicrobial therapy for 3–6 weeks plus catheter re- Sarcoptes scabiei. moval and debridement of the infected soft tissue [211].

Infections Related to Iatrogenic Procedures Acknowledgments Many iatrogenic procedures disrupt the integumentary barrier and increase the risk of infection for immunocompromised Potential conflicts of interest. D.L.S. has received research funding from Wyeth, Lederle, Pfizer, Amgen, Roche, and Cubist and has served as patients. Vascular-access devices are the most common iatro- a consultant for Schering Plough, Pfizer and Arpida. A.L.B. has served as genic factor that predisposes patients to skin and soft-tissue a consultant for Merck, Cubist, Pharmacia, and Schering Plough. H.F.C. infections, but many patients with intravenous catheters also has received grant or research support from Ortho-McNeil and Cubist, have additional factors (e.g., neutropenia, cellular immuno- has served as a consultant for or on the advisory board of Otho-McNeil and Osmotics, and has received honoraria from Basilea. P.D. has received deficiency, or humoral immunodeficiency) that increase their grants for clinical research from, served on the advisory board of, and/or risk of infection. Intravenous vascular-access devices are almost lectured for honoraria from GlaxoSmithKline, Bayer, Eli Lily, Merck, Wy- universal for patients, such as blood, marrow, and solid organ eth-Ayerst, Bristol-Myers Squibb, AstraZeneca, Pfizer, Aventis, Hoffman– La Roche, Arrow, Ortho-McNeil, Perke-Davis, Abbot, ICOS, Immunex, transplant recipients, who are undergoing cancer therapy or in Chiron, Searle, Cubist, Virucon, InterMune, Peninsula, Johnson & Johnson, need of intensive care. These vascular devices allow adminis- and BRAHMS. E.J.C.G. has served as a consultant for, on the speakers’ tration of multiagent therapy, blood products, prolonged an- bureaus of, and/or has received research support from Merck, Aventis, Cubist, Bayer, Schering Plough, GlaxoSmithKline, Ortho-McNeil, and Vi- timicrobial treatment, intravenous nutrition, and withdrawal curon and has served on the scientific advisory board of Merck, Bayer, and of blood for monitoring and microbial evaluation. Many of Schering Plough. J.G.M. has served on the speakers’ bureaus of Merck, these catheters remain in place for prolonged periods, and the Pfizer, Enzon, Aventis, and Schering Plough. All other authors: no conflicts. risk of cutaneous infections varies with the device, the duration of catheter placement, and the severity of immune suppression. References Cutaneous infections associated with catheter placement in- clude the entry site infection (inflammation from the entry site 1. Wong CH, Khin LW, Heng KS, Tan KC, Low CO. The LRINEC to the first subcutaneous cuff), a tunnel infection (inflammation (Laboratory risk indicator for necrotizing fasciitis) score: a tool for distinguishing necrotizing fasciitis from other soft tissue infections. involving the skin and soft tissues that surround the catheter Crit Care Med 2004; 32:1535–41. tunnel from the catheter cuff to the venous entrance), or vas- 2. Thorell E, Jackson MA, Bratcher D, Swanson DS, Selvaragan R. An- cular port-pocket infection. Tunnel and port-pocket infections timicrobial resistance of Staphylococcus aureus from Kansas City chil- dren: what is the appropriate current therapy for pediatric staphy- are frequently accompanied by positive blood culture results lococcal infections [abstract 252]? In: Proceedings and abstracts of (30%–40% of episodes), whereas blood culture results are rarely the 42nd Annual Meeting of the Infectious Diseases Society of America

1400 • CID 2005:41 (15 November) • Stevens et al. (Boston). Alexandria, VA: Infectious Diseases Society of America, aerobic and aerobic bacteriology and outpatient management. Ann 2004:81. Intern Med 1977; 87:145–9. 3. Ruhe JJ, Monson TP. Use of tetracyclines for infections caused by 25. Ghoneim AT, McGoldrick J, Blick PW, Flowers MW, Marsden AK, methicillin-resistant Staphylococcus aureus [abstract 516]. In: Pro- Wilson DH. Aerobic and anaerobic bacteriology of subcutaneous ab- ceedings and abstracts of the 42nd Annual Meeting of the Infectious scesses. Br J Surg 1981; 68:498–500. Diseases Society of America (Boston). Alexandria, VA: Infectious Dis- 26. Brook I, Frazier EH. Aerobic and anaerobic bacteriology of wounds eases Society of America, 2004:139. and cutaneous abscesses. Arch Surg 1990; 125:1445–51. 4. Van Beneden CA, Facklam R, Lynfield R, Glennen A, Beall B, Whitney 27. Leach RD, Eykyn SJ, Phillips I, Corrin B, Taylor EA. Anaerobic axillary C. Erythromycin resistance among invasive group A streptococcal abscess. Br Med J 1979; 2:5–7. infections, United States, 1999–2001 [abstract 345]. In: Proceedings 28. Whitehead SM, Leach RD, Eykyn SJ, Phillips I. The aetiology of scrotal and abstracts of the 42nd Annual Meeting of the Infectious Diseases sepsis. Br J Surg 1982; 69:729–30. Society of America (Boston). Alexandria, VA: Infectious Diseases So- 29. Edmiston CE Jr, Walker AP, Krepel CJ, Gohr C. The nonpuerperal ciety of America, 2004:102. breast infection: aerobic and anaerobic microbial recovery from acute 5. Yun HJ, Lee SW, Yoon GM, et al. Prevalence and mechanisms of low- and chronic disease. J Infect Dis 1990; 162:695–9. and high-level mupirocin resistance in staphylococci isolated from a 30. Whitehead SM, Leach RD, Eykyn SJ, Phillips I. The aetiology of Korean hospital. J Antimicrob Chemother 2003; 51:619–23. perirectal sepsis. Br J Surg 1982; 69:166–8. 6. Committee on Infectious Diseases, American Academy of Pediatrics. 31. Diven DG, Dozier SE, Meyer DJ, Smith EB. Bacteriology of inflamed Antimicrobial agents and related therapy. In: Pickering LK, ed. Red and uninflamed epidermal inclusion cysts. Arch Dermatol 1998; 134: book 2003 report of the Committee on Infectious Diseases. 26th ed. 49–51. Elk Grove Village, IL: American Academy of Pediatrics, 2003:693–4. 32. Llera JL, Levy RC. Treatment of cutaneous abscess: a double-blind 7. Miller LG, Perdreau-Remington F, Rieg G, et al. Necrotizing fasciitis clinical study. Ann Emerg Med 1985; 14:15–9. caused by community-associated methicillin-resistant Staphylococcus 33. Macfie J, Harvey J. The treatment of acute superficial abscesses: a aureus in Los Angeles. N Engl J Med 2005; 352:1445–53. prospective clinical trial. Br J Surg 1977; 64:264–6. 8. Ferrieri P, Dajani AS, Wannamaker LW, Chapman SS. Natural history 34. Decker MD, Lybarger JA, Vaughn WK, Hutcheson RH Jr, Schaffner of impetigo. 1. Site sequence of acquisition and familial patterns of W. An outbreak of staphylococcal skin infections among river rafting spread of cutaneous streptococci. J Clin Invest 1972; 51:2851–62. guides. Am J Epidemiol 1986; 124:969–76. 9. Adams BB. Dermatologic disorders of the athlete. Sports Med 2002;32: 35. Sosin DM, Gunn RA, Ford WL, Skaggs JW. An outbreak of furun- 309–21. culosis among high school athletes. Am J Sports Med 1989; 17:828–32. 10. Fehrs LJ, Flanagan K, Kline S, Facklam RR, Quackenbush K, Foster 36. Zimakoff J, Rosdahl VT, Petersen W, Scheibel J. Recurrent staphy- LR. Group A beta-hemolytic streptococcal skin infections in a US lococcal furunculosis in families. Scand J Infect Dis 1988; 20:403–5. meat-packing plant. JAMA 1987; 258:3131–4. 37. Hedstrom SA. Recurrent staphylococcal furunculosis: bacteriological 11. Hirschmann JV. Impetigo: etiology and therapy. Curr Clin Top Infect findings and epidemiology in 100 cases. Scand J Infect Dis 1981; 13: Dis 2002; 22:42–51. 115–9. 12. Darmstadt GL, Lane AT. Impetigo: an overview. Pediatr Dermatol 38. Raz R, Miron D, Colodner R, Staler Z, Samara Z, Keness Y. A 1-year 1994; 11:293–303. trial of nasal mupirocin in the prevention of recurrent staphylococcal 13. Demidovich CW, Wittler RR, Ruff ME, Bass JW, Browning WC. Im- nasal colonization and skin infection. Arch Intern Med 1996; 156: petigo: current etiology and comparison of penicillin, erythromycin, 1109–12. and cephalexin therapies. Am J Dis Child 1990; 144:1313–5. 39. Lipsky BA, Pecoraro RE, Ahroni JH, Peugeot RL. Immediate and long- 14. Kaplan EL, Anthony BF, Chapman SS, Ayoub EM, Wannamaker LW. term efficacy of systemic antibiotics for eradicating nasal colonization The influence of the site of infection on the immune response to with Staphylococcus aureus. Eur J Clin Microbiol Infect Dis 1992; 11: group A streptococci. J Clin Invest 1970; 49:1405–14. 43–7. 15. Bisno AL, Nelson KE, Waytz P, Brunt J. Factors influencing serum 40. Klempner MS, Styrt B. Prevention of recurrent staphylococcal skin antibody response in streptococcal pyoderma. J Lab Clin Med 1973;81: infections with low-dose oral clindamycin therapy. JAMA 1988; 260: 410–20. 2682–5. 16. Kaplan EL, Wannamaker LW. Suppression of the anti–streptolysin O response by cholesterol and by lipid extracts of rabbit skin. J Exp 41. Bisno AL, Stevens DL. Streptococcal infections in skin and soft tissues. Med 1976; 144:754–67. N Engl J Med 1996; 334:240–5. 17. Derrick CW Jr, Dillon HC Jr. Impetigo contagiosa. Am Fam Physician 42. Chartier C, Grosshans E. Erysipelas. Int J Dermatol 1990; 29:459–67. 1971; 4:75–81. 43. Chartier C, Grosshans E. Erysipelas: an update. Int J Dermatol 18. Ferrieri P, Dajani AS, Wannamaker LW. A controlled study of peni- 1996; 35:779–81. cillin prophylaxis against streptococcal impetigo. J Infect Dis 1974; 44. Swartz MN. Clinical practice: cellulitis. N Engl J Med 2004; 350: 129:429–38. 904–12. 19. Dagan R, Bar-David Y. Comparison of amoxicillin and clavulanic acid 45. Bernard P, Plantin P, Roger H, et al. Roxithromycin versus penicillin (augmentin) for the treatment of nonbullous impetigo. Am J Dis in the treatment of erysipelas in adults: a comparative study. Br J Child 1989; 143:916–8. Dermatol 1992; 127:155–9. 20. Barton LL, Friedman AD. Impetigo: a reassessment of etiology and 46. Martin JM, Green M, Barbadora KA, Wald ER. Erythromycin-resistant therapy. Pediatr Dermatol 1987; 4:185–8. group A streptococci in schoolchildren in Pittsburgh. N Engl J Med 21. Barton LL, Friedman AD, Sharkey AM, Schneller DJ, Swierkosz EM. 2002; 346:1200–6. Impetigo contagiosa III: comparative efficacy of oral erythromycin 47. York MK, Gibbs L, Perdreau-Remington F, Brooks GF. Characteri- and topical mupirocin. Pediatr Dermatol 1989; 6:134–8. zation of antimicrobial resistance in Streptococcus pyogenes isolates 22. Britton JW, Fajardo JE, Krafte-Jacobs B. Comparison of mupirocin from the San Francisco Bay area of northern California. J Clin Mi- and erythromycin in the treatment of impetigo. J Pediatr 1990; 117: crobiol 1999; 37:1727–31. 827–9. 48. Dupuy A, Benchikhi H, Roujeau JC, et al. Risk factors for erysipelas 23. Weinstein L, Le Frock J. Does antimicrobial therapy of streptococcal of the leg (cellulitis): case-control study. BMJ 1999; 318:1591–4. pharyngitis or pyoderma alter the risk of glomerulonephritis? J Infect 49. Dan M, Heller K, Shapira I, Vidne B, Shibolet S. Incidence of ery- Dis 1971; 124:229–31. sipelas following venectomy for coronary artery bypass surgery. In- 24. Meislin HW, Lerner SA, Graves MH, et al. Cutaneous abscesses: an- fection 1987; 15:107–8.

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1401 50. Baddour LM, Bisno AL. Recurrent cellulitis after saphenous venec- days) treatment for uncomplicated cellulitis. Arch Intern Med tomy for coronary bypass surgery. Ann Intern Med 1982; 97:493–6. 2004; 164:1669–74. 51. Simon MS, Cody RL. Cellulitis after axillary lymph node dissection 75. Bergkvist PI, Sjobeck K. Antibiotic and prednisolone therapy of er- for carcinoma of the breast. Am J Med 1992; 93:543–8. ysipelas: a randomized, double blind, placebo-controlled study. Scand 52. Baddour LM. Breast cellulitis complicating breast conservation ther- J Infect Dis 1997; 29:377–82. apy. J Intern Med 1999; 245:5–9. 76. Bergkvist PI, Sjobeck K. Relapse of erysipelas following treatment with 53. Bouma J, Dankert J. Recurrent acute leg cellulitis in patients after prednisolone or placebo in addition to antibiotics: a 1-year follow- radical vulvectomy. Gynecol Oncol 1988; 29:50–7. up. Scand J Infect Dis 1998; 30:206–7. 54. Dankert J, Bouma J. Recurrent acute leg cellulitis after hysterectomy 77. Babb RR, Spittell JA Jr, Martin WJ, Schirger A. Prophylaxis of re- with pelvic lymphadenectomy. Br J Obstet Gynaecol 1987; 94:788–90. current lymphangitis complicating lymphedema. JAMA 1966; 195: 55. Perl B, Gottehrer NP, Raveh D, Schlesinger Y, Rudensky B, Yinnon 871–3. AM. Cost-effectiveness of blood cultures for adult patients with cel- 78. Kremer M, Zuckerman R, Avraham Z, Raz R. Long-term antimicrobial lulitis. Clin Infect Dis 1999; 29:1483–8. therapy in the prevention of recurrent soft-tissue infections. J Infect 56. Kielhofner MA, Brown B, Dall L. Influence of underlying disease 1991; 22:37–40. process on the utility of cellulitis needle aspirates. Arch Intern Med 79. Sjoblom AC, Eriksson B, Jorup-Ronstrom C, Karkkonen K, Lindqvist 1988; 148:2451–2. M. Antibiotic prophylaxis in recurrent erysipelas. Infection 1993; 21: 57. Hook EW III, Hooton TM, Horton CA, Coyle MB, Ramsey PG, Turck 390–3. M. Microbiologic evaluation of cutaneous cellulitis in adults. Arch 80. Wang JH, Liu YC, Cheng DL, et al. Role of benzathine penicillin G Intern Med 1986; 146:295–7. in prophylaxis for recurrent streptococcal cellulitis of the lower legs. 58. Sachs MK. The optimum use of needle aspiration in the bacteriologic Clin Infect Dis 1997; 25:685–9. diagnosis of cellulitis in adults. Arch Intern Med 1990; 150:1907–12. 81. Kasseroller R. Sodium selenite as prophylaxis against erysipelas in 59. Leppard BJ, Seal DV, Colman G, Hallas G. The value of bacteriology secondary lymphedema. Anticancer Res 1998; 18:2227–30. and serology in the diagnosis of cellulitis and erysipelas. Br J Dermatol 82. Groom AV, Wolsey DH, Naimi TS, et al. Community-acquired meth- 1985; 112:559–67. icillin-resistant Staphylococcus aureus in a rural American Indian com- 60. Sigurdsson AF, Gudmundsson S. The etiology of bacterial cellulitis munity. JAMA 2001; 286:1201–5. as determined by fine-needle aspiration. Scand J Infect Dis 1989; 21: 83. Herold BC, Immergluck LC, Maranan MC, et al. Community-ac- 537–42. quired methicillin-resistant Staphylococcus aureus in children with no 61. Newell PM, Norden CW. Value of needle aspiration in bacteriologic identified predisposing risk. JAMA 1998; 279:593–8. diagnosis of cellulitis in adults. J Clin Microbiol 1988; 26:401–4. 84. Centers for Disease Control and Prevention. Outbreaks of commu- nity-associated methicillin-resistant Staphylococcus aureus skin infec- 62. Lebre C, Girard-Pipau F, Roujeau JC, Revuz J, Saiag P, Chosidow O. tions—Los Angeles County, California, 2002–2003. MMWR Morb Value of fine-needle aspiration in infectious cellulitis. Arch Dermatol Mortal Wkly Rep 2003; 52:88. 1996; 132:842–3. 85. Ma XX, Ito T, Tiensasitorn C, et al. Novel type of staphylococcal 63. Lutomski DM, Trott AT, Runyon JM, Miyagawa CI, Staneck JL, Rivera cassette chromosome mec identified in community-acquired methi- JO. Microbiology of adult cellulitis. J Fam Pract 1988; 26:45–8. cillin-resistant Staphylococcus aureus strains. Antimicrob Agents Che- 64. Duvanel T, Auckenthaler R, Rohner P, Harms M, Saurat JH. Quan- mother 2002; 46:1147–52. titative cultures of biopsy specimens from cutaneous cellulitis. Arch 86. Okuma K, Iwakawa K, Turnidge JD, et al. Dissemination of new Intern Med 1989; 149:293–6. methicillin-resistant Staphylococcus aureus clones in the community. 65. Eriksson B, Jorup-Ronstrom C, Karkkonen K, Sjoblom AC, Holm SE. J Clin Microbiol 2002; 40:4289–94. Erysipelas: clinical and bacteriologic spectrum and serological aspects. 87. Dufour P, Gillet Y, Bes M, et al. Community-acquired methicillin- Clin Infect Dis 1996; 23:1091–8. resistant Staphylococcus aureus infections in France: emergence of a 66. Bernard P, Toty L, Mounier M, Denis F, Bonnetblanc JM. Early de- single clone that produces Panton-Valentine leukocidin. Clin Infect tection of streptococcal group antigens in skin samples by latex par- Dis 2002; 35:819–24. ticle agglutination. Arch Dermatol 1987; 123:468–70. 88. Methicillin-resistant Staphylococcus aureus infections in correctional 67. Bernard P, Bedane C, Mounier M, Denis F, Catanzano G, Bonnetblanc facilities—Georgia, California, and Texas, 2001–2003. MMWR Morb JM. Streptococcal cause of erysipelas and cellulitis in adults: a mi- Mortal Wkly Rep 2003; 52:992–6. crobiologic study using a direct immunofluorescence technique. Arch 89. Methicillin-resistant Staphylococcus aureus infections among compet- Dermatol 1989; 125:779–82. itive sports participants—Colorado, Indiana, Pennsylvania, and Los 68. Baddour LM, Bisno AL. Recurrent cellulitis after coronary bypass Angeles County, 2000–2003. MMWR Morb Mortal Wkly Rep 2003;52: surgery. Association with superficial fungal infection in saphenous 793–5. venectomy limbs. JAMA 1984; 251:1049–52. 90. Stevens DL, Smith LG, Bruss JB, et al. Randomized comparison of 69. Semel JD, Goldin H. Association of athlete’s foot with cellulitis of linezolid (PNU-100766) versus oxacillin-dicloxacillin for treatment of the lower extremities: diagnostic value of bacterial cultures of ipsi- complicated skin and soft tissue infections. Antimicrob Agents Che- lateral interdigital space samples. Clin Infect Dis 1996; 23:1162–4. mother 2000; 44:3408–13. 70. Eriksson BK. Anal colonization of group G b-hemolytic streptococci 91. Stevens DL, Herr D, Lampiris H, Hunt JL, Batts DH, Hafkin B. in relapsing erysipelas of the lower extremity. Clin Infect Dis 1999; Linezolid versus vancomycin for the treatment of methicillin-resistant 29:1319–20. Staphylococcus aureus infections. Linezolid MRSA Study Group. Clin 71. Burman WJ, Cohn DL, Reves RR, Wilson ML. Multifocal cellulitis Infect Dis 2002; 34:1481–90. and monoarticular arthritis as manifestations of Helicobacter cinaedi 92. Markowitz N, Quinn EL, Saravolatz LD. Trimethoprim-sulfameth- bacteremia. Clin Infect Dis 1995; 20:564–70. oxazole compared with vancomycin for the treatment of Staphylo- 72. Kirsner RS, Pardes JB, Eaglstein WH, Falanga V. The clinical spectrum coccus aureus infection. Ann Intern Med 1992; 117:390–8. of lipodermatosclerosis. J Am Acad Dermatol 1993; 28:623–7. 93. Ahrenholz DH. Necrotizing soft-tissue infections. Surg Clin North 73. Jorup-Ronstrom C, Britton S, Gavlevik A, Gunnarsson K, Redman Am 1988; 68:199–214. AC. The course, costs, and complications of oral versus intravenous 94. Lewis RT. Necrotizing soft-tissue infections. Infect Dis Clin North penicillin therapy of erysipelas. Infection 1984; 12:390–4. Am 1992; 6:693–703. 74. Hepburn MJ, Dooley DP, Skidmore PJ, Ellis MW, Starnes WF, Ha- 95. Rea WJ, Wyrick WJ Jr. Necrotizing fasciitis. Ann Surg 1970; 172: sewinkle WC. Comparison of short-course (5 days) and standard (10 957–64.

1402 • CID 2005:41 (15 November) • Stevens et al. 96. Giuliano A, Lewis F Jr, Hadley K, Blaisdell FW. Bacteriology of nec- 121. Inglesby TV, Henderson DA, Bartlett JG, et al. Anthrax as a biological rotizing fasciitis. Am J Surg 1977; 134:52–7. weapon: medical and public health management. Working Group on 97. Stevens DL, Tanner MH, Winship J, et al. Reappearance of scarlet Civilian Biodefense. JAMA 1999; 281:1735–45. fever toxin A among streptococci in the Rocky Mountain West: severe 122. Dixon TC, Meselson M, Guillemin J, Hanna PC. Anthrax. N Engl J group A streptococcal infections associated with a toxic shock-like Med 1999; 341:815–26. syndrome. N Engl J Med 1989; 321:1–7. 123. Update: investigation of bioterrorism-related anthrax and interim 98. Chelsom J, Halstensen A, Haga T, Hoiby EA. Necrotising fasciitis due guidelines for exposure management and antimicrobial therapy, Oc- to group A streptococci in western Norway: incidence and clinical tober 2001. MMWR Morb Mortal Wkly Rep 2001; 50:909–19. features. Lancet 1994; 344:1111–5. 124. Bass JW, Freitas BC, Freitas AD, et al. Prospective randomized double 99. Zimbelman J, Palmer A, Todd J. Improved outcome of clindamycin blind placebo-controlled evaluation of azithromycin for treatment of compared with beta-lactam antibiotic treatment for invasive Strep- cat-scratch disease. Pediatr Infect Dis J 1998; 17:447–52. tococcus pyogenes infection. Pediatr Infect Dis J 1999; 18:1096–100. 125. Reboli AC, Farrar WE. Erysipelothrix rhusiopathiae: an occupational 100. Mulla ZD, Leaverton PE, Wiersma ST. Invasive group A streptococcal pathogen. Clin Microbiol Rev 1989; 2:354–9. infections in Florida. South Med J 2003; 96:968–73. 126. Venditti M, Gelfusa V, Tarasi A, Brandimarte C, Serra P. Antimicrobial 101. Stevens DL. Dilemmas in the treatment of invasive Streptococcus py- susceptibilities of Erysipelothrix rhusiopathiae. Antimicrob Agents ogenes infections. Clin Infect Dis 2003; 37:341–3. Chemother 1990; 34:2038–40. 102. Kaul R, McGeer A, Norrby-Teglund A, et al. Intravenous immuno- 127. Srinivasan A, Kraus CN, DeShazer D, et al. Glanders in a military globulin therapy for streptococcal toxic shock syndrome: a compar- research microbiologist. N Engl J Med 2001; 345:256–8. ative observational study. Clin Infect Dis 1999; 28:800–7. 128. Perry RD, Fetherston JD. Yersinia pestis—etiologic agent of plague. 103. Darenberg J, Ihendyane N, Sjolin J, et al. Intravenous immunoglob- Clin Microbiol Rev 1997; 10:35–66. ulin G therapy in streptococcal toxic shock syndrome: a European 129. Inglesby TV, Dennis DT, Henderson DA, et al. Plague as a biological randomized, double-blind, placebo-controlled trial. Clin Infect Dis weapon: medical and public health management. Working Group on 2003; 37:333–40. Civilian Biodefense. JAMA 2000; 283:2281–90. 104. Sissolak D, Weir WR. Tropical pyomyositis. J Infect 1994; 29:121–7. 130. Enderlin G, Morales L, Jacobs RF, Cross JT. Streptomycin and alter- 105. Laucks SS. Fournier’s gangrene. Surg Clin North Am 1994; 74: native agents for the treatment of tularemia: review of the literature. 1339–52. Clin Infect Dis 1994; 19:42–7. 106. Eke N. Fournier’s gangrene: a review of 1726 cases. Br J Surg 2000; 131. Johansson A, Berglund L, Gothefors L, Sjostedt A, Tarnvik A. Cip- 87:718–28. rofloxacin for treatment of tularemia in children. Pediatr Infect Dis 107. Altemeier WA, Fullen WD. Prevention and treatment of gas gangrene. J 2000; 19:449–53. JAMA 1971; 217:806–13. 132. Chocarro A, Gonzalez A, Garcia I. Treatment of tularemia with cip- 108. Stevens DL, Laine BM, Mitten JE. Comparison of single and com- rofloxacin. Clin Infect Dis 2000; 31:623. bination antimicrobial agents for prevention of experimental gas gan- 133. Perez-Castrillon JL, Bachiller-Luque P, Martin-Luquero M, Mena- grene caused by Clostridium perfringens. Antimicrob Agents Che- Martin FJ, Herreros V. Tularemia epidemic in northwestern Spain: mother 1987; 31:312–6. clinical description and therapeutic response. Clin Infect Dis 2001; 109. Stevens DL, Maier KA, Laine BM, Mitten JE. Comparison of clin- 33:573–6. damycin, rifampin, tetracycline, metronidazole, and penicillin for ef- 134. Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events ficacy in prevention of experimental gas gangrene due to Clostridium and negligence in hospitalized patients: results of the Harvard Medical perfringens. J Infect Dis 1987; 155:220–8. Practice study I. N Engl J Med 1991; 324:370–6. 110. Stevens DL, Bryant AE, Adams K, Mader JT. Evaluation of hyperbaric 135. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guideline oxygen therapy for treatment of experimental Clostridium perfringens for prevention of surgical site infection, 1999: Hospital Infection Con- infection. Clin Infect Dis 1993; 17:231–7. trol Practices Advisory Committee. Infect Control Hosp Epidemiol 111. Talan DA, Citron DM, Abrahamian FM, Moran GJ, Goldstein EJ. 1999; 20:250–78. Bacteriologic analysis of infected dog and cat bites. Emergency Med- 136. Gaynes RP, Culver DH, Horan TC, Edwards JR, Richards C, Tolson icine Animal Bite Infection Study Group. N Engl J Med 1999; 340: JS. Surgical site infection (SSI) rates in the United States, 1992–1998: 85–92. 112. Goldstein EJ, Citron DM, Wield B, et al. Bacteriology of human and the National Nosocomial Infections Surveillance System basic SSI risk animal bite wounds. J Clin Microbiol 1978; 8:667–72. index. Clin Infect Dis 2001; 33(Suppl 2):S69–S77. 113. Goldstein EJ. New horizons in the bacteriology, antimicrobial sus- 137. Dellinger EP. Approach to the patient with postoperative fever. In: ceptibility and therapy of animal bite wounds. J Med Microbiol Gorbach SL, Bartlett JG, Blacklow NR, eds. Infectious diseases in 1998; 47:95–7. medicine and surgery. Philadelphia: W. B. Saunders, 1998:903–9. 114. Goldstein EJ, Reinhardt JF, Murray PM, Finegold SM. Outpatient 138. Burke JF. The effective period of preventive antibiotic action in ex- therapy of bite wounds: demographic data, bacteriology, and a pro- perimental incisions and dermal lesions. Surgery 1961; 50:161–8. spective, randomized trial of amoxicillin/clavulanic acid versus pen- 139. Classen DC, Evans RS, Pestotnik SL, Horn SD, Menlove RL, Burke icillin +/Ϫ dicloxacillin. Int J Dermatol 1987; 26:123–7. JP. The timing of prophylactic administration of antibiotics and the 115. Talan DA, Abrahamian FM, Moran GJ, Citron DM, Tan JO, Goldstein risk of surgical-wound infection. N Engl J Med 1992; 326:281–6. EJ. Clinical presentation and bacteriologic analysis of infected human 140. Stone HH, Haney BB, Kolb LD, Geheber CE, Hooper CA. Prophy- bites in patients presenting to emergency departments. Clin Infect lactic and preventive antibiotic therapy: timing, duration and eco- Dis 2003; 37:1481–9. nomics. Ann Surg 1979; 189:691–9. 116. Transmission of HIV by human bite. Lancet 1987; 2:522. 141. Dellinger EP, Gross PA, Barrett TL, et al. Quality standard for anti- 117. Vidmar L, Poljak M, Tomazic J, Seme K, Klavs I. Transmission of microbial prophylaxis in surgical procedures. Infectious Diseases So- HIV-1 by human bite. Lancet 1996; 347:1762. ciety of America. Clin Infect Dis 1994; 18:422–7. 118. Dusheiko GM, Smith M, Scheuer PJ. Hepatitis C virus transmitted 142. Bratzler DW, Houck PM. Antimicrobial prophylaxis for surgery: an by human bite. Lancet 1990; 336:503–4. advisory statement from the National Surgical Infection Prevention 119. Davis LG, Weber DJ, Lemon SM. Horizontal transmission of hepatitis Project. Clin Infect Dis 2004; 38:1706–15. B virus. Lancet 1989; 1:889–93. 143. McDonald M, Grabsch E, Marshall C, Forbes A. Single- versus mul- 120. Fiumara NJ, Exner JH. Primary following a human bite. Sex tiple-dose antimicrobial prophylaxis for major surgery: a systematic Transm Dis 1981; 8:21–2. review. Aust N Z J Surg 1998; 68:388–96.

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1403 144. Bartlett P, Reingold AL, Graham DR, et al. Toxic shock syndrome to viridans streptococci: a case-control study of predisposing factors. associated with surgical wound infections. JAMA 1982; 247:1448–50. Clin Infect Dis 1992; 14:1201–7. 145. Raab MG, O’Brien M, Hayes JM, Graham DR. Postoperative toxic 171. European Organization for Research and Treatment of Cancer shock syndrome. Am J Orthop 1995; 24:130–6. (EORTC) International Antimicrobial Therapy Cooperative Group 146. Huizinga WK, Kritzinger NA, Bhamjee A. The value of adjuvant and the National Cancer Institute of Canada-Clinical Trials Group. systemic antibiotic therapy in localised wound infections among hos- Vancomycin added to empirical combination antibiotic therapy for pital patients: a comparative study. J Infect 1986; 13:11–6. fever in granulocytopenic cancer patients. J Infect Dis 1991; 163: 147. Mandell GL, Bennett JE, Dolin R. Mandell, Douglas and Bennett’s 951–8. principles and practice of infectious diseases. New York: Churchill 172. Edmond MB, Ober JF, Dawson JD, Weinbaum DL, Wenzel RP. Van- Livingstone, 1995. comycin-resistant enterococcal bacteremia: natural history and at- 148. Howard RJ. Surgical infectious diseases. Norwalk: Appleton & Lange, tributable mortality. Clin Infect Dis 1996; 23:1234–9. 1988. 173. Martin MA, Pfaller MA, Wenzel RP. Coagulase-negative staphylo- 149. Fry DE. Surgical infections. Boston: Little, Brown and Company, 1995. coccal bacteremia: mortality and hospital stay. Ann Intern Med 150. Greenfield LJ, Mulholland MW, Oldham KT, Zelenock GB. Surgery: 1989; 110:9–16. scientific principles and practice. Philadelphia: J. B. Lippincott Com- 174. Wenzel RP. Perspective: attributable mortality—the promise of better pany, 1993. antimicrobial therapy. J Infect Dis 1998; 178:917–9. 151. Townsend CM Jr, Beauchamp RD, Evers BM, Mattox KL, Sabiston 175. Brown AE, Kiehn TE, Armstrong D. Bacterial resistance in the patient S. Textbook of surgery: the biologic basis of modern surgical practice. with neoplastic disease. Infect Dis Clin Pract 1995; 4(Suppl 3): Philadelphia: W. B. Saunders and Company, 2001. S136–44. 152. Wilmore DW, Cheung LY, Harken AH, Holcroft JW, Meakins JL, 176. Rubinstein E, Cammarata S, Oliphant T, Wunderink R. Linezolid Soper NJ. ACS surgery: principles and practice. New York: WebMD, (PNU-100766) versus vancomycin in the treatment of hospitalized 2002. patients with nosocomial pneumonia: a randomized, double-blind, 153. Gorbach SL, Bartlett JG, Blacklow NR. Infectious diseases. Philadel- multicenter study. Clin Infect Dis 2001; 32:402–12. phia: W. B. Saunders and Company, 1998. 177. Ozer H, Armitage JO, Bennett CL, et al. 2000 Update of recommen- 154. Dellinger EP. Nosocomial infection. In: Wilmore DW, Cheung LY, dations for the use of hematopoietic colony-stimulating factors: evi- Harken AH, Holcroft JW, Meakins JL, Soper NJ, eds. ACS surgery: dence-based, clinical practice guidelines. American Society of Clinical principles and practice. New York: WebMD, 2002:1221–38. Oncology Growth Factors Expert Panel. J Clin Oncol 2000; 18: 155. Cruse PJE. Wound infections: epidemiology and clinical character- 3558–85. istics. In: Howard RJ, Simmons RL, eds. Surgical infectious diseases. 178. Cometta A, Kern WV, De Bock R, et al. Vancomycin versus placebo Norwalk: Appleton & Lange, 1988:319–29. for treating persistent fever in patients with neutropenic cancer re- 156. Bobrow BJ, Pollack CV Jr, Gamble S, Seligson RA. Incision and drain- ceiving piperacillin-tazobactam monotherapy. Clin Infect Dis 2003; age of cutaneous abscesses is not associated with bacteremia in afebrile 37:382–9. adults. Ann Emerg Med 1997; 29:404–8. 179. Pizzo PA, Robichaud KJ, Gill FA, Witebsky FG. Empiric antibiotic 157. Meislin HW. Pathogen identification of abscesses and cellulitis. Ann and antifungal therapy for cancer patients with prolonged fever and Emerg Med 1986; 15:329–32. granulocytopenia. Am J Med 1982; 72:101–11. 158. Dellinger EP. Postoperative wound infection. In: Schlossberg D, ed. 180. EORTC International Antimicrobial Therapy Cooperative Group. Current therapy of infectious disease. St. Louis: Mosby, 1996:337–9. Empiric antifungal therapy in febrile granulocytopenic patients. Am 159. Wade JC. Management of infection in patients with acute leukemia. J Med 1989; 86:668–72. Hematol Oncol Clin North Am 1993; 7:293–315. 181. Walsh TJ, Teppler H, Donowitz GR, et al. Caspofungin versus lipo- 160. Pizzo PA. Management of fever in patients with cancer and treatment- somal amphotericin B for empirical antifungal therapy in patients induced neutropenia. N Engl J Med 1993; 328:1323–32. with persistent fever and neutropenia. N Engl J Med 2004; 351: 161. Donowitz GR. Fever in the compromised host. Infect Dis Clin North 1391–402. Am 1996; 10:129–48. 182. Walsh TJ, Pappas P, Winston DJ, et al. Voriconazole compared with 162. Wolfson JS, Sober AJ, Rubin RH. Dermatologic manifestations of liposomal amphotericin B for empirical antifungal therapy in patients infections in immunocompromised patients. Medicine (Baltimore) with neutropenia and persistent fever. N Engl J Med 2002; 346:225–34. 1985; 64:115–33. 183. Goodrich JM, Reed EC, Mori M, et al. Clinical features and analysis 163. Wolfson JS, Sober AJ, Rubin RH. Dermatologic manifestations of of risk factors for invasive candidal infection after marrow trans- infection in the compromised host. Annu Rev Med 1983; 34:205–17. plantation. J Infect Dis 1991; 164:731–40. 164. Hughes WT, Armstrong D, Bodey GP, et al. 2002 Guidelines for the 184. Rex JH, Walsh TJ, Sobel JD, et al. Practice guidelines for the treatment use of antimicrobial agents in neutropenic patients with cancer. Clin of candidiasis. Infectious Diseases Society of America. Clin Infect Dis Infect Dis 2002; 34:730–51. 2000; 30:662–78. 165. NCCN practice guidelines for fever and neutropenia. National Com- 185. Wingard JR, Merz WG, Saral R. Candida tropicalis: a major pathogen prehensive Cancer Network. Oncology (Williston Park) 1999; 13: in immunocompromised patients. Ann Intern Med 1979; 91:539–43. 197–257. 186. Jarowski CI, Fialk MA, Murray HW, et al. Fever, rash, and muscle 166. Lopez FA, Sanders CV. Dermatologic infections in the immunocom- tenderness: a distinctive clinical presentation of disseminated candi- promised (non-HIV) host. Infect Dis Clin North Am 2001; 15: diasis. Arch Intern Med 1978; 138:544–6. 671–702, xi. 187. Walsh TJ, Newman KR, Moody M, Wharton RC, Wade JC. Trichos- 167. Wingard JR, Santos GW, Saral R. Differences between first and sub- poronosis in patients with neoplastic disease. Medicine (Baltimore) sequent fevers during prolonged neutropenia. Cancer 1987; 59:844–9. 1986; 65:268–79. 168. Chatzinikolaou I, Abi-Said D, Bodey GP, Rolston KV, Tarrand JJ, 188. Patterson TF, Kirkpatrick WR, White M, et al. Invasive aspergillosis: Samonis G. Recent experience with Pseudomonas aeruginosa bacter- disease spectrum, treatment practices, and outcomes. I3 Aspergillus emia in patients with cancer: retrospective analysis of 245 episodes. Study Group. Medicine (Baltimore) 2000; 79:250–60. Arch Intern Med 2000; 160:501–9. 189. Ascioglu S, Rex JH, de Pauw B, et al. Defining opportunistic invasive 169. Glauser M. Empiric therapy of bacterial infections in patients with fungal infections in immunocompromised patients with cancer and severe neutropenia. Diagn Microbiol Infect Dis 1998; 31:467–72. hematopoietic transplants: an international consensus. Clin 170. Elting LS, Bodey GP, Keefe BH. Septicemia and shock syndrome due Infect Dis 2002; 34:7–14.

1404 • CID 2005:41 (15 November) • Stevens et al. 190. Lin SJ, Schranz J, Teutsch SM. Aspergillosis case-fatality rate: system- 212. Wallace RJ Jr, Brown BA, Onyi GO. Skin, soft tissue, and bone in- atic review of the literature. Clin Infect Dis 2001; 32:358–66. fections due to Mycobacterium chelonae chelonae: importance of prior 191. Kontoyiannis DP, Sumoza D, Tarrand J, Bodey GP, Storey R, Raad corticosteroid therapy, frequency of disseminated infections, and re- II. Significance of aspergillemia in patients with cancer: a 10-year sistance to oral antimicrobials other than clarithromycin. J Infect Dis study. Clin Infect Dis 2000; 31:188–9. 1992; 166:405–12. 192. Allo MD, Miller J, Townsend T, Tan C. Primary cutaneous aspergillosis 213. Bennett C, Vardiman J, Golomb H. Disseminated atypical mycobac- associated with Hickman intravenous catheters. N Engl J Med terial infection in patients with hairy cell leukemia. Am J Med 1986;80: 1987; 317:1105–8. 891–6. 193. Walmsley S, Devi S, King S, Schneider R, Richardson S, Ford-Jones 214. Patel R, Roberts GD, Keating MR, Paya CV. Infections due to non- L. Invasive Aspergillus infections in a pediatric hospital: a ten-year tuberculous mycobacteria in kidney, heart, and liver transplant re- review. Pediatr Infect Dis J 1993; 12:673–82. cipients. Clin Infect Dis 1994; 19:263–73. 194. Gartenberg G, Bottone EJ, Keusch GT, Weitzman I. Hospital-acquired 215. Gaviria JM, Garcia PJ, Garrido SM, Corey L, Boeckh M. Nontuber- mucormycosis (Rhizopus rhizopodiformis) of skin and subcutaneous culous mycobacterial infections in hematopoietic stem cell transplant tissue: epidemiology, mycology and treatment. N Engl J Med 1978; recipients: characteristics of respiratory and catheter-related infec- 299:1115–8. tions. Biol Blood Marrow Transplant 2000; 6:361–9. 195. Dennis JE, Rhodes KH, Cooney DR, Roberts GD. Nosocomical Rhi- 216. Berkey P, Bodey GP. Nocardial infection in patients with neoplastic zopus infection (zygomycosis) in children. J Pediatr 1980; 96:824–8. disease. Rev Infect Dis 1989; 11:407–12. 196. Anaissie E. Opportunistic mycoses in the immunocompromised host: 217. Simpson GL, Stinson EB, Egger MJ, Remington JS. Nocardial infec- experience at a cancer center and review. Clin Infect Dis 1992; tions in the immunocompromised host: a detailed study in a defined 14(Suppl 1):S43–53. population. Rev Infect Dis 1981; 3:492–507. 197. Krcmery V Jr, Jesenska Z, Spanik S, et al. Fungaemia due to Fusarium 218. Smego RA Jr, Moeller MB, Gallis HA. Trimethoprim-sulfamethoxa- spp. in cancer patients. J Hosp Infect 1997; 36:223–8. zole therapy for Nocardia infections. Arch Intern Med 1983; 143: 198. Boutati EI, Anaissie EJ. Fusarium, a significant emerging pathogen in 711–8. patients with hematologic malignancy: ten years’ experience at a can- 219. Dimino-Emme L, Gurevitch AW. Cutaneous manifestations of dis- cer center and implications for management. Blood 1997; 90: seminated cryptococcosis. J Am Acad Dermatol 1995; 32:844–50. 999–1008. 220. Neuville S, Dromer F, Morin O, Dupont B, Ronin O, Lortholary O. 199. Sheehan DJ, Hitchcock CA, Sibley CM. Current and emerging azole Primary cutaneous cryptococcosis: a distinct clinical entity. Clin Infect antifungal agents. Clin Microbiol Rev 1999; 12:40–79. Dis 2003; 36:337–47. 200. Wald A, Leisenring W, van Burik JA, Bowden RA. Epidemiology of 221. Anderson DJ, Schmidt C, Goodman J, Pomeroy C. Cryptococcal dis- Aspergillus infections in a large cohort of patients undergoing bone ease presenting as cellulitis. Clin Infect Dis 1992; 14:666–72. marrow transplantation. J Infect Dis 1997; 175:1459–66. 222. Saag MS, Graybill RJ, Larsen RA, et al. Practice guidelines for the 201. Viscoli C, Girmenia C, Marinus A, et al. Candidemia in cancer pa- management of cryptococcal disease. Infectious Diseases Society of tients: a prospective, multicenter surveillance study by the Invasive America. Clin Infect Dis 2000; 30:710–8. Fungal Infection Group (IFIG) of the European Organization for 223. Shuttleworth D, Philpot CM, Knight AG. Cutaneous cryptococcosis: Research and Treatment of Cancer (EORTC). Clin Infect Dis 1999; treatment with oral fluconazole. Br J Dermatol 1989; 120:683–7. 28:1071–9. 224. Davies SF, Sarosi GA, Peterson PK, et al. Disseminated histoplasmosis 202. Stevens DA, Kan VL, Judson MA, et al. Practice guidelines for diseases in renal transplant recipients. Am J Surg 1979; 137:686–91. caused by Aspergillus. Infectious Diseases Society of America. Clin 225. Wheat J, Sarosi G, McKinsey D, et al. Practice guidelines for the Infect Dis 2000; 30:696–709. management of patients with histoplasmosis. Infectious Diseases So- 203. Deresinski SC, Stevens DA. Caspofungin. Clin Infect Dis 2003; 36: ciety of America. Clin Infect Dis 2000; 30:688–95. 1445–57. 226. Locksley RM, Flournoy N, Sullivan KM, Meyers JD. Infection with 204. Wingard JR, Kubilis P, Lee L, et al. Clinical significance of nephro- varicella-zoster virus after marrow transplantation. J Infect Dis toxicity in patients treated with amphotericin B for suspected or 1985; 152:1172–81. proven aspergillosis. Clin Infect Dis 1999; 29:1402–7. 227. Arvin AM, Pollard RB, Rasmussen LE, Merigan TC. Cellular and 205. Walsh TJ, Hiemenz JW, Seibel NL, et al. Amphotericin B lipid complex humoral immunity in the pathogenesis of recurrent herpes viral in- for invasive fungal infections: analysis of safety and efficacy in 556 fections in patients with lymphoma. J Clin Invest 1980; 65:869–78. cases. Clin Infect Dis 1998; 26:1383–96. 228. Balfour HH Jr, Bean B, Laskin OL, et al. Acyclovir halts progression 206. Walsh TJ, Finberg RW, Arndt C, et al. Liposomal amphotericin B for of herpes zoster in immunocompromised patients. N Engl J Med empirical therapy in patients with persistent fever and neutropenia. 1983; 308:1448–53. National Institute of Allergy and Infectious Diseases Mycoses Study 229. Meyers JD, Flournoy N, Thomas ED. Infection with herpes simplex Group. N Engl J Med 1999; 340:764–71. virus and cell-mediated immunity after marrow transplant. J Infect 207. Herbrecht R, Denning DW, Patterson TF, et al. Voriconazole versus Dis 1980; 142:338–46. amphotericin B for primary therapy of invasive aspergillosis. N Engl 230. Tyring S, Barbarash RA, Nahlik JE, et al. Famciclovir for the treatment J Med 2002; 347:408–15. of acute herpes zoster: effects on acute disease and postherpetic neu- 208. Cagnoni PJ, Walsh TJ, Prendergast MM, et al. Pharmacoeconomic ralgia. A randomized, double-blind, placebo-controlled trial. Collab- analysis of liposomal amphotericin B versus conventional amphoter- orative Famciclovir Herpes Zoster Study Group. Ann Intern Med icin B in the empirical treatment of persistently febrile neutropenic 1995; 123:89–96. patients. J Clin Oncol 2000; 18:2476–83. 231. Cowan FM, Johnson AM, Ashley R, Corey L, Mindel A. Relationship 209. Marr KA, Seidel K, White TC, Bowden RA. Candidemia in allogeneic between antibodies to herpes simplex virus (HSV) and symptoms of blood and marrow transplant recipients: evolution of risk factors after HSV infection. J Infect Dis 1996; 174:470–5. the adoption of prophylactic fluconazole. J Infect Dis 2000; 181: 232. Johnson RE, Nahmias AJ, Magder LS, Lee FK, Brooks CA, Snowden 309–16. CB. A seroepidemiologic survey of the prevalence of herpes simplex 210. Ichiki Y, Hirose M, Akiyama T, Esaki C, Kitajima Y. Skin infection virus type 2 infection in the United States. N Engl J Med 1989; 321: caused by Mycobacterium avium. Br J Dermatol 1997; 136:260–3. 7–12. 211. Sanderson TL, Moskowitz L, Hensley GT, Cleary TJ, Penneys N. Dis- 233. Fleming DT, McQuillan GM, Johnson RE, et al. Herpes simplex virus seminated Mycobacterium avium-intracellulare infection appearing as type 2 in the United States, 1976 to 1994. N Engl J Med 1997; 337: a panniculitis. Arch Pathol Lab Med 1982; 106:112–4. 1105–11.

Guidelines for Skin and Soft-Tissue Infections • CID 2005:41 (15 November) • 1405 234. Balfour HH Jr, Benson C, Braun J, et al. Management of acyclovir- 236. Mermel LA, Farr BM, Sherertz RJ, et al. Guidelines for the manage- resistant herpes simplex and varicella-zoster virus infections. J Acquir ment of intravascular catheter-related infections. Clin Infect Dis Immune Defic Syndr 1994; 7:254–60. 2001; 32:1249–72. 235. Toome BK, Bowers KE, Scott GA. Diagnosis of cutaneous cytomeg- alovirus infection: a review and report of a case. J Am Acad Dermatol 1991; 24:860–7.

Note added in proof. Since this article was accepted for publication, the Food and Drug Administration has approved dalbavancin (Seltzer E, Dorr MB, Goldstein BP, et al. Once-weekly dalbavancin versus standard-of-care antimicrobial regimens for treatment of skin and soft-tissue infections. Clin Infect Dis 2003;37:1298–303) and tigecycline (Ellis-Grosse EJ, Babinchak T, Dartois N, et al. The efficacy and safety of tigecycline in the treatment of skin and skin-structure infections: results of 2 double-blind phase 3 comparison studies with vancomycin-aztreonam. Clin Infect Dis 2005;41[Suppl 5]:S341–53) for treatment of skin and soft-tissue infections, including those caused by methicillin-resistant Staphylococcus aureus. Dalbavancin was compared with the standard-of-care regimen, and cure rates and adverse effects were similar between study groups. Tigecycline was compared with vancomycin-aztreonam, and outcomes were similar between study groups. Interestingly, the incidence of nausea and vomiting was higher among patients in the tigecycline arm, and transaminase levels were higher in the vancomycin-aztreonam arm.

1406 • CID 2005:41 (15 November) • Stevens et al.

The new england journal of medicine

original article

Methicillin-Resistant S. aureus Infections among Patients in the Emergency Department

Gregory J. Moran, M.D., Anusha Krishnadasan, Ph.D., Rachel J. Gorwitz, M.D., M.P.H., Gregory E. Fosheim, M.P.H., Linda K. McDougal, M.S., Roberta B. Carey, Ph.D., and David A. Talan, M.D., for the EMERGEncy ID Net Study Group*

Abstract

Background From the Department of Emergency Med- Methicillin-resistant Staphylococcus aureus (MRSA) is increasingly recognized in infec- icine (G.J.M., A.K., D.A.T.) and the Division tions among persons in the community without established risk factors for MRSA. of Infectious Diseases (G.J.M., D.A.T.), Olive View–UCLA Medical Center, Sylmar, Calif.; and the Division of Healthcare Methods Quality Promotion, National Center for We enrolled adult patients with acute, purulent skin and soft-tissue infections pre- Infectious Diseases, Centers for Disease Control and Prevention, Atlanta (R.J.G., senting to 11 university-affiliated emergency departments during the month of G.E.F., L.K.M., R.B.C.). Address reprint re- August 2004. Cultures were obtained, and clinical information was collected. Available quests to Dr. Moran at the Department of S. aureus isolates were characterized by antimicrobial-susceptibility testing, pulsed-field Emergency Medicine, Olive View–UCLA Medical Center, 14445 Olive View Dr., gel electrophoresis, and detection of toxin genes. On MRSA isolates, we performed North Annex, Sylmar, CA 91342, or at typing of the staphylococcal cassette chromosome mec (SCCmec), the genetic element [email protected]. that carries the mecA gene encoding methicillin resistance. *Members of the EMERGEncy ID Net Study Group are listed in the Appendix. Results S. aureus was isolated from 320 of 422 patients with skin and soft-tissue infections N Engl J Med 2006;355:666-74. Copyright © 2006 Massachusetts Medical Society. (76 percent). The prevalence of MRSA was 59 percent overall and ranged from 15 to 74 percent. Pulsed-field type USA300 isolates accounted for 97 percent of MRSA iso- lates; 74 percent of these were a single strain (USA300-0114). SCCmec type IV and the Panton–Valentine leukocidin toxin gene were detected in 98 percent of MRSA isolates. Other toxin genes were detected rarely. Among the MRSA isolates, 95 per- cent were susceptible to clindamycin, 6 percent to erythromycin, 60 percent to fluo- roquinolones, 100 percent to rifampin and trimethoprim–sulfamethoxazole, and 92 percent to tetracycline. Antibiotic therapy was not concordant with the results of susceptibility testing in 100 of 175 patients with MRSA infection who received anti- biotics (57 percent). Among methicillin-susceptible S. aureus isolates, 31 percent were USA300 and 42 percent contained pvl genes.

Conclusions MRSA is the most common identifiable cause of skin and soft-tissue infections among patients presenting to emergency departments in 11 U.S. cities. When antimicrobial therapy is indicated for the treatment of skin and soft-tissue infections, clinicians should consider obtaining cultures and modifying empirical therapy to provide MRSA coverage.

666 n engl j med 355;7 www.nejm.org august 17, 2006

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. mrsa infections among patients in the emergency department

ethicillin-resistant staphylococ- geles; Minneapolis; New Orleans; New York; Phila- cus aureus (MRSA) emerged in the 1960s delphia; Phoenix, Ariz.; and Portland, Oreg. These as a cause of infection among patients departments had a combined approximate total of M 1 14 exposed to the bacteria in health care settings. 900,000 visits per year. The study was approved by More recently, MRSA infections have been report- the institutional review board at each site. ed among persons without such exposure (com- Patients 18 years of age or older presenting in munity-associated MRSA).2,3 Community-associ- August 2004 with purulent skin and soft-tissue ated outbreaks of MRSA infection have occurred infections of less than one week’s duration (ex- among prisoners, intravenous-drug users, athletes, cluding perirectal abscesses) were enrolled in the military trainees, and men who have sex with study. Consent was obtained in writing at eight men.4-6 Community-associated MRSA has primar- sites and orally with provision of an information ily been described as a cause of skin and soft-tissue sheet at three sites. Information on demographic infections, but it has also been associated with sep- characteristics, clinical presentation, potential risk sis and necrotizing pneumonia.7-9 As compared factors for MRSA infection, and treatments pro- with health care–associated MRSA isolates, com- vided was collected by emergency department munity-associated MRSA isolates tend to be re- physicians using standardized forms. Manage- sistant to fewer antibiotics, to produce different ment decisions were made on an individual basis toxins,10 and to have different types of the gene by physicians in the emergency department. Fol- complex known as staphylococcal cassette chro- low-up data were obtained by telephone approxi- mosome mec (SCCmec); this complex contains the mately two to three weeks after enrollment. mecA gene that confers methicillin resistance.10 Specimens were obtained from the single larg- Pulsed-field gel electrophoresis (PFGE) and other est area of infection with the use of sterile Dacron methods have identified a small number of molec- swabs and were processed and cultured at hospital ular types that have accounted for most commu- laboratories according to standard techniques.15 nity-associated MRSA isolates characterized in the Each laboratory determined the antimicrobial sus- United States.11 ceptibility of S. aureus isolates to the panel of Some institutions have a high prevalence of agents routinely tested at that laboratory.16 Avail- MRSA isolated from patients with sporadic skin able S. aureus isolates were forwarded to the Cen- and soft-tissue infections that are not associated ters for Disease Control and Prevention (CDC) for with an outbreak.12,13 However, data are limited further characterization. The inducibility of clinda- regarding the prevalence of MRSA as a cause of mycin resistance was determined by the D-zone skin and soft-tissue infections among patients in disk-diffusion test.17 The presence of genes for several communities throughout the United States staphylococcal enterotoxins A through E and H, and the S. aureus isolates associated with these in- toxic shock syndrome toxin 1 (TSST-1), and Pan- fections. Therefore, we determined the prevalence ton–Valentine leukocidin (pvl) and the type of of MRSA as a cause of skin infections among adult SCCmec were determined by the polymerase chain patients presenting to emergency departments in reaction. All isolates were typed by PFGE with several geographically diverse, metropolitan areas the use of SmaI restriction endonuclease. Addi- in the United States. We also determined the bac- tional methods are described in detail in the Sup- teriologic characteristics of S. aureus isolated from plementary Appendix (available with the full text skin and soft-tissue infections and evaluated fac- of this article at www.nejm.org). tors potentially associated with MRSA infections of Descriptive statistics were used to summarize skin and soft tissue. the characteristics of the patients and the preva- lence of MRSA. To identify potential risk factors Methods for MRSA infection among patients with skin and soft-tissue infections, we calculated adjusted odds We conducted a prospective prevalence study in- ratios and 95 percent confidence intervals. Vari- volving adult patients with skin and soft-tissue infec- ables associated with MRSA infection in bivariate tions who presented to hospitals in the EMERGEncy analyses were explored further with the use of ID Net, a network of university-affiliated emer- multivariate logistic regression. gency departments in 11 U.S. cities: Albuquerque; Audits of emergency department and laboratory Atlanta; Charlotte, N.C.; Kansas City, Mo.; Los An- logs for patients with a discharge diagnosis of ab-

n engl j med 355;7 www.nejm.org august 17, 2006 667

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

scess, cellulitis, or wound infection were conduct- The prevalence of MRSA ranged from 15 to 74 ed at all study sites (except Atlanta) to determine percent, and MRSA was the most common identi- the proportion of patients meeting eligibility cri- fiable cause of skin and soft-tissue infections in teria who were enrolled in the study. Demograph- 10 of 11 emergency departments. MRSA was iso- ic and clinical characteristics of enrolled patients lated from 61 percent of abscesses, 53 percent of were compared with those of unenrolled patients. purulent wounds, and 47 percent of cases of cel- lulitis with purulent exudate. Other organisms Results isolated from 1 percent or more of infections included 71 isolates of methicillin-susceptible A total of 422 patients with skin and soft-tissue S. aureus (MSSA) (17 percent); 30 isolates of strep- infections were enrolled. The median age was 39 tococcus species (7 percent) including 6 group B years (range, 18 to 79; interquartile range, 28 to streptococcus, 2 group A streptococcus, 3 non– 47), and 62 percent were men. Race or ethnic group group A and non–group B β-hemolytic streptococ- was determined by the clinicians: 49 percent of pa- cus, 4 anaerobic or microaerophilic streptococcus, tients were non-Hispanic blacks, 25 percent were and 15 viridans group streptococcus; 12 isolates non-Hispanic whites, 22 percent were Hispanic, of coagulase-negative staphylococci (3 percent); and 4 percent belonged to other groups. Infections and 6 isolates of Proteus mirabilis (1 percent). Cul- were located on the upper extremities in 29 percent tures from 31 patients were polymicrobial; 10 of of patients, lower extremities in 27 percent, torso these patients had MRSA. No microorganism was in 17 percent, perineum in 14 percent, and head isolated from 38 patients (9 percent). and neck in 13 percent. Infections were classified A total of 218 MRSA isolates (88 percent) and as an abscess in 81 percent of patients, an infected 55 MSSA isolates (77 percent) from 10 emergency wound in 11 percent, and as cellulitis with purulent departments were sent to the CDC for genetic and exudate in 8 percent. phenotypic characterization. The pulsed-field types S. aureus was isolated from skin and soft-tissue of 216 of the MRSA isolates (99 percent) tested infections in 320 patients (76 percent); 249 of the were characterist ic of communit y-associated M R SA: S. aureus isolates (78 percent) were MRSA. MRSA 212 were type USA300, 2 were type USA400, and was isolated from 59 percent of patients (Table 1). 2 were type USA1000.11,18 Of 212 MRSA isolates

Table 1. Bacterial Isolates from Purulent Skin and Soft-Tissue Infections in 11 U.S. Emergency Departments.*

No. of Patients Enrolled MRSA MSSA Other Bacteria No Bacterial Growth Site (N = 422) (N = 249)† (N = 71) (N = 64)‡ (N = 38) number (percent) Albuquerque 42 25 (60) 10 (24) 3 (7) 4 (10) Atlanta 32 23 (72) 4 (12) 3 (9) 2 (6) Charlotte, N.C. 25 17 (68) 0 4 (16) 4 (16) Kansas City, Mo. 58 43 (74) 6 (10) 4 (7) 5 (9) Los Angeles 47 24 (51) 6 (13) 8 (17) 9 (19) Minneapolis 28 11 (39) 4 (14) 9 (32) 4 (14) New Orleans 69 46 (67) 11 (16) 9 (13) 3 (4) New York 20 3 (15) 8 (40) 5 (25) 4 (20) Philadelphia 58 32 (55) 12 (21) 12 (21) 2 (3) Phoenix, Ariz. 30 18 (60) 8 (27) 4 (13) 0 Portland, Oreg. 13 7 (54) 2 (15) 3 (23) 1 (8)

* A total of 31 cultures, including 10 cultures from which MRSA was isolated, were polymicrobial. Because of rounding, percentages may not total 100. MSSA denotes methicillin-resistant Staphylococcus aureus. † P<0.001 for the test for homogeneity of MRSA prevalence across sites. ‡ Other bacteria isolated were as follows: MSSA (17 percent), streptococcus species (7 percent), coagulase-negative staphylococci (3 percent), and Proteus mirabilis (1 percent).

668 n engl j med 355;7 www.nejm.org august 17, 2006

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. mrsa infections among patients in the emergency department characterized as type USA300, 156 (74 percent) teria (Table 2) included antibiotic use in the month had a single pulsed-field pattern (strain USA300- before enrollment, the presence of an abscess or 0114). SCCmec type IV, characteristic of commu- a lesion attributed to a spider bite at enrollment, nity-associated MRSA,19 was found in 214 (98 history of MRSA infection, and a recent history of percent) of the MRSA isolates, and pvl toxin genes close contact with someone with a similar skin were present in 213 (98 percent). Genes for staph- infection. The presence of an underlying illness ylococcal enterotoxins A, B, C, D, E, and H and and characterization as belonging to the “other” TSST-1 were identified in five or fewer MRSA iso- category of race or ethnic background were nega- lates. Eight S. aureus isolates collected at the At- tively associated with the isolation of MRSA. In lanta site in April 2005 were similar to other study multivariate logistic-regression analyses, all these isolates with regard to PFGE and toxin character- factors were associated with MRSA infection, with istics. the exception of the presence of an abscess (Table USA300 was also the most common pulsed- 3). Black race was independently associated with field type among MSSA isolates, accounting for 17 MRSA infection. Controlling for study site did not of 55 MSSA isolates (31 percent) sent to the CDC. affect the association between any of these factors Among these type USA300 isolates, 8 (47 percent) and MRSA infection. Among 64 patients with had a pulsed-field pattern closely related to that of none of these factors, 31 (48 percent) were infected the MRSA strain USA300-0114. In addition, pvl with MRSA. The only factor that was significantly genes were detected in 23 MSSA isolates (42 per- associated with isolation of MRSA, as compared cent), including 17 (100 percent) of the isolates with MSSA, was the presence of abscess at enroll- characterized as USA300. ment (odds ratio, 2.3; 95 percent confidence inter- MRSA susceptibilities were as follows: 100 val, 1.2 to 4.4). percent were susceptible to trimethoprim–sulfa- Complete information about treatment was methoxazole (217 of 217) and to rifampin (186 of available for 406 of the 422 patients (96 percent). 186); 95 percent were susceptible to clindamycin Of these, 79 (19 percent) were treated with incision (215 of 226), 92 percent to tetracycline (207 of and drainage alone, 39 (10 percent) received anti- 226), 60 percent to fluoroquinolones (106 of 176), biotics alone, 267 (66 percent) were treated with and 6 percent to erythromycin (13 of 226). Al- both incision and drainage and antibiotics, and 21 though the proportion of all S. aureus isolates (5 percent) neither underwent incision and drain- (MRSA and MSSA) that were resistant to clindamy- age nor received antibiotics. Of 400 patients for cin was less than 15 percent at 10 of the study whom information about the outcome was avail- sites, 6 of 10 S. aureus isolates from New York City able, 59 (15 percent) were admitted to the hospital. (60 percent) were resistant to clindamycin. Among An antistaphylococcal penicillin or cephalospo- the isolates that were sent to the CDC, 11 of 218 rin was given to 198 of 311 patients who received MRSA isolates (5 percent) and 7 of 55 MSSA iso- antibiotics (64 percent). In 100 of 175 MRSA in- lates (13 percent) were not susceptible to clinda- fections for which antibiotic treatment was pro- mycin, including 4 (2 percent) MRSA isolates and vided (57 percent), antibiotic therapy was not con- 5 (9 percent) MSSA isolates with inducible clinda- cordant with the results of susceptibility testing. mycin resistance detected by an antimicrobial- Of the 422 patients, 248 (59 percent) were con- susceptibility D-zone disk-diffusion test. Sixty-six tacted for follow-up 15 to 21 days (median, 17) patients with MRSA infections (27 percent) had after their visit to the emergency department, and one or more established risk factors for health 238 (96 percent) of those patients who were con- care–associated MRSA; these included 43 pa- tacted for follow-up reported that their infection tients who had been hospitalized within the past had resolved or improved. There were no signifi- year, 28 with a history of MRSA infection, 2 who cant differences in the outcome between patients resided in a long-term care facility, and 1 who infected with MRSA and those infected with other was undergoing dialysis. Isolates from 55 of these bacteria or between patients in whom the infect- patients were evaluated at the CDC, and 54 (98 ing MRSA isolate was resistant and those in whom percent) had pulsed-field types characteristic of the isolate was susceptible to the prescribed anti- community-associated MRSA. biotic. Baseline characteristics were similar for Features associated with the isolation of MRSA patients with and those without follow-up infor- as compared with the isolation of any other bac- mation.

n engl j med 355;7 www.nejm.org august 17, 2006 669

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

Table 2. Potential Risk Factors for Infection with MRSA, as Compared with Other Bacteria, in Patients with Purulent Skin and Soft-Tissue Infections in 11 U.S. Emergency Departments.*

MRSA Other Bacteria Odds Ratio Risk Factor (N = 249) (N = 135) (95% CI)† no./total no. (%) Race or ethnic group Non-Hispanic white 61/247 (25) 37/135 (27) 1.0‡ Non-Hispanic black 134/247 (54) 53/135 (39) 1.6 (0.9–2.6) Hispanic 48/247 (19) 36/135 (27) 0.8 (0.4–1.5) Other 4/247 (2) 9/135 (7) 0.3 (0.1–0.9)§ Intravenous-drug user 26/244 (11) 11/135 (8) 1.3 (0.6–3.0) Taken any antibiotic in past mo 84/245 (34) 24/134 (18) 2.4 (1.4–4.1)§ Taking antibiotic for the infection 61/244 (25) 24/134 (18) 1.5 (0.9–2.7) Prison or jail in past yr 53/244 (22) 19/135 (14) 1.7 (0.9–3.1) Competitive sports involving contact in past mo 21/246 (9) 8/135 (6) 1.5 (0.6–3.8) Homeless in past yr 40/246 (16) 17/135 (13) 1.3 (0.7–2.6) Abscess 203/243 (84) 97/131 (74) 1.8 (1.0–3.1)§ Spontaneous infection (no apparent precipitating factor) 100/248 (40) 49/135 (36) 1.2 (0.8–1.9) Reported spider bite 71/248 (29) 17/135 (13) 2.8 (1.5–5.3)§ Underlying illness 25/249 (10) 34/135 (25) 0.3 (0.2–0.6)§ HIV infection 11/249 (4) 4/135 (3) 1.5 (0.5–4.9) History of MRSA infection 28/242 (12) 5/132 (4) 3.3 (1.2–10.1)§ Hospitalized in past yr 43/247 (17) 32/135 (24) 0.7 (0.4–1.2) Resident in long-term care facility 2/244 (1) 2/134 (1) 0.6 (0.0–7.6) Health care worker 17/244 (7) 5/135 (4) 1.9 (0.7–6.9) Homosexual male contact 7/132 (5) 5/79 (6) 0.8 (0.2–3.4) Close contact with person with similar infection 43/245 (18) 8/135 (6) 3.4 (1.5–8.1)§ Household contact with MRSA infection 15/245 (6) 3/131 (2) 2.8 (0.8–15.2) Household contact with health care exposure¶ 46/249 (18) 30/135 (22) 0.8 (0.5–1.4) Household contact with jail exposure 42/243 (17) 15/131 (11) 1.6 (0.8–3.2) Household contact with intravenous-drug use 27/245 (11) 11/133 (8) 1.4 (0.6–3.1)

* CI denotes confidence interval, and HIV human virus. † Missing data were excluded in these calculations. Information regarding race was missing for two patients. ‡ This group served as the reference group for other race variables. § P<0.05. ¶ This category includes hospitalization or residence in a long-term care facility in past year, employment as a health care worker, receipt of dialysis, or presence of an indwelling catheter or tube.

Case-finding audits revealed that approximate- from whom wound cultures were obtained (57 ly 42 percent of eligible patients were enrolled. As percent). compared with enrolled patients, unenrolled pa- tients were similar in terms of age (mean, 38 years; Discussion range, 18 to 82), sex (63 percent were male), and race or ethnic group (57 percent were white non- MRSA has emerged as the most common identi- Hispanic or Hispanic, 39 percent were black, and fiable cause of skin and soft-tissue infections in 4 percent were in other groups). MRSA was iso- several metropolitan areas across the United States. lated in 135 of 236 eligible but unenrolled patients Although more than 80 percent of patients with

670 n engl j med 355;7 www.nejm.org august 17, 2006

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. mrsa infections among patients in the emergency department skin and soft-tissue infections associated with Table 3. Results of Multivariate Logistic-Regression Analyses to Identify Potential MRSA in this study received empirical antimicro- Risk Factors for MRSA Infection.* bial therapy for their infection, the infecting iso- late was resistant to the agent prescribed for 57 Adjusted Odds Ratio Characteristic (95% CI)* percent of these patients. This finding suggests a need to reconsider empirical antimicrobial choic- Race es for skin and soft-tissue infections in areas where Non-Hispanic white 1.0† MRSA is prevalent in the community. Non-Hispanic black 1.9 (1.1–3.4) Our findings are consistent with the dramatic Other 0.3 (0.1–1.0) trend of increasing reports of outbreaks and in- Use of any antibiotic in past mo (vs. no use) 2.4 (1.3–4.3) creased prevalence of community-associated MRSA Reported spider bite (vs. other cause of infection) 3.0 (1.6–5.7) during the past few years. MRSA was uncommon in community-acquired skin and soft-tissue infec- Underlying illness (vs. no underlying infection) 0.3 (0.2–0.6) tions before 2000 and accounted for only 3 percent History of MRSA infection (vs. absence of such history) 3.4 (1.1–10) of staphylococcal isolates submitted to Minnesota Close contact with person with similar infection 3.8 (1.6–9.3) laboratories in 2000.20 Between 2001 and 2004, (vs. no close contact) the prevalence of MRSA among patients with * Estimates were adjusted for all other variables in the table. CI denotes confi- skin and soft-tissue infections at our Los Ange- dence interval. les institution increased from 29 percent to 64 † This group served as the reference group for other race variables. percent.13 Virtually all (99 percent) MRSA strains isolated consistent with other recent reports.24,25 S. aureus from skin and soft-tissue infections in this study strains containing pvl genes have been associated had pulsed-field types characteristic of commu- with spontaneous skin and soft-tissue infections nity-associated MRSA, even though more than and necrotizing pneumonia; however, the role of 25 percent of patients had established risk fac- pvl toxin in the pathogenesis of S. aureus skin and tors for health care–associated MRSA. A single soft-tissue infections has not been fully eluci- pulsed-field type (USA300) accounted for 97 per- dated.26 cent of MRSA isolates and 31 percent of MSSA Most patients in our study were treated with isolates. A single strain (USA300-0114) associated β-lactam agents such as cephalexin and dicloxa- with previously reported community outbreaks5 cillin, to which MRSA isolates are not suscepti- accounted for 72 percent of MRSA isolates, and ble. Although we had limited follow-up informa- a closely related strain was the single most com- tion, we found no association between patients’ mon MSSA strain identified. Pulsed-field type outcomes and the susceptibility of the pathogen USA300 has been linked to community-associ- to the prescribed antimicrobial agents. This ab- ated MRSA outbreaks throughout the country5,11 sence of an association, which is consistent with and represents the leading cause of community- previous reports,3,27 suggests that most simple associated MRSA in single-center prevalence stud- skin abscesses, even when caused by MRSA, can ies.12,21 USA300 has rapidly replaced other pulsed- be cured with adequate drainage alone. Nonethe- field types to become predominant among centers less, when antibiotics are clinically indicated and that have conducted longitudinal studies.22 Our MRSA is prevalent in the community, it is diffi- finding of the genetic similarity of MRSA and MSSA cult to justify empirical use of agents known to isolates from community-associated infections be inactive against MRSA. The susceptibility of a is consistent with previous reports.7,8,21,23 This given pathogen to prescribed antimicrobial agents similarity suggests the acquisition of SCCmec by may be more likely to affect the outcome among S. aureus strains established in the community or patients with cellulitis or purulent wounds. Un- the loss of SCCmec by community-associated MRSA fortunately, there were insufficient numbers of strains. The predominance of isolates from one these patients with follow-up information in our genetic background may be related to virulence or study to assess this relationship. Although we transmissibility factors that confer unusual fitness. identified several clinical and epidemiologic fac- Ninety-eight percent of MRSA isolates and tors associated with MRSA infection, it does not more than 40 percent of MSSA isolates in our appear that the presence or absence of these fac- study contained pvl genes; these findings are tors would be useful to guide decisions about the

n engl j med 355;7 www.nejm.org august 17, 2006 671

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

use of empirical antibiotics. Most patients with- agent such as vancomycin for MRSA coverage re- out MRSA had at least one of these factors, and mains appropriate. One study reported that clinda- almost half of those without any of these factors mycin therapy was successful in children with in- were found to have MRSA. vasive community-associated MRSA infections.29 Various antimicrobial agents, such as clindamy- In recent randomized clinical trials, newer agents cin, trimethoprim–sulfamethoxazole, and doxycy- with MRSA activity had efficacy similar (dapto- cline, have been recommended for outpatient em- mycin and tigecycline) or superior (linezolid) to pirical treatment of community-associated skin that of vancomycin for the treatment of compli- and soft-tissue infections that may be attributable cated skin and soft-tissue infections or skin and to MRSA.28-30 More than 90 percent of MRSA iso- soft-tissue infections associated with MRSA.35-37 lates in our study were susceptible to each of these As compared with patients with other bacte- agents. Likewise, 100 percent of the MRSA isolates rial infections of the skin, patients with MRSA were susceptible to rifampin. Although resistance infection were more likely to report a spider bite to rifampin monotherapy has occurred rapidly, the as the reason for their skin lesion, perhaps be- combination of rifampin plus trimethoprim–sul- cause of the propensity for MRSA strains circu- famethoxazole has been shown to eradicate MRSA lating in the community to cause painful lesions colonization and has been suggested for the treat- in the absence of previous skin trauma. Thus, cli- ment of MRSA infection in the community.31 Re- nicians should consider the possibility of MRSA sistance to macrolides and fluoroquinolones was infection in patients who report spider bites. Eigh- prevalent among MRSA isolates in this and other teen percent of patients with skin and soft-tissue studies.3,12,13 infections associated with MRSA reported close Although the prevalence of clindamycin resis- contact with a person who had a similar infection. tance, including inducible resistance, was low This finding highlights the importance of educat- overall, it varied geographically. Clindamycin has ing patients about methods to prevent further been used successfully in the treatment of infec- transmission of infection, including keeping le- tions with MRSA isolates possessing inducible sions covered with clean, dry bandages; practicing resistance.29,32 However, clinical treatment failures good hand hygiene; and avoiding the sharing of have also been reported.33 Therefore, if clindamy- contaminated items. cin therapy is being considered, S. aureus isolates The high prevalence of MRSA among patients with the potential for inducible clindamycin re- with community-associated skin and soft-tissue sistance (i.e., isolates resistant to erythromycin infections has implications for hospital policies but susceptible to clindamycin on initial testing) regarding infection control. Standard precautions should be evaluated for inducible resistance by (including the use of gowns and gloves by health D-zone disk-diffusion testing.17 care workers for contact with wound drainage) Patients with nonpurulent cellulitis were not should be used for all patients. Contact precau- included in our study. Previous studies have shown tions, which include the use of gowns and gloves that a large proportion of cellulitis may be attrib- for all contact with patients or their environment, utable to group A streptococcus.34 In contrast, have been recommended for patients in acute care among infections characterized as cellulitis with inpatient facilities who are known to be infected purulent drainage in our study, MRSA was isolated or colonized with MRSA.38 Our results suggest from 47 percent and group A streptococcus was that strategies used for patients with confirmed rarely isolated. Although generally susceptible in MRSA infections should be considered for all pa- vitro to clindamycin, most group A streptococci tients with purulent skin and soft-tissue infections are resistant to trimethoprim–sulfamethoxazole. in areas with a high prevalence of MRSA. To provide coverage for streptococcal infection, the In many U.S. cities, MRSA is now the most use of clindamycin or a combination of a β-lac- common pathogen isolated in the emergency tam plus trimethoprim–sulfamethoxazole may be department from patients with skin and soft-tis- preferable for nonpurulent cellulitis. sue infections. Clinicians should consider obtain- Optimal empirical therapy for severely ill hos- ing cultures from patients with skin and soft-tis- pitalized patients with complicated skin and soft- sue infections and modifying standard empirical tissue infections has not been established; however, therapy to provide MRSA coverage when antibiot- broad-spectrum intravenous therapy including an ics are indicated. Further studies are needed to

672 n engl j med 355;7 www.nejm.org august 17, 2006

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. mrsa infections among patients in the emergency department determine the extent of this infection in other lo- ports having received consulting fees from or having served on advisory boards for Pfizer and Ortho-McNeil; lecture fees from cations, follow trends in antimicrobial suscepti- Schering-Plough; and research support from Pfizer and Aventis. bility, and identify optimal therapy. No other potential conflict of interest relevant to this article was Supported by a cooperative agreement (U50/CCU912342) with reported. the CDC. We are indebted to Daniel Jernigan, Sigrid McAllister, Jean Pa- The findings and conclusions of this report are those of the tel, David Lonsway, George Killgore, Brandi Limbago, and Laura authors and do not necessarily represent the views of the fund- Jevitt at the CDC for providing technical and laboratory assistance; ing agency. to the following site research coordinators: Ricky Amii, Cynthia Dr. Moran reports having received consulting fees from or Nguyen, Deborah Sibley, Mehr Merabodi, Carolyn Oakes, Marlena having served on advisory boards for Schering-Plough and Pfiz- Wald, Tove Ryman, Yvonne Sanchez, Joni Kopitzke, Karen Pfaff, er; lecture fees from Schering-Plough, Pfizer, Aventis, and Cub- Sara Newton, Mary Mulrow, and Carol Von Hofen; and to the ist; and research support from Pfizer and Aventis. Dr. Talan re- physicians at our study sites who assisted with data collection.

appendix The following investigators participated in the EMERGEncy ID Net Study Group: Olive View–University of California at Los Angeles Medical Center, Sylmar: F.M. Abrahamian; Hennepin County Medical Center, Minneapolis: M. Biros; University of New Mexico Health Sciences Center, Albuquer- que: P.R. Cheney; Bellevue Hospital Center, New York: W.K. Chiang; Louisiana State University Health Science Center, New Orleans: L.M. Dunbar; Maricopa Medical Center, Phoenix, Ariz.: E. Gross; Emory University School of Medicine, Atlanta: K.L. Heilpern; Oregon Health Sciences University, Portland: J. Jui; Temple University School of Medicine, Philadelphia: D.J. Karras; University of Missouri–Kansas City, Kansas City: M.T. Steele; Carolinas Medical Center, Charlotte, N.C.: M. Sullivan.

References

1. Barrett FF, McGehee RF Jr, Finland M. resistant Staphylococcus aureus carrying the tional supplement. M100-S16. Wayne, Pa.: Methicillin-resistant Staphylococcus aureus at Panton-Valentine leukocidin genes. Clin Clinical and Laboratory Standards Insti- Boston City Hospital: bacteriologic and epi- Infect Dis 2005;40:100-7. tute, 2006. demiologic observations. N Engl J Med 10. Naimi TS, LeDell KH, Como-Sabetti K, 18. Pan ES, Diep BA, Charlebois ED, et al. 1968;279:441-8. et al. Comparison of community- and Population dynamics of nasal strains of 2. Herold BC, Immergluck LC, Maranan health care-associated methicillin-resis- methicillin-resistant Staphylococcus aureus MC, et al. Community-acquired methicil- tant Staphylococcus aureus infection. JAMA — and their relation to community-asso- lin-resistant Staphylococcus aureus in children 2003;290:2976-84. ciated disease activity. J Infect Dis 2005; with no identified predisposing risk. JAMA 11. McDougal LK, Steward CD, Killgore 192:811-8. 1998;279:593-8. GE, Chaitram JM, McAllister SK, Tenover 19. Daum RS, Ito T, Hiramatsu K, et al. 3. Fridkin SK, Hageman JC, Morrison M, FC. Pulsed-field gel electrophoresis typ- A novel methicillin-resistance cassette in et al. Methicillin-resistant Staphylococcus au- ing of oxacillin-resistant Staphylococcus au- community-acquired methicillin-resistant reus disease in three communities. N Engl reus isolates from the United States: estab- Staphylococcus aureus isolates of diverse ge- J Med 2005;352:1436-44. [Erratum, N Engl lishing a national database. J Clin Microbiol netic backgrounds. J Infect Dis 2002;186: J Med 2005;352:2362.] 2003;41:5113-20. 1344-7. 4. Outbreaks of community-associated 12. Frazee BW, Lynn J, Charlebois ED, 20. Naimi TS, LeDell KH, Boxrud DJ, et methicillin-resistant Staphylococcus aureus Lambert L, Lowery D, Perdreau-Reming- al. Epidemiology and clonality of commu- skin infections — Los Angeles County, ton F. High prevalence of methicillin- nity-acquired methicillin-resistant Staphy- California, 2002–2003. MMWR Morb Mor- resistant Staphylococcus aureus in emergency lococcus aureus in Minnesota, 1996-1998. Clin tal Wkly Rep 2003;52:88. department skin and soft tissue infec- Infect Dis 2001;33:990-6. 5. Kazakova SV, Hageman JC, Matava M, tions. Ann Emerg Med 2005;45:311-20. 21. Mishaan AM, Mason EO Jr, Martinez- et al. A clone of methicillin-resistant Staph- 13. Moran GJ, Amii RN, Abrahamian FM, Aguilar G, et al. Emergence of a predomi- ylococcus aureus among professional foot- Talan DA. Methicillin-resistant Staphylo- nant clone of community-acquired Staphy- ball players. N Engl J Med 2005;352:468- coccus aureus in community-acquired skin lococcus aureus among children in Houston, 75. infections. Emerg Infect Dis 2005;11:928- Texas. Pediatr Infect Dis J 2005;24:201-6. 6. Zinderman CE, Conner B, Malakooti 30. 22. Chavez-Bueno S, Bozdogan B, Katz K, MA, LaMar JE, Armstrong A, Bohnker BK. 14. Talan DA, Moran GJ, Mower W, et al. et al. Inducible clindamycin resistance and Community-acquired methicillin-resistant EMERGEncy ID NET: an emergency de- molecular epidemiologic trends of pedi- Staphylococcus aureus among military re- partment-based emerging infections sen- atric community-acquired methicillin- cruits. Emerg Infect Dis 2004;10:941-4. tinel network. Ann Emerg Med 1998;32: resistant Staphylococcus aureus in Dallas, Tex- 7. Gonzalez BE, Martinez-Aguilar G, 703-11. as. Antimicrob Agents Chemother 2005; Hulten KG, et al. Severe staphylococcal sep- 15. Bannerman TL. Staphylococci and 49:2283-8. sis in adolescents in the era of community- other catalase positive cocci that grow 23. Adem PV, Montgomery CP, Husain acquired methicillin-resistant Staphylococcus aerobically. In: Murray PR, Baron EJ, Jo- AN, et al. Staphylococcus aureus sepsis and aureus. Pediatrics 2005;115:642-8. gensen JH, eds. Manual of clinical micro- the Waterhouse–Friderichsen syndrome in 8. Mongkolrattanothai K, Boyle S, Kahana biology. 8th ed. Washington, D.C.: ASM children. N Engl J Med 2005;353:1245-51. MD, Daum RS. Severe Staphylococcus aureus Press, 2003:384-404. 24. Yamasaki O, Kaneko J, Morizane S, et infections caused by clonally related com- 16. Performance standards for antimicro- al. The association between Staphylococcus munity-acquired methicillin-susceptible bial susceptibility testing: 14th informa- aureus strains carrying Panton-Valentine and methicillin-resistant isolates. Clin In- tional supplement. M1000-S14. Wayne, leukocidin genes and the development of fect Dis 2003;37:1050-8. Pa.: National Committee for Clinical Lab- deep-seated follicular infection. Clin In- 9. Francis JS, Doherty MC, Lopatin U, et oratory Standards, 2004. fect Dis 2005;40:381-5. al. Severe community-onset pneumonia 17. Performance standards for antimicro- 25. Issartel B, Tristan A, Lechevallier S, et in healthy adults caused by methicillin- bial susceptibility testing: 16th informa- al. Frequent carriage of Panton-Valentine

n engl j med 355;7 www.nejm.org august 17, 2006 673

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. mrsa infections among patients in the emergency department

leucocidin genes by Staphylococcus aureus coccus aureus in children. Pediatr Infect Dis lulitis in adults: a microbiologic study us- isolates from surgically drained abscess- J 2003;22:593-8. ing a direct immunofluorescence tech- es. J Clin Microbiol 2005;43:3203-7. 30. Rybak MJ, LaPlante KL. Community- nique. Arch Dermatol 1989;125:779-82. 26. Lina G, Piemont Y, Godail-Gamot F, associated methicillin-resistant Staphylococ- 35. Weigelt J, Itani K, Stevens D, et al. Li- et al. Involvement of Panton-Valentine cus aureus: a review. Pharmacotherapy 2005; nezolid versus vancomycin in treatment leukocidin-producing Staphylococcus aureus 25:74-85. of complicated skin and soft tissue infec- in primary skin infections and pneumo- 31. Chambers HF. Treatment of infection tions. Antimicrob Agents Chemother 2005; nia. Clin Infect Dis 1999;29:1128-32. and colonization caused by methicillin- 49:2260-6. 27. Lee MC, Rios AM, Aten MF, et al. resistant Staphylococcus aureus. Infect Con- 36. Carpenter CF, Chambers HF. Dapto- Management and outcome of children trol Hosp Epidemiol 1991;12:29-35. mycin: another novel agent for treating with skin and soft tissue abscesses caused 32. Drinkovic D, Fuller ER, Shore KP, infections due to drug-resistant gram-pos- by community-acquired methicillin-resis- Holland DJ, Ellis-Pegler R. Clindamycin itive pathogens. Clin Infect Dis 2004;38: tant Staphylococcus aureus. Pediatr Infect treatment of Staphylococcus aureus express- 994-1000. Dis J 2004;23:123-7. ing inducible clindamycin resistance. J An- 37. Nathwani D. Tigecycline: clinical evi- 28. Kaplan SL. Treatment of community- timicrob Chemother 2001;48:315-6. dence and formulary positioning. Int J An- associated methicillin-resistant Staphylo- 33. Siberry GK, Tekle T, Carroll K, Dick J. timicrob Agents 2005;25:185-92. coccus aureus infections. Pediatr Infect Dis J Failure of clindamycin treatment of meth- 38. Garner JS. Guidelines for isolation pre- 2005;24:457-8. icillin-resistant Staphylococcus aureus ex- cautions in hospitals. Part I. Evolution of 29. Martinez-Aguilar G, Hammerman pressing inducible clindamycin resistance isolation practices. Am J Infect Control WA, Mason EO Jr, Kaplan SL. Clindamy- in vitro. Clin Infect Dis 2003;37:1257-60. 1996;24:24-31. cin treatment of invasive infections caused 34. Bernard P, Bedane C, Mounier M, De- Copyright © 2006 Massachusetts Medical Society. by community-acquired, methicillin-resis- nis F, Catanzano G, Bonnetblanc JM. tant and methicillin-susceptible Staphylo- Streptococcal cause of erysipelas and cel-

PERSONAL ARCHIVES IN THE JOURNAL ONLINE Individual subscribers can store articles and searches using a feature on the Journal’s Web site (www.nejm.org) called “Personal Archive.” Each article and search result links to this feature. Users can create personal folders and move articles into them for convenient retrieval later.

674 n engl j med 355;7 www.nejm.org august 17, 2006

The New England Journal of Medicine Downloaded from nejm.org on August 8, 2012. For personal use only. No other uses without permission. Copyright © 2006 Massachusetts Medical Society. All rights reserved. Journal of Antimicrobial Chemotherapy Advance Access published November 3, 2009

Journal of Antimicrobial Chemotherapy doi:10.1093/jac/dkp386

Polyclonal multiply antibiotic-resistant methicillin-resistant Staphylococcus aureus with Panton–Valentine leucocidin in England

Matthew J. Ellington1*, Mark Ganner1, Marina Warner2, Barry D. Cookson1 and Angela M. Kearns1

1Laboratory of Healthcare Associated Infections, Centre for Infections, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK; 2Antibiotic Resistance Monitoring and Reference Laboratory, Centre for Infections, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK

Received 10 July 2009; returned 2 September 2009; revised 24 September 2009; accepted 3 October 2009 Downloaded from

Objectives: Community-associated methicillin-resistant Staphylococcus aureus (MRSA) including those encoding Panton-Valentine leucocidin (PVL) are often described as more susceptible to a range of antibiotics than their hospital-associated counterparts. Recent scattered reports of the emergence of multiresistant PVL-MRSA have highlighted the potential for resistance to emerge. Here we detail http://jac.oxfordjournals.org/ polyclonal multiply antibiotic-resistant PVL-MRSA occurring in England. Methods: PVL-MRSA from community-based and hospitalized patients located across England were identified by PCR. Isolates were characterized via MIC determinations, toxin gene profiling, PFGE, SCCmec, spa and agr typing. Multilocus sequence typing (MLST) was performed on selected isolates. Patient demographic and available disease data were retained for analysis. Results: Seventy-six PVL-MRSA isolates resistant to three further classes of antibiotic were identified between 2005 and 2008 from centres in each of the Health Protection Agency’s geographic regions in

England. Patient demographics were typical for PVL-MRSA, and some travel associations were ident- by guest on August 8, 2012 ified along with clonal spread. One instance of familial transmission in the community was detected. PVL-MRSA belonging to MLST clonal complex (CC) 1 (sequence type 772) were consistently highly resistant; multiply antibiotic-resistant representatives of CCs 5, 8, 22, 59 and 80 were also identified. Ciprofloxacin resistance was common amongst the study isolates (51 of 76 isolates). Conclusions: Genetically diverse multiply antibiotic-resistant PVL-MRSA were identified, and included representatives of a recently emerged multiresistant clone (dubbed the Bengal Bay clone). Risk factors and disease presentations were typical for PVL-MRSA infections. This work highlights the diminishing utility of ciprofloxacin susceptibility for putative identification of PVL-MRSA.

Keywords: PVL, community, MRSA

Introduction and soft tissue infections in young adults (18–40 years), but also severe diseases such as necrotizing pneumonia associated Worldwide, healthcare-associated methicillin-resistant with high mortality rates.2 Staphylococcus aureus (HA-MRSA) are commonly resistant to CA-MRSA are often more genetically diverse within a multiple classes of antibiotic with pandemic lineages including locale, and less antibiotic resistant than HA-MRSA,3 with many the New York–Japan clone [sequence type (ST)5-MRSA- being susceptible to most antimicrobials excepting b-lactams, SCCmecII] and the Brazilian clone (ST239-SCCmecIII).1 although this situation is evolving.4 In North America the Recently emerged community-associated strains of MRSA predominant CA-MRSA clone, USA300, which encodes (CA-MRSA) with enhanced transmissibility and/or virulence Panton–Valentine leucocidin (PVL), has caused an epidemic of have become increasingly recognized worldwide. CA-MRSA are infections in community settings and has now infiltrated and associated with a significant burden of disease, most often skin become endemic in some US healthcare establishments.5

...... *Corresponding author. Tel: þ44-20-8327-7259; Fax: þ44-20-8200-7449; E-mail: [email protected]

...... Page 1 of 5 # The Author 2009. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected] Ellington et al.

Recently, a multiply antibiotic-resistant variant of USA300 has by PFGE of total DNA restricted with SmaI, as described pre- emerged in San Francisco with resistance to mupirocin, erythro- viously.7 Banding patterns were analysed using BioNumerics soft- mycin and clindamycin, and variably to ciprofloxacin and ware (Applied Maths, Ghent, Belgium). Isolates with banding tetracycline.4 Additionally, multiply antibiotic-resistant represen- profiles .80% similar were considered closely related. As described 7 tatives of the South-East Asian clone (ST59-MRSA- previously, PCR to determine agr, SCCmec and ccr type, in SCCmecVII/IV) with PVL have also been reported.6 addition to spa sequence typing, was used, and for selected isolates, Nevertheless, multiply antibiotic-resistant PVL-MRSA remain multilocus sequence typing (MLST) was performed to determine unusual, for the time being at least. the ST and clonal complex (CC). The Health Protection Agency’s national Staphylococcus Reference Unit has reported 11 lineages of PVL-MRSA in England. Herein we report the widespread geographic distribution Results of genetically diverse ciprofloxacin and multiply drug-resistant PVL-MRSA, and the detection of a newly emerged multiply Seventy-six multiresistant PVL-MRSA belonging to six distinct antibiotic-resistant clone of PVL-MRSA in England. genetic lineages were identified amongst isolates referred to the national reference laboratory, and comprised 3 in 2005, 9 in 2006, 27 in 2007 and 37 in 2008. The isolates were referred Methods from each of the Health Protection Agency’s nine regions in England, with no more than three highly resistant PVL-MRSA Bacterial isolates and susceptibility testing of the same lineage referred by each centre, indicating geo-

MICs for S. aureus isolates from suspected PVL-related disease and graphic dissemination (Table 1). Patient demographics were Downloaded from close contact screenings, submitted to the national Staphylococcus typical for individuals with PVL-MRSA: the male/female ratio reference laboratory for England, between 2005 and 2008 that tested was almost equal and patients had a mean age of 32 years positive for the mecA and lukSF-PV (PVL) genes were determined (median 31 and mode of 27 years; data from 75 individuals were by Etest (AB BIODISK, Solna, Sweden), or agar dilution using available). Available clinical data (from 62 individuals) indi- Iso-Sensitest agar (Oxoid, Basingstoke, UK) according to the BSAC cated that the majority (53/76) presented with skin and soft method. The antimicrobials tested were: penicillin, oxacillin, cipro- tissue infections, four individuals were asymptomatic (identified http://jac.oxfordjournals.org/ floxacin, tetracycline, doxycycline, minocycline, tigecycline, ery- as a result of outbreak control screening) and five patients had thromycin, gentamicin, neomycin, fusidic acid, clindamycin, invasive PVL-associated disease (two of whom subsequently rifampicin, quinupristin/dalfopristin, teicoplanin, vancomycin, died). The isolates from the five patients with invasive disease daptomycin, mupirocin and linezolid. Inducible resistance to clinda- represented four distinct lineages. Direct strain transmission mycin was tested by the BSAC D-disc test. Seventy-six PVL-MRSA between two community-based family members was detected. resistant to three or more classes of commonly used antimicrobials Various antibiotic susceptibility profiles were identified within [fluoroquinolones, aminoglycosides, macrolides/lincosamides (ery- each of five genetic lineages of highly resistant PVL-MRSA thromycin and clindamycin), tetracyclines, fusidic acid and detected (see below). Heterogeneous oxacillin MICs (range: by guest on August 8, 2012 mupirocin] were retained for study. Wherever possible, patient demographic and clinical data were collected from referral forms, 1–128 mg/L) were observed, and ciprofloxacin resistance was returned clinical questionnaires and telephone follow-up. detected in 49 of 76 isolates, representing MLST CCs 8 (ST8), 22 (ST22), 80 (ST80), 1 (ST772) and 5 (ST866). All isolates tested remained susceptible to rifampicin, quinupristin/ Genetic characterization of isolates dalfopristin, linezolid and vancomycin. The characteristics of In addition to multiplex PCR, which was used to detect mecA, the multiresistant PVL-MRSA identified in England are sum- lukSF-PV and 13 other toxin genes, isolate relatedness was analysed marized below.

Table 1. Regional distribution of multiply antibiotic-resistant PVL-MRSA detected in England between 2005 and 2008

MLST CC

HPA regionsa CC8 CC22 CC59 CC80 CC1 (ST772-like) CC5 (ST866-like) Total

East of England 1 1 2 1 2 7 East Midlands 2 1 3 6 London 3 1 3 5 12 North East 2 2 1 1 6 North West 2 2 4 1 9 South East 1 4 8 1 14 South West 2 2 3 7 West Midlands 2 1 4 1 8 Yorkshire and Humber 2 2 2 1 7 Total 7 12 10 11 29 7 76 aSee: http://www.hpa.org.uk/web/HPAweb&HPAwebStandard/Page/1169747331532.

Page 2 of 5 Emerging multiresistant PVL-MRSA

CC1 (ST772-like) PVL-MRSA Discussion Thirty multiply antibiotic-resistant isolates, referred from 23 Community-associated (including PVL-positive) MRSA are different centres were related to known ST772 isolates (CC1; broadly considered less resistant than their HA-MRSA counter- single locus variant of ST1) by spa and SCCmec typing and parts, but do appear to be evolving resistance.4 Using the other genetic characteristics (Table 2). Seventeen of the 30 cases criterion of MRSA resistant to three or more further classes had evidence of a travel or familial link with the Bengal Bay of antibiotic (including fluoroquinolones, aminoglycosides, area or the Indian subcontinent. All 30 isolates were ciprofloxa- macrolides, tetracyclines, fusidic acid, mupirocin), we report cin resistant and a further ‘core’ resistance pattern of erythromy- increasing numbers of multiply antibiotic-resistant PVL-MRSA cin, gentamicin and trimethoprim resistance was observed in 26/ identified amongst isolates referred to the national 30 isolates; tetracycline resistance was also noted in two Staphylococcus reference laboratory. isolates. Globally disseminated PVL-MRSA clones were represented amongst the six lineages of multiply antibiotic-resistant PVL-MRSA identified; no single lineage was more associated CC5 (ST866-like) isolates with invasive disease than another. Notwithstanding reference Six isolates were genetically similar and grouped by PFGE with laboratory bias, the geographic distribution and the partial travel isolates known to be ST866 (CC5; double locus variant of ST5) association of the cases suggest that these multiply antibiotic- (Table 2). The six isolates had highly variable antibiotypes: four resistant PVL-MRSA were disseminated across England, albeit were resistant to gentamicin and three were resistant to erythro- at a low level, during the study period; however, their numbers mycin (clindamycin inducible). Resistance to tetracycline, did build annually across the four study years. Previously, Downloaded from mupirocin and ciprofloxacin was variable. ST772 S. aureus with PVL have been detected in India,8 Bangladesh,9 as well as Malaysia where common travel to/from Bangladesh was noted.10 The ST772 PVL-MRSA cases detected CC8 isolates in England in this study were often linked with India and Bangladesh (n¼17/30). The consistently high level of antimi- Seven isolates, from four geographically distinct centres, crobial resistance, clonality and geographic dissemination of the http://jac.oxfordjournals.org/ arcA included six ACME- -positive USA300-like isolates isolates detected emphasizes the potential transmissibility of this (Table 2). All seven isolates were resistant to erythromycin and lineage (herein dubbed the ‘Bengal Bay clone’). Multiply ciprofloxacin. Resistance to gentamicin, tetracycline, fusidic acid antibiotic-resistant representatives of other international and mupirocin was variable. The two most resistant isolates CA-MRSA clones included USA300, but in contrast to the situ- were resistant to four antibiotic classes including fluoroquino- ation in the USA,4 mupirocin, erythromycin/clindamycin, cipro- lones, macrolides and mupirocin with variable aminoglycoside floxacin and tetracycline resistance did not occur in the same or fusidic acid resistance. UK USA300 isolate in this study. The convergence of ciproflox-

acin resistance in multiple PVL-MRSA lineages highlights the by guest on August 8, 2012 diminishing utility of the identification of ciprofloxacin suscepti- CC22 isolates bility in MRSA as a putative marker for PVL-MRSA. This, Twelve genetically similar isolates, from 11 geographically distinct alongside reference laboratory bias/non-systematic collection of centres, were grouped with known ST22 (CC22) PVL-MRSA isolates, suggests a current under-ascertainment of the totality based upon PFGE banding profiles. Trimethoprim and gentamicin of multiply antibiotic-resistant PVL-MRSA and, by inference, resistance was observed in 11/12 CC22-PVL-MRSA. Eight isolates multiply antibiotic-resistant CA-MRSA in the UK, and under- were ciprofloxacin resistant (Table 2). lines the need for continued and enhanced surveillance of antibiotic susceptibility amongst them. Guidance for the treatment of PVL S. aureus-related infec- CC59 isolates tions in England11 recommends agents such as linezolid, clinda- mycin and rifampicin for the treatment of serious disease and Ten multiresistant CC59 isolates harbouring either SCCmecIV mupirocin for decolonization purposes. Whilst resistance to line- or VII referred from 10 different centres were otherwise geneti- zolid and rifampicin was not detected amongst the isolates cally similar (Table 2). All 10 isolates were erythromycin, clin- tested, resistance to clindamycin (constitutive or inducible) and damycin, neomycin and tetracycline resistant; a single isolate mupirocin was detected in multiple lineages of PVL-MRSA. was also resistant to ciprofloxacin. Similarly, doxycycline is one of the agents recommended for treatment of skin and soft tissue infections, and here we ident- ified tetracycline resistance (with a concomitant decrease in dox- CC80 isolates ycycline susceptibility), in multiple lineages of multiply Eleven genetically similar multiply antibiotic-resistant isolates antibiotic-resistant PVL-MRSA suggesting these compounds to (each from different centres) similar to known European clone be of diminishing efficacy against some PVL-MRSA. Together, CC80 PVL-MRSA were detected. All 11 isolates were resistant these data highlight the need for (i) improved ongoing surveil- to tetracycline and fusidic acid in addition to one or more of ery- lance via prospective studies, (ii) the choice of therapeutic thromycin (with inducible clindamycin resistance), gentamicin regimen to be backed up by susceptibility data and (iii) the refer- or trimethoprim. Ciprofloxacin resistance was detected in one ral of isolates to reference laboratories for PVL testing to be multiply antibiotic-resistant CC80 PVL-MRSA isolate. based on clinical suspicion rather than susceptibility data.

Page 3 of 5 Table 2. Characteristics of multiresistant PVL-MRSA

Isolates MLST spa SCCmec (n) CC ST Antibiotic resistances type spa repeat succession type ccr agr Toxin gene profile

24 1 772 OXA, (CHL), CIP, ERY, GEN, TMP t657 r26r23r13r21r17r34r33r34 VII C2 2 sea,(sec), seg, sei, lukSF-PV t345 r26r23r13r21r17r34r34r33r34 3 1 OXA, (CHL), CIP, ERY, (GEN), TET, TMP t657 r26r23r13r21r17r34r33r34 VII C2 2 sea, seg, sei, lukSF-PV t345 r26r23r13r21r17r34r34r33r34 3 1 OXA, (CHL), CIP, GEN, (TET), (TMP) t657 r26r23r13r21r17r34r33r34 VII C2 2 sea,(sec), seg, sei, lukSF-PV 6 5 866 OXA, ERY, CLI (I), TET, DOX, (GEN), (TMP), t002 r26r23r17r34r17r20r17r12r17r16 IV 2 2 (sea), (sec), seg, sei, tst, lukSF-PV (CIP) 10 59 59 OXA, (CIP), ERY, CLI, NEO, TET, DOX t437 r04r20r17r20r17r25r34 VII C2 1 (seb), lukSF-PV t3590 r04r02r17r20r17r25r34 Ellington ae4o 5 of 4 Page 4 80 80 OXA, NEO, TET, FUS, t044 r07r23r12r34r34r33r34 IV 2 3 seh,(etd), lukSF-PV 4 80 OXA, ERY, CLI (I), (NEO), TET, DOX, FUS t044 r07r23r12r34r34r33r34 IV 2 3 seh,(etd), lukSF-PV t131 r07r23r12r34r33r34 tal et 3 80 OXA, (CIP), (GEN), TET, (TMP), FUS t044 r07r23r12r34r34r33r34 IV 2 3 seh,(etd), lukSF-PV

3 22 22 OXA, CIP, ERY, GEN, TMP t852 r07r23r13r23r31r05r17r25r17r25r16r28 IV 2 1 seg, sei, lukSF-PV . 4 22 OXA, CIP, GEN, TMP t005 r26r23r13r23r31r05r17r25r17r25r16r28 IV 2 1 seg, sei, lukSF-PV t2816 r26r23r13r23r31r31r05r17r25r17r25r16r28 t1516 r26r25r17r25r16r28 4 22 OXA, ERY, CLI (I), GEN, TMP, (MUP) t005 r26r23r13r23r31r05r17r25r17r25r16r28 IV 2 1 seg, sei, lukSF-PV 1 22 OXA, CIP, ERY, CLI (I), FUS t1790 r26r23r23r29r17r25r17r25r28 IV 2 1 sec, seg, sei, lukSF-PV 3 8 8 OXA, CIP, ERY, CLI (I), TET t008 r11r19r12r21r17r34r24r34r22r25 IV 2 1 lukSF-PV t121 r11r19 ...r21r17r34r24r34r22r25 2 8 OXA, CIP, ERY, CLI (I), FUS, (MUP) t008 r11r19r12r21r17r34r24r34r22r25 IV 2 1 lukSF-PV 2 8 OXA, CIP, ERY, CLI (I), GEN, (MUP) t008 r11r19r12r21r17r34r24r34r22r25 IV 2 1 lukSF-PV

OXA, oxacillin; CHL, chloramphenicol; CIP, ciprofloxacin; ERY, erythromycin; GEN, gentamicin, TMP, trimethoprim; TET, tetracycline; CLI (I), clindamycin (inducible); DOX, doxycycline; NEO, neomycin; FUS, fusidic acid; MUP, mupirocin. All isolates tested for toxin genes sea-e, seg-j, eta, etb, etd, tst and lukSF-PV. Bracketed data denote variable findings.

All isolates tested were susceptible to quinupristin/dalfopristin, minocycline, daptomycin, tigecycline, vancomycin, teicoplanin and linezolid.

Downloaded from from Downloaded http://jac.oxfordjournals.org/ by guest on August 8, 2012 8, August on guest by Emerging multiresistant PVL-MRSA

Acknowledgements 4. Diep BA, Chambers HF, Graber CJ et al. Emergence of multidrug-resistant, community-associated, methicillin-resistant We would like to thank microbiologists and Health Protection Staphylococcus aureus clone USA300 in men who have sex with men. Units across England. Additional thanks to Clare Martin for pro- Ann Intern Med 2008; 148: 249–57. viding database support and Marjorie Ganner for technical 5. Kourbatova EV, Halvosa JS, King MD et al. Emergence of assistance. community-associated methicillin-resistant Staphylococcus aureus USA 300 clone as a cause of health care-associated infections among patients with prosthetic joint infections. Am J Infect Control 2005; 33: 385–91. Funding 6. Takano T, Higuchi W, Otsuka T et al. Novel characteristics of community-acquired methicillin-resistant Staphylococcus aureus strains This work was supported by the Health Protection Agency, UK. belonging to multilocus sequence type 59 in Taiwan. Antimicrob Agents Chemother 2008; 52: 837–45. 7. Ellington MJ, Perry C, Ganner M et al. Clinical and molecular epidemiology of ciprofloxacin-susceptible MRSA encoding PVL in Transparency declarations England and Wales. Eur J Clin Microbiol Infect Dis 2009; 28: 1113–21. None to declare. 8. Goering RV, Shawar RM, Scangarella NE et al. Molecular epidemiology of methicillin-resistant and methicillin-susceptible Staphylococcus aureus isolates from global clinical trials. J Clin Microbiol 2008; 46: 2842–7.

References 9. Afroz S, Kobayashi N, Nagashima S et al. Genetic characteriz- Downloaded from ation of Staphylococcus aureus isolates carrying Panton–Valentine 1. Witte W, Cuny C, Klare I et al. Emergence and spread of leukocidin genes in Bangladesh. Jpn J Infect Dis 2008; 61: 393–6. antibiotic-resistant Gram-positive bacterial pathogens. Int J Med 10. Neela V, Ehsanollah GR, Zamberi S et al. Prevalence of Microbiol 2008; 298: 365–77. Panton–Valentine leukocidin genes among carriage and invasive 2. Morgan MS. Diagnosis and treatment of Panton–Valentine leu- Staphylococcus aureus isolates in Malaysia. Int J Infect Dis 2009; 13: kocidin (PVL)-associated staphylococcal pneumonia. Int J Antimicrob e131–2. http://jac.oxfordjournals.org/ Agents 2007; 30: 289–96. 11. Health Protection Agency. Guidance on the Diagnosis and 3. Kluytmans-Vandenbergh MF, Kluytmans JA. Community- Management of PVL-Associated Staphylococcus aureus Infections acquired methicillin-resistant Staphylococcus aureus: current perspec- (PVL-SA) in the UK. http://www.hpa.org.uk/webc/HPAwebFile/ tives. Clin Microbiol Infect 2006; 12 Suppl 1: 9–15. HPAweb_C/1218699411960 (17 September 2009, date last accessed). by guest on August 8, 2012

Page 5 of 5 ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Apr. 2007, p. 1515–1519 Vol. 51, No. 4 0066-4804/07/$08.00ϩ0 doi:10.1128/AAC.01201-06 Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Effect of Antibiotics on Staphylococcus aureus Producing Panton-Valentine Leukocidinᰔ Oana Dumitrescu, Sandrine Boisset, Cedric Badiou, Michele Bes, Yvonne Benito, Marie-Elisabeth Reverdy, Franc¸ois Vandenesch, Jerome Etienne, and Gerard Lina* INSERM, U851, Lyon F-69008, France, and Universite´Lyon 1, Centre, National de re´fe´rence des Staphylocoques, Faculte´Laennec, Lyon F-69008, France

Received 25 September 2006/Returned for modification 19 October 2006/Accepted 10 January 2007

We examined the capacity of Staphylococcus aureus strains to release Panton-Valentine leukocidin (PVL) in the presence of antibiotics. No PVL was detected when S. aureus was incubated at inhibitory concentrations, while subinhibitory concentrations of oxacillin enhanced the PVL level; clindamycin, linezolid, and fusidic acid were inhibitory; and vancomycin had roughly no effect.

Staphylococcus aureus is an important human pathogen. It medium, bacterial inoculum, and growth conditions in order to expresses a variety of exoproteins, including Panton-Valentine be able to extrapolate our results to the clinical setting (20). leukocidin (PVL) (31). While Voyich et al. could not establish The PVL concentration was determined in culture superna- clear differences in virulence between isogenic pairs of PVL- tants by using a specific solid-phase sandwich -linked positive/negative strains (29), Labandeira-Rey et al. clearly immunosorbent assay (ELISA) as recommended by the man- demonstrated the role of PVL as a major determinant of vir- ufacturer (Agro-Bio; bioMe´rieux). Unfortunately, when using ulence in an acute pneumonia mouse model using other sets of Mueller-Hinton (MH) medium and CSLI conditions, the PVL isogenic strains for PVL (13) and thus confirmed the results of level was close to the detection limit in the absence of antibi- the princeps experiments showing that PVL is a virulence fac- otics (data not shown). MH medium was thus replaced by tor (15). The apparent discrepancy between these studies ba- casein hydrolysate and yeast extract (CCY) medium, which sically comes from the choice of the experimental models and increases PVL levels (32). The PVL level was 50 times higher the choice of the strains. in CCY than in MH medium (data not shown), while the MICs PVL is now frequently detected in clinical practice, as it is obtained with the two media were of the same order (Table 2), produced by community-acquired methicillin-resistant S. aureus except with gentamicin, which dramatically increased (632- (CA-MRSA) clones currently spreading throughout the world fold). CCY medium was therefore used in the rest of the study, (27). PVL has been linked to specific human S. aureus infections and gentamicin was excluded. such as primary skin and soft tissue disease and severe necrotizing pneumonia, where the mortality rate is about 75% (10, 14). Sev- eral lines of evidence incriminate PVL in necrotizing pneumonia TABLE 1. Strains, plasmid, and phage pathogenesis, including the strong epidemiological link with PVL- Strain, plasmid, Reference producing S. aureus isolates (10) and the immunodetection of Description PVL in the lung (9) and that solely PVL-producing S. aureus or phage or source isolates are able to reproduce necrotizing pneumonia in experi- S. aureus strains mental models (13). Antibiotics that inhibit PVL production RN6390 22 Laboratory strain that maintains its may be more appropriate for the treatment of severe necro- hemolytic activity when propagated on sheep erythrocyte agar tizing pneumonia, by analogy with their use in streptococcal (parental strain) and staphylococcal toxic shock syndrome (8, 25, 26). LUG855 This study RN6390 phiSLT We examined the effect of antibiotics on PVL release by LUG1124 This study RN6390 carrying pLUG547 ϩ ϩ methicillin-sensitive S. aureus and CA-MRSA strains in vitro. HT20010734 18 ST1; agr3 mecA lukS-PV lukF-PV HT20020488 27 ST80; agr3 mecAϩ lukS-PV lukF-PVϩ We chose a reference strain lysogenized by phage phiSLT ϩ ϩ HT20030203 24 ST8; agr1 mecA lukS-PV lukF-PV (encoding luk-PV) and five isolates representing the main HT20040332 30 ST59; agr1 mecAϩ lukS-PV lukF-PVϩ PVL-producing CA-MRSA clones (Table 1) (23, 24, 27, 30). HT20041010 23 ST80; agr3 mecAϩ lukS-PV lukF-PVϩ We intended to use experimental procedures as close as pos- sible to Clinical Laboratory Standards Institute (CSLI) recom- Plasmid pLUG547 This study Derivative of pTCV-lac containing mendations for MIC determinations in terms of the culture the lukS lukF-PV promoter (nucleotide Ϫ480 to the start codon) fused to lacZ * Corresponding author. Mailing address: Centre National de Re´f- ´rencee des Staphylocoques, INSERM E0230, 7 rue Guillaume Para- Phage din, 69372 Lyon Cedex 08, France. Phone: 33 472 11 07 75. Fax: 33 478 phiSLT 19 lukS lukF-PVϩ-containing phage 11 07 64. E-mail: [email protected]. isolated from A980470 ᰔ Published ahead of print on 22 January 2007.

1515 1516 NOTES ANTIMICROB.AGENTS CHEMOTHER.

TABLE 2. MICs of selected antibiotics for MSSA and CA-MRSA isolates in MH and CCY media

MIC (␮g/ml) in MH medium/MIC in CCY mediuma

Antibiotic LUG855 HT20010734 HT20020488 HT20030203 HT20040332 HT20041010 (RN6390) Oxacillin 0.12/0.06 32/16 8/8 32/16 16/8 32/16 Clindamycin 0.06/0.12 0.12/0.25 Ͼ128/Ͼ128 0.12/0.25 Ͼ128/Ͼ128 Ͼ128/Ͼ128 Linezolid 1/2 0.5/2 0.5/2 1/2 1/2 1/2 Vancomycin 1/4 0.5/4 0.5/4 1/4 1/4 2/4 Fusidic acid Ͻ0.03/Ͻ0.03 Ͻ0.03/Ͻ0.03 4/4 0.12/0.06 Ͻ0.03/Ͻ0.03 Ͼ128/128

a MICs were determined with a standard microdilution method recommended by the CSLI in MH broth and CCY medium inoculated with 5 ϫ 105 CFU/ml (20).

To examine the influence of antibiotics on PVL release, lin and was unmodified using sub-MIC concentrations of van- PVL was quantified in the culture supernatant of S. aureus comycin. As LUG855 is highly sensitive to fusidic acid, the LUG855 incubated with various concentrations of oxacillin, latter antibiotic was not tested for its effect on PVL release. vancomycin, clindamycin, and linezolid for 24 h. As shown in To confirm these results, experiments were reproduced us- Fig. 1A, no PVL was detected when bacteria were incubated ing five different CA-MRSA isolates (Table 1 and Fig. 1B). with inhibitory concentrations of oxacillin, clindamycin, fusidic Linezolid induced a concentration-dependent decrease in the acid, linezolid, or vancomycin. This could be explained by the PVL level from one-eighth the MIC (four of five isolates) and fact that PVL production requires bacterial growth (3). from one-quarter the MIC (all isolates) to the MIC. Clinda- PVL is associated with intense necrosis in vivo, possibly mycin, tested on the two susceptible strains, induced a strong leading to poor antibiotic diffusion and suboptimal concentra- concentration-dependent decrease in PVL levels from one- tions at sites of infection (4). We therefore examined the effect eighth the MIC to the MIC. Again, PVL release by the CA- of subinhibitory antibiotic concentrations on PVL. PVL levels MRSA isolates increased in the presence of all subinhibitory released by LUG885 depended on the antibiotic and the con- oxacillin concentrations, by 2- to 6.5-fold. With vancomycin, centration used (Fig. 1A). Clindamycin and linezolid induced PVL levels were unmodified except with HT20020488 at one- concentration-dependent decreases in PVL levels from one- quarter and one-eighth the MIC and HT20010734 at one-half eighth the MIC, while it was significantly increased (up to the MIC, which decreased and increased PVL release, respec- threefold) at one-eighth and one-quarter the MIC with oxacil- tively. Fusidic acid, tested on isolates with MICs higher than

FIG. 1. Effect of antibiotics on PVL. S. aureus LUG855 (A) and S. aureus strains HT20010734, HT20020488, HT20030203, HT20040332, and HT20041010 (B) were incubated in CCY medium with or without antibiotics (at 1, [1/2], [1/4], and [1/8] MIC), according to standard CSLI procedures, for 25 h at 37°C without shaking. Samples were taken for bacterial counting (plate counting of colonies from diluted broth) and PVL ␮ quantification by ELISA. Results are ratios of g of PVL/log10 CFU of bacteria cultured with the indicated concentrations of antibiotic by means ␮ Ϯ of g of PVL/log10 CFU of bacteria cultured without antibiotic and expressed as percentage values. Values are means standard deviations (five statistically different from the control (corresponding isolate grown ,ء .(different experiments in panel A and three different experiments in panel B without antibiotic), with a P value of Ͻ0.05, by one-way analysis of variance followed by a posteriori Dunnett’s test. ND, not determined. VOL. 51, 2007 NOTES 1517

FIG. 1—Continued.

0.03 ␮g/ml, induced a concentration-dependent decrease in the ␤-galactosidase activity. Samples were adjusted to an optical PVL level. density at 600 nm of 1 before cell lysis with the FastPrep To examine the effect of antibiotics on PVL gene transcrip- instrument (QBiogen). Protein concentrations and ␤-galacto- tion, S. aureus LUG1124 (containing the plasmid-borne luk-PV sidase activity were determined in the lysates by using the promoter fused to the lacZ gene described in Table 1) was Bradford method (1) and the Beta-Glo system (Promega), cultured with or without oxacillin, vancomycin, or linezolid at respectively. As shown in Fig. 2, ␤-galactosidase activity was one-eighth, one-quarter, and one-half the MIC and assayed for significantly enhanced, by 3- to 20-fold, by oxacillin at sub-MIC 1518 NOTES ANTIMICROB.AGENTS CHEMOTHER.

lide resistance gene, and the strain was too sensitive to fusidic acid. In conclusion, subinhibitory concentrations of oxacillin en- hanced PVL levels by all the isolates through PVL promoter activation as previously observed for S. aureus alpha-hemolysin (21). How oxacillin enhances luk-PV transcription remains to be determined. We could hypothesize the involvement of SOS pathway stimulation by ␤-lactams (16) and those of response regulatory pathways engaged in peptidoglycan synthesis (7, 12). By contrast, subinhibitory concentrations of clindamycin, linezolid, and fusidic acid significantly reduced PVL release. This was not explained by the impact of these antibiotics on bacterial growth because PVL was detectable at the cell den- sity achieved (data not shown). These antibiotics have previ- ously been shown to reduce the production of several other toxins (2, 5, 6, 8, 25, 26, 28), possibly through their impact on bacterial protein synthesis and transcription (11, 17). These data showing that subinhibitory antibiotic concentra- tions can either up-regulate or down-regulate PVL release by S. aureus may have therapeutic implications. It provides a log- ical basis for future studies to examine whether linezolid, clin- damycin, or fuscidic acid administration could improve the outcome of severe infections due to PVL-producing S. aureus strains.

We thank Christine Courtier, Christine Cardon, Ce´line Spinelli, Caroline Bouveyron, Martine Rougier, Annie Martra, and Florence Couzon for their technical advice and David Young for editorial guidance. The laboratory received a research grant from Pfizer.

REFERENCES 1. Ausubel, F., R. Brent, R. Kingston, B. Moore, J. Seidman, J. Smith, and K. Struhl. 1987. Current protocols in molecular biology. Wiley Interscience, New York, NY. 2. Bernardo, K., N. Pakulat, S. Fleer, A. Schnaith, O. Utermohlen, O. Krut, S. Muller, and M. Kronke. 2004. Subinhibitory concentrations of linezolid reduce Staphylococcus aureus virulence factor expression. Antimicrob. Agents Chemother. 48:546–555. 3. Bronner, S., P. Stoessel, A. Gravet, H. Monteil, and G. Prevost. 2000. Vari- able expressions of Staphylococcus aureus bicomponent leucotoxins semi- quantified by competitive reverse transcription-PCR. Appl. Environ. Micro- FIG. 2. Variation of lukS-PV lukF-PV gene transcription induced biol. 66:3931–3938. 4. Cars, O. 1990. Pharmacokinetics of antibiotics in tissues and tissue fluids: a by subinhibitory concentrations of antibiotics. S. aureus LUG1124 con- review. Scand. J. Infect. Dis. Suppl. 74:23–33. taining a plasmid-carried luk-PV promoter-lacZ fusion was grown dur- 5. Coyle, E. A., R. Cha, and M. J. Rybak. 2003. Influences of linezolid, peni- ing 24 h at 37°C with or without one-eighth, one-quarter, and one-half cillin, and clindamycin, alone and in combination, on streptococcal pyrogenic the MIC of oxacillin, vancomycin, and linezolid and assayed for ␤-ga- exotoxin A release. Antimicrob. Agents Chemother. 47:1752–1755. lactosidase activity. ␤-Galactosidase activity is expressed as a ratio of 6. Dickgiesser, N., and U. Wallach. 1987. Toxic shock syndrome toxin-1 (TSST- arbitrary units per milligram of bacterial protein cultured with the 1): influence of its production by subinhibitory antibiotic concentrations. indicated concentration of antibiotic by arbitrary units per milligram of Infection 15:351–353. bacterial protein cultured without antibiotic. Values are means Ϯ 7. Gardete, S., S. W. Wu, S. Gill, and A. Tomasz. 2006. Role of VraSR in antibiotic resistance and antibiotic-induced stress response in Staphylococcus standard deviations (three different experiments). *, statistically dif- aureus. Antimicrob. Agents Chemother. 50:3424–3434. ferent from control (LUG1124 grown without antibiotics), with a P 8. Gemmell, C. G., and C. W. Ford. 2002. Virulence factor expression by value of Ͻ0.05, by one-way analysis of variance followed by a posteriori gram-positive cocci exposed to subinhibitory concentrations of linezolid. J. Dunnett’s test. Antimicrob. Chemother. 50:665–672. 9. Genestier, A. L., M. C. Michallet, G. Prevost, G. Bellot, L. Chalabreysse, S. Peyrol, F. Thivolet, J. Etienne, G. Lina, F. M. Vallette, F. Vandenesch, and L. Genestier. 2005. Staphylococcus aureus Panton-Valentine leukocidin di- concentrations, reflecting luk-PV promoter activation. It was rectly targets mitochondria and induces Bax-independent apoptosis of hu- higher than expected with ELISA quantification, but we used a man neutrophils. J. Clin. Investig. 115:3117–3127. ␤ 10. Gillet, Y., B. Issartel, P. Vanhems, J. C. Fournet, G. Lina, M. Bes, F. nonlinear luminometric method to quantify -galactosidase Vandenesch, Y. Piemont, N. Brousse, D. Floret, and J. Etienne. 2002. Asso- activity. By contrast, LacZ expression was strongly reduced by ciation between Staphylococcus aureus strains carrying gene for Panton- linezolid at one-half the MIC, indicating repression of the Valentine leukocidin and highly lethal necrotising pneumonia in young im- munocompetent patients. Lancet 359:753–759. luk-PV promoter, and was not modified by lower concentra- 11. Herbert, S., P. Barry, and R. P. Novick. 2001. Subinhibitory clindamycin tions of linezolid or vancomycin. Clindamycin and fusidic acid differentially inhibits transcription of exoprotein genes in Staphylococcus aureus. Infect. Immun. 69:2996–3003. were not examined upon luk-PV transcription, because the 12. Kuroda, M., K. Kuwahara-Arai, and K. Hiramatsu. 2000. Identification of plasmid carrying the PVL::lacZ fusion also harbored a macro- the up- and down-regulated genes in vancomycin-resistant Staphylococcus VOL. 51, 2007 NOTES 1519

aureus strains Mu3 and Mu50 by cDNA differential hybridization method. 23. Ramdani-Bouguessa, N., M. Bes, H. Meugnier, F. Forey, M. E. Reverdy, G. Biochem. Biophys. Res. Commun. 269:485–490. Lina, F. Vandenesch, M. Tazir, and J. Etienne. 2006. Detection of methi- 13. Labandeira-Rey, M., F. Couzon, S. Boisset, E. L. Brown, M. Bes, Y. Benito, cillin-resistant Staphylococcus aureus strains resistant to multiple antibiotics E. M. Barbu, V. Vazquez, M. Ho¨o¨k, J. Etienne, F. Vandenesch, and M. G. and carrying the Panton-Valentine leukocidin genes in an Algiers hospital. Bowden. 2007. Staphylococcus aureus Panton Valentine leukocidin causes Antimicrob. Agents Chemother. 50:1083–1085. necrotizing pneumonia. Science, in press. 24. Said-Salim, B., B. Mathema, K. Braughton, S. Davis, D. Sinsimer, W. 14. Lina, G., Y. Piemont, F. Godail-Gamot, M. Bes, M. O. Peter, V. Gauduchon, Eisner, Y. Likhoshvay, F. R. Deleo, and B. N. Kreiswirth. 2005. Differential F. Vandenesch, and J. Etienne. 1999. Involvement of Panton-Valentine distribution and expression of Panton-Valentine leucocidin among commu- leukocidin-producing Staphylococcus aureus in primary skin infections and nity-acquired methicillin-resistant Staphylococcus aureus strains. J. Clin. Mi- pneumonia. Clin. Infect. Dis. 29:1128–1132. crobiol. 43:3373–3379. 15. Lina, G., F. Vandenesch, and J. Etienne. 2006. A brief history of Staphylo- 25. Schlievert, P. M., and J. A. Kelly. 1984. Clindamycin-induced suppression of coccus aureus Panton Valentine leucocidin. In V. L. Yu (ed.), Antimicrobial toxic-shock syndrome-associated exotoxin production. J. Infect. Dis. 149:471. therapy and vaccines: microbes, 2nd ed., vol. I. ESun Technologies, LLC, 26. Sriskandan, S., A. McKee, L. Hall, and J. Cohen. 1997. Comparative effects Pittsburgh, PA. of clindamycin and ampicillin on superantigenic activity of Streptococcus 16. Miller, C., L. E. Thomsen, C. Gaggero, R. Mosseri, H. Ingmer, and S. N. pyogenes. J. Antimicrob. Chemother. 40:275–277. Cohen. 2004. SOS response induction by beta-lactams and bacterial defense 27. Vandenesch, F., T. Naimi, M. C. Enright, G. Lina, G. R. Nimmo, H. against antibiotic lethality. Science 305:1629–1631. Heffernan, N. Liassine, M. Bes, T. Greenland, M. E. Reverdy, and J. Etienne. 17. Mukhtar, T. A., and G. D. Wright. 2005. Streptogramins, oxazolidinones, and 2003. Community-acquired methicillin-resistant Staphylococcus aureus car- other inhibitors of bacterial protein synthesis. Chem. Rev. 105:529–542. rying Panton-Valentine leukocidin genes: worldwide emergence. Emerg. In- 18. Naimi, T. S., K. H. LeDell, K. Como-Sabetti, S. M. Borchardt, D. J. Boxrud, fect. Dis. 9:978–984. J. Etienne, S. K. Johnson, F. Vandenesch, S. Fridkin, C. O’Boyle, R. N. 28. van Langevelde, P., J. T. van Dissel, C. J. Meurs, J. Renz, and P. H. Danila, and R. Lynfield. 2003. Comparison of community- and health care- Groeneveld. 1997. Combination of flucloxacillin and gentamicin inhibits toxic associated methicillin-resistant Staphylococcus aureus infection. JAMA 290: shock syndrome toxin 1 production by Staphylococcus aureus in both loga- 2976–2984. rithmic and stationary phases of growth. Antimicrob. Agents Chemother. 19. Narita, S., J. Kaneko, J. Chiba, Y. Piemont, S. Jarraud, J. Etienne, and Y. 41:1682–1685. Kamio. 2001. Phage conversion of Panton-Valentine leukocidin in Staphy- 29. Voyich, J. M., M. Otto, B. Mathema, K. R. Braughton, A. R. Whitney, D. lococcus aureus: molecular analysis of a PVL-converting phage, phiSLT. Welty, R. D. Long, D. W. Dorward, D. J. Gardner, G. Lina, B. N. Kreiswirth, Gene 268:195–206. and F. R. DeLeo. 2006. Is Panton-Valentine leukocidin the major virulence 20. National Committee for Clinical Laboratory Standards. 2004. Performance determinant in community-associated methicillin-resistant Staphylococcus standards for antimicrobial susceptibility testing. Approved standard M100- aureus disease? J. Infect. Dis. 194:1761–1770. S14. National Committee for Clinical Laboratory Standards, Wayne, PA. 30. Wannet, W. J., M. E. Heck, G. N. Pluister, E. Spalburg, M. G. van Santen, 21. Ohlsen, K., W. Ziebuhr, K. P. Koller, W. Hell, T. A. Wichelhaus, and J. X. W. Huijsdans, E. Tiemersma, and A. J. de Neeling. 2004. Panton-Valen- Hacker. 1998. Effects of subinhibitory concentrations of antibiotics on alpha- tine leukocidin positive MRSA in 2003: the Dutch situation. Eur. Surveill. toxin (hla) gene expression of methicillin-sensitive and methicillin-resistant 9:28–29. Staphylococcus aureus isolates. Antimicrob. Agents Chemother. 42:2817– 31. Ward, P. D., and W. H. Turner. 1980. Identification of staphylococcal 2823. Panton-Valentine leukocidin as a potent dermonecrotic toxin. Infect. Im- 22. Peng, H. L., R. P. Novick, B. Kreiswirth, J. Kornblum, and P. Schlievert. mun. 28:393–397. 1988. Cloning, characterization, and sequencing of an accessory gene regu- 32. Woodin, A. M. 1959. Fractionation of a leucocidin from Staphylococcus lator (agr)inStaphylococcus aureus. J. Bacteriol. 170:4365–4372. aureus. Biochem. J. 73:225–237. BMJ 2012;344:e2400 doi: 10.1136/bmj.e2400 (Published 11 April 2012) Page 1 of 4

Endgames

ENDGAMES

CASE REPORT

Fever and rash in a returning traveller

1 2 Neelam Kumar core medical trainee , David J Lewis professor of infectious diseases

1Clinical Infection Unit, St George’s Hospital, London SW17 0QT, UK; 2St George’s Vaccine Institute, St George’s University of London, London, UK

A previously fit and well 26 year old male student of Chinese origin attended the emergency department with a five day history Answers of a febrile illness. He had returned to the United Kingdom from a three week visit to Malaysia and Singapore the day before the 1 What are the most likely clinical diagnosis illness began. His main symptoms were fever, headache, and and important differential diagnosis? nausea. He had also noted a faint widespread rash the day before Short answer admission. On the basis of the history and clinical picture, the diagnosis is On examination his temperature was 39.7°C, his blood pressure probably dengue infection. Malaria is an important differential was 134/77 mm Hg, and he had a regular heart rate of 75 diagnosis and should be excluded with serial blood film beats/min. He had a respiratory rate of 14 breaths/min and examination or direct antigen testing (or both). oxygen saturations of 98% on room air. The only relevant clinical finding was a confluent erythematous macular rash Long answer across his chest, back, arms, and legs, with multiple small discrete areas of sparing. Dengue infection classically presents with an abrupt onset of fever, headache, severe myalgia (so called breakbone fever), Full blood count showed haemoglobin 163 g/L (120-180), and arthralgia. A rash occurs in about half of cases and is platelets 86×109/L (130-400×109), white cell count (WCC) 9 9 9 9 typically faint, erythematous, and macular (figure), with discrete 3.4×10 /L (4.5-11.0×10 ), neutrophils 1.7×10 /L (2.0-7.5×10 ; areas of sparing (white islands in a sea of red). Nausea, vomiting, 40-75% WCC), lymphocytes 1.4×109/L (1.3-3.5×109; 20-45% 9 abdominal pain, and minor mucosal bleeding are also common. WCC), and no eosinophilia (0.04-0.44×10 /L; 1-6% WCC). A The tourniquet test is a clinical test of capillary fragility that blood film was negative for malaria parasites and confirmed can help make the clinical diagnosis of dengue infection; it has thrombocytopenia with atypical lymphocytes. Renal function, a specificity of 84-91% but a sensitivity of only 33-34%.1 The liver function, and clotting tests were unremarkable and C test is performed by inflating a blood pressure cuff above venous reactive protein was less than 4 mg/L. pressure for five minutes and then deflating it. The test is Questions positive if there are more than two petechiae distal to the cuff. Typical laboratory abnormalities include leucopenia with a 1 What are the most likely clinical diagnosis and important relative lymphocytosis and thrombocytopenia. Liver function differential diagnosis? may be mildly deranged. If the disease progresses, the 2 How would you confirm this diagnosis? haematocrit value may increase, the albumin concentration may 3 What are the potential complications of this disease? fall, and clotting abnormalities may develop. Most infections are mild and self limiting, however. The incubation period is 4 How is this disease treated? three to 14 days (typically five to seven), and it is endemic to 5 How can this disease be prevented? South East Asia, the western Pacific, and the Americas.

Correspondence to: N Kumar [email protected]

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;344:e2400 doi: 10.1136/bmj.e2400 (Published 11 April 2012) Page 2 of 4

ENDGAMES

be admitted to hospital and closely monitored for progression to severe dengue. It is unclear why severe dengue occurs in some cases but not in others. The most popular theory is that of antibody dependent enhancement.11 There are four dengue serotypes and it is thought that primary infection with one serotype leads to transient protection against all serotypes. As levels of neutralising antibodies fall, infection with a different serotype may result in non-neutralising antibodies forming a complex with the virus. Such complexes may lead to enhanced phagocytosis by mononuclear cells, enhanced replication, and an increase in viral load and cytokine production. Primary dengue infection can result in severe disease, however, and disease severity in secondary infections varies greatly. The pathogenesis of severe dengue is complex, encompassing factors related to the both host and the pathogen. Several other diseases present in a similar manner, including malaria, chikungunya, rickettsial infections, enteric fever, and 4 How is this disease treated? acute HIV seroconversion. Non-infectious causes for the rash such as drug reactions should also be considered. Short answer No treatment is available for dengue infection, only supportive 2 How would you confirm this diagnosis? measures. Short answer Dengue infection can be confirmed by the detection of antiviral Long answer IgM antibodies, the detection of specific nucleic acid using No effective treatment is available for dengue infection. reverse transcriptase polymerase chain reaction, viral isolation Treatment is based on assessing patients for the presence of in cell culture, or antigen detection. warning signs or severe dengue and admission and supportive treatment if present. The warning signs reflect the risk of Long answer entering a critical phase—characterised by an increase in capillary permeability and plasma leakage—after the acute In the UK dengue infection is most commonly confirmed using febrile phase (box 2). This usually occurs around the third or tests for the detection of IgM antibodies against the flavivirus. fourth day of the infection. Some patients may then enter a IgM is detectable on the fifth day of illness in 80% of people recovery phase, during which fluid is gradually reabsorbed from with dengue fever and by the sixth day in 93-99%. Such tests the extravascular space; fluid overload is possible if fluid are cheap, widely available, and have a sensitivity and specificity resuscitation has been excessive. Other patients may progress of more than 93% if performed on or after the fifth day.2 Results to severe dengue with shock from plasma leakage, haemorrhage are typically available in 48-72 hours, depending on laboratory from disseminated intravascular coagulation, or organ organisation. In the first few days the sensitivity of reverse dysfunction from hypoperfusion. Such patients should be transcriptase polymerase chain reaction assays for viral nucleic managed in a high dependency setting. acid is higher, although the test is more expensive and less widely available in UK hospitals.3 The virus can also be isolated During the recovery phase, discharge can be considered once the patient has been afebrile for 48 hours and clinical status and in cell culture; this test is highly specific but of limited 10 availability. Rapid antigen tests have been developed to detect laboratory indices have improved. In the UK and Europe, most cases are mild and limited to the acute febrile phase, especially the viral glycoprotein non-structural protein 1; these tests are 12 13 highly sensitive and specific, cheap, and provide almost in adults. In such cases, outpatient treatment may be immediate results.4 5 Currently they are not widely available in considered if a clinical diagnosis of dengue is made and other NHS trusts. serious differential diagnoses, such as malaria, are excluded. Criteria for outpatient treatment include the absence of any 3 What are the potential complications of this warning sign or serious comorbidity (including pregnancy) in disease? a patient between the ages of 16 and 65 years who can be discharged under the supervision of a responsible adult.10 Short answer Dengue infection can progress to severe dengue fever 5 How can this disease be prevented? complicated by shock, haemorrhage, or organ dysfunction. Short answer Long answer No effective vaccine is available, and prevention is aimed at controlling and avoiding exposure to the mosquito vector Aedes On the basis of a seminal cohort study of children in Thailand, aegypti. complications of dengue infection have traditionally been classified as dengue fever, dengue haemorrhagic fever, and Long answer dengue shock syndrome.6 7 More recent studies have shown that such case definitions underdiagnose cases of severe dengue The virus is transmitted by the mosquito Aedes aegypti and infection.8 9 In view of this, the World Health Organization prevention is aimed at controlling and avoiding this vector. It published updated guidelines in 2009 that advise classifying breeds in stagnant pools of water in artificial containers abundant the disease as dengue infection with or without warning signs in household rubbish, so it thrives particularly well in busy and severe dengue (box 1).10 Patients with warning signs should cities. Travellers to endemic areas should be advised of the risk of exposure and symptoms of dengue infection and to take

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;344:e2400 doi: 10.1136/bmj.e2400 (Published 11 April 2012) Page 3 of 4

ENDGAMES

Box 1 World Health Organization suggested case classification of dengue 200910 Probable dengue Patient lives in or has recently travelled to a dengue endemic area Fever and two of the following criteria: • Nausea • Rash • Aches and pains • Positive tourniquet test • Leucopenia • Any warning sign

Warning signs Abdominal pain and tenderness Persistent vomiting Clinical fluid accumulation Mucosal bleed Lethargy, restlessness Liver enlargement >2 cm Laboratory: increase in the haematocrit value with a rapid decrease in platelet count

Severe dengue Severe plasma leakage leading to: • Shock • Fluid accumulation with respiratory distress Severe bleeding Severe organ involvement: • Liver: aspartate aminotransferase or alanine aminotransferase ≥100 U/L • Central nervous system: impaired consciousness • Heart and other organs

Box 2 Phases of dengue infection and their associated clinical problems10 Febrile phase (days 1-3) Dehydration; high fever may cause neurological disturbances and febrile seizures in young children

Critical phase (days 3-6) Shock from plasma leakage, severe haemorrhage, organ impairment

Recovery phase (days 6-10) Hypervolaemia (only if intravenous fluid therapy has been excessive or extended into this period) necessary precautions. Insect repellents and protective clothing (ELISA) test. The patient made a full recovery with no long are effective means of reducing mosquito bites. Aedes term sequelae. mosquitoes prefer to feed in the early morning and late afternoon. Travellers should therefore be reminded that Competing interests: Both authors have completed the ICMJE uniform mosquitoes that transmit dengue generally bite during the day, disclosure form at www.icmje.org/coi_disclosure.pdf (available on whereas those that transmit malaria bite between dusk and dawn. request from the corresponding author) and declare: no support from Antimosquito precautions may need to be used day and night any organisation for the submitted work; no financial relationships with in areas where the transmission of these two diseases overlaps. any organisations that might have an interest in the submitted work in Currently, there is no effective vaccine against dengue infection, the previous three years; no other relationships or activities that could although advanced clinical trials of a recombinant live vaccine appear to have influenced the submitted work. based on a yellow fever virus backbone are under way.14 Patient consent obtained. Provenance and peer review: Not commissioned; externally peer Patient outcome reviewed.

A presumptive diagnosis of dengue infection was made on the 1 Mayxay M, Phetsouvanh R, Moore CE, Chansamouth V, Vongsouvath M, Sisouphone basis of the clinical picture. The main differential diagnosis of S, et al. Predictive diagnostic value of the tourniquet test for the diagnosis of dengue infection in adults. Trop Med Int Health 2011;16:127-33. malaria was excluded with a negative direct antigen test and 2 Souza VA, Tateno AF, Oliveira RR, Dominguesb RB, Araujo ES, Kuster GW, et al. serial blood film examination. Because the patient had no Sensitivity and specificity of three ELISA-based assays for discriminating primary from secondary acute dengue virus infection. J Clin Virol 2007;39:230-3. warning signs and he lived with a responsible adult he was 3 Kong YY, Thay CH, Tin TC, Devi S. Rapid detection, serotyping and quantitation of dengue discharged home. An outpatient appointment was made for four viruses by TaqMan real-time one-step RT-PCR. J Virol Methods 2006;138:123-30. weeks later. Dengue infection was confirmed retrospectively 4 Dussart P, Labeau B, Lagathu G, Louis P, Nunes MR, Rodrigues SG, et al. Evaluation of an enzyme immunoassay for detection of dengue virus NS1 antigen in human serum. with a positive IgM enzyme linked immunosorbent assay Clin Vaccine Immunol 2006;13:1185-9.

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;344:e2400 doi: 10.1136/bmj.e2400 (Published 11 April 2012) Page 4 of 4

ENDGAMES

5 Libraty DH, Young PR, Pickering D, Endy TP, Kalayanarooj S, Green S, et al. High 10 WHO. Dengue. Guidelines for diagnosis, treatment, prevention and control. New edition. circulating levels of the dengue virus nonstructural protein NS1 early in dengue illness 2009. http://whqlibdoc.who.int/publications/2009/9789241547871_eng.pdf. correlate with the development of dengue hemorrhagic fever. J Infect Dis 2002;186:1165-8. 11 Halstead SB. Neutralization and antibody-dependent enhancement of dengue viruses. 6 WHO. Dengue haemorrhagic fever: diagnosis, treatment, prevention and control. 2nd ed. Adv Virus Res 2003;60:421-67. 1997. www.who.int/csr/resources/publications/dengue/itoviii.pdf. 12 Allwinn R, Hofknecht N, Doerr HW. Dengue in travelers is still underestimated. Intervirology 7 Cohen SN, Halstead SB. Shock associated with dengue infection. I: clinical and physiologic 2008;51:96-100. manifestations of dengue hemorrhagic fever in Thailand, 1964. J Pediatr 1966;68:448-56. 13 Stephenson I, Roper J, Fraser M, Nicholson K, Wiselka M. Dengue fever in febrile returning 8 Srikiathachorn A, Gibbons R, Green S, Libraty DH, Thomas SJ, Endy TP, et al. Dengue travelers to a UK regional infectious diseases unit. Travel Med Infect Dis 2003;1:89-9. haemorrhagic fever: the sensitivity and specificity of the World Health Organisation 14 Guy B, Barrere B, Malinowski C, Saville M, Teyssou R, Lang J. From research to phase definitions for identification of severe cases of dengue in Thailand, 1994-2005. Clin Infect III: preclinical, industrial and clinical development of the Sanofi Pasteur tetravalent dengue Dis 2010;50:1135-43. vaccine. Vaccine 2011;29:7229-41. 9 Phuong CX, Nhan NT, Kneen R, Thuy PT, van Thien C, Nga NT, et al. Clinical diagnosis and assessment of severity of confirmed dengue infections in Vietnamese children: is the world health organization classification system helpful? Am J Trop Med Hyg 2004;70:172-9. Cite this as: BMJ 2012;344:e2400 © BMJ Publishing Group Ltd 2012

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe Journal of Antimicrobial Chemotherapy Advance Access published October 12, 2010

J Antimicrob Chemother doi:10.1093/jac/dkq362

Severity assessment of skin and soft tissue infections: cohort study of management and outcomes for hospitalized patients

Charis Marwick 1*, Janice Broomhall 1, Colin McCowan 1, Gabby Phillips 2, Sebastian Gonzalez-McQuire 3†, Kasem Akhras 4‡, Sanjay Merchant 4, Dilip Nathwani 5 and Peter Davey 1

1Quality, Safety and Informatics Research Group, Division of Clinical and Population Sciences and Education, Mackenzie Building, Kirsty Semple Way, Dundee DD2 4BF, Scotland, UK; 2Department of Medical Microbiology, Ninewells Hospital, Dundee DD1 9SY, Scotland, UK; 3Janssen, 50–100 Holmers Farm Way, High Wycombe, Buckinghamshire HP12 4EG, UK; 4Johnson and Johnson Pharmaceutical Services, LLC, Raritan, NJ, USA; 5Department of Infection and Immunodeficiency, Ninewells Hospital, Dundee DD1 9SY, Scotland, UK

*Corresponding author. Tel: +44-1382-420000; Fax: +44-1382-420010; E-mail: [email protected] †Present address: GlaxoSmithKline, Stockley Park West, Uxbridge UB11 1BU, UK. Downloaded from ‡Present address: Pfizer Oncology Medical Group, Inc., 235 42nd Street, New York, NY 10017, USA.

Received 5 February 2010; returned 22 March 2010; revised 27 August 2010; accepted 28 August 2010

Background: Skin and soft tissue infections (SSTIs) are caused by bacterial invasion of the skin and underlying

soft tissues and can present with a wide spectrum of signs, symptoms and illness severity. They are a common http://jac.oxfordjournals.org/ indication for antimicrobial therapy. However, there are few data on treatment outcomes or the validity of clini- cal severity scores. Methods: Two hundred and five adult patients admitted to Ninewells Hospital, Scotland in 2005, and treated with antibiotics for SSTI, were identified. They were stratified into four classes of severity (class IV¼most severe) based on sepsis, co-morbidity and their standardized early warning score (SEWS). Empirical antimicro- bial therapy by severity class was compared with the recommendations of a UK guideline.

Results: Thirty-five different empirical antimicrobial regimens were prescribed. Overall, 43% of patients were by guest on August 8, 2012 over-treated, this being particularly common in the lowest severity class I (65% patients). Thirty-day mortality was 9% (18/205) and 17 patients (8%) died during their index admission. Mortality (30 day) and inadequate therapy increased with severity class: I, no sepsis or co-morbidity (45% patients, 1% mortality, 14% therapy inadequate); II, significant co-morbidity but no sepsis (32% patients, 11% mortality, 39% therapy inadequate); III, sepsis but SEWS ,4 (17% of patients, 17% mortality, 39% therapy inadequate); and IV, sepsis plus SEWS ≥4 (6% of patients, 33% mortality, 92% therapy inadequate). Conclusions: SSTI in hospital is associated with significant mortality. Choice of empirical therapy is not evidence based, with significant under-treatment of severely ill patients.

Keywords: cellulitis, sepsis, SSTIs, Staphylococcus aureus, staphylococcal infections, MRSA

Introduction community-acquired pneumonia where guidelines are founded on a large evidence base from prospective cohort studies,4 Skin and soft tissue infections (SSTIs) are clinical entities of vari- there is little evidence about the use of clinical scores to assess able presentation, aetiology and severity that involve microbial severity, predict outcome or to guide therapy. invasion of the layers of the skin and underlying soft tissues. Practice guidelines for SSTI from the USA2 and the UK,5 and SSTIs range from mild infections, such as pyoderma, to serious an evidence review of cellulitis and erysipelas,6 did not identify 1 life-threatening infections, such as necrotizing fasciitis. These any cohort studies of management or outcome of patients hos- 2 infections lead to considerable morbidity and are the indication pitalized with SSTI, other than erysipelas or necrotizing fascii- for a significant proportion of antimicrobials prescribed in hospi- tis.7 –11 More recently, one large retrospective cohort study tal. In a point prevalence survey of antibiotic use in 20 hospitals looked at failure of antimicrobial therapy in SSTIs.12 The UK across Europe in 2006, SSTI was the second commonest guidelines (CREST Guidelines on the Management of Cellulitis in indication for antibiotic treatment, after lower respiratory tract Adults)5 advise on appropriate site of care, choice of antimicro- 3 infection (LRTI). However, in comparison with LRTI and bial and route of drug administration, depending on the clinical

# The Author 2010. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]

1of11 Marwick et al.

severity, stratified by Eron score,13 which was devised from the (ii) Patients with the above ICD10 discharge diagnoses but who did not recommendations of an expert panel. have blood cultures taken. To our knowledge, there is no validated severity grading (iii) Patients who had blood cultures taken but did not have the above system that predicts clinical outcome in SSTIs, along the lines SSTI discharge codes (because we did not know how sensitive our of that for community-acquired pneumonia.4 There is no pre- discharge code group would be). viously published clinical study of outcomes in relation to severity (iv) Additional cases were identified opportunistically during screening for another study of patients aged ≥65, who had microbiology samples classification. A new algorithm has recently been proposed to taken during an admission to hospital in Tayside in 2005 with at least assess severity of SSTIs and guide therapy, based on clinical one ICD10 discharge code indicating infection (from a predefined list, presentation, co-morbidity, and the anatomical region and size which included the above SSTI codes). of the affected body area.1 It seems potentially useful but requires further study and validation. The aim of this study was to devise an operational, clarified Data extraction modification of the Eron classification13 and to apply it to a cohort of hospitalized patients with SSTI. The objectives were Case notes were reviewed to extract evidence of SSTI (a written diagnosis to compare antimicrobial management with the recommen- of SSTI in the case notes plus antimicrobial treatment), the data required dations in the UK guideline5 and to assess the relationship to determine illness severity and the prescribed antimicrobial therapy. Co-morbidity data were also extracted to allow calculation of a Charlson between severity (as stratified by our modification of the Eron 16

index for each patient, and to record the presence of diabetes mellitus Downloaded from classification), management and outcomes. because some guidelines list diabetes as a factor that may influence outcome.2 Results of microbiology tests were retrieved from Central Vision, the Methods electronic data management system used in NHS Tayside. These included Ethics statement culture and susceptibility testing of blood and wound swab cultures

taken during the index admission. http://jac.oxfordjournals.org/ Ethics committee approval was not required for this study as data were from electronic and paper hospital records. The protocol was approved by the Caldicott Guardian for hospital inpatients. Data linkages Patient data were linked by unique community health index (CHI) number across three electronic databases, with all other data extracted from hos- Study population and design pital case notes. The study population was residents of Tayside aged ≥18 in 2005 who (i) CHI number database. Patient demographics (age, gender, place of had a new admission to Ninewells Hospital between 1 January and residence, socio-economic status) were extracted. 31 December 2005. The design was a retrospective cohort study of by guest on August 8, 2012 (ii) Scottish Morbidity Record set of databases. Those relevant to this patients who received antibiotic treatment for SSTI while in hospital. study were the SMR 01 (General acute inpatient and day case dis- Patients for whom treatment of the SSTI was not the primary reason charges) and GRO(S) (death registrations) databases. for admission (e.g. admitted due to co-existing illness or difficult social (iii) Central Vision data management system of all laboratory tests per- circumstances) were included. All hospitalizations of patients aged ≥18 formed in NHS Tayside. Microbiology data were extracted from were examined from the Scottish Morbidity Record of hospital admissions Central Vision. Details included date of test, sample type, anatomical (SMR01) database over the period January–December 2005 along with site and results of culture and susceptibility testing. records from other linked datasets.

Case identification Case definitions for severity classification The original aim was to extract clinical data from a balanced sample of Severity classification in the study was based on a modification of the 200 cases of SSTI, made up of 100 patients with blood cultures taken Eron classification, which was used in the CREST guidelines. It was felt, plus 100 patients without. Previous studies suggest that patients with as has been previously recognized, that the descriptions of clinical pre- 1 blood cultures taken are more likely to have sepsis and worse outcome sentations in the Eron classification system are ambiguous. This could compared with those patients without blood cultures.14,15 It was make classification of a patient’s illness severity unclear and subjective expected, therefore, that including 100 of each would provide enough so, for this study, the definitions of severity class were modified to patients, distributed across the severity classes, to allow comparison of include only objective criteria that made the classes mutually exclusive. their treatment and outcomes. The ICD10 discharge code group used An internationally recognized definition of sepsis, consisting of the for SSTI was based on an ICD9 code list used in a previous study con- systemic inflammatory response syndrome, SIRS (¼ two or more of the 3 ducted by Johnson and Johnson, and is also similar to an ICD9 code following criteria: white blood cell count ,4or.12/mm , temperature list in a published study.12 These included discharge codes for post- ,36 or .388C, heart rate .90 beats/min, respiratory rate .20 operative infections, but not for necrotizing fasciitis (M72.6), in line with breaths/min), plus infection was used in the modified severity classifi- 17 a previous study,12 as this is often considered a distinct clinical entity. cation. Poorly defined, outdated clinical descriptions such as the 17 Patients in the study were identified from four mutually exclusive ‘sepsis syndrome’, which defined Eron class IV, were excluded from groups. For groups (i)–(iii), random samples from all patients in the data- the study definitions. ‘Sepsis syndrome’ is not a term used in medical base who met the criteria were selected for case-note review. documentation in the study hospital, and does not have a specific defi- nition, so an alternative definition for class IV was devised. A standar- (i) Patients with a study discharge diagnosis for SSTI (ICD10 codes L03X, dized early warning score (SEWS)18 of ≥4 in a patient with sepsis was L030–3, L038–9, L97X, M608, M860–9, R02X, S913, T131, T814, used to indicate the most severely unwell patients (class IV). The SEWS T874) who had blood cultures taken. is calculated, on the patient’s routine clinical observations chart, by

2of11 Management and outcomes of SSTI in hospital JAC

Table 1. SEWS parameters and scoring systems (adapted from Jorup-Ronstrom et al.20)

Score

Parameter 3 210123

Respiratory rate (breaths/min) ≤8 9–20 21–30 31–35 ≥36 Oxygen saturation (%) ,85 85–89 90–92 ≥93 Temperature (8C) ,34 34–34.9 35–35.9 36–37.9 38–38.4 ≥38.5 Systolic blood pressure (mmHg) ≤69 70–79 80–99 100–199 ≥200 Heart rate (beats/min) ≤29 30–39 40–49 50–99 100–109 110–129 ≥130 AVPU response (stimulus required to induce response) unresponsive pain verbal alert

AVPU, alert, verbal, pain, unresponsive (category of stimulus required to generate patient response). Case example, patient with SSTI and sepsis: respiratory rate¼26; oxygen saturation¼94%; temperature¼38.4; blood pressure¼84/62; heart rate¼107; AVPU response¼alert; SEWS¼4; and urgent medical review is mandated. Downloaded from summing any ‘scores’ allocated for abnormal clinical measurements regimen for that severity class, based on the most likely causative (Table 1). A cut-off of ≥4 was chosen as this score mandated urgent organisms. medical review of the patient, across the study NHS trust, and has Class I: appropriate¼oral therapy active against Streptococcus pyo- been demonstrated to predict in-hospital mortality and length of stay genes and Staphylococcus aureus; over-treatment¼any intravenous http://jac.oxfordjournals.org/ for general medical patients.18 therapy OR oral therapy with unnecessarily broad-spectrum cover; and under-treatment¼oral therapy not sufficiently active against S. pyogenes and S. aureus. Definitions of severity classes used in CREST Class II: appropriate¼intravenous therapy active against S. pyogenes and in the study and S. aureus; over-treatment¼intravenous therapy with unnecessarily broad-spectrum cover; and under-treatment¼any oral therapy OR intra- Class I: CREST definition, patients have no signs of systemic toxicity and venous therapy that is not sufficiently active against S. pyogenes and no uncontrolled co-morbidities; and study definition, no recorded signifi- S. aureus. cant co-morbidity (peripheral vascular disease, chronic venous insuffi- Class III: appropriate¼intravenous therapy active against S. pyogenes

, by guest on August 8, 2012 ciency or morbid obesity), no sepsis and SEWS 4. and S. aureus; over-treatment¼intravenous therapy with unnecessarily Class II: CREST definition, patients are either systemically ill or systemi- broad-spectrum cover; and under-treatment¼any oral therapy OR cally well but with co-morbidity such as peripheral vascular disease, intravenous therapy that is not sufficiently active against S. pyogenes chronic venous insufficiency or morbid obesity, which may complicate and S. aureus. or delay resolution of their infection; and study definition, documentation Class IV: appropriate¼intravenous therapy active against S. pyogenes, of one or more significant co-morbidities (peripheral vascular disease, S. aureus, Gram-negative bacteria and anaerobic bacteria that included a chronic venous insufficiency or morbid obesity) but no sepsis and drug that reduces toxin production by S. pyogenes (clindamycin or linezo- , SEWS 4. lid);2 and under-treatment¼any oral therapy OR intravenous therapy not Class III: CREST definition, patients may have a significant systemic sufficiently active against the full range of potential pathogens and their upset such as acute confusion, tachycardia, tachypnoea or hypotension toxins.2 or may have unstable co-morbidities that may interfere with response Appropriateness of first-line antimicrobial therapy by clinical severity, to therapy or have a limb-threatening infection due to vascular compro- using the above rationale, was assessed independently by two Infectious , mise; and study definition, sepsis but SEWS 4. Diseases specialists (P. D. and C. M.) and agreement on each item of data Class IV: CREST definition, patients have sepsis syndrome or severe was recorded. Disagreements were resolved by discussion. life-threatening infection such as necrotizing fasciitis; and study defi- In addition to clinical appropriateness, the appropriateness of anti- ≥ nition, sepsis with SEWS 4. microbial therapy according to the results of microbiological tests was classified. This analysis was restricted to the organisms implicated as the cause of the SSTI. First, only isolates from blood culture or from Classification of antibiotic treatment swabs of infected wounds, including ulcers, were considered. Secondly, The initial antimicrobial therapy prescribed for each patient was com- only samples taken within 7 days before to 7 days after starting anti- pared with the recommendations of the CREST Guidelines on the Man- microbial therapy for the SSTI were included. Thirdly, only clinically signifi- agement of Cellulitis in Adults, published by the Department of Health cant isolates were included, defined as any organism other than and Social Security (DHSS), Northern Ireland (June 2005).5 These guide- coagulase-negative staphylococci or diphtheroids from blood cultures, lines are based on the Eron classification of severity13 and are used and only S. aureus or streptococci from wound swabs. across the UK. They were in use in the study hospital in 2005. In this Full susceptibility data for implicated organisms (meeting the above study, the modified severity classes I–IV (above) were used in place of criteria) were obtained from Medical Microbiology. The results on Eron severity classes I–IV. Therapy was classified as appropriate, under- Central Vision are reported selectively; for example, the susceptibility of treatment or over-treatment according to the recommendations of the S. aureus to ciprofloxacin is not reported for all isolates on Central CREST guidelines. Where the specific antimicrobials prescribed did not Vision. Initial antimicrobial therapy was coded as microbiologically appro- feature anywhere in the CREST guidelines, the following rationale was priate if all significant bacterial isolates from a patient were susceptible to used to classify the appropriateness of the prescribed antibiotic at least one of the antimicrobials in their initial therapy.

3of11 Marwick et al.

Data analysis Table 2. Patient demography for 205 hospital inpatients with SSTI by The distribution of continuous variables was checked for skewness and, blood culture status when appropriate, mean and standard deviations were reported. Other- wise, median and inter-quartile ranges were stated. Differences in pro- SSTI with no blood SSTI and blood P value portions between groups were examined using the appropriate x2 tests culture, N (%) culture, N (%) (x2 test) or tests for trend. Differences in continuous variables used independent Student’s t-test or ANOVA, after ensuring any assumptions had been No. of patients 107 98 met. Inter-rater reliability between the Infection specialists for assigning appropriateness of first-line therapy by infection severity was examined Age group with the two-way kappa test.19 Demographic and clinical variables, ,40 10 (9) 11 (11) 0.811 along with severity class, were evaluated for effect on 30 day mortality, 40–49 13 (12) 7 (7) using univariate and multivariable logistic regression. All analyses were 50–59 16 (15) 13 (13) performed using Stata version 9.0 or Microsoft Excel. 60–69 17 (16) 14 (14) 70–79 22 (21) 25 (26) ≥80 29 (27) 28 (29) Results Male 47 (44) 46 (47) 0.665 Case identification Female 60 (56) 52 (53) Downloaded from In total, 205 valid cases of SSTI were identified from the four Died in hospital 9 (8) 8 (8) 0.949 mutually exclusive patient groups described above. Eighty-two Died within 30 daysa 11 (10) 7 (7) 0.428 (76%) of 108 patients screened in the group with SSTI discharge Died within 90 daysa 14 (13) 14 (14) 0.802 codes and blood cultures taken were valid cases. A similar b number were identified from the group with SSTI discharge SIMD codes but without blood cultures taken, where 84 (81%) of 1 (most affluent) 12 (12) 17 (20) 0.360 http://jac.oxfordjournals.org/ 104 cases screened were valid. Screening of 112 sets of case 2 29 (30) 16 (19) notes from the third group, without SSTI discharge codes but 3 17 (17) 18 (21) with blood cultures taken, only identified seven (6%) valid 4 20 (20) 16 (19) cases. Thirty-two additional cases, meeting the study definition 5 (most deprived) 20 (20) 19 (22) for SSTI, were opportunistically identified during screening of Charlson index case notes for another study. Five of these patients had ICD10 0 48 (45) 37 (38) 0.386 codes in the study SSTI code group and 13 had other codes indi- 1–2 39 (36) 45 (46)

cating infection or inflammation of the skin, not included in our by guest on August 8, 2012 3+ 20 (19) 16 (16) original SSTI group. Fourteen had no discharge codes indicating SSTI, but six of these had blood cultures taken and two had SIMD, Scottish index of multiple deprivation. non-site-specific infection codes, for e.g. infection following a a procedure. Death within 30 or 90 days was calculated from the start date of anti- biotic therapy for the 189 patients with full antibiotic prescription data Of the final 205 valid cases of SSTI, 107 had no blood culture available, and from the date of admission for the remaining 16 patients. and 98 had one or more blood cultures taken (Table 2). Full data bSIMD data were available for 98 patients with no blood culture and for about the initial antimicrobial regimen were not available for 16 86 patients with blood cultures. patients so analysis of the appropriateness of treatment was done on 189 patients. Antibiotic treatment was started in the first 2 days of hospitalization in 161 (85%) of these patients. In trauma surgery (n¼16, 8%) or specialist surgical wards (n¼20, the remaining patients antibiotic treatment was started a 10%). Length of stay ranged from 1 to 226 days with a highly median of 10 days after admission, range 3–75 days. skewed distribution: mean 18 days, median 2 days. Of the 205 study patients, 92 (45%) patients were in severity class I, 66 (32%) in class II, 35 (17%) in class III and 12 (6%) in Patient demographics and clinical description class IV. Unadjusted mortality at 30 days significantly increased There were no statistically significant differences in demographic with severity class but readmission did not vary. Crude 30 day characteristics between patients with and without blood cultures mortality rates were 1% (1/92) in class I, 11% (7/66) in class (Table 2). In-hospital mortality was 8% (17/205) for the cohort, II, 17% (6/35) in class III and 33% (4/12) in class IV (x2 test, and 30 day mortality was 9% (18/205). There was no significant P,0.001). difference in mortality between those patients with and without Thirty-eight patients (19%) had diabetes mellitus. In compari- blood cultures taken (Table 2). son with class I patients, there were substantially more patients The majority of patients (81%) were admitted from their own with diabetes in class II (28/66, 42% compared with 2/92, 2%). home. Others were admitted from residential or nursing homes However, the proportion of diabetics in classes I and II combined (11%) or transferred from other hospitals (5%). The source of (30/158, 19%) was similar to the proportion in classes III and IV admission for five patients (2%) was unknown. Most patients combined (8/47, 17%). Logistic regression, with diabetes as an were admitted to General Medicine (n¼100, 49%) or General independent risk factor, did not show any significant effect of Surgery (n¼51, 25%). The remaining patients were admitted diabetes on 30 day mortality, either unadjusted (odds ratio to specialist medical wards (n¼28, 14%) orthopaedics and 1.13, 95% CI 0.98–1.31), or adjusted for age, sex, Charlson

4of11 Management and outcomes of SSTI in hospital JAC

index of co-morbidity, severity class and treatment (odds ratio whereas 33% (4/12) of patients in class IV were initially 1.13, 95% CI 0.92–1.40). treated with flucloxacillin alone.

Antimicrobial therapy Inter-rater agreement on appropriateness The 189 patients for whom full prescription data were available received a total of 35 different antibiotic regimens (45 if route (by CREST guidelines) of administration is also considered), resulting in 79 different Agreement between raters about the appropriateness of each combinations of severity class and antibiotic regimen. Most combination of antibiotic regimen and severity class was 85% patients (133, 70%) received a b-lactam antibiotic. The initial (67/79), two-way kappa¼0.61 (95% CI 0.41–0.80, P,0.01) indi- antibiotic regimen was monotherapy in 143 (76%), two drugs cating substantial agreement. In the 12 instances of disagree- in 43 (23%) and three drugs in 3 (2%) patients. Full details of ment, three were under-treatment versus appropriate, and all regimens prescribed, by severity class, are in Table 3.In nine were over-treatment versus appropriate. The three regimens general, the spectrum of initial therapy increased with severity. considered under-treatment, for that severity class, by one rater However, only 52 (29%) of 177 patients in classes I–III received were: intravenous erythromycin for class II, clindamycin and flucloxacillin alone as recommended in the CREST guidelines,5 metronidazole for class IV, and intravenous clarithromycin for Downloaded from

Table 3. Details of all prescribed antibiotic regimens by severity class, including final consensus decisions on appropriateness and whether there was any initial disagreement on appropriateness

No. of Consensus Initial patients Antibiotic 1 Route 1 Antibiotic 2 Route 2 Antibiotic 3 Route 3 appropriate disagreement http://jac.oxfordjournals.org/

Severity class I 5 clindamycin po appropriate 0 2 flucloxacillin po penicillin V po appropriate 0 12 flucloxacillin po appropriate 0 9 benzylpenicillin iv flucloxacillin iv over 0 1 ceftazidime iv metronidazole iv over 0 2 ceftriaxone iv over 0

2 cefuroxime iv metronidazole iv over 0 by guest on August 8, 2012 2 cefuroxime iv over 0 1 ciprofloxacin po clindamycin iv over 0 2 ciprofloxacin po clindamycin po over 0 5 clindamycin iv over 0 1 co-amoxiclav iv flucloxacillin iv over 0 1 co-amoxiclav iv metronidazole iv over 0 5 co-amoxiclav po over 1 9 co-amoxiclav iv over 0 16 flucloxacillin iv over 0 1 piperacillin/tazobactam iv over 0 3 cefalexin po under 0 1 ciprofloxacin iv under 0 7 ciprofloxacin po under 0 1 metronidazole po under 0

Severity class II 2 benzylpenicillin iv flucloxacillin iv appropriate 1 1 ceftriaxone iv appropriate 0 2 clindamycin iv appropriate 0 2 clindamycin po appropriate 0 12 flucloxacillin iv appropriate 0 1 vancomycin iv appropriate 0 1 ceftriaxone iv flucloxacillin iv metronidazole iv over 0 2 cefuroxime iv over 1 1 ciprofloxacin po clindamycin po rifampicin po over 0 1 ciprofloxacin iv clindamycin iv over 0

Continued

5of11 Marwick et al.

Table 3. Continued

No. of Consensus Initial patients Antibiotic 1 Route 1 Antibiotic 2 Route 2 Antibiotic 3 Route 3 appropriate disagreement

3 ciprofloxacin po clindamycin po over 1 2 ciprofloxacin po flucloxacillin iv over 1 1 clindamycin iv flucloxacillin iv over 0 2 co-amoxiclav iv over 1 1 gentamicin iv piperacillin/tazobactam iv over 0 1 amoxicillin po under 0 1 ciprofloxacin po rifampicin po under 0 2 ciprofloxacin po under 0 1 clarithromycin po under 0 1 co-amoxiclav po flucloxacillin po under 0 1 co-amoxiclav po metronidazole po under 0 1 co-amoxiclav po under 0 Downloaded from 1 erythromycin iv under 1 1 flucloxacillin po penicillin V po under 0 7 flucloxacillin po under 0 1 fusidic acid po under 0 1 gentamicin iv under 0

1 norfloxacin po under 0 http://jac.oxfordjournals.org/ 2 rifampicin po trimethoprim po under 0

Severity class III 1 benzylpenicillin iv flucloxacillin iv appropriate 0 1 ceftriaxone iv appropriate 0 1 clarithromycin iv appropriate 1 1 clindamycin iv appropriate 0 5 flucloxacillin iv appropriate 0 1 vancomycin iv appropriate 0 by guest on August 8, 2012 1 cefuroxime iv metronidazole iv over 0 1 ciprofloxacin po flucloxacillin iv over 1 1 co-amoxiclav iv flucloxacillin iv over 1 6 co-amoxiclav iv over 1 1 piperacillin/tazobactam iv over 0 1 benzylpenicillin iv metronidazole iv under 0 1 benzylpenicillin iv under 0 1 ciprofloxacin po under 0 1 clarithromycin po under 0 4 clindamycin po under 0 1 co-amoxiclav po metronidazole po under 0 2 co-amoxiclav po under 0 1 levofloxacin iv under 0 1 metronidazole iv under 0

Severity class IV 1 clindamycin iv levofloxacin iv metronidazole iv appropriate 0 1 ciprofloxacin po under 0 1 clarithromycin iv co-amoxiclav iv under 0 1 clindamycin iv metronidazole iv under 1 1 clindamycin po under 0 1 co-amoxiclav iv under 0 4 flucloxacillin iv under 0 1 metronidazole iv under 0 1 vancomycin iv under 0 189 12 iv, intravenous; over, over-treatment; po, oral; under, under-treatment.

6of11 Management and outcomes of SSTI in hospital JAC

class III. The resolution was to code the first two regimens as patients with MRSA versus 3/20 (15%) patients with MSSA, under-treatment and the third as appropriate treatment. odds ratio 5.67 (95% CI 1.07–30.09). The patients with MSSA Eight of the nine regimens rated as over-treatment by one rater that was resistant to initial therapy received benzylpenicillin involved patients in classes I–III who received co-amoxiclav, (n¼1), clarithromycin (n¼1) or ciprofloxacin (n¼1). The patients ciprofloxacin or cefuroxime, with or without the addition of flu- with MRSA that was resistant to initial therapy received cloxacillin. The resolution was to code all eight as over-treatment. co-amoxiclav (n¼3), clindamycin (n¼2) or cefuroxime (n¼1). The final case of disagreement about over-treatment involved a The patients with MRSA that was susceptible to empirical patient in class II who received intravenous flucloxacillin plus therapy received gentamicin (n¼2), ciprofloxacin plus clindamy- benzyl penicillin. The resolution was to code this as appropriate. cin (n¼1), fusidic acid (n¼1), trimethoprim plus rifampicin Consensus decisions on the appropriateness of each prescribed (n¼1) or vancomycin (n¼1). antibiotic regimen by severity class, and whether or not there The number of patients with MRSA and MSSA in each severity was initial disagreement, are detailed in Table 3. class (and percentage of total patients in that class) was: class I: 4 (4%) MRSA, 13 (14%) MSSA; class II: 7 (11%) MRSA, 5 (8%) MSSA; class III: 1 (3%) MRSA, 2 (6%) MSSA. None of the patients Appropriateness (by CREST guidelines) of therapy in class IV had significant bacterial isolates. by severity class Patients in whom antibiotic treatment for an SSTI was started Overall, only 26% of patients received antibiotics classified as .2 days after admission to hospital were more likely than those appropriate (Table 4). The prevalence of over-treatment was inver- treated in the first 2 days to have MRSA [5/28 (17.9%) compared Downloaded from sely related to and the prevalence of under-treatment was posi- with 7/161 (4.3%), x2 test P¼0.007]. There was no significant tively associated with increasing severity class (Table 4). Only difference in MSSA rate between these two groups [2/28 one of 12 (8%) patients in class IV received empirical treatment (7.1%) compared with 18/161 (11.2%), x2 test P¼0.522]. that was likely to cover all potential pathogens and toxins (clinda-

mycin plus levofloxacin plus metronidazole). Seven patients http://jac.oxfordjournals.org/ received regimens with significant gaps in antimicrobial spectrum Multivariable analysis of mortality (narrow-spectrum penicillins alone, metronidazole alone, cipro- None of the demographic variables [age, sex, Charlson index of floxacin alone or vancomycin alone), and one patient received co-morbidity, and Scottish index of multiple deprivation (SIMD)] intravenous co-amoxiclav, which we did not regard as adequate had any significant effect on 30 day mortality, either unadjusted therapy for severe sepsis with SSTI. In addition, two patients or adjusted, in this cohort of 189 SSTI patients (Table 5). There received oral therapy. Even in class I infections 10% of patients was an increase in the odds ratio for death within 30 days with received oral therapy that is not sufficiently active against S. pyo- each increase in severity class. Once adjusted for demographic genes and S. aureus (ciprofloxacin alone or metronidazole alone). variables and appropriateness of first-line therapy, having infec- by guest on August 8, 2012 tions in severity class III and IV was significantly associated with Microbiology results and resistance increased mortality (Table 5). The very wide 95% confidence intervals reflect the low numbers of deaths, but there is a clear There were 43 clinically significant bacterial isolates from speci- association between increasing clinical severity class and mens taken within 7 days of start of antimicrobial therapy mortality. The appropriateness of treatment had no significant from 33 patients. Three isolates were from blood cultures effect on the likelihood of death, particularly once adjusted for (S. aureus n¼2, Bacteroides fragilis n¼1). All of the remaining other variables (Table 5). isolates were S. aureus from wound swabs. There were 20 The number of SIRS criteria a patient met did not influence patients with one or more isolates of methicillin-susceptible 30 day mortality, either unadjusted or adjusted for demographic S. aureus (MSSA) and 12 patients with one or more isolates of variables and for appropriateness of therapy (Table 5). Because methicillin-resistant S. aureus (MRSA). Patients with MRSA were there is overlap between the SIRS criteria, the factors comprising more likely to receive microbiologically inappropriate therapy, SEWS and the study definitions of severity class, these three vari- with isolates being resistant to initial therapy in 6/12 (50%) ables (severity class, SIRS and SEWS) were considered in separate regression models, adjusting the effect of each for the demo- Table 4. Appropriateness of initial antibiotic therapy by severity class for graphic variables and treatment appropriateness. 189 hospital inpatients with SSTI Increasing SEWS had a stepwise and significant effect on mortality, which was even greater once other factors had been Treatment classification by comparison with adjusted for. There were insufficient numbers of outcome events UK guidance, N (%)5 to look separately at the effect of each individual clinical variable that contributes to SEWS (pulse, blood pressure, etc.; Table 1). Severity Total no. appropriate Even the odds ratios for the effect of total SEWS on 30 day mor- class patients treatment over-treatment under-treatment tality had very wide 95% confidence intervals (Table 5).

I 88 19 (22) 57 (65) 12 (14) II 56 20 (36) 14 (25) 22 (39) Discussion III 33 10 (30) 10 (30) 13 (39) IV 12 1 (8) 0 11 (92) SSTIs are common, with Hospital Episode statistics in the UK All 189 50 (26) 81 (43) 58 (31) recording 69576 admissions with a diagnosis of cellulitis in 2004–05.6 Therefore, it is surprising that there is very little

7of11 Marwick et al.

Table 5. Multivariable analysis for death within 30 days of start of treatment for SSTI (all odds ratios were calculated with data from the 189 patients with complete information about antibiotic treatment)

Unadjusted odds ratio 95% CI P value Adjusteda odds ratio 95% CI P value

Age group ,60 1.0 1.0 60–69 4.76 0.42–54.56 0.210 2.21 0.14–34.98 0.575 70–79 10.10 1.17–86.86 0.035 9.89 0.81–120.91 0.073 80+ 6.63 0.76–58.52 0.088 6.04 0.50–73.59 0.158

Male 1.0 1.0 Female 3.27 0.88–12.08 0.076 3.14 0.60–16.49 0.176

Charlson index 0–2 1.0 1.0 3+ 1.31 0.35–4.94 0.694 0.30 0.04–2.62 0.274

SIMD quintile Downloaded from 1 or 2 (more deprived) 1.0 1.0 3 1.46 0.39–5.55 0.577 0.69 0.11–4.28 0.693 4 0.32 0.04–2.80 0.305 0.18 0.01–2.33 0.190 5 (more affluent) 0.94 0.22–4.00 0.938 1.73 0.29–10.23 0.546

Severity class http://jac.oxfordjournals.org/ I 1.0 1.0 II 4.33 0.44–42.61 0.209 6.51 0.51–83.12 0.149 III 18.83 2.18–162.91 0.008 32.39 2.80–374.49 0.005 IV 45.50 4.53–457.20 0.001 167.88 5.30–5319.54 0.004

Initial treatment appropriate 1.0 1.0 over-treatment 0.60 0.12–3.11 0.545 1.02 0.14–7.72 0.981 under-treatment 2.51 0.63–10.02 0.194 0.57 0.09–3.51 0.545 by guest on August 8, 2012

SIRS criteria 0 1.0 1.0 1 2.68 0.62–11.65 0.189 1.98 0.41–9.65 0.399 2 4.62 0.97–21.95 0.055 5.09 0.83–31.23 0.79 3 or 4 3.16 0.49–20.21 0.225 3.79 0.36–39.74 0.266

SEWS 0 1.0 1.0 1 11.70 1.17–116.51 0.036 14.47 1.29–162.16 0.030 2 13.00 1.12–150.83 0.040 16.09 1.16–222.62 0.038 3 42.54 4.37–414.63 0.001 72.63 5.12–1029.51 0.002 4+ 31.2 3.27–297.88 0.003 107.90 5.00–2328.62 0.003 aAdjusted values are reported based on a logistic regression model including age group, sex, Charlson index, initial treatment, Scottish index of multiple deprivation (SIMD) quintile and severity class. Separate regression models were created for SIRS criteria and for SEWS with the same co-variates, with the exception of severity class. None of the co-variates in these two additional models had statistically significant odds ratios (data not shown). evidence to inform practice guidelines, which are instead based relationship between empirical treatment and a severity classifi- on expert consensus.2,5,13 cation in SSTIs. The study definitions for each severity class were Uncertainty among doctors about appropriate empirical adapted from those published in a set of UK guidelines5 to make antimicrobial treatment of SSTIs is demonstrated by the 35 dif- the criteria more objective and to ensure that the four classes ferent antibiotic regimens prescribed in this cohort study. This were mutually exclusive. is only slightly less practice variation than the 46 different For class II, the CREST guidance recommends intravenous empirical regimens reported in a retrospective cohort study of therapy due to the presence of specified co-morbidities that 47219 patients with SSTI from .400 hospitals in the USA.12 To ‘may complicate or delay resolution of their infection’. However, our knowledge this is the first published study to evaluate the our data showed little difference between classes I and II in

8of11 Management and outcomes of SSTI in hospital JAC

terms of antimicrobial therapy, over half of which was intravenous bacteraemic patients, the relationship between ‘appropriate’ in both of these classes, or adjusted outcome. Likewise, a random- therapy and outcome in staphylococcal infections is unclear.24 ized trial of intravenous versus oral therapy in 60 patients with ery- Further difficulties in assessing appropriateness arise as a result sipelas without complications showed no clinical benefit of of the changing clinical and political climate. During the study intravenous therapy as evaluated by fever duration, hospital stay period, our antibiotic policy recommended including ciprofloxacin and sick leave.20 for severity class IV patients, but this has now changed because These data question whether classes I and II could actually of increasing concerns about Clostridium difficile. Consistency be merged, with severity assessment and treatment decisions and transparency in evaluation of appropriate therapy in trials based on clinical parameters (SIRS and SEWS), rather than also of therapy-related outcomes are essential.25 considering co-morbidity. Accordingly, the SSTI management Under-treatment, according to clinical severity, was associated guidelines in our NHS trust have been changed due to the findings with greater mortality but the association was not significant. of this study, and now include only three severity classes, with oral There are two potential explanations. First, in 34% (20/58) of therapy recommended for the least severe class (which includes patients coded as under-treated this was because they received patients who would previously have been in class II). oral therapy for class II, which may actually be as effective as The presence of sepsis in association with SSTIs, even without intravenous treatment.20 Second, our study data do not allow more severe physiological upset (i.e. severity class III), was assessment of outcome by treatment in classes III and IV associated with worse outcome than having no sepsis (classes because only one patient in class IV received appropriate anti- I and II), consistent with existing evidence.17,21 While the microbial therapy. Consequently, more data are required from Downloaded from actual number of SIRS criteria met did not appear to have an prospective studies to fully define the relationship between effect on mortality, meeting two or more criteria, which in the empirical therapy and outcome. context of infection means sepsis (and in this study means A conservative approach to the microbiology of SSTI was severity class III) is associated with increased mortality (Table 5). taken, and only blood culture isolates (other than coagulase-

Severity class IV should include patients with severe sepsis, negative staphylococci) and S. aureus isolates from wound http://jac.oxfordjournals.org/ the CREST definition for class IV being ‘sepsis syndrome or swabs were included (there were no streptococci isolated from severe life-threatening infection’. The accepted definition of study patients in the defined time period). Even with this conser- severe sepsis is sepsis complicated by organ dysfunction,22 but vative approach the proportion of patients with bacteria resistant different systems can be used to define organ dysfunction22 to empirical therapy was high (9/33, 27%) because of the local and, particularly in a retrospective observational study, these prevalence of MRSA, which is similar to other UK centres. It is can be difficult to apply. In this study, the SEWS system perhaps surprising, therefore, that CREST guidance does not rec- (Table 1), which was designed to identify patients at risk of ommend MRSA cover for even the most severely ill patients.5 This cardiac arrest and death in hospital,18 was used to classify the is partly explained by the lower prevalence of MRSA when and most severely ill patients (class IV) as those with a SEWS of where (Northern Ireland) the guidance was written but, even in by guest on August 8, 2012 ≥4. A previous study demonstrated that a SEWS of ≥4 on admis- the USA where, particularly community-acquired, MRSA is more sion to hospital predicted in-hospital mortality and length of stay prevalent, the evidence for empirical coverage is not conclusive. for general medical patients.18 SEWS charts were in use across One published study has found a non-significant trend towards our NHS trust at the time of the study so these data were avail- poorer outcome in SSTI when MRSA was not covered at the able for all patients. Also, there was an agreed hospital-wide outset.12 policy that mandated urgent medical review for any patient Most MRSA in the UK is healthcare associated. Many patients with a SEWS of ≥4, 24 h a day, 7 days a week. The significantly admitted to hospital have significant previous healthcare greater mortality in our class IV patients, and the stepwise exposure and the factors required for risk stratification for increase in mortality with increase in SEWS (Table 5), suggest MRSA26 –28 may not be immediately available on presentation that SEWS could have a role in identifying the highest risk to hospital. While it seems intuitive that early appropriate patients and guiding the use of intensive therapy in SSTIs. therapy improves outcome, and therefore that patients with While we had insufficient outcome events to examine the SSTI and sepsis should receive empirical coverage of MRSA, the effect of the individual components of the SEWS, the total evidence for this approach comes from other disease entities score is potentially more clinically useful, although limits gener- and no such studies exist for SSTI.29 alization to other early warning scoring systems. The relative Strengths of our study are that we used an electronic data- importance of each SEWS component, and the cut-off values base to randomly sample patients for screening. We were also that best predict outcome in different clinical contexts, is the able to obtain additional data from case notes, including indi- subject of a recent investigation.23 Again, a prospective study cators of severity of infection and data from microbiology speci- in SSTI patients is required to evaluate this, and to assess any mens collected within 7 days of the start of antimicrobial added value of risk scores based on laboratory tests.11 therapy. These data have enabled analysis of appropriateness The level of agreement between two independent raters on of initial therapy adjusted for severity of illness, which was not appropriateness of treatment by clinical severity achieved in possible in a large retrospective study of SSTI from an adminis- this study was high at 85%, much higher than we have found trative database.12 in previous work.19 This is most likely due to the fact that we sti- The principal weaknesses of our study are common to any pulated a priori what would and would not be considered appro- retrospective study and include errors of omission in coding priate therapy, based on CREST guideline recommendations, for and the consequent risks of sampling bias as well as the inability each severity class. What constitutes appropriate therapy in a to adjust for confounding by variables that are not reliably col- given clinical situation can be very difficult to define. Even in lected in case notes.12 Also, we included patients who did not

9of11 Marwick et al. meet the original sampling criteria, but whose case notes were Acknowledgements screened for a different study and met the SSTI study definition. These findings, in an earlier version, were presented as a free paper at the This introduces further risk of sampling bias. Nineteenth European Congress of Clinical Microbiology and Infectious Significant bacterial isolates were only cultured from 33 (16%) Diseases, Helsinki, 2009 (Abstract O301, http://www.blackwellpublishing. of all 205 patients, and only 3 (6%) of the 47 more severely com/eccmid19/abstract.asp?id=73995). ill patients in classes III and IV, so we could not analyse We thank Alison Bell (Health Informatics Centre, University of Dundee) outcome by pathogen. This is common to studies of SSTI and Ross Martin (Department of Medical Microbiology, Ninewells where cultures are often of low yield, so treatment has to be Hospital). based on the likely pathogens rather than the individual patient’s microbiology results.1,30 Other weaknesses are that we did not study any potential effects of time to antibiotic administration, or of antibiotic dosing, on patient outcome. Time to antibiotics Funding can influence outcome in severely ill septic patients31 so is a This work was supported by an unrestricted research grant from potential source of confounding in relation to class IV. Under- Janssen-Cilag. dosing could also adversely affect outcome and, although drug dosing tends to be standard across our hospital, we did not specifically look at any effect of dose variation on outcome.

Transparency declarations Downloaded from Also, we did not record potential adverse effects of antibiotic C. Marwick has been supported in conference attendance by therapy, such as C. difficile-associated diarrhoea, or the isolation Janssen-Cilag. J. B.’s salary has been supported in part by funding from of resistant pathogens following treatment. Janssen-Cilag. C. McCowan has no conflicts of interest to declare. G. P. This study has added to the case for clinical trials to compare has no conflicts of interest to declare. S. G.-M. was the Anti-Infectives selection of antimicrobial drug and route of administration in Outcomes Research Manager for Janssen-Cilag UK at the time of

SSTI, which was the conclusion of a recent review of the evi- study and is currently employed by GlaxoSmithKline Mexico as Health http://jac.oxfordjournals.org/ dence.6 Two research questions could potentially be addressed Economics Manager; both companies produce anti-infectives. K. A. was in a single trial. First, is intravenous therapy given first-line Director of Health Economics and Pricing at Johnson & Johnson more effective than oral therapy for patients without systemic Pharmaceutical Services, LLC at the time of the study and is now inflammatory response? Second, is empirical therapy with Senior Director of Global Outcomes Research at Pfizer Oncology Medical activity against MRSA more effective than flucloxacillin for Group; both companies produce anti-infectives. S. M. is an employee patients with systemic inflammatory response? of Johnson & Johnson Pharmaceutical Services, LLC, who produce To inform the design of a clinical trial, data from prospective anti-infectives. D. N. has received honoraria for serving on advisory boards or speaking at symposia supported by Wyeth, Pfizer, Jonhson & cohort studies in targeted clinical settings, e.g. acute medical Johnson and Novartis, all of which produce anti-infectives; he has by guest on August 8, 2012 admissions or acute surgical admissions, are required in order received no fees or financial support in undertaking work related to this to assess likely recruitment rates and perform power calculations manuscript. P. D. has been paid honoraria, consultancy fees or speaker’s for a randomized trial. Prospective data on microbial pathogens fees by Johnson & Johnson, who are the manufacturers of various anti- and validation of proposed clinical severity scoring systems biotics and anti-infective products. would also be required. In particular, evidence is required about the predictive value in SSTI of laboratory tests that have 8,10,11 been shown to predict outcome in necrotizing fasciitis or References severe sepsis.21 The measurement of clinical outcome in trials of therapy in 1 Ki V, Rotstein C. Bacterial skin and soft tissue infections in adults: a review of their epidemiology, pathogenesis, diagnosis, treatment and SSTI requires careful consideration given the heterogeneity of site of care. Can J Infect Dis Med Microbiol 2008; 19: 173–84. clinical presentation, and correspondingly different associated mortality rates.30,32 A recent review of historical literature from 2 Stevens DL, Bisno AL, Chambers HF et al. Practice guidelines for the diagnosis and management of skin and soft-tissue infections. Clin the immediately pre- and post-antibiotic eras described the Infect Dis 2005; 41: 1373–406. untreated mortality of cellulitis/erysipelas as 11%, and major 32 3 Ansari F, Erntell M, Goossens H et al. The European surveillance abscess as 6%, with the recommendation that measures of of antimicrobial consumption (ESAC) point-prevalence survey of outcome in trials should be weighted according to the specific antibacterial use in 20 European hospitals in 2006. Clin Infect Dis 2009; 30,32 type of SSTI. Also, given that the mortality associated 49: 1496–504. 32 with SSTIs is low in the antibiotic era, other measures of 4 Fine MJ, Auble TE, Yealy DM et al. A prediction rule to identify low-risk clinical failure of therapy must be defined and agreed for more patients with community-acquired pneumonia. New Engl J Med 1997; informative comparison of antimicrobial agents in clinical 336: 243–50. trials.30,33 5 CREST (Clinical Resource Efficiency Support Team). CREST (Clinical In conclusion we have shown that patients with SSTIs and a Resource Efficiency Support Team) Guidelines On The Management Of systemic inflammatory response have significantly worse out- Cellulitis In Adults. In: CREST, ed: DHSS Northern Ireland, 2005; 1–31. comes than other patients with SSTIs. The majority of these 6 Morris A. Cellulitis and erysipelas. BMJ Clin Evid 2003; 12: 1708. patients receive clinically inadequate antimicrobial therapy with 7 Bosshardt TL, Henderson VJ, Organ CH Jr. Necrotizing soft-tissue the currently recommended treatment for class IV being broad- infections. Arch Surg 1996; 131: 846–52; discussion, 852–4. spectrum and intravenous. The optimal choice of antimicrobial 8 Elliott DC, Kufera JA, Myers RA. Necrotizing soft tissue infections. Risk therapy, and whether this includes coverage for MRSA, requires factors for mortality and strategies for management. Ann Surg 1996; review and definitive study in clinical trials. 224: 672–83.

10 of 11 Management and outcomes of SSTI in hospital JAC

9 Eriksson B, Jorup-Ronstrom C, Karkkonen K et al. Erysipelas: clinical and 21 GaoF,MelodyT,DanielsDFet al. The impact of compliance with the bacteriologic spectrum and serological aspects. Clin Infect Dis 1996; 23: 6-hour and 24-hour sepsis bundles on hospital mortality in patients with 1091–8. severe sepsis: a prospective observational study. Crit Care 2005; 9:764–70. 10 Wall DB, de Virgilio C, Black S et al. Objective criteria may assist in 22 Levy MM, Fink MP, Marshall JC et al. 2001 SCCM/ESICM/ACCP/ATS/SIS distinguishing necrotizing fasciitis from nonnecrotizing soft tissue International Sepsis Definitions Conference. Intensive Care Med 2003; infection. Am J Surg 2000; 179: 17–21. 29: 530–8. 11 Wong CH, Khin LW, Heng KS et al. The LRINEC (Laboratory Risk 23 Cuthbertson BH, Boroujerdi M, Prescott G. The use of combined Indicator for Necrotizing Fasciitis) score: a tool for distinguishing physiological parameters in the early recognition of the deteriorating necrotizing fasciitis from other soft tissue infections. Crit Care Med acute medical patient. J R Coll Physicians Edinb 2010; 40: 19–25. 2004; 32: 1535–41. 24 Ammerlaan H, Seifert H, Harbarth S et al. Adequacy of antimicrobial 12 Edelsberg J, Berger A, Weber DJ et al. Clinical and economic treatment and outcome of Staphylococcus aureus bacteremia in 9 consequences of failure of initial antibiotic therapy for hospitalized Western European countries. Clin Infect Dis 2009; 49: 997–1005. patients with complicated skin and skin-structure infections. Infect 25 Schwaber MJ, Carmeli Y. The effect of antimicrobial resistance on Control Hosp Epidemiol 2008; 29: 160–9. patient outcomes: importance of proper evaluation of appropriate 13 Eron LJ, Lipsky BA, Low DE et al. Managing skin and soft tissue therapy. Critical Care 2009; 13: 106–7. infections: expert panel recommendations on key decision points. 26 Furuno JP, McGregor JC, Harris AD et al. Identifying groups at high risk J Antimicrob Chemother 2003; 52 Suppl 1: i3–17. for carriage of antibiotic-resistant bacteria. Arch Intern Med 2006; 166: 14 Jones GR. A re-evaluation of blood cultures. Rev Med Microbiol 1995; 580–5. Downloaded from 6: 109–18. 27 Lodise TP, McKinnon PS, Rybak M. Prediction model to identify patients 15 Jones GR, Lowes JA. The systemic inflammatory response as a with Staphylococcus aureus bacteremia at risk for methicillin resistance. predictor of bacteraemia and outcome from sepsis. Q J Med 1996; 89: Infect Control Hosp Epidemiol 2003; 24: 655–61. 515–22. 28 Sax H, Harbrath S, Gavazzi G et al. Prevalence and prediction of previously unknown MRSA carriage on admission to a geriatric hospital. 16 Charlson ME, Pompei P, Ales KL. A new method of classifying Age Ageing 2005; 34: 456–62. prognostic comorbidity in longitudinal studies: development and http://jac.oxfordjournals.org/ 29 Nathwani D. New antibiotics for the management of complicated skin validation. J Chronic Dis 1987; 40: 373–83. and soft tissue infections: are they any better?. Int J Antimicrob Agents 17 Bone RC, Balk RA, Cerra FB et al. Definitions for sepsis and organ failure 2009; 34 Suppl 1: S24–9. and guidelines for the use of innovative therapies in sepsis. The ACCP/ 30 Stevens DL. Antimicrobial agents for complicated skin and skin- SCCM Consensus Conference Committee. American College of Chest structure infections: noninferiority margins, placebo-controlled trials, Physicians/Society of Critical Care Medicine. Chest 1992; 101: 1644–55. and the complexity of clinical trials. Clin Infect Dis 2009; 49: 392–4. 18 Paterson R, MacLeod DC, Thetford D et al. Prediction of in-hospital 31 Kumar A, Roberts D, Wood KE et al. Duration of hypotension before mortality and length of stay using an early warning scoring system: initiation of effective antimicrobial therapy is the critical determinant of clinical audit. Clin Med 2006; 6: 281–4. survival in human septic shock. Crit Care Med 2006; 34: 1589–96. by guest on August 8, 2012 19 Marwick C, Watts E, Evans J et al. Quality of care in sepsis 32 Spellberg B, Talbot GH, Boucher HW et al. Antimicrobial agents management: development and testing of measures for improvement. for complicated skin and skin-structure infections: justification of J Antimicrob Chemother 2007; 60: 694–7. noninferiority margins in the absence of placebo-controlled trials. Clin 20 Jorup-Ronstrom C, Britton S, Gavlevik A et al. The course, costs and Infect Dis 2009; 49: 383–91. complications of oral versus intravenous penicillin therapy of erysipelas. 33 Garcı´a MS. Early antibiotic treatment failure. Int J Antimicrob Agents Infection 1984; 12: 390–4. 2009; 34: S14–9.

11 of 11 Journal of Antimicrobial Chemotherapy (2008) 61, 976–994 doi:10.1093/jac/dkn096 Advance Access publication 13 March 2008 Guidelines for UK practice for the diagnosis and management of methicillin-resistant Staphylococcus aureus (MRSA) infections presenting in the community

Dilip Nathwani1*, Marina Morgan2, Robert G. Masterton3, Matthew Dryden4, Barry D. Cookson5, Gary French6 and Deirdre Lewis7 on behalf of the British Society for Antimicrobial Chemotherapy Working Party on Community-onset MRSA Infections

1Infection Unit, Ninewells Hospital and Medical School, Dundee DD1 9SY, UK; 2Royal Devon & Exeter NHS 3

Foundation Trust, Exeter EX2 5DW, UK; Lothian University Hospitals NHS Trust, Edinburgh EH3 9YW, UK; Downloaded from 4Royal Hampshire County Hospital, Winchester SO22 5DG, UK; 5Laboratory of Healthcare-Associated Infection, HPA Centre for Infections, London 9W9 5HT, UK; 6St Thomas’ Hospital, Kings’ College, London SE1 7EH, UK; 7HPA Southwest, Stonehouse, Gloucester GL10 3RF, UK

These guidelines have been developed by a Working Party convened on behalf of the British Society http://jac.oxfordjournals.org/ for Antimicrobial Chemotherapy. Their aim is to provide general practitioners and other community- and hospital-based healthcare professionals with pragmatic advice about when to suspect MRSA infection in the community, when and what cultures should be performed and what should be the man- agement options, including the need for hospitalization.

Keywords: community onset, case scenarios, MRSA infection, diagnosis, management by guest on August 8, 2012

Contents 6.1.2 When to suspect PVL+ CA-MRSA 6.1.2.1 Clinical presentation 1. Introduction 6.1.2.2 Treatment factors 2. Background and definitions 6.1.2.3 Other risk factors 2.1 Definitions 6.1.2.4 Prevalence of CA-MRSA 2.1.1 MRSA (methicillin-resistant S. aureus) 2.1.2 CA-MRSA (community-associated MRSA) 6.1.3 When should specimens be sent for culture? 2.1.3 HA-MRSA (healthcare-associated MRSA) 6.1.4 When should specimens for culture not be taken? 6.1.5 Treatment principles 3. Guideline remit 6.1.6 Empirical antibiotic therapy 4. Guideline development and methodology 5. General principles of diagnosis and management 6.2 Community-associated and community-onset pneumo- 5.1 Clinical assessment nia suspected to be due to PVL-producing MRSA 5.2 Laboratory investigation 6.2.1 What is the likely clinical diagnosis? 5.2.1 Clinical samples 6.2.2 When to suspect PVL+ CA-MRSA 5.2.2 Testing for PVL 6.2.2.1 Clinical presentation 5.2.3 Antimicrobial susceptibility testing 6.2.2.2 Treatment factors 6.2.2.3 Other risk factors 6. Clinical case scenarios 6.2.2.4 Prevalence of CA-MRSA 6.1 Community-associated and community-onset MRSA skin and soft tissue infection 6.2.3 When should specimens be sent for culture? 6.1.1 What is the likely clinical diagnosis? 6.2.4 Radiological investigation

...... *Corresponding author. Tel: þ44-1382-660111; Fax: þ44-1382-816178; E-mail: [email protected]

...... 976 # The Author 2008. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected] Review

6.2.5 Treatment principles 6.2.6 Empirical antibiotic therapy 6.2.7 Adjunctive therapy options 6.2.7.1 Immunoglobulin therapy (IVIG) 6.2.7.2 Other adjunctive therapy

6.3 Healthcare-associated community-onset MRSA skin and soft tissue infection 6.3.1 What is the likely clinical diagnosis? 6.3.2 When to suspect HA-MRSA 6.3.2.1 Clinical presentation 6.3.2.2 Risk factors

6.3.3 When should specimens be sent for culture? 6.3.4 Treatment principles 6.3.5 Empirical antibiotic therapy Downloaded from 1. Introduction This guidance aims to complement existing guidance on preven- tion and treatment of infections caused by methicillin-resistant Staphylococcus aureus (MRSA) and focuses on typical common and less common community-onset infections with an emphasis Figure 1. PVL-related disease: microbiology algorithm. http://jac.oxfordjournals.org/ on community-associated MRSA (CA-MRSA). An important issue covered by these guidelines is the management of serious staphylococci is currently undergoing consultation.3 infection caused by MRSA arising in the community. Although Whether to decolonize infected patients or screen and such infections are rare at present, they usually affect young, decolonize contacts if positive is a key issue that is dis- previously healthy people and may have a rapid and devastating cussed in the PVL guidance4 and will not be repeated here. course. Specific guidance is given on the management of staphy- However, the evidence base supporting many of the rec- lococcal pneumonia, although other serious manifestations of ommendations related to screening and decolonization is these infections are emerging. Serious S. aureus infections can by guest on August 8, 2012 poor and worthy of further investigation. be caused by strains that are methicillin-resistant or -susceptible † Antibiotic treatment of MRSA in general is discussed by and which may or may not express the pathogenic Panton– Gemmell et al.4 This guidance is currently being revised. Valentine leucocidin (PVL) toxin. The role of the general prac- titioner (GP) is to recognize that the patient is seriously unwell and needs to be managed in hospital (see Section 6, Appendix 1 All of these documents can be accessed via the Health and the algorithm in Figure 1). Protection Agency (HPA) web site.5 A summary of the commonest MRSA clinical problems [skin and soft tissue infections (SSTIs); serious and deep seated infec- tions] presenting to GPs and guidance on their treatment is pre- sented in Appendix 1. Practical advice on the isolation of 2. Background and definitions MRSA in the urine is also provided in the Appendix but not covered in the main body of the guidelines because the evidence S. aureus is the major bacterial cause of skin, soft tissue and base for the management of this situation is poor. The rec- bone infections, and one of the commonest causes of ommendations could be implemented, for example, by their inte- healthcare-associated bacteraemia. About one-quarter of healthy gration into new or existing care pathways. people carry one or more strains asymptomatically at any given Guidelines on various aspects of the management and control time and infections are commonly endogenous being caused by of MRSA are available and are revised regularly. the patient’s colonizing strain.6 Antibiotics and surgical drainage are the basis of treatment of staphylococcal infections, but the † This guidance is for the diagnosis and management of emergence of multiple resistance to penicillin, methicillin and MRSA infections that arise in the community. Much of other agents has compromised therapy. Methicillin resistance this is relevant to GPs. was first detected in S. aureus in 1961,7 shortly after the agent † Guidance for the control of MRSA in healthcare facilities was introduced clinically, and over the last four decades, there is given by Coia et al.1 has been a global epidemic of MRSA.8,9 † Laboratory diagnosis of MRSA is discussed by Brown MRSA is usually acquired during exposure to hospitals and et al.2 other healthcare facilities, and causes a variety of serious † Revision of previous interim guidance on the diagnosis healthcare-associated infections. A number of UK and USA and management of infection due to PVL-producing guidelines have been produced on the prevention, control,

977 Review diagnosis and treatment of MRSA infections in hospitals and increasing, particularly in the USA, Canada and Australia. In other healthcare facilities.2,10– 12 However, there has been an some areas of the USA, a significant proportion of serious increase in MRSA infections presenting in the community that S. aureus infections presenting in community practice or at acci- has not been properly addressed by existing guidelines.4 dent and emergency departments is now due to CA-MRSA Many (and in the UK at the present time, most) MRSA infec- types.15,24– 26 There are also emerging reports of CA-MRSA tions that appear to have a community onset occur in patients from Europe, including Scandinavian countries that have, until who are found to have had direct or indirect contact with hospi- now, been almost free of HA-MRSA.27 – 30 There have been rela- tals, care homes or other healthcare facilities.13,14 These MRSA tively few reports of CA-MRSA from the UK,31 – 33 but experi- strains are typical of the local healthcare-associated MRSA ence elsewhere suggests that these are likely to increase in the (HA-MRSA) and may be carried asymptomatically by patients future. for months after discharge. However, new strains of MRSA have The Department of Health asked the HPA to lead in the pro- recently emerged that cause infections in community patients duction of guidance for PVL-related disease.3 In 2007, the who have no previous history of direct or indirect healthcare Specialist Advisory Committee on Antimicrobial Resistance also contact. These strains have been designated CA-MRSA.15 identified the need to produce some specific guidance around CA-MRSA strains are genetically and phenotypically distinct diagnosis and treatment of infections caused by PVL-positive from HA-MRSA. They typically resemble some strains of staphylococci. The British Society of Antimicrobial methicillin-susceptible S. aureus (MSSA) in being susceptible to Chemotherapy (BSAC) on the other hand independently wished a wider range of anti-staphylococcal antibiotics (some are resist- to consider producing guidance with the broader remit of diag- ant only to b-lactams), and often produce PVL, a toxin that nosis and management of community-onset MRSA including destroys white blood cells and is a staphylococcal virulence infections acquired in the healthcare setting but which would Downloaded from factor.16,17 Differences between CA- and HA-MRSA are sum- include PVL-related disease. HPA members with expertise in marized in Table 1. PVL-producing strains of CA-MRSA appear this area were included in the BSAC Working Party and this gui- to be associated with increased risk of transmission, compli- dance recognizes HPA and other related guidance where appro- cations and hospitalization. For example, in one large commu- priate. It is hoped that this will raise general awareness about the nity outbreak of CA-MRSA, 23% of patients required epidemiology and pathological significance of CA-MRSA. It is http://jac.oxfordjournals.org/ hospitalization.18 PVL has a clear role in the pathogenesis of hoped that the early implementation of effective diagnosis, man- severe necrotizing pneumonia17,19,20 and is associated with agement, prevention and control of these new infections will greater pulmonary and bone-related complications. Its role in prevent some of the present difficulties with HA-MRSA34 devel- skin infections is less certain, although PVL is a potent dermato- oping with CA-MRSA. necrotic toxin.21 In the UK, the overall prevalence of S. aureus strains that carry the gene for PVL production is believed to be ,2%, and 2.1 Definitions 22 these are mainly MSSA. Although the overall prevalence of The internationally agreed definitions of HA-MRSA, CA-MRSA CA-MRSA is also presently low worldwide (thought to be and other S. aureus strains and their limitations are given by guest on August 8, 2012 23 ,0.5% of all MRSA), there is clear evidence that this is below.35 The definitions were originally based on

Table 1. HA-MRSA versus CA-MRSA

Parameter HA-MRSA CA-MRSA

Typical patient elderly, debilitated and/or critically or chronically ill young, healthy people; students, professional athletes and military service personnel Infection site often bacteraemia with no obvious infection focus. predilection for skin and soft tissue, producing cellulitis and Also surgical wounds, open ulcers, IV lines and abscesses. May cause necrotising community acquired catheter urines. May cause ventilator associated pneumonia, septic shock or bone and joint infections pneumonia Transmission within healthcare settings; little spread among community-acquired. May spread in families and sports household contacts teams Clinical setting of in an inpatient setting, but increasingly HA-MRSA in an outpatient or community setting diagnosis infections in soft tissue and urine are occurring in primary care Medical history history of MRSA colonization, infection, recent no significant medical history or healthcare contact surgery; admission to a hospital or nursing home, antibiotic use; dialysis, permanent indwelling catheter Virulence of community spread is limited, PVL genes usually community spread occurs easily. PVL genes often present, infecting strain absent predisposing to necrotising soft tissue or lung infection Antibiotic often multiresistant with result that choice of agents generally susceptible to more antibiotics than HA-MRSA susceptibility often very limited

978 Review

epidemiological features but, for the reasons outlined below, principles of diagnosis and management of MRSA infections in microbiological characteristics are now also important. the community. Infection prevention guidance is kept to a minimum as other guidance exists in this area. Our guidance 2.1.1 MRSA (methicillin-resistant S. aureus) covers: . Strains of S. aureus that are resistant to the isoxazoyl penicillins (1) Patients whose infections were acquired in the hospital but such as methicillin, oxacillin and flucloxacillin. MRSA are who present in the community (hospital or cross-resistant to all currently licensed b-lactam antibiotics. healthcare-associated and community-onset MRSA infec- tions). This is the commonest MRSA-related community 2.1.2 CA-MRSA (community-associated MRSA) problem in the UK. GPs may expect to encounter this on a . MRSA strains isolated from patients in an outpatient or commu- daily basis in patients who are chronically ill, elderly, dia- nity setting (community onset), or within 48 h of hospital admis- betic with open skin lesions, have had recent surgery or in sion (hospital onset). Patients also typically have no previous patients who are regular hospital attendees. history of MRSA infection or colonization, hospitalization, (2) Infections acquired in the community in patients with no surgery, dialysis or residence in a long-term care facility within hospital contact or other related risk factors. These may the previous year, and absence of indwelling catheters or percu- present in the community (community-associated and taneous devices at the time of culture (Table 2). community-onset, for example the skin infection or pneumo- nia scenarios) or, rarely, in hospital (community-associated 2.1.3 HA-MRSA (healthcare-associated MRSA) and hospital-onset).

. MRSA strains that are transmitted to and circulate between indi- Downloaded from viduals who have had contact with healthcare facilities. These Infections due to hospital or healthcare-associated S. aureus infections can present in the hospital or healthcare setting (hos- strains presenting in hospital (hospital-onset and pital or healthcare onset) or in the community (community hospital-acquired) are excluded, since they are already covered onset), for example after hospital discharge. in disease-specific guidance (e.g. ventilator-associated pneumo- However, the boundaries between HA-MRSA and nia, endocarditis, surgical site infection). The guidance on anti- http://jac.oxfordjournals.org/ CA-MRSA are becoming blurred due to the movement of microbial treatment and prophylaxis of MRSA infections patients and infections between hospitals and the community, produced by the Joint Working Party of the BSAC, the Hospital and to nosocomial outbreaks of CA-MRSA following admission 4 36 Infection Society and the Infection Control Nurses Association of colonized or infected patients. In the USA, where is due to be updated in 2008. CA-MRSA is now common, it is becoming increasingly difficult The Working Party recognizes that in the UK, PVL toxin- to distinguish between CA- and HA-MRSA on clinical and epi- 37 producing staphylococci are presently uncommon and when they demiological grounds. Since HA-MRSA and CA-MRSA cause infection they are more frequently MSSA. Interim gui- strains are often genotypically and phenotypically different dance for the diagnosis and management of PVL-associated by guest on August 8, 2012 (Table 1), the microbiological characteristics of the S. aureus S. aureus infections in the UK has been published by a group isolates may help distinguish between HA- and CA-infections. convened by the HPA.3 However, in light of the global emer- gence of severe infections and outbreaks due to PVL-producing CA-MRSA,38 infections with PVL-producing MRSA in the UK 3. Guideline remit are likely to increase. This guideline provides recommendations based on evidence obtained from existing guidance on best practice in the 4. Guideline development and methodology Table 2. Populations at increased risk of community-associated The Working Party was convened under the auspices of the MRSA (adapted from references 20, 32–34 and 38) BSAC and comprised microbiologists, an infectious disease physician and public health physicians, all of whom had an Risk groups interest in MRSA infections and had previous experience in guideline development. Advice regarding guideline format and Children ,2 years old content from a GP and nurse were sought prior to guideline Athletes (mainly contact-sport participants) development. Injection drug users The Working Party adopted a ‘common clinical scenario’ Men who have sex with men approach to the diagnosis and recognition of these infections Military personnel which they hope will be useful for the community prac- Inmates of correctional facilities, residential homes or shelters titioner and broader healthcare team. The evidence base for Vets, pet owners and pig farmers the guidance in these scenarios has been provided by a systema- Patients with post-flu-like illness and/or severe pneumonia tic appraisal of existing guidelines or consensus documents Patients with concurrent SSTI using the approach developed by the ADAPTE Group.39 This History of colonization or recent infection with CA-MRSA adaptation process involves six steps: (1) searching for existing History of antibiotic consumption in the previous year, particularly guidelines; (2) assessment of guideline quality; (3) assessment quinolones or macrolides of applicability and adaptation of recommendations to target setting; (4) literature update; (5) adaptation of guideline format;

979 Review and (6) implementation. We felt that this was more appropriate governmental and regional expert groups.43,44 Although none of than a new systematic literature review, since recent guidelines these documents gave evidence of a systematic process of produ- from Canada40 included a thorough systematic review of much cing evidence-based recommendations, they provided useful of the relevant CA-MRSA literature. Furthermore, other gui- information for some areas of our guidance. Most of our rec- dance on community-onset healthcare-acquired MRSA exists but ommendations based on existing guidance, primarily the is not presented in the context of ‘a community onset’ infection. Canadian guidelines, were SIGN grade D and arose from evi- The levels of evidence and the strength of recommendations dence in non-controlled studies and expert opinion. Only in a were categorized using the Scottish Intercollegiate Guidelines few cases did evidence come from the results of case–control or Network (SIGN) evidence statements and grades of recommen- cohort studies, reflecting the evolving nature of the disease. In dations (SIGN Guideline 50—a guideline developers’ handbook many areas, our recommendations are simply recommendations at www.sign.ac.uk), the details of which are shown in Tables S1 for good medical practice and reflect the experience and and S2 [Supplementary data available at JAC Online (http://jac. opinions of the Working Party. oxfordjournals.org/)]. On occasion, guideline development The Working Party assessed all the contributions and agreed groups find that there is an important practical point that they on the key recommendations before submitting the draft for wish to emphasize but for which there is not, nor is there likely external peer review. Changes suggested by the external review to be, any research evidence. This will typically be where some were incorporated before finalizing its adoption, endorsement aspect of treatment is regarded as being of such sound clinical and implementation. The BSAC agreed that this guideline would practice that nobody is likely to question it. These are marked in be developed in close collaboration and consultation with a the guideline as Good Practice Points, and are indicated by the Working Group of the HPA which was revising guidance on acronym ‘GPP’. It must be emphasized that these are not an PVL-associated staphylococcal infections in England. Other Downloaded from alternative to evidence-based recommendations and have only stakeholders are the Royal College of Nursing, the Infection been used where there was no alternative means of highlighting Prevention Society (formerly the Infection Control Nurses the issue. Association), the Hospital Infection Society, the British An English language only literature search was undertaken Infection Society, the Royal College of General Practitioners, for existing guidance related to clinical presentation, diagnosis the Royal College of Physicians, the Royal College of Surgeons, http://jac.oxfordjournals.org/ and treatment of CA-MRSA infection with particular reference the College of Emergency Medicine, the Royal College of to SSTIs and pneumonia. We searched Medline and Google/ Paediatricians and the UK Pharmacy Association. Google Scholar as well as the various guideline databases with the following keywords: ‘community-acquired or -onset S. aureus’, ‘Panton–Valentine Leucocidin’ and ‘methicillin- resistant S. aureus’. The following guideline databases were 5. General principles of diagnosis and management searched in detail: Guideline International Network, National Institute for Health and Clinical Excellence, National Guideline 5.1 Clinical assessment Clearing, SIGN, as well as a recent Canadian guideline.40 This by guest on August 8, 2012 was deemed sufficiently important to be worthy of more detailed When presented with a patient in the community who the clini- appraisal using the AGREE instrument.41 The scope and cian thinks may have a staphylococcal infection, the decision- purpose of the Canadian guideline, stakeholder involvement, making process related to diagnosis and treatment includes clarity and presentation were judged to be of good quality, answering the following questions: although the guideline provided few support tools for its appli- (1) What is the severity of illness? cation. However, the rigour of development was judged moder- (2) Is distinction between MSSA or MRSA infection possible? ate to poor because there was no consistent explicit linking of (3) If MRSA is suspected, is it likely to be CA-MRSA or the evidence to the recommendation, no tables of evidence, no HA-MRSA? supporting evidence for many recommendations, references were (4) Should further microbiological assessment/investigation be often missing and there were no details of how consensus state- undertaken and, if so, how? ments were developed for many of the statements. The basis for (5) Does empirical antibiotic therapy (or definitive therapy if the Canadian guidelines’ recommendations and strength of evi- microbiology results are available) need to be started? dence was less explicit than with the SIGN methodology and (6) Does adjunct therapy (for example surgical drainage) need more generous in making high-level or strong recommendations to be considered? based on grade 3 evidence (such as expert opinion, case studies (7) Does the patient need to be admitted to hospital? and so on). Other weaknesses include failure to identify the (8) What advice should be given to direct contacts and house- applicability of the guideline to an organization or healthcare hold members? system, no appreciation or mention of potential cost implications of the guideline and no criteria for monitoring quality improve- ment. The guideline group appeared to have editorial indepen- This clinical risk-assessment is detailed for each of the dence although there were no conflict of interest statements. We ‘typical’ clinical scenarios below. Many factors in these have attempted to address some of these weaknesses in the risk-assessments are similar for all the scenarios, including dis- present guidance. tinguishing between CA-MRSA and HA-MRSA strains The other documents looked at included other published (Table 1) and risk factors for CA-MRSA based on clinical, epi- reviews of the subject.15,42 Zetola et al.15 were particularly demiological, laboratory and treatment factors (Tables 1 and 2). thorough in defining search strategy and selection criteria, and The predictive value of these risk factors for distinguishing included papers in Spanish. There were also statements from between different strains, healthcare associated as opposed to

980 Review

community associated, is poor. Nevertheless, the scenarios pre- a moderate amount of surrounding erythema. It has not sented provide guidance for best practice. responded to topical antiseptic cream. He gives a history of three previous ‘skin infections’ over the last year, treated with several courses of oral antibiotics. The current infection 5.2 Laboratory investigation appears to have been precipitated by a ‘minor’ injury to his Because of the changing epidemiology of community-onset thigh during football at school. MRSA and the need to optimize antibiotic therapy, clinical samples should be sent for microbiological testing. 6.1.1 What is the likely clinical diagnosis? 5.2.1 Clinical samples . The clinical diagnosis of an SSTI presenting as an abscess is . Appropriate clinical samples (e.g. pus, swabs from lesions, sputa straightforward. Pain, tenderness, erythema and swelling are common in SSTIs and offer around a 93% to 97% sensitivity etc.) from suspected cases should be submitted to the local 46 microbiology department for analysis. Accident and emergency (95% CI 83–100) in the clinical diagnosis of cellulites. The staff, GPs and other healthcare practitioners should be alerted to most likely microbial cause of this is S. aureus, although the importance of taking specimens when incising and draining pyogenic streptococci (e.g. b-haemolytic streptococci such as abscesses. Samples should be cultured on non-selective media Streptococcus pyogenes, and group C or G streptococci) are (e.g. blood agar) for the recovery of potential pathogens, includ- other possibilities. ing S. aureus.

þ Downloaded from 5.2.2 Testing for PVL 6.1.2 When to suspect PVL CA-MRSA (currently rare) 6.1.2.1 Clinical presentation. . Genes encoding PVL may be carried by both MSSA and The spectrum of disease caused by MRSA. PVL-positive MSSA display variable antimicrobial sus- CA-MRSA appears to be similar to that caused by CA-MSSA. ceptibility profiles, whereas the majority of PVL-positive MRSA Furuncles (abscesses in hair follicles, or ‘boils’), carbuncles (coa- 45 lesced masses of furuncles with deeper tissue involvement) and

currently found in the UK are susceptible to ciprofloxacin. http://jac.oxfordjournals.org/ S. aureus (MSSA or ciprofloxacin-susceptible MRSA) recovered other abscesses appear to be the most frequently reported clinical from suspected cases should be referred to the HPA manifestations. They may or may not have accompanying celluli- Staphylococcus Reference Laboratory for toxin gene profiling tis. Erythematous papules and nodules, folliculitis and/or impetigo including PVL testing. This is a PCR-based assay that can be are less common presentations. completed within a working day. Figure 1 is a suggested One specific presentation appears to be typical of cutaneous algorithm. CA-MRSA infections. This is the spontaneous appearance of a raised tender red lesion, which may progress to develop a necrotic centre. This may lead to the suspicion of a ‘spider bite’ 5.2.3 Antimicrobial susceptibility testing where such occurrences are common, e.g. North America or by guest on August 8, 2012 . This should be performed by the laboratory’s routine methods. Australia. Most reports of such lesions have come from the USA However, if the organism is erythromycin-resistant by disc and have not been as frequently reported from other countries. testing, inducible clindamycin resistance should be tested for by In the UK, where spider bites are rare, these ‘dermatonecrotic’ the D-zone test, which involves placement of erythromycin and lesions increase the likelihood of a diagnosis of CA-MRSA but clindamycin discs in close proximity on an agar plate inoculated are not pathognomic. They can also be found in infections due with a standardized suspension of the isolate. Flattening of the to PVL-positive MSSA strains. clindamycin zone of inhibition in the area between the two discs (resulting in a D-shaped zone of inhibition) indicates the pre- Recommendation 1 sence of inducible clindamycin resistance in the presence of ery- thromycin resistance (positive D-zone test). † If ‘spider bite’ lesions are present, the possibility of CA-MRSA or PVL-positive MSSA infection should be considered and appropriate investigation and manage- 3 6. Clinical case scenarios ment instituted. [D ].

In the following section, three case scenarios are presented. The Many other severe cutaneous complications of CA-MRSA key questions related to clinical assessment are presented and have been reported and include extensive cellulitis, necrotizing guidance in the form of recommendations (in bold) are pro- fasciitis and purpura fulminans.47 Involvement of adjacent struc- vided. Additional guidance is provided on isolation of MRSA in tures, either by direct spread or bacteraemia, such as septic the urine. thrombophlebitis, pyomyositis, septic arthritis and osteomyelitis, has all been described.15,40,42 No particular patterns of clinical 6.1 Community-associated and community-onset MRSA presentation have yet emerged to allow differentiation from MSSA infections. skin and soft tissue infection Anecdotal reports suggest that recurrent (two or more in 6 A 14-year-old school boy with a skin infection is brought by his months) furuncles or abscesses, or clusters of infections within a mother to their GP. He is not systemically unwell but has a red, household may indicate PVL-positive CA-MRSA. [D3]. However, tender, fluctuant discharging skin lesion over his thigh. There is this pattern can also be seen in PVL-positive MSSA infections.

981 Review

Recommendation 2 and almost half of those without any of these factors were found to have MRSA.24 † If there is a history of recurrent abscesses or household CA-MRSA has been linked to the specific risk factors out- clusters of infection, the possibility of CA-MRSA or PVL- lined in Table 2. However, the value of these risk factors in dif- positive MSSA infection should be considered and appro- 3 ferentiating between MRSA and MSSA and predicting priate investigation and management instituted. [D ]. successful antimicrobial therapy is uncertain and currently should not be relied upon. [C22]. In UK practice, many of these 6.1.2.2 Treatment factors. In the absence of an undrained abscess risk factors are probably not relevant, except for (1) history of or focus of infection, a poor response to existing therapy (which travel to an endemic area such as North America and (2) recent in most cases will be an isoxazole penicillin such as fluclox- colonization or contact with CA-MRSA. 3 acillin) increases the likelihood of CA-MRSA infection [D ]. Recommendation 5 Recommendation 3 † If any of the risk factors outlined in Table 2 are † If there has been a prior poor response to b-lactam present, especially recent travel to an endemic area therapy, the possibility of CA-MRSA or PVL-positive such as North America and recent colonization or MSSA infection should be considered and appropriate contact with CA-MRSA, the possibility of CA-MRSA investigation and management instituted. [D 3]. infection should be considered and appropriate investi- gation and management instituted. [C 2–].

Exposure to one or more antibiotics in the past year (as opposed Downloaded from to no use) and use of quinolones or macrolides are potential treatment-related risk factors for CA-MRSA infection. [D2]. 6.1.2.4 Prevalence of CA-MRSA. The threshold where the preva- In a large English case–control study, there was a significant lence of CA-MRSA in the community becomes an important relationship between exposure to increasing numbers of anti- consideration in determining the choice of initial empirical anti- biotic therapy is uncertain. In the UK, the prevalence of biotics and diagnosis of CA-MRSA. This paper made no distinc- http://jac.oxfordjournals.org/ tion related to the PVL status of these organisms. For exposure CA-MRSA infection is very low or unknown. Although there is to 1, 2–3 or 4 antibiotics in the previous year, the odds ratios evidence of emerging PVL-related infection in certain parts of (ORs) were, respectively, 1.57 (CI 1.36–1.80), 2.46 (CI 2.15– the country, prevalence of infection or colonization on its own 2.83) and 6.24 (CI 5.43–7.17). Receipt of quinolones [OR 3.37 currently cannot be regarded as a reliable indicator of the likeli- (CI 2.80–4.09)] and macrolides [OR 2.50 (CI 2.14–2.91)] in hood of MRSA infection in an individual patient and therefore the previous year was specific associations.48 However, it is not is not a guide to antibiotic choice. [GPP]. clear in this study whether these cases were due to PVL-positive Recommendation 6 or -negative strains of S. aureus. In another large prospective study of adult patients in the USA with SSTIs presenting to the † Because the prevalence of CA-MRSA in the UK is pre- by guest on August 8, 2012 emergency department, results of a multi-regression analyses sently very low, suspected community staphylococcal showed that use of an antibiotic in the past month (as opposed infections should not be treated as if they were to no use) had a 2.4-fold (95% CI 1.4–4.3) greater risk of infec- CA-MRSA in the absence of significant specific risk tion with CA-MRSA.24 factors. This is an evolving situation and the local prevalence of CA-MRSA should be monitored and Recommendation 4 advice modified as necessary. † If there is a history of exposure to one or more anti- biotics in the past year, especially quinolones or macro- lides, the possibility of CA-MRSA infection should be 6.1.3 When should specimens be sent for culture? considered and appropriate investigation and manage- 2 ment instituted. [D ]. Recommendation 7 † Cultures should be taken from septic sites if: 6.1.2.3 Other risk factors. A recent prospective USA cohort (1) CA-MRSA is suspected because of the risk assessment study found that clinical and epidemiological risk factors in based on clinical presentation, treatment factors and persons hospitalized for CA-MRSA infection cannot distinguish other risk factors outlined in Tables 1 and 2. [D3]. reliably between MRSA and MSSA.37 Indeed, in this study, post (2) There are recurrent furuncles or abscesses (two or hoc modelling revealed that in a country where the prevalence more in 6 months). [D 3]. of MRSA in the community is ,10%, as is presently the case (3) There is a history of spread in the family or to others, in the UK, patients lacking the three strongest risk factors would e.g. sporting contacts (the information may be avail- only have a 7% post-test probability of MRSA. In another pro- able from the public health/infection control team). spective prevalence study of MRSA infections among patients in [D 4]. the emergency department, the presence or absence of these (4) If there is severe infection (extensive or progressive clinical and epidemiological risk factors was not useful in disease with evidence of systemic sepsis), the patient guiding decisions about the use of antibiotic therapy. Most should be hospitalized and a skin/abscess culture and patients without MRSA had at least one of these risk factors, blood culture should be taken. [GPP].

982 Review

Cultures will determine the prevalence of these infections in 6.1.6 Empirical antibiotic therapy a particular setting or community and will allow appropriate . There is conflicting evidence that inappropriate empirical therapy rationalization of antibiotic treatment. for SSTIs at the time of initial clinical presentation leads to an inferior outcome. Four studies18,24,50,51 compared susceptibility of 6.1.4 When should specimens for culture not be taken? the pathogen to the prescribed antimicrobial agent with clinical outcome. The patients involved mainly had abscesses. There was no difference in outcomes, suggesting that these infections, even Recommendation 8 when caused by MRSA, can be cured with drainage only. † Do not take cultures routinely from patients presenting However, in one recent retrospective analysis of the impact of with minor SSTIs and without a history of previous antimicrobial therapy on outcome for uncomplicated community- MRSA. [D 4]. onset SSTIs, there was a small but significant (OR 2.80; 95% CI † Do not routinely aspirate material for culture from cel- 1.26–6.22; P ¼ 0.01) increase in treatment failures (defined as lulitis in the absence of discharge or broken skin. worsening of signs of infection, requiring additional incision and [GPP]. drainage, subsequent hospital admission, microbiological failure or new culture proven lesions during antimicrobial therapy), in those who received initial therapy with agents ineffective in 6.1.5 Treatment principles vitro.52 Use of ineffective agents was the only independent predic- tor of treatment failure in multivariate analysis. Recommendation 9A The discrepancy between the results in this study and those

of previous reports could be due to a variety of reasons. They Downloaded from † Do not give systemic antibiotics to patients with minor 52 3 include: the larger size of the study population, which may be SSTIs or small abscesses (<5 cm). [D ]. more likely to detect a true difference in outcome compared † Incise and drain small abscesses without cellulitis and 12 with smaller studies; different patient populations, such as adults do not give antibiotic therapy. [A ]. from one healthcare system compared with children from † After incision and drainage start empirical or culture- another; and differences in underlying health status.53 guided systemic antibiotic therapy for larger abscesses http://jac.oxfordjournals.org/ 3 Furthermore, since most of these patients had non-life- or if there are infections in other family members. [D ]. threatening cutaneous infections, it is difficult to extrapolate these data to more severe infections. Additional support for Recommendation 9B empirical therapy that is active against the isolated pathogen comes from a retrospective chart review of 399 sequential, † Guidance regarding topical antibiotic treatment, deco- culture-confirmed community-onset S. aureus SSTIs, of which lonization and screening is beyond the remit of this 227 were due to MRSA. Use of an effective agent in empirical guideline although the group recommend that existing therapy was associated with an increased odds ratio [OR 5.91 evidence for need and effectiveness is poor. For further (95% CI 3.14–11.13)] of clinical resolution when controlled by guest on August 8, 2012 information, refer to guidance provided by the HPA.3 for incision, drainage and HIV status.54 This study confirmed the effectiveness of trimethoprim/sulfamethoxazole as the empirical therapy in this setting. A further prospective random- A recent randomized controlled trial, in a single centre with ized trial of empirical therapy of trimethoprim/sulfamethoxazole large numbers of injecting drug users, compared treatment with or doxycycline for outpatient SSTIs in an area of high MRSA cefalexin, which lacks activity against MRSA, with placebo in prevalence showed the equivalent effectiveness of either 166 adult patients who had surgically drained abscesses. therapy, although the treatment failure rates were 9% in the Although MRSA was isolated from most patients, more than 85% trimethoprim/sulfamethoxazole arm compared with none in of patients in both arms were cured without the need for the other.55 additional therapy at a 1 week follow-up visit, suggesting that Epidemiological profiling of CA-MRSA strains in 49 most of these infections resolve without antimicrobial therapy. England and Wales over a 2 year period 2004–05 suggests Another prospective study of otherwise healthy children with skin that all were susceptible to clindamycin, trimethoprim, and soft tissue abscesses concluded that incision and drainage of vancomycin, linezolid and mupirocin. 56 CA-MRSA abscesses with a diameter of less than a 5 cm without CA-MRSA strains that are erythromycin-resistant (by posses- 50 antibiotics was adequate. However, these findings should be sion of the erm gene) and are initially susceptible to clindamy- balanced by reports of recurrence or worsening of infections not cin can potentially develop resistance to clindamycin during 51 treated with systemic antibiotics effective against MRSA. therapy. The global reported rates of such inducible resistance Recommendation 9C vary from 2% to 94%. A double disc diffusion test (D-test) can be used to determine whether clindamycin-susceptible † In compromised patients or those with severe disease, CA-MRSA strains harbour inducible resistance.57 The local give systemic antibiotic therapy based on clinical assess- laboratory should perform a D-test. ment and local susceptibilities of strains while awaiting In severe infections with features of toxic shock or necrotiz- definitive susceptibility results. [GPP]. ing fasciitis, there is a theoretical case for using two or three † Ensure that empirical treatment also provides cover agents such as linezolid combined with clindamycin and rifam- against S. pyogenes. [GPP]. Oral flucloxacillin and clin- picin. This is based on in vitro synergy58 and the ability of line- damycin have activity against S. pyogenes, whereas zolid and clindamycin to inhibit toxin production.58,59 [GPP]. A tetracycline and trimethoprim often do not. D-test should be performed if clindamycin is used. Existing

983 Review

MRSA treatment guidance does not recommend three agents but breath. Nobody else in the family has been unwell. She was pre- rather linezolid with rifampicin as initial therapy.4 viously healthy, but had a flu-like illness 5 days prior to her rapid deterioration. She had been given amoxicillin the previous Recommendation 10 day by her GP. She is admitted to hospital with the working diagnosis of community-acquired pneumonia (CAP). Her temp- † Because of the absence of evidence of rising prevalence erature is 39.58C, respiratory rate is 40/min, white cell count is of CA-MRSA in the UK and the lack of unequivocal 3.7109/L and C-reactive protein (CRP) 360 mg/L. Chest X-ray evidence that inappropriate antimicrobial therapy reveals some increased shadowing in both lungs, with fluffy alters outcome, there is no need to change existing opacities. empirical therapy recommendations (below) for non- severe presumed S. aureus infections. [C3]. 6.2.1 What is the likely clinical diagnosis? † In the UK, the recommended community treatment for suspected MSSA infections is oral flucloxacillin 500– This clinical diagnosis is of a community-acquired lower respir- 1000 mg 6 hourly (or oral clindamycin 300–450 mg 6 atory tract infection, with the unusual feature of severe haemop- hourly in penicillin allergic patients); 5–7 days of treat- tysis. Severe CAP with haemoptysis following a flu-like illness ment is normally sufficient. could also indicate primary influenza pneumonia, although the † If the patient is known to be MRSA-positive OR lesion absence of family members with flu-like symptoms is against cultures yield MRSA alone, then community treatment this. The most likely microbial cause of the pneumonia is bac- should be either oral doxycycline (contra-indicated in terial and S. aureus should be considered as a potential pathogen because of the combination of high CRP and low white cell children <12 years) 100 mg 12 hourly, or fusidic acid Downloaded from 500 mg 8 hourly, or trimethoprim 200 mg 12 hourly, count in a young patient with haemoptysis. Streptococcus pneu- each combined with rifampicin 300 mg 12 hourly (see moniae and S. pyogenes may also present with a similar picture. Appendix). Fusidic acid and rifampicin should NOT The fact that she has worsened on amoxicillin therapy also sup- be used as monotherapy because of the danger of resist- ports a diagnosis of staphylococcal rather than streptococcal ance emergence. All these agents can be used in penicil- pneumonia. Pulmonary embolus is unlikely because the features lin allergic patients. of sepsis predominate, but this should be excluded if in doubt. http://jac.oxfordjournals.org/ † Trimethoprim (combined with sulfamethoxazole) or doxycycline without rifampicin is also effective for 6.2.2 When to suspect CA-MRSA ambulatory therapy of MRSA SSTIs.51 [A11]. Oral 6.2.2.1 Clinical presentation. There is nothing in this history to linezolid 600 mg twice daily is an alternative option for suggest infection with CA-MRSA rather than with CA-MSSA. use ‘under expert guidance’, but because of its high There are features, however, consistent with PVL-associated sta- cost it should be reserved for patients who are not able phylococcal disease, including haemoptysis, high respiratory rate to take or tolerate the above regimens. Linezolid may and a low white cell count (leucopenia) in the presence of a

not be readily available in primary care pharmacies. high CRP and systemic sepsis in a previously healthy individ- by guest on August 8, 2012 † If Group A streptococcal (GAS) infection is suspected, ual.20,60 The presence of blood in sputum should alert the clini- oral therapy should include an agent active against this cian to the possibility of PVL production. Assessing the severity organism (b-lactam or clindamycin). of pneumonia in children or young adults should not include † For severe infections with known or suspected age-dependent scoring systems such as CURB-65 as the score CA-MRSA, start treatment in hospital with parenteral will be misleadingly low. vancomycin, teicoplanin, daptomycin (but not for pneu- monia) or linezolid. Tigecycline may also offer broader Recommendation 11 polymicrobial cover if required. There is no evidence † Consider a diagnosis of lower respiratory tract infection that one agent is superior to another. [GPP]. caused by PVL-producing S. aureus in rapidly progress- † In severe infections with features of toxic shock or ive pneumonia evolving into acute respiratory distress necrotizing fasciitis, there is a theoretical case for using syndrome. A fever of >3988888C, respiratory rate of >40 two or three agents such as linezolid combined with breaths per minute, a tachycardia of >140 beats per clindamycin and rifampicin. minute with haemoptysis and hypotension make the diagnosis likely. The presence of significant haemoptysis 21 More detailed guidance in this area is available.4 A good and hypotension usually confirms the diagnosis. [C ]. example of a care pathway for SSTIs has recently been pub- lished from Washington, USA (http://www.metrokc.gov/health/ Additional questions should be asked to elicit whether any providers/epidemiology/MRSA-guidelines.pdf). family members have a history of skin sepsis, any contact with healthcare facilities or are known MRSA carriers. 6.2 Community-associated and community-onset Recommendation 12 pneumonia suspected to be due to PVL-producing MRSA † Although relatively few patients developing necrotizing (this would apply to PVL-producing MSSA also) pneumonia due to CA-MRSA have a previous history of A 37-year-old mother of two young children is admitted to hos- skin sepsis themselves,60 – 64 consider the possibility of pital very unwell with a 48 h history of pleuritic chest pain, CA-MRSA infection if there is a history of recurrent cough with frank haemoptysis and increasing shortness of (two or more in 6 months) furuncles or abscesses in

984 Review

family members or clustering of infections within the CURB-65 score of three or above. However, if a diagno- household. [D3]. sis of CAP has been definitely confirmed, and a patient has no severity indicators or co-morbid disease, then blood cultures may be omitted. [A2].66 6.2.2.2 Treatment factors. Use of any antibiotic in the past month (versus no use) has been identified as a potential treatment-related risk factor for CA-MRSA SSTI. [D2]. Taking cultures will help to estimate the prevalence of However, there are no such data for CA-MRSA pneumonia. CA-MRSA and rationalize antibiotic treatment. Fewer than 25% of patients with CA-MRSA pneumonia have had positive blood Recommendation 13 cultures.64,67– 76 Until further evidence is available, do not discount the possibility of CA-MRSA infection in severe pneumonia on 6.2.4 Radiological investigation the basis of a lack of history of recent antibiotic therapy. Multilobular alveolar infiltrates are still usual in pneumonia due to PVL-producing staphylococci. Compared with pneumonia 6.2.2.3 Other risk factors. due to HA-MRSA, they are more likely to cavitate and produce effusions.61 However, more commonly, acute PVL-related infec- Recommendation 14 tions produce initially few, if any, chest X-ray changes, leading clinicians to misdiagnose infections as simple exacerbations of † If any of the risk factors outlined in Table 2 are bronchitis or asthma.77,78 Radiological changes develop rapidly, present, especially (in the UK) recent travel to an

with single or multiple opacities ,3 cm diameter being sugges- Downloaded from endemic area such as North America and recent coloni- tive of staphylococcal infection. Cavitation may appear on serial zation or contact with CA-MRSA, the possibility of X-rays and may be detected earlier with ultrasound. However, CA-MRSA infection should be considered and appro- 22 computed tomography (CT) scanning or magnetic resonance priate investigation and management instituted. [C ]. imaging allows the best evaluation of the ongoing pathology, particularly with cystic changes.74,79 The classical multilobular 6.2.2.4 Prevalence of CA-MRSA. The threshold where the preva- infiltrates and diffuse multilobar opacities followed by cavity http://jac.oxfordjournals.org/ lence of CA-MRSA in the community becomes an important formation may develop after only a few days and are best con- 80,81 consideration in determining the choice of empirical antibiotic firmed with CT. Overall, the incidence of complicated pneu- therapy is uncertain. In the UK, the prevalence of this infection monia caused by PVL-producing S. aureus is far higher than is very low or unknown. Therefore, local variations in preva- with non-PVL-producing strains. lence currently have very little influence on the likelihood of an Recommendation 17 individual patient having CA-MRSA infection. † Consider the possibility of PVL-producing CA-MRSA Recommendation 15 pneumonia if there are suggestive chest radiological fea- by guest on August 8, 2012 † Consider the possibility of CA-MRSA infection in tures such as multilobular alveolar infiltrates, cavitation 3 severe community pneumonia regardless of the local and pleural effusions. [D ]. prevalence of CA-MRSA. [GPP]. 6.2.5 Treatment principles Clinical Practice Point Owing to the small numbers of cases reported, there is not yet In this patient, the absence of previous recurrent infections clear evidence that early appropriate antibiotic therapy will does not lessen the probability that the infection is due to improve outcomes in PVL-producing staphylococcal pneumonia. PVL-producing S. aureus. Previous exposure to antibiotics has However, with an expected mortality approaching 75%,60,82 and been recognized as a potential risk factor for HA-MRSA but is in line with data from other types of CAP, hospital-acquired not yet identified clearly as a risk for CA-MRSA.65 pneumonia, ventilator-associated pneumonia and severe sepsis, early intensive care support and appropriate antimicrobial 6.2.3 When should specimens be sent for culture? therapy are essential. Recommendation 18 Recommendation 16 † Refer patients with suspected CA-MRSA pneumonia to † Gram’s stain and culture should be performed immedi- intensive care as soon as possible. Basic principles of ately on admission. The Gram stain result may point to resuscitation should be followed and ventilatory support the identity of the likely infecting organism. Relative implemented when clinically necessary. [GPP]. paucity of neutrophils in the Gram stain in a patient † Pending antibiotic susceptibility results implement with advanced pneumonia, severe haemoptysis and a empirical antibiotic therapy that covers CA-MRSA as low white count is supportive of PVL-associated staphy- soon as possible. [GPP]. lococcal pneumonia. [GPP]. † The 2004 update of the British Thoracic Society (BTS) guidelines recommends taking blood cultures from 6.2.6 Empirical antibiotic therapy patients with severe CAP preferably before antibiotic According to the BTS guidance,66 patients with severe CAP therapy is commenced in patients with a severity should receive antibiotics (co-amoxiclav or cefuroxime or

985 Review cefotaxime plus a macrolide) within 2 h of hospital admission. Working Party recommends that consideration be given to using [C]. Many current hospital antibiotic policies recommend avoid- this combination for initial therapy.90 The combination can be ing cephalosporin use because of the association between these started empirically in a patient with the clinical features listed antibiotics and Clostridium difficile colitis. The Working Party above AND Gram-positive cocci in clusters in respiratory recommends that the CURB-65 severity scoring tool for CAP secretions. Some experts suggest adding rifampicin 600 mg should not be applied to young people or children who initially twice daily, based on synergistic activity and intracellular clear- may appear to only have a mild respiratory illness. [GPP]. ance of staphylococci.91 The potential role of tigecycline (a gly- Unsuspected MRSA caused fatal pneumonia in four cylcycline) or tetracyclines in severe necrotizing PVL-associated Minnesotan children who were initially treated empirically with pneumonia is as yet unclear and daptomycin is not indicated as cephalosporins.61 Conventional doses of vancomycin may it is inactivated by lung surfactant.92 produce inadequate lung concentrations for MRSA infection and, despite high trough serum levels, breakthrough continuous Recommendation 19 bacteraemia has been reported days into glycopeptide † The CURB-65 severity scoring tool for CAP should not therapy.79,82 be applied to young people or children who initially Antimicrobials effective against MRSA that also decrease may appear to only have a mild respiratory illness. exotoxin production, such as clindamycin and linezolid, are [GPP]. theoretically desirable. Clindamycin decreases production of † If a risk assessment suggests the possibility of toxic shock syndrome toxin 1 by 95% in stationary-phase cul- CA-MRSA, then empirical therapy for CAP should tures83 and stops the normal peak of a-toxin production during 59 include cover for CA-MRSA. Consider adding linezolid the late exponential phase of growth. Clindamycin and line- Downloaded from 600 mg 12 hourly and high dose clindamycin 1.2–1.8 g zolid both markedly suppress PVL production as staphylococci 6 hourly. If the organism is isolated, check that it is approach stationary phase and there may be no PVL detectable susceptible to clindamycin and the D-test is negative. 12 h after starting treatment.59 Flucloxacillin is bactericidal, but [GPP]. Some authors have also suggested the routine the low subinhibitory concentrations achievable in vivo in addition of rifampicin 600 mg twice daily if necrotic tissue may further augment PVL toxin and 3,90 PVL-positive staphylococcal pneumonia is suspected. http://jac.oxfordjournals.org/ a-haemolysin production.59 Subinhibitory concentrations of [GPP]. The roles of ceftobiprole (a cephalosporin with clindamycin, linezolid and fusidic acid all induce a anti-MRSA properties) and tigecycline in this setting concentration-dependent decrease of PVL concentration, have yet to be determined. whereas with low concentrations of oxacillin, the concentration of PVL increases up to 3-fold.84 Various combinations of vancomycin, clindamycin, linezolid, rifampicin and co-trimoxazole have been used in differing doses 6.2.7 Adjunctive therapy options and combinations in PVL pneumonia cases, with varying 6.2.7.1 Immunoglobulin therapy (IVIG). IVIG neutralizes the degrees of success.20,79,82– 85 Linezolid treatment successes have cytopathic effect and pore-formation of PVL in vitro, the inhi- by guest on August 8, 2012 been reported by several authors.79 –81,86 –88 Three of four bition being concentration dependent.93 No randomized con- patients with necrotizing pneumonia clinically failing vancomy- trolled trials have been performed to assess the role of cin therapy responded to a change to linezolid and rifampicin.79 immunoglobulin therapy in this setting. It has been used in Decreased vancomycin susceptibility discouraged the use of several patients with PVL-associated pneumonia.63,77,86,94 The vancomycin in two cases.63,81 A PVL-positive USA300 MRSA optimal dosage of IVIG is uncertain; 2 g/kg is recommended for strain for which the vancomycin MIC was 2–4 mg/L responded streptococcal toxic syndrome,95 repeated at 48 h if there is still to a combination of linezolid, teicoplanin and rifampicin, evidence of sepsis or failure to respond. The combination of although the infected patient was hospitalized for 6 weeks.81 linezolid and IVIG was effective in a boy with septic arthritis Predicting the susceptibility of staphylococci is becoming and pneumonia, who was discharged from intensive care to the increasingly difficult and depends largely on the geographical general ward on day 5.86 location and clonality of the circulating strains. Most isolates of strain USA300 are resistant only to b-lactams and macrolides, Recommendation 20 but recently mupirocin, tetracycline, clindamycin and fluoroqui- † Although most supporting evidence is anecdotal, the 89 nolone resistances have been reported. Twelve of 123 isolates use of immunoglobulin (IVIG) should be considered in examined contained tetK and ermC genes, but remained suscep- severe sepsis and necrotizing pneumonia known or sus- tible to doxycycline and minocycline. S. aureus isolates resistant pected to be due to S. aureus. IVIG should be given at a to erythromycin but sensitive to clindamycin must be ‘D-tested’ dose of 2 g/kg, repeating the dose if improvement is not to exclude inducible clindamycin resistance. Combining clinda- satisfactory. [GPP]. mycin with linezolid is synergistic in vitro.58 In the UK, pneumo- nia due to clindamycin-resistant CA-MRSA has not been reported so far and the majority of CA-MRSA are PVL-negative (Angela 6.2.7.2 Other adjunctive therapy. Anecdotal reports of the usage Kearns, Laboratory of Healthcare-Associated Infection, HPA, per- of granulocyte colony-stimulating factor96 and extra-corporeal sonal communication). membrane oxygenation64,94,97 suggest that they are largely Although there is as yet no unequivocal clinical evidence to unsuccessful. Although of theoretical benefit in very early support the combination of linezolid (intravenous, 600 mg 12 sepsis, once active haemorrhage has occurred, activated protein hourly) plus clindamycin (intravenous, 1.2–1.8 g 6 hourly), C should not be used. It therefore has no role in PVL-associated because of the life-threatening nature of this disease, the pneumonia.

986 Review Downloaded from http://jac.oxfordjournals.org/ by guest on August 8, 2012

Figure 2. Management of patient with suspected staphylococcal pneumonia in the healthcare setting. IVIG, intravenous immunoglobulin.

Recommendation 21 6.3 Healthcare-associated community-onset MRSA skin and soft tissue infection † Based on current available evidence and experience, the Although we have chosen the example below, another very Working Party cannot provide guidance on the value of common scenario is a recently discharged hospitalized patient the above adjunct therapies. who presents with an ‘infected leg ulcer’ from which MRSA is isolated. The generic guidance and decision-making process pro- Figure 2 provides an algorithm that may help clinicians vided here should also apply to this scenario. implement the recommendations related to suspected staphylo- The district nurse requests a GP to do a home visit on a coccal pneumonia. 73-year-old man. She has been attending the patient regularly

987 Review over the last 3 weeks to dress a pressure sore that had developed history of frequent hospital admissions or admission to hospital over the sacrum. The man had recently been in hospital for a within the last 6 months); surgery or admission to an intensive revision of his left hip prosthesis. The procedure was compli- care unit within the last 6 months; exposure to invasive devices cated by an episode of healthcare-acquired pneumonia and of all types, especially central venous catheters; previous MRSA rehabilitation had been slow. He had spent 4 weeks in hospital colonization/infection or exposure to an MRSA-colonized before discharge and during this time he had developed a 6 cm patient; the presence of extensive wounds and/or burns; and irregular pressure sore over the sacrum. In the 2 days before the exposure to antimicrobial drugs, especially cephalosporins, attendance of the GP, the lesion had become slightly larger with fluoroquinolones and macrolides.47 an increase in foul-smelling exudate. On the day of the visit, The ability of these risk factors to predict the likelihood of there was now a spreading area of redness (erythema) around HA-MRSA is uncertain. In recently hospitalized (,24 h) the site. Although there was now some pain at the lesion, there patients with staphylococcal bacteraemia, one case-controlled were no other systemic signs of infection. study identified previous MRSA infection or colonization as the strongest predictor of HA-MRSA. In this model, previous 6.3.1 What is the likely clinical diagnosis? MRSA infection or colonization, presence of a central venous catheter, documentation of a skin ulcer or cellulitis at hospital The in this case would be defined as Grade 3 as it admission were all independently associated with HA-MRSA involves full-thickness skin loss with damage or necrosis of the bacteraemia. The model identified 75% of the cases correctly subcutaneous tissues, extending down to, but not penetrating with a sensitivity of 76% and specificity of 73%.14 In another through, the underlying fascia.98 All such ulcers are colonized cohort study of inpatients with S. aureus bacteraemia, infection with a mixture of organisms. A chronic non-healing ulcer may with HA-MRSA was associated with prior antibiotic exposure Downloaded from reflect underlying osteomyelitis. The diagnosis of active infec- (within 60 days of the bacteraemic episode) [OR 9.2 (95% CI tion involving a pressure ulcer must therefore depend upon clini- 4.8–17.9)]; presence of decubitus ulcers [OR 2.5 (95% CI 1.2– cal assessment rather than microbiology culture results. The 4.9)], hospital onset [OR 3.0 (95% CI 1.9–4.9)] and history of European Wound Management Association has defined the cri- hospitalization within 6 months of episode of infection [OR 2.5 teria for recognizing early wound infection and implementing an (95% CI 1.5–3.8)]. In this model, the clinicians first had to escalating therapeutic strategy.99 The overt signs of local infec- http://jac.oxfordjournals.org/ identify the bacteraemia as community or hospital onset, then, tion (Stage 3 involvement) include the discharge of pus with depending on the risk factors of either prior antibiotic exposure swelling, pain, erythema and local warmth. On examination, the and/or decubitus ulcer, an estimate of the risk of MRSA could surrounding tissues may appear unhealthy and deteriorating. The be made. Thus, if the infection was community onset, the principal pathogens associated with active wound infections of patient had recent antibiotic exposure and had a decubitus ulcer, pressure ulcers are S. aureus, Streptococcus species, anaerobes the likelihood of MRSA is 91%. This model is limited to a and Pseudomonas aeruginosa.100 The likely diagnosis here is specific cohort of patients from one site, needs broader vali- therefore of a locally infected pressure ulcer where S. aureus is dation and is applicable only to hospitalized patients with sta-

one of the most frequently associated pathogens. by guest on August 8, 2012 phylococcal bacteraemia. However, these principles may prove helpful in assessing patients in the community. 6.3.2 When to suspect HA-MRSA Recommendation 23 6.3.2.1 Clinical presentation. There are no specific clinical appearances or relationships that reliably identify a SSTI lesion † When managing a patient with a community-onset as being infected with S. aureus or any particular type of SSTI, the possible involvement of HA-MRSA should be 44,53 S. aureus. Three types of S. aureus are recognized as causing considered and the appropriate risk assessment done. disease in community settings, namely MSSA, CA-MRSA and Recent hospitalization, previous MRSA infection or HA-MRSA. The spectrum of disease caused is similar, with SSTI colonization, previous antibiotic exposure and the pre- being the commonest related condition. However, CA-MRSA is sence of a decubitus ulcer or wound should alert clini- more strongly associated with SSTI than HA-MRSA and is more cians to the possibility of HA-MRSA. [D 3]. likely to occur in younger patients.101 Although the appearances of SSTI with these types of pathogen are very similar, CA-MRSA is more often associated with red, raised lesions with a central area of necrosis (see case scenario 1). 6.3.3 When should specimens be sent for culture? The early diagnosis and treatment of infection in patients with a Recommendation 22 Stage 3 infected pressure ulcer reduces the risk of complications † There are no clinical characteristics that allow differen- and leads to improved patient outcomes. Knowing the identity tiation between different strains of MRSA infection. and antimicrobial susceptibility of organisms infecting pressure [GPP]. ulcers is of help when an antimicrobial treatment has failed, when the presence of a resistant pathogen is suspected or when a patient requires screening for a specific organism.99 Because the presence of MRSA will alter the choice of antimicrobial 6.3.2.2 Risk factors. The risk factors associated with HA-MRSA therapy, it is sensible to culture for MRSA when there is a sig- are well defined and include:44,102,103 advanced age; underlying nificant risk and infection is sufficiently severe to warrant sys- co-morbidities and severity of illness; inter-institutional transfer temic therapy. If systemic therapy is not indicated, then there is of the patient, especially from a nursing home or residence in a little benefit from taking a sample except for infection control long-term care facility; prolonged hospitalization (including a purposes.

988 Review

Recommendation 24 over an 18 month period of MRSA infection among adult patients harbouring MRSA has been shown to be 29%,104 † In an episode of SSTI involving an infected pressure with 28% of them complicated by bacteraemia and 56% by sore or another type of wound infection, a culture pneumonia, soft tissue infection, osteomyelitis or septic arthritis. should be taken if the condition warrants systemic anti- On subsequent admission to hospital, MRSA carriage is associ- microbial treatment or for infection control purposes. ated with an increased risk of sepsis. One study showed that [GPP]. 19% of the patients who were MRSA culture positive from an admission nasal sample subsequently developed infection with MRSA during their hospital stay, compared with 1.5% of cases 6.3.4 Treatment principles colonized with MSSA and 2.0% of uncolonized patients.105 The principles of antibiotic therapy are as outlined in scenario 1 Hence, MRSA colonization at admission significantly increased and in the Appendix. the risk of subsequent S. aureus infection compared with MSSA colonization. There are also data to show that morbidity and † Give appropriate systemic antibiotic treatment guided by mortality are increased with MRSA when compared with MSSA laboratory susceptibility results. infections, even when controlling for co-morbidities and other † Use appropriate topical antimicrobial and dressing care. risk factors.106,107 This may be related to the increased risk of † Implement appropriate surgical adjunctive management. initial therapy not covering MRSA. A significant increase in † Consider the possibility of underlying complications such mortality has been found to be associated with MRSA when as osteomyelitis. compared with MSSA bacteraemia [OR 1.93; 95% CI 1.54– 106 2.42; P , 0.001]. MRSA infection may also carry an increase Downloaded from in morbidity related to additional days of treatment required 6.3.5 Empirical antibiotic therapy resulting in increased hospital stay.107,108 No strong relationships In the majority of such patients, MRSA will have been isolated have been identified with other morbidity outcomes.109 from a recent microbiological specimen at the time of the con- Thus, HA-MRSA carriage is associated with an increased risk of infection and morbidity and mortality. However, no such

sultation and appropriate MRSA treatment can be started. This http://jac.oxfordjournals.org/ therapy should be guided by susceptibility results. data are available for CA-MRSA. To date, there are no data to support the use of agents to eliminate S. aureus colonization in Recommendation 25 relation to community MSSA or CA-MRSA SSTIs. † In cases where empirical therapy is required and there Recommendation 26 are significant risk factors for HA-MRSA, the Working Group recommends starting one of the regimens ident- † Standard infection control advice should be given to ified in the Appendix after taking appropriate microbio- patients with SSTIs due to HA-MRSA. This should logical specimens. This therapy should be rationalized in include the importance of dressings management, hand light of microbiological culture results. [GPP]. decontamination and the avoidance of transfer of infec- by guest on August 8, 2012 † For severe or more progressive SSTIs, there is as yet no tion by, for example, sharing razors, contact sports etc. conclusive evidence that empirical therapy covering MRSA leads to improved outcome. However, in severe Acknowledgements infections, where a risk assessment suggests the likeli- hood of HA-MRSA, high-dose empirical therapy The review of this guideline was initiated by the Specialist against MRSA should be used. [GPP]. Advisory Committee on Antimicrobial Resistance (SACAR), the † Patients with severe infections should be admitted to Department of Health and the HPA. It was undertaken indepen- hospital where skin and blood cultures should be taken, dently by a Working Party of the BSAC in close collaboration collections drained, tissues debrided as necessary and with the HPA. parenteral antibiotic therapy started. Appropriate infection control measures should be instigated and urgent consultation made with a local infection special- Transparency declarations ist. [GPP]. † For severe infections where HA-MRSA might be D. N. declares that during the preparation of this document, he involved, the Working Party recommends starting the was not in the employment of any pharmaceutical firm with treatment recommended in the Appendix. [GPP]. interest in the content of the guidelines although he did accept Wound care should be carried out in the community appointments to the advisory boards of Pfizer, Wyeth, Johnson under strict aseptic technique. Hand hygiene is necess- & Johnson and Novartis, and has spoken at symposia supported ary before and after direct patient contact. by them. † For further information relating to isolation, screening M. D. has received honoraria for consultation and speaking and decolonization, refer to the guidelines for the symposia from Bayer, Novartis, Arpida and Pfizer. control and prevention of MRSA in healthcare B. D. C. accepted appointments to the advisory boards of facilities. 2 GlaxoSmithKline, Wyeth and Merck Sharp and Dohme, and has spoken at symposia supported by 3M. G. F. has accepted appointments to the advisory boards of HA-MRSA carriage constitutes a greater risk for the develop- Pfizer, Wyeth and Novartis, and has spoken at symposia sup- ment of S. aureus infection than MSSA carriage.103 The risk ported by them.

989 Review

R. G. M. has received speaker honoraria from AstraZeneca 12. Siegel JD, Rhinehart E, Jackson M et al. Management of and Wyeth. multidrug-resistant organisms in healthcare settings. http://www.cdc. M. M. and D. L. have nothing to declare relevant to this gov/ncidod/dhqp/pdf/ar/mdroGuideline2006.pdf (6 February 2008, date publication. last accessed). 13. Adedeji A, Weller TM, Gray JW. MRSA in children presenting to hospitals in Birmingham, UK. J Hosp Infect 2007; 65: 29–34. 14. Tacconelli E, Venkataraman L, De Girolami PC et al. Comment on editorial process Methicillin-resistant Staphylococcus aureus bacteraemia diagnosed at hospital admission: distinguishing between community-acquired versus This Working Party Report was put out for consultation on healthcare-associated strains. J Antimicrob Chemother 2004; 53: 11 October 2007 (consultation period closed on 8 November 474–9. 2007) and amended in light of the comments prior to its sub- 15. Zetola N, Francis JS, Nuermberger EL et al. mission to the journal. This consultation exercise replaced the Community-acquired meticillin-resistant Staphylococcus aureus:an journal’s peer review process. emerging threat. Lancet Infect Dis 2005; 5: 275–86. 16. Vandenesch F, Naimi T, Enright MC et al. Community-acquired methicillin resistant Staphylococcus aureus carrying Panton–Valentine Supplementary data leukocidin genes: worldwide emergence. Emerg Infect Dis 2003; 9: 978–84. Tables S1 and S2 show the SIGN system used to grade the evi- 17. Genestier AL, Michallet MC, Prevost G et al. Staphylococcus dence and recommendations and are available as Supplementary aureus Panton–Valentine leukocidin directly targets mitochondria and data at JAC Online (http://jac.oxfordjournals.org/). induces Bax-independent apoptosis of human neutrophils. J Clin Invest Downloaded from 2005; 115: 3117–27. 18. Fridkin SK, Hageman JC, Morrison M et al. Methicillin-resistant References Staphylococcus aureus disease in three communities. N Engl J Med 2005; 352: 1436–44. 1. Coia JE, Duckworth GJ, Edwards DI et al. Guidelines for the 19. Labandeira-Rey M, Couzon F, Boisset S et al. Staphylococcus http://jac.oxfordjournals.org/ control and prevention of methicillin-resistant Staphylococcus aureus aureus Panton–Valentine leukocidin causes necrotizing pneumonia. (MRSA) in healthcare facilities. J Hosp Infect 2006; 63 Suppl 1: Science 2007; 315: 1130–3. S1–44. 20. Gillet Y, Issartel B, Vanhems P et al. Association between 2. Brown DF, Edwards DI, Hawkey PM et al. Guidelines for the Staphylococcus aureus strains carrying gene for Panton–Valentine laboratory diagnosis and susceptibility testing of methicillin-resistant leukocidin and highly lethal necrotising pneumonia in young immuno- Staphylococcus aureus (MRSA). J Antimicrob Chemother 2005; 56: competent patients. Lancet 2002; 359: 753–9. 1000–18. 21. Voyich JM, Otto M, Mathema B et al. Panton–Valentine leuko- 3. Health Protection Agency. Guidance on the diagnosis and man- cidin the major virulence determinant in community-associated agement of PVL-associated Staphylococcus aureus infections methicillin-resistant Staphylococcus aureus disease? J Infect Dis 2006; by guest on August 8, 2012 (PVL-SA) in the UK. Currently undergoing consultation. http://www. 194: 1761–70. hpa.org.uk/consultations/2008/pvl.htm (6 February 2008, date last 22. Holmes A, Ganner M, McGuane T et al. Staphylococcus accessed). aureus isolates carrying Panton–Valentine leucocidin genes in 4. Gemmell CG, Edwards DI, Fraise AP et al. Guidelines for the England and Wales: frequency, characterization, and association with prophylaxis and treatment of methicillin-resistant Staphylococcus clinical disease. J Clin Microbiol 2005; 43: 2384–90. aureus (MRSA) infections in the UK. J Antimicrob Chemother 2006; 57: 589–608. 23. Salgado CD, Farr BM, Calfee DP. Community-acquired methicillin-resistant Staphylococcus aureus: a meta-analysis of preva- 5. Health Protection Agency. Topics A–Z/Staphylococuus aureus/ lence and risk factors. Clin Infect Dis 2003; 36: 131–9. Guidelines. http://www.hpa.org.uk/infections/topics_az/staphylo/ staphylo_guidelines.htm (6 February 2008, date last accessed). 24. Moran GJ, Krishnadasan A, Gorwitz RJ et al. Methicillin-resistant S. aureus infections among patients in the emer- 6. von Eiff C, Becker K, Machka K et al. Nasal carriage as a gency department. N Engl J Med 2006; 355: 666–74. source of Staphylococcus aureus bacteremia. Study Group. N Engl J Med 2001; 344:11–6. 25. Frazee BW, Lynn J, Charlebois ED et al. High prevalence of 7. Jevons MP. Celbenin-resistant staphylococci. BMJ 1961; 1: methicillin-resistant Staphylococcus aureus in emergency department 124–5. skin and soft tissue infections. Ann Emerg Med 2005; 45: 311–20. 8. Wenzel RP, Nettleman MD, Jones RN et al. Methicillin-resistant 26. King MD, Humphrey BJ, Wang YF et al. Emergence of Staphylococcus aureus: implications for the 1990s and effective control community-acquired methicillin-resistant Staphylococcus aureus measures. Am J Med 1991; 91: 221–7S. USA300 clone as the predominant cause of skin and soft-tissue infec- 9. Grundmann H, Aires-de-Sousa M, Boyce J et al. Emergence tions. Ann Intern Med 2006; 144: 309–17. and resurgence of methicillin-resistant Staphylococcus aureus as a 27. Nuno A, Faria NA, Oliveira DC et al. Epidemiology of emerging public-health threat. Lancet 2006; 368: 874–85. methicillin-resistant Staphylococcus aureus (MRSA) in Denmark: a 10. Guidelines for the control and prevention of methicillin-resistant nationwide study in a country with low prevalence of MRSA infection. Staphylococcus aureus (MRSA) in hospitals. http://www.bsac.org.uk/ J Clin Microbiol 2005; 43: 1836–42. _db/_documents/mrsa_INFECTION_JUNE_05.pdf (6 February 2008, 28. Wannet WJ, Spalburg E, Heck ME et al. Emergence of virulent date last accessed). methicillin-resistant Staphylococcus aureus strains carrying Panton– 11. Revised methicillin-resistant Staphylococcus aureus (MRSA) Valentine leucocidin genes in The Netherlands. J Clin Microbiol 2005; infection control guidelines for hospitals. http://www.his.org.uk/_db/ 43: 3341–5. _documents/MRSA_Guidelines_final_(revised)_aug98.doc (6 February 29. Fossum AE, Bukholm G. Increased incidence of methicillin- 2008, date last accessed). resistant Staphylococcus aureus ST80, novel ST125 and SCCmecIV in

990 Review the south-eastern part of Norway during a 12-year period. Clin clinical features, management, and prevention. Int J Dermatol 2007; Microbiol Infect 2006; 12: 627–33. 46: 1–11. 30. Larsen AR, Stegger M, Goering RV et al. Emergence and dis- 48. Schneider-Lindner V, Delaney J, Dial A et al. Antimicrobial semination of the methicillin resistant Staphylococcus aureus USA300 drugs and community-acquired methicillin-resistant Staphylococcus clone in Denmark (2000–2005). Euro Surveill 2007; 12: [Epub ahead aureus, United Kingdom. Emerg Infect Dis 2007; 13: 994–1000. of print] http://www.eurosurveillance.org/em/v12n02/1202-222.asp. 49. Rajendran PM, Young DM, Maurer T et al. Treatment of 31. Robinson DA, Kearns AM, Holmes A et al. Re-emergence of abscesses in the era of methicillin-resistant Staphylococcus aureus— early pandemic Staphylococcus aureus as a community-acquired are antibiotics necessary? In: Program and Abstracts of the 92nd methicillin-resistant clone. Lancet 2005; 365: 1256–8. Clinical Congress of the American College of Surgeons (Chicago). 32. Anon. Hospital-associated transmission of Panton–Valentine 2006: S59. American College of Surgeons, Chicago, USA. leukocidin (PVL) positive community-assocaited MRSA in the West 50. Lee MC, Rios AM, Aten MF et al. Management and outcome of Midlands. CDR Wkly 2006; 16: 50. children with skin and soft tissue abscesses caused by 33. David MD, Kearns AM, Gossain S et al. Community-associated community-acquired methicillin-resistant Staphylococcus aureus. methicillin-resistant Staphylococcus aureus: nosocomial transmission Pediatr Infect Dis J 2004; 23: 123–7. in a neonatal unit. J Hosp Infect 2006; 64: 244–50. 51. Miller LG, Spellberg B. Treatment of community-associated 34. Voss A. Preventing the spread of MRSA. BMJ 2004; 329: 521. methicillin-resistant Staphylococcus aureus skin and soft tissue infec- 35. Millar BC, Loughrey A, Elborn JS et al. Proposed definitions of tions with drainage but no antibiotic therapy. Pediatr Infect Dis J 2004; community-associated methicillin-resistant Staphylococcus aureus 23: 795. author reply 795-6. (CA-MRSA). J Hosp Infect 2007; 67: 109–13. 52. Ruhe JJ, Smith N, Bradsher RW et al. Community-onset 36. Otter JA, French GL. Nosocomial transmission of methicillin-resistant Staphylococcus aureus skin and soft-tissue infec- community-associated methicillin-resistant Staphylococcus aureus:an tions: impact of antimicrobial therapy on outcome. Clin Infect Dis 2007; Downloaded from emerging threat. Lancet Infect Dis 2006; 6: 753–5. 44: 777–84. 37. Miller LG, Perdreau-Remington F, Bayer AS et al. Clinical and 53. Gorwitz RJ. The role of ancillary antimicrobial therapy for treat- epidemiologic characteristics cannot distinguish community-associated ment of uncomplicated skin infections in the era of methicillin-resistant Staphylococcus aureus infection from methicillin- community-associated methicillin-resistant Staphylococcus aureus. susceptible S. aureus infection: a prospective investigation. Clin Infect Clin Infect Dis 2007; 44: 785–7. Dis 2007; 44: 471–82. 54. Szumowski JD, Cohen DE, Kanaya F et al. Treatment and out- http://jac.oxfordjournals.org/ 38. Ellis MW, Hospenthal DR, Dooley DP et al. Natural history of comes of infections by methicillin-resistant Staphylococcus aureus at community-acquired methicillin-resistant Staphylococcus aureus coloni- an ambulatory clinic. Antimicrob Agents Chemother 2007; 51: 423–8. zation and infection in soldiers. Clin Infect Dis 2004; 39: 971–9. 55. Cenizal MJ, Skiest D, Luber S et al. Prospective randomized 39. Fervers B, Burgers JS, Haugh MC et al. Adaptation of clinical trial of empiric therapy with trimethoprim-sulfamethoxazole or doxycy- guidelines: literature review and proposition for a framework and pro- cline for outpatient skin and soft tissue infections in an area of high cedure. Int J Qual Health Care 2006; 18: 167–76. prevalence of methicillin-resistant Staphylococcus aureus. Antimicrob 40. Barton-Forbes M, Hawkes M, Moore D et al. Guidelines for the Agents Chemother 2007; 51: 2628–30. 56. Kearns AM, Ganner M, Hill RLR et al. Community-associated

prevention and management of community-associated methicillin by guest on August 8, 2012 resistant Staphyococcus aureus (CA-MRSA): a perspective for MRSA ST8-SCCmecIVa (USA-300): experience in England and Canadian healthcare practitioners. Can J Infect Dis Med Microbiol Wales. Clin Microbiol Infect 2007; 13 Suppl 1: S27. 2006; 17 Suppl C: 1B–24B. 57. Fiebelkorn KR, Crawford SA, McElmeel ML et al. Practical disk 41. AGREE (Appraisal of guidelines research & evaluation). http:// diffusion method for detection of inducible clindamycin resistance in www.agreecollaboration.org/ (6 February 2008, date last accessed). Staphylococcus aureus and coagulase-negative staphylococci. J Clin 42. Kowalski TJ, Berbari EF, Osmon DR. Epidemiology, treatment, Microbiol 2003; 41: 4740–4. and prevention of community-acquired methicillin-resistant 58. Ohlsen K, Ziebuhr W, Koller KP et al. Effects of sub-inhibitory Staphylococcus aureus infections. Mayo Clin Proc 2005; 80: 1201–7. concentration of antibiotics on a-toxin (hla) gene expression of 43. Gorwitz RJ, Jernigan DB, Powers JH et al. Strategies for clini- methicillin-sensitive and methicillin-resistant Staphylococcus aureus cal management of MRSA in the community: Summary of an experts’ isolates. Antimicrob Agents Chemother 1998; 42: 2817–23. meeting convened by the Centres for Diseases Control and 59. Stevens DL, Ma Y, Salmi DB et al. Impact of antibiotics on Prevention. 2006. http://www.cdc.gov/NCIDOD/DHQP/pdf/ar/ expression of virulence-associated exotoxin genes in methicillin- CAMRSA_ExpMtgStrategies.pdf (6 February 2008, date last sensitive and methicillin-resistant Staphylococcus aureus. J Infect Dis accessed). 2007; 195: 202–11. 44. Deillit T, Duchin J. Guidelines for evaluation and management 60. Gillet Y, Vanhems P, Lina G et al. Factors predicting mortality of community-associated methicillin-resistant Staphylococcus aureus in necrotizing community-acquired pneumonia caused by skin and soft tissue infections in the outpatient settings. http://www. Staphylococcus aureus containing Panton–Valentine leukocidin. Clin metrokc.gov/health/providers/epidemiology/MRSA-guidelines.pdf (4 Infect Dis 2007; 45: 315–21. February 2008, date last accessed). 61. Anon. Four paediatric deaths from community acquired methi- 45. Hill R, Kearns M, Ellington M et al. Algorithmic features to pre- cillin resistant Staphylococcus aureus—Minnesota and North Dakota sumptively identify PVL-positive MRSA in individual cases or survey 1997–1999. MMWR 1999; 48: 707–10. trends in the UK. In: Abstracts of the 47th Interscience Conference on 62. Torell E, Molin D, Tano E et al. Community-acquired pneumo- Antimicrobial Agents and Chemotherapy, Chicago, 17–20 September nia and bacteraemia in a healthy young woman caused by methicillin- 2007. Abstract K-1083, p. 349. American Society for Microbiology, resistant Staphylococcus aureus (MRSA) carrying the genes encoding Washington, DC, USA. Panton–Valentine leukocidin (PVL). Scand J Infect Dis 2005; 37: 46. Newell PM, Norden CW. Value of needle aspiration in bacterio- 902–4. logic diagnosis of cellulitis in adults. J Clin Microbiol 1988; 26: 401–4. 63. Salliot C, Zeller V, Puechal X et al. Panton–Valentine 47. Cohen PR. Community-acquired methicillin-resistant leukocidin-producing Staphylococcus aureus infections: report of 4 Staphylococcus aureus skin infections: a review of epidemiology, French cases. Scand J Infect Dis 2006; 38: 192–5.

991 Review

64. Pogue M, Burton S, Kreyling P et al. Severe methicillin- carrying the Panton–Valentine leukocidin genes. Clin Infect Dis 2005; resistant Staphylococcus aureus community-acquired pneumonia 40: 1376–8. associated with influenza, Louisiana and Georgia, December 2006– 83. van Langevelde P, van Dissel JT, Meurs CJ et al. Combination January 2007. MMWR 2007; 56: 325–9. of flucloxacillin and gentamicin inhibits toxic shock syndrome toxin 1 65. Baggett HC, Hennessy TW, Rudolph K et al. Community-onset production by Staphylococcus aureus in both logarithmic and stationary methicillin-resistant Staphylococcus aureus associated with antibiotic phases of growth. Antimicrob Agents Chemother 1997; 41: 1682–5. use and the cytotoxin Panton–Valentine leukocidin during a furunculo- 84. Dumitrescu O, Boisset S, Badiou C et al. Effect of antibiotics sis outbreak in rural Alaska. J Infect Dis 2004; 189: 1565–73. on Staphylococcus aureus producing Panton–Valentine leukocidin. 66. British Thoracic Society. BTS guidelines for the management Antimicrob Agents Chemother 2007; 51: 1515–9. of community acquired pneumonia in adults—2004 update. http://www. 85. Francis JS, Carroll K, Nuermberger E et al. Authors reply. Clin brit-thoracic.org.uk/c2/uploads/MACAPrevisedApr04.pdf (6 February Infect Dis 2005; 40: 1378–9. 2006, date last accessed). 86. Hampson FG, Hancock SW, Primhak RA. Disseminated sepsis 67. Al-Tawfiq JA, Aldaabil RA. Community-acquired MRSA bac- due to a Panton–Valentine leukocidin producing strain of community teremic necrotizing pneumonia in a patient with scrotal ulceration. acquired methicillin resistant Staphylococcus aureus and use of intra- J Infect 2005; 51: 241–3. venous immunoglobulin therapy. Arch Dis Child 2006; 91: 201. 68. Garnier F, Tristan A, Francois B et al. Pneumonia and new 87. Martin BT, Palasanthiran P, Gosbell IB et al. Severe childhood methicillin-resistant Staphylococcus aureus clone. Emerg Infect Dis pneumonitis caused by the Queensland strain of community-acquired 2006; 12: 498–500. methicillin-resistant Staphylococcus aureus. Med J Aust 2005; 182: 69. Gonzalez BE, Teruya J, Mahoney DH et al. Venous thrombosis 249. associated with staphylococcal osteomyelitis in children. Pediatrics 88. Lopez-Aguilar C, Perez-Roth E, Moreno A et al. Association 2006; 117: 1673–9. Downloaded from between the presence of the Panton–Valentine leukocidin-encoding 70. Jeyaratnam D, Reid C, Kearns A et al. Community associated gene and a lower rate of survival among hospitalized pulmonary MRSA: an alert to paediatricians. Arch Dis Child 2006; 91: 511–2. patients with staphylococcal disease. J Clin Microbiol 2007; 45: 274–6. 71. Jones TF, Creech CB, Erwin P et al. Family outbreaks of inva- 89. Han LL, McDougal LK, Gorwitz RJ et al. High frequencies of sive community-associated methicillin-resistant Staphylococcus aureus clindamycin and tetracycline resistance in methicillin-resistant infection. Clin Infect Dis 2006; 42:76–8.

Staphylococcus aureus pulsed-field type USA300 isolates collected at http://jac.oxfordjournals.org/ 72. Ma XX, Galiana A, Pedreira W et al. Community-acquired a Boston ambulatory health center. J Clin Microbiol 2007; 45: 1350–2. methicillin-resistant Staphylococcus aureus, Uruguay. Emerg Infect Dis 90. Morgan MS. Diagnosis and treatment of Panton–Valentine leu- 2005; 11: 973–6. kocidin (PVL)-associated staphylococcal pneumonia. Int J Antimicrob 73. Mongkolrattanothai K, Boyle S, Kahana MD et al. Severe Agents 2007; 30: 289–96. Staphylococcus aureus infections caused by clonally related 91. Yamaoka T. The bactericidal effects of anti-MRSA agents with community-acquired methicillin-susceptible and methicillin-resistant iso- rifampicin and sulphamethoxazole-trimethoprim against intracellular lates. Clin Infect Dis 2003; 37: 1050–8. phagocytosed MRSA. J Infect Chemother 2007; 13: 141–6. 74. Moyaasita T, Schimamoto Y, Mishiya H et al. Destructive pul- 92. Silverman JA, Mortin LI, Vanpraagh AD et al. Inhibition of dap- monary embolism in a patient with community acquired staphylococcal tomycin by pulmonary surfanct: in vitro modelling and impact. J Infect by guest on August 8, 2012 bacteraemia. J Infect Chemother 2002; 8: 99–102. Dis 2005; 191: 2149–52. 75. Nimmo GR, Playford EG. Community-acquired MRSA bacter- 93. Gauduchon V, Cozon G, Vandenesch F et al. Neutralization of aemia: four additional cases including one associated with severe Staphylococcus aureus Panton–Valentine leukocidin by intravenous pneumonia. Med J Aust 2003; 178: 245. immunoglobulin in vitro. J Infect Dis 2004; 189: 346–53. 76. Peleg AY, Munckhof WJ, Kleinschmidt SL et al. Life-threatening community-acquired methicillin-resistant 94. Agwu A, Brady KM, Ross T et al. Cholera-like diarrhea and Staphylococcus aureus infection in Australia. Eur J Clin Microbiol Infect shock associated with community-acquired methicillin-resistant Dis 2005; 24: 384–7. Staphylococcus aureus (USA400 clone) pneumonia. Pediatr Infect Dis J 2007; 26: 271–3. 77. Banthia S, Meka VG, Pillai SK et al. A fatal case of necrotizing pneumonia caused by community-associated methicillin-resistant 95. Darenberg J, Soderquist B, Normark BH et al. Differences in Staphylococcus aureus. Infect Dis Clin Pract 2005; 13: 132–8. potency of intravenous polyspecific immunoglobulin G against strepto- coccal and staphylococcal superantigens: implications for therapy of 78. Honarpour N, Mao JT. A case of fatal community-acquired toxic shock syndrome. Clin Infect Dis 2004; 38: 836–42. necrotizing pneumonia caused by Panton-Valentine leukocidin positive methicillin-sensitive Staphylococcus aureus. Curr Resp Med Rev 2007; 96. Torbin E, Andersson O, Kasemo AU et al. PVL-positiva Staph 3: 49–51. Aureus gav nekrotiserande pneumoni. Lakartidningen 2007; 7: 79. Micek ST, Dunne M, Kollef MH. Pleuropulmonary complications 509–13. of Panton–Valentine leukocidin-positive community-acquired 97. Adem PV, Montgomery CP, Husain AN et al. Staphylococcus methicillin-resistant Staphylococcus aureus: importance of treatment aureus sepsis and the Waterhouse Friedrichsen syndrome in children. with antimicrobials inhibiting exotoxin production. Chest 2005; 128: N Engl J Med 2005; 353: 271–2. 2732–8. 98. Grey JE, Enoch S, Harding KG. Pressure ulcers. BMJ 2006; 80. Monaco M, Antonucci R, Palange P et al. Methicillin-resistant 332: 472–5. Staphylococcus aureus necrotizing pneumonia. Emerg Infect Dis 2005; 99. Moore Z, Romanelli M. Topical management of infected grade 11: 1647–8. 3 and 4 pressure ulcers. European Wound Management Association 81. Tronci M, Parisi G, Pantosti A et al. CA-MRSA strain with (EWMA). Position Document: Management of wound infection. decreased vancomycin susceptibility as a cause of serious invasive London: MEP Ltd, 2006. infection in an immunocompetent adolescent. Clin Microbiol Infect 100. Sanada H, Nakagami H, Romanelli M. Identifying criteria for 2007; 13 Suppl 1: S449. pressure ulcer infection. European Wound Management Association 82. Wargo KA, Eiland EH. Appropriate antimicrobial therapy for (EWMA). Position Document: Identifying criteria for wound infection. community-acquired methicillin-resistant Staphylococcus aureus London: MEP Ltd, 2005.

992 Review

101. Naimi TS, LeDell KH, Como-Sabetti K et al. Comparison of community- and healthcare-associated methicillin-resistant *When rifampicin is used please consider drug interactions, e.g. Staphylococcus aureus infection. JAMA 2003; 290: 2976–84. warfarin, methadone, hormone contraceptives, theophylline etc. 102. Safdar N, Maki DG. The commonality of risk factors for noso- †Linezolid is an expensive alternative and may not be readily comial colonization and infection with antimicrobial-resistant available at community pharmacies. Staphylococcus aureus, enterococcus, Gram-negative bacilli, Note: if GAS infection is suspected, oral therapy should Clostridium difficile, and Candida. Ann Intern Med 2002; 136: 834–44. include an active agent against this organism (b-lactam or clinda- 103. Kluytmans J, van Belkum A, Verbrugh H. Nasal carriage of mycin). Tetracyclines and trimethoprim, although active against Staphylococcus aureus: epidemiology, underlying mechanisms, and associated risks. Clin Microbiol Rev 1997; 10: 505–20. MRSA, are not recommended for suspected GAS infections. 104. Huang SS, Platt R. Risk of methicillin-resistant Staphylococcus aureus infection after previous infection or colonization. Clin Infect Dis B. Serious and deep-seated MRSA infections 2003; 36: 281–5. 105. Davis KA, Stewart JJ, Crouch HK et al. Methicillin-resistant The Working Party recommends that suspected serious and Staphylococcus aureus (MRSA) nares colonization at hospital admis- deep-seated MRSA infections are assessed and treated in hospi- sion and its effect on subsequent MRSA infection. Clin Infect Dis 2004; tal. This includes suspected bacteraemia and staphylococcal 39: 776–82. pneumonia as in Scenario 2. Refer to MRSA treatment guide- 106. Cosgrove SE, Sakoulas G, Perencevich EN et al. Comparison lines for more detail.4 of mortality associated with methicillin-resistant and methicillin- susceptible Staphylococcus aureus bacteremia: a meta-analysis. Clin

Infect Dis 2003; 36:53–9. B1. For MRSA pneumonia (þ/2 PVL), treat with: Downloaded from 107. Engemann JJ, Carmeli Y, Cosgrove SE et al. Adverse clinical and economic outcomes attributable to methicillin resistance among † Linezolid (600 mg intravenous 12 hourly) PLUS clindamy- patients with Staphylococcus aureus surgical site infection. Clin Infect cin (1.2–1.8 g intravenous 8 hourly) þ/2 rifampicin Dis 2003; 36: 592–8. (600 mg intravenous 12 hourly). 108. Hardy KJ, Hawkey PM, Gao F et al. Methicillin resistant

Staphylococcus aureus in the critically ill. Br J Anaesth 2004; 92: http://jac.oxfordjournals.org/ 121–30. B2. For other deep-seated MRSA infections, such as 109. Graffunder EM, Venezia RA. Risk factors associated with noso- bacteraemia, osteomyelitis, abscesses, endocarditis, and comial methicillin-resistant Staphylococcus aureus (MRSA) infection including previous use of antimicrobials. J Antimicrob Chemother 2002; including those infections caused by PVL-producing 49: 999–1005. CA-MRSA, treat with: 110. Scottish Intercollegiate Guidelines Network (SIGN). Guideline † 88. Management of suspected bacterial urinary tract infections in First-line: Either teicoplanin (400–800 mg intravenous adults. http://www.sign.ac.uk/guidelines/published/numlist.html (6 every 24 h (following loading) or vancomycin (1 g intrave- February 2008, date last accessed). nous 12 hourly) PLUS one of the following: gentamicin by guest on August 8, 2012 Appendix. Summary of treatment recommendations (5–7 g/kg intravenous once daily), rifampicin (300 mg po twice daily) or sodium fusidate (500 mg po thrice daily). for MRSA infections in the community † Second-line: Linezolid (600 mg intravenous/po 12 hourly) † Before treating, clinicians may wish to seek advice of a local Alternative: Daptomycin (4 mg/kg intravenous once daily). microbiologist. Licensed for SSTIs, and for bacteraemia and right-sided endocarditis due to S. aureus. † Alternative: Tigecycline (100 mg loading dose followed by A. Patients with SSTIs 50 mg intravenous twice daily). Licensed for complicated SSTIs. (a) Follow local guidelines for treating SSTIs, for example flu- cloxacillin or clindamycin for minor SSTI without systemic Notes: upset. If the patient is febrile, appears unwell or is toxic with SSTI, consider assessment in hospital. (a) Assessment in hospital likely to be required. (b) Swab the lesion if purulent exudate is present. (b) Bone and joint infections may require a prolonged course of (c) If MRSA is suspected because of previous colonization/iso- treatment. lation, or surgical/healthcare-related, it is very important to (c) Monitor serum vancomycin/teicoplanin and gentamicin collect a microbiology sample. concentrations (for example adjust doses to achieve trough (d) If MRSA is isolated or strongly suspected, treat with: concentrations for teicoplanin of 10–20 mg/L and for vancomycin of 10–15 mg/L). W Rifampicin* (300 mg po twice daily) PLUS sodium fusi- date (500 mg thrice daily) OR doxycycline (100 mg po twice daily) (doxycycline not recommended for paedia- C. MRSA in urine tric use) for 5–7 days. W Rifampicin (300 mg po twice daily) PLUS trimethoprim The Working Party believes that while this area is not covered (200 mg po twice daily) for 5–7 days. in a clinical scenario in the text, it is often a frequently encoun- W Linezolid† (600 mg po twice daily) following discussion tered clinical problem. The Working Party felt that recommen- with Consultant Microbiologist or Infectious Disease dations are warranted to provide guidance for practice in the physician. community.

993 Review

† Antibiotics are unlikely to clear MRSA in the presence of likely to require systemic antibiotic treatment. [GPP]. The a urinary catheter. There is no good evidence that catheter Working Party recommends that in patients with normal changes need to be covered with appropriate antibiotic renal function (children excluded), doxycycline (100 mg prophylaxis to prevent catheter-related urinary tract twice daily) or tetracycline (250–500 mg 6 hourly) should infections.110 be used. Trimethoprim (200 mg 12 hourly) or nitrofuran- † A significant MRSA urinary tract infection with systemic toin (50–100 mg four times a day for 5–7 days) could be symptoms and the presence of white cells in the urine is alternatives.4 Downloaded from http://jac.oxfordjournals.org/ by guest on August 8, 2012

994

BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 1 of 8

Clinical Review

CLINICAL REVIEW

Diagnosis and management of cellulitis

1 Gokulan Phoenix core surgical trainee year 1 (London Deanery) , Saroj Das consultant vascular 2 3 surgeon , Meera Joshi core surgical trainee year 1 (Oxford Deanery)

1Department of General Surgery, Chelsea and Westminster Hospital, London SW10 9NH; 2Department of General Surgery, the Hillingdon Hospitals, London; 3Department of General Surgery, Wexham Park Hospital, Oxford

Cellulitis is an acute, spreading, pyogenic inflammation of the A multicentre study of 11 US hospitals reported a prevalence lower dermis and associated subcutaneous tissue. It is a skin of MRSA ranging from 15% to 74% (59% overall).15 A recent and soft tissue infection that results in high morbidity and severe review reports an increase in CA-MRSA rates in Europe.16 financial costs to healthcare providers worldwide. Cellulitis is managed by several clinical specialists including primary care Who is at risk of cellulitis? physicians, surgeons, general medics, and dermatologists. We assess the most recent evidence in the diagnosis and No link with age or gender has been established. However, a management of cellulitis. recent prospective case controlled study comprising 150 patients with cellulitis and 300 controls found white people to be at What is the extent of the problem? higher risk.17 Alcohol intake and smoking have been disproved 18 In 2008-9 there were 82 113 hospital admissions in England as risk factors in case-control studies. and Wales lasting a mean length of 7.2 days1; an estimated Commonly identified risk factors are listed in box 1. General £133m (€170m; $209m) was spent on bed stay alone.2 Cellulitis systemic risk factors include venous insufficiency, regarded to accounted for 1.6% of emergency hospital admissions during be the most frequent19; lymphoedema, both a predisposing factor 2008-9.3 and a complication of cellulitis20; peripheral vascular disease; 9 In Australia, hospital admissions for cellulitis have risen to 11.5 diabetes mellitus; and obesity. Local factors include tinea pedis, 9 people per 10 000 (2001-2) with the average admission lasting ulcers, trauma, and insect bites. 5.9 days.4 In the US more than 600 000 hospitalisations were recorded in 2010,5 representing 3.7% of all emergency Can cellulitis be prevented in those at 6 admissions. In all, 14.2 million Americans visited primary care risk? physicians, hospital outpatient departments, and emergency services with skin and soft tissue infections in 2005, an increase Besides the management of lymphoedema, there is no evidence from 321 to 481 visits per 100 000 (50% increase; P=0.003) to support the active management of other risk factors including since 1997. Over 95% of this change was attributed to abscesses diabetes mellitus, peripheral vascular disease, and tinea pedis. and cellulitis. Hospital visits for abscesses and cellulitis have In lymphoedema, decongestive lymphatic therapy, consisting increased from 173 to 325 per 1000 population (88% increase; 7 of manipulation of the lymphatic system through massage, has P<0.001). been associated with reduced recurrence of cellulitis. In a prospective study of 299 people who underwent decongestive What causes cellulitis? lymphatic therapy the incidence of cellulitis infections decreased from 1.10 to 0.65 infections per person per year.21 Cellulitis is caused by a wide range of organisms (see table 1⇓). The majority of cases are caused by Streptococcus pyogenes or Staphylococcus aureus. A review of prospective and How is the diagnosis of cellulitis made? retrospective laboratory studies found that S aureus accounted Clinical diagnosis for 51% of all aspiration and punch biopsy cultures positive for cellulitis, and Streptococcus accounted for 27%.8 Cellulitis most commonly affects the lower extremities, and often presents as an acute, tender, erythematous, and swollen A prospective study demonstrated that the majority of S aureus area of skin. In severe cases blisters, ulcers, oedema, associated infections in the US are now meticillin resistant; among 389 14 lymphangitis, and lymphadenopathy may be present. blood culture isolates of S aureus, 63% (244) were CA-MRSA. Constitutional features include fever and malaise. In the late

Correspondence to: G Phoenix [email protected]

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 2 of 8

CLINICAL REVIEW

Summary points

Cellulitis episodes in the United States, the United Kingdom, and Australia have risen over the past decade, with an increase in community acquired meticillin resistant Staphylococcus aureus (CA-MRSA) cases of cellulitis in the US, and to a lesser extent the UK and Australia. Antibiotic resistant strains of CA-MRSA are already emerging Diagnosis is based on clinical findings with investigations lending weight to confirm or refute diagnosis Existing guidelines need revision, taking into consideration CA-MRSA and other emerging strains as well as using new clinical classification systems such as the Dundee criteria Use outpatient parenteral antibiotic therapy if available More randomised control trials assessing the management of predisposing factors and long term therapy for recurrent cellulitis are required

Sources and selection criteria

We searched PubMed and the Cochrane library for recent and clinically relevant cohort studies and randomised controlled trials on cellulitis, using the search terms “cellulitis”, “erysipelas”, “diagnosis”, “investigation”, “recurrence”, “complications” and “management”. For position statements and guidelines we consulted the British Lymphology Society (BLS), National Health Service Clinical Knowledge Summaries (CKS), Clinical Resource Efficiency Support Team (CREST), and Infectious Disease Society of America (IDSA).

Box 1 Predisposing risk factors for lower limb cellulitis9 17 General Non-modifiable—pregnant; white race Modifiable—venous insufficiency; lymphoedema; peripheral arterial disease; immunosuppression; diabetes

Local Non-modifiable—trauma; animal and insect bites; tattoos Modifiable—ulcers; eczema; athlete’s foot (tinea pedis); burns stages widespread features of sepsis including hypotension and whether a blood culture was taken or not. The cost of negative tachycardia may also be present. blood cultures was $34 950 (£22 255; €28 560) and the cost for Other conditions can masquerade as cellulitis. Several the 11 positive cultures was $1100, amounting to an excess cost of $36 050. The authors concluded blood cultures were neither differential diagnoses (see table 2⇓), especially in the lower 20 limbs, can present with similar signs and symptoms. In a recent clinically effective or cost effective. National guidelines, prospective study of 145 patients, 28% of patients were including the Northern Ireland Clinical Resource Efficiency incorrectly diagnosed with lower limb cellulitis. The diagnosis Support Team (CREST) 2005 guidelines on the management most commonly mistaken as cellulitis was (venous) stasis of cellulitis in adults, recommend taking blood cultures only in dermatitis (37%).22 patients that have significant systemic upset including pyrexia (>38°C).10 In view of the potential for misdiagnosis on clinical observation alone, investigations are sometimes recommended to help In a prospective study of 50 patients with cellulitis, cultures confirm or refute the diagnosis. from skin biopsies and aspirations that showed true positives were found to be 20% and 10% respectively.25 CREST Blood investigations guidelines suggest the use of skin biopsies and aspirations in only selected patients, where the diagnosis of cellulitis is in In a prospective study of 150 people admitted to the emergency doubt.10 department that examined the feasibility of using C reactive In regard to wound swabs, a multicentre prospective study from protein level and white cell count as indicators of bacterial France that analysed wound swab samples from 214 patients infections including cellulitis, white cell counts had a specificity with lower limb cellulitis identified 183 (85.5%) positive of 84.5% and a sensitivity of 43.0% and C reactive protein had cultures; S aureus and Streptococcus being the most frequently a sensitivity of 67.1%, specificity of 94.8% (positive predictive 23 isolated micro-organisms (56% and 21% respectively). value 94.6% and negative predictive value 67.9%). An elevated Sensitivities from the swabs showed resistance to the empirical level of C reactive protein is a better indicator of bacterial antibiotics that had initially been used, prompting a change in infection than an elevated white cell count but a normal level antibiotics.26 CREST guidelines suggest the use of swabs on of C reactive protein cannot rule out an infection. Blood open cellulitis wounds.10 investigations do not appear to be clinically useful for diagnosis.

Microbiology Imaging Imaging techniques are useful when there is a suspicion of an Prospective studies have shown true positive rates from blood 24 25 underlying abscess associated with cellulitis, necrotising cultures in those with suspected cellulitis are between 2-4%. fasciitis, or when the diagnosis of cellulitis is uncertain. In a In a retrospective study of 757 people admitted to a medical retrospective study of 542 emergency department patients for centre with cellulitis, blood cultures were performed for 553 whom the clinical diagnosis of cellulitis was in doubt, 109 (17%) people (73%)—only 11 (2%) were positive. Eight of 11 patients were found to have a deep vein thrombosis on Doppler with positive blood cultures were changed from empirical ultrasound.27 treatment with cefazolin to penicillin. Furthermore, all those in the study, including those with systemic toxicity, recovered,

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 3 of 8

CLINICAL REVIEW

In a prospective observational study of 216 adult emergency When should a person be admitted to hospital department patients with a clinical diagnosis of lower limb for intravenous antibiotics? cellulitis, an ultrasonography scan changed the management in The Cochrane review from 2010 also states the need for further 71 patients (56%) in regard to the need for drainage of evaluation of oral versus intravenous antibiotics as well as the underlying abscesses. In the pre-test group that were believed efficacy of outpatient parenteral antibiotic therapy (OPAT).30 not to need drainage of any underlying abscess, ultrasonography resulted in a change in management in 32 of 44 patients (73%), In a prospective study of 205 consecutive adults admitted to a including 16 in whom drainage was eliminated. In the pre-test Scottish hospital for cellulitis, 43% were found to be overtreated based on CREST guidelines. The study suggests they possibly group that was believed to not need further drainage, 33 ultrasonography changed the management in 39 of 82 (48%), could have been managed as outpatients on oral antibiotics. with 33 receiving drainage and six receiving further diagnostic The CREST guidelines determine route of administration based imaging. Ultrasound may therefore guide management of on the Eron clinical classification system, taking into cellulitis by detection of occult abscess, prevention of invasive consideration the presence of systemic toxicity and procedures, and providing guidance for further imaging or comorbidities. consultation.28 Eron classification v Dundee classification Other imaging studies, such as MRI (magnetic resonance imaging) may be useful in those with an equivocal diagnosis of The Eron classification is based on expert opinions, and is cellulitis or with suspicion of necrotising fasciitis. According among the most widely used classification systems for diagnosis 34 to CREST guidelines, the physician should be alert to the and treatment of cellulitis. The Eron classification is possibility of necrotising fasciitis upon presentation of tense summarised in table 3⇓. oedema, skin necrosis, crepitus, paraesthesia with an elevated However, new criteria such as the 2011 Dundee classification white cell count greater than 14×109/L, and in the are also available33—a comparison between the two is provided haemodynamically stable patient an MRI scan is warranted.10 in table 4⇓. Seventy per cent of people that, based on Eron In a prospective study of 36 patients with a clinical diagnosis recommendations, would be treated with inpatient stay and of acute infectious cellulitis, MRI demonstrated necrotising intravenous antibiotics meet the criteria for outpatient fasciitis in 16 people, all of whom underwent surgical management based on the Dundee criteria.33 Further validation debridement. Distinct MRI features were found in people with of the Dundee criteria is required. necrotising soft tissue infections, including hyper-attenuating signals on T2 weighted images at the deep fasciae and poorly Outpatient parenteral antibiotic therapy defined areas of hyper-intense signal on T2 weighted images (OPAT) within muscles. In cellulitis, signal intensity abnormalities are 29 A prospective study on 344 episodes of treatment administered only within the subcutaneous fat. by a UK OPAT service showed that 87% of patients were cured, readmission rate was 6.3%, and patient satisfaction was high. What is the treatment of cellulitis? OPAT costs 41% of inpatient costs when calculated using General measures include rest, elevation of any affected limbs, conservative cost measurements. The authors of the study concluded that clinicians should use OPAT where available35; and analgesia. The area of cellulitis should be clearly marked 10 and reviewed daily for progression or regression to assess the this is supported by CREST guidelines. efficacy of the antibiotic regimen.10 When should a switch to oral antibiotics be However, there is still uncertainty regarding the optimal made? antibiotic choice, duration, and route of antibiotic therapy, and the use of corticosteroids. A recent Cochrane review could not CREST guidelines suggest indications for a switch to oral draw any definitive conclusions on the optimal antibiotics, therapy are apyrexia (<37.8°C) for 48 hours, regression of duration, or route of administration from an analysis of 25 cellulitis from a clearly marked area (on daily review), and a randomised controlled trials, as no two trials had compared the falling C reactive protein level.10 same antibiotics.30 A summary of the main antibiotics that are currently recommended in US and UK national guidelines, as When to seek further advice? well as in large prospective studies, are provided in table 1. CREST and NHS Clinical Knowledge Summaries guidelines CREST guidelines still recommend amoxicillin or flucloxacillin suggest that if there is doubt in the diagnosis, atypical for the majority of cases of cellulitis caused by S aureus, presentations, or no improvement in clinical symptoms and Streptococcus, or when the organism has not been identified,10 signs after 48 hours, then advice from a dermatologist or but clinicians should take into account the rise in CA-MRSA microbiologist or both should be sought.29 rates. The 2011 Infectious Diseases Society of America national guidelines have now recommended patients with pus forming Can recurrence be prevented? cellulitis to be treated with antibiotics that target CA-MRSA.11 The efficacy of other agents that target CA-MRSA has been Several prospective and retrospective studies suggest a high proportion of cellulitis sufferers develop recurrent episodes, studied. One retrospective cohort study has shown doxycycline 9 20 or minocycline to be effective in 95% of patients (n=276) with especially in those with untreated predisposing factors. One 12 retrospective study reported 47% recurrence in a cohort of 171 CA-MRSA. Clindamycin is also therapeutic, with susceptibility 20 in isolates as high as 93%. However, the development of people who had suffered one prior episode. resistance is not uncommon and as it associated with cases of Clostridium difficile, it should be discontinued on the Antibiotic prophylaxis development of diarrhoea.31 In those with severe cellulitis The Dermatology Clinical Trials Networks PATCH II trial requiring admission to hospital, linezolid and vancomycin were (prophylactic antibiotics for the treatment of cellulitis at home found to have good efficacy.32

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 4 of 8

CLINICAL REVIEW

II) was a large, multicentre, randomised trial in the UK that methicillin-resistant Staphylococcus aureus infections in adults and children. Clin Infect Dis 2011;52:e18-e55. assessed the efficacy of 6 months of penicillin V prophylaxis 12 Ruhe JJ, Menon A. Tetracyclines as an oral treatment option for patients with community in reducing recurrence. A total of 123 participants were onset skin and soft tissue infections caused by methicillin-resistant Staphylococcus aureus. Antimicrob Agents Chemother 2007;51:3298-303. randomised into those treated with penicillin (n=60) versus 13 UK Dermatology Clinical Trials Network’s PATCH Trial Team, Thomas K, Crook A, Foster placebo (n=63); recurrence rates were 20% and 33% respectively K, Mason J, Chalmers J, et al. Prophylactic antibiotics for the prevention of cellulitis (erysipelas) of the leg: results of the UK Dermatology Clinical Trials Network’s PATCH II (hazard ratio 0.53, 95% confidence interval 0.26 to 1.07, P=0.08) trial. Br J Dermatol 2012;166:169-78. with no difference in the number of adverse effects between 14 King MD, Humphrey BJ, Wang YF, Kourbatova EV, Ray SM, Blumberg HM. Emergence 13 of community-acquired methicillin-resistant Staphylococcus aureus USA 300 clone as both groups. The authors of this study conclude that there is the predominant cause of skin and soft-tissue infections. Ann Intern Med 2006;144:309-17. no statistical significance seen in the reduction of cellulitis rates 15 Moran GJ, Krishnadasan A, Gorwitz RJ, Fosheim GE, McDougal LK, Carey RB, et al. for penicillin V for prophylaxis, but there are promising results Methicillin-resistant S aureus infections among patients in the emergency department. N Engl J Med 2006;355:666-74. and longer term prophylaxis (for one year) may be required. 16 Otter JA, French GL. Molecular epidemiology of community-associated meticillin-resistant The PATCH I trial, which assesses one year penicillin V Staphylococcus aureus in Europe. Lancet Infect Dis 2010;10:227-39. 13 17 Halpern J, Holder R, Langford NJ. Ethnicity and other risk factors for acute lower limb prophylaxis, is under way. CREST guidelines advise antibiotic cellulitis: a UK-based prospective case-control study. Br J Dermatol 2008;158:1288-92. prophylaxis with penicillin V or erythromycin for 1 to 2 years 18 Dupuy A, Benchikhi H, Roujeau JC, Bernard P, Vaillant L, Chosidow O, et al. Risk factors 10 for erysipelas of the leg (cellulitis): case-control study. BMJ 1999;318:1591-4. in patients with two or more previous episodes of cellulitis. 19 Jorup-Rönström C, Britton S. Recurrent erysipelas: predisposing factors and cost of prophylaxis. Infection 1987;15:105-6. Contributors: SD planned and initiated the manuscript. GP planned and 20 Keeley VL. Lymphoedema and cellulitis: chicken or egg? Br J Dermatol 2008;158:1175-6. 21 Ko DS, Lerner R, Klose G, Cosimi AB. Effective treatment of lymphedema of the contributed to the manuscript and provided figures. MJ contributed to extremities. Arch Surg 1998;133:452-8. the manuscript and provided figures. SD critically revised drafts of the 22 David CV, Chira S, Eells SJ, Ladrigan M, Papier A, Miller LG, et al. Diagnostic accuracy in patients admitted to hospitals with cellulitis. Dermatol Online J 2011;17:1. article and approved the content of the final version to be published. 23 Chan YL, Liao HC, Tsay PK, Chang SS, Chen JC, Liaw SJ. C-reactive protein as an SD is guarantor. indicator of bacterial infection of adult patients in the emergency department. Chang Gung Med J 2002;25:437-45. Competing interests: All authors have completed the ICMJE uniform 24 Perl B, Gottehrer NP, Ravesh D, Schlesinger Y, Rudensky B, Yinnon AM. disclosure form at www.icmje.org/coi_disclosure.pdf (available on Cost-effectiveness of blood cultures for adult patients with cellulitis. Clin Infect Dis 1999;29:1483-8. request from the corresponding author) and declare: no support from 25 Hook EW 3rd, Hooton TM, Horton CA, Coyle MB, Ramsey PG, Turck M. Microbiologic any organisation for the submitted work; no financial relationships with evaluation of cutaneous cellulitis in adults. Arch Intern Med 1986;146:295-7. 26 Holzapfel L, Jacquet-Francillon T, Rahmani J, Achard P, Marcellin E, Joffre T, et al. any organisations that might have an interest in the submitted work in Microbiological evaluation of infected wounds of the extremities in 214 adults. J Accid the previous three years; no other relationships or activities that could Emerg Med 1999;16:32-4. 27 Rabuka CE, Azoulay LY, Kahn SR. Predictors of a positive duplex scan in patients with appear to have influenced the submitted work. a clinical presentation compatible with deep vein thrombosis or cellulitis. Can J Infect Dis Provenance and peer review: Not commissioned; externally peer 2003;14:210-4. 28 Tayal VS, Hasan N, Norton HJ, Tomaszewski CA. The effect of soft-tissue ultrasound on reviewed. the management of cellulitis in the emergency department. Acad Emerg Med 2006;13:384-8. 1 Department of Health. Hospital episode statistics. Primary diagnosis 2008-2009. NHS 29 Rahmouni A, Chosidow O, Mathieu D, et al. MR imaging in acute infectious cellulitis. Radiology Information Centre, 2010. www.hesonline.nhs.uk. 1994;192:493-6. 2 NHS. Institute for innovation and improvement. Quality and service improvement tools. 30 Kilburn SA, Featherstone P, Higgins B, Brindle R. Interventions for cellulitis and erysipelas. Cochrane Database Syst Rev 2008. www.institute.nhs.uk/quality_and_service_improvement_tools/quality_and_service_ 2010;6:CD004299. improvement_tools/length_of_stay.html. 31 Forcade NA, Parchman ML, Jorgensen JH, Du LC, Nyren NR, Treviño LB, et al. 3 Blunt I, Bardsley M, Dixon J. Trends in emergency admissions in England 2004-9. The Prevalence, severity, and treatment of community-acquired methicillin-resistant Nuffield Trust, 2010. www.nuffieldtrust.org.uk/sites/files/nuffield/Trends_in_emergency_ Staphylococcus aureus (CA-MRSA) skin and soft tissue infections in 10 medical clinics J Am Board admissions_REPORT.pdf. in Texas: a South Texas Ambulatory Research Network (STARNet) Study. Fam Med 4 Australian Institute of Health and Welfare. Australian hospital statistics 2001-02. 2003. 2011;24:543-50. www.aihw.gov.au/publication-detail/?id=6442467479. 32 Stevens DL, Bisno AL, Chambers HF, Everett ED, Dellinger P, Goldstein EJ, et al. Practice Clin Infect 5 Agency for Healthcare Research and Quality. HCUP Databases. Healthcare Cost and guidelines for the diagnosis and management of skin and soft-tissue infections. Dis Utilization Project (HCUP). Overview of the Nationwide Inpatient Sample (NIS). June 2005;41:1373-1406. 2012. www.hcup-us.ahrq.gov/nisoverview.jsp. 33 Marwick C, Broomhall J, McCowan C, Phillips G, Gonzalez-McQuire S, Akhras K, et al. 6 National Hospital Ambulatory Medical Care Survey: 2008 Emergency Department Severity assessment of skin and soft tissue infections: cohort study of management and J Antimicrob Chemother Summary. www.cdc.gov/nchs/fastats/ervisits.htm. outcomes for hospitalized patients. 2011;66:387-97. Clin 7 Hersh AL, Chambers HF, Maselli JH, Gonzales R. National trends in ambulatory visits 34 Eron L. J. Infections of skin and soft tissues: outcome of a classification scheme. Infect Dis and antibiotic prescribing for skin and soft-tissue infections. Arch Intern Med 2000;31:287. 2008;168:1585-91. 35 Chapman AL, Dixon S, Andrews D, Lillie PJ, Bazaz R, Patchett JD. Clinical efficacy and 8 Chira S, Miller LG. Staphylococcus aureus is the most common identified cause of cellulitis: cost-effectiveness of outpatient parenteral antibiotic therapy (OPAT): a UK perspective. J Antimicrob Chemother a systematic review. Epidemiol Infect 2010;138:313-7. 2009;64:1316-24. 9 Cox NH, Colver GB, Paterson WD. Management and morbidity of cellulitis of the leg. J Accepted: 12 July 2012 R Soc Med 1998;91:634-7. 10 Clinical Resource Efficiency Support Team (2005) Guidelines on the management of cellulitis in adults. Crest, Belfast. http://www.acutemed.co.uk/docs/Cellulitis%20guidelines, Cite this as: BMJ 2012;345:e4955 %20CREST,%2005.pdf. 11 Liu C, Bayer A, Cosgrove SE, Daum RS, Fridkin SK, Gorwitz RJ, et al. Clinical practice © BMJ Publishing Group Ltd 2012 guidelines by the Infectious Diseases Society of America for the treatment of

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 5 of 8

CLINICAL REVIEW

Ongoing research

UK Dermatology Clinical Trials Network. Prophylactic antibiotics for the treatment of cellulitis at home (PATCH) I study—a multicentre randomised control trial in the UK assessing the efficacy of a one year course of penicillin V prophylaxis versus placebo in patients with recurrent cellulitis

Additional educational resources Resources for healthcare professionals Kilburn SA, Featherstone P, Higgins B, Brindle R. Interventions for cellulitis and erysipelas. Cochrane Database Syst Rev 2010;6:CD004299—the most recent review of randomised controlled trials on various antimicrobial options for cellulitis, with evidence for the most commonly used antibiotics Thomas K, Crook A, Foster K, Mason J, Chalmers J, et al; for the UK Dermatology Clinical Trials Network’s PATCH Trial Team. Prophylactic antibiotics for the prevention of cellulitis (erysipelas) of the leg: results of the UK Dermatology Clinical Trials Network’s PATCH II trial. Br J Dermatol 2012;166:169-78—a recent randomised controlled trial that assessed the efficacy of long term prophylactic penicillin V for recurrent cellulitis. No statistical significance was seen in the reduction in rates of recurrence with penicillin V Chronic oedema and lymphoedema BMJ learning module. http://learning.bmj.com/learning/module-intro/lymphoedema-.html? moduleId=10029385—BMJ module on the diagnosis, investigation, and treatment of lymphoedema associated with cellulitis

Information resources for patients Cellulitis Support Group. www.mdjunction.com/cellulitis—forums for cellulitis sufferers to discuss various treatment options (free registration required)

Tables

Table 1| Treatment recommendations for cellulitis based on organisms9 10 11 12 13

Clinical presentation Organism Antibiotic

Typical cellulitis Streptococcus pyogenes Amoxicillin or flucloxacillin Typical cellulitis—pus forming Staphylococcus aureus Flucloxacillin Typical cellulitis in the US—pus forming CA-MRSA, HA-MRSA Doxycycline or minocycline or clindamycin or vancomycin Penicillin allergy NA Erythromycin or clarithromycin or clindamycin Cat or dog bite Pasteurella multocida Co-amoxiclav; if allergic to penicillin: doxycycline and metronidazole Freshwater exposure Aeromonas hydrophila Ciprofloxacillin Saltwater exposure Vibrio vulnificus Doxycycline Necrotising fasciitis Clostridium perfringens Benzylpenicillin, ciprofloxacillin, and clindamycin Butchers and fish handlers Erysipelothrix Ciprofloxacillin

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 6 of 8

CLINICAL REVIEW

Table 2| Common differential diagnoses for cellulitis with defining characteristics10

Differential Defining characteristics

Stasis dermatitis Absence of pain or fever; circumferential; bilateral Acute arthritis Involvement of joint; pain on movement Pyoderma gangrenosum Ulcerations on the legs; history of inflammatory bowel disease Hypersensitivity/drug reaction Exposure to allergen or drug; pruritus; absence of fever; absence of fever or pain Deep vein thrombosis Absence of skin changes or fever Necrotising fasciitis Severe pain, swelling and fever; rapid progression; pain out of proportion; systemic toxicity; skin crepitus; necrosis; ecchymosis

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 7 of 8

CLINICAL REVIEW

Table 3| Eron clinical classification system34

Class Systemic toxicity Comorbidities Oral v intravenous antibiotics Outpatient v hospital admission I No sign None Oral Outpatient II May or may not have systemic illness Peripheral vascular disease, Intravenous Hospital admission for 48 hours then obesity, venous insufficiency outpatient parenteral antibiotic therapy III Significant systemic toxicity—confusion, Unstable Intravenous Hospital tachycardia, tachypnoea, hypotension IV Sepsis syndrome/necrotising fasciitis Unstable Intravenous with or without surgical Hospital debridement

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2012;345:e4955 doi: 10.1136/bmj.e4955 (Published 7 August 2012) Page 8 of 8

CLINICAL REVIEW

Table 4| Eron classification v Dundee classification15 33

Parameter Eron (2003) Dundee (2010)

Strength of evidence Expert opinion Retrospective study of 205 consecutive patients Incorporated into guidelines? CREST and NHS acute trusts NA Validated? Yes Yes Criteria Comorbidities including obesity and peripheral vascular The importance of comorbidities disease Systemic toxicity: pyrexia (>38ºc), hypotension, tachypnoea, Obesity and peripheral vascular disease not counted towards hospital and tachycardia admission Up to date definition of systemic inflammatory response syndrome (SIRS) Standardised and validated early warning scores

For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe HPA Colindale Botulism Duty Doctor Protocol

Health Protection Agency Centre for Infections

Duty Doctor Botulism Protocol

Version Updated November 2011

Botulinum Antitoxins from different producers have differing compositions and dosages. · Always check with the issuing centre which type is being issued and read the product information sheet for dosages and administration · Ensure the product information sheet accompanies the antitoxin, for use by the hospital doctors caring for the patient

For medical advice: Dr Fortune Ncube [email protected]

For laboratory advice: Dr Kathie Grant [email protected]

Version 6.0 1 November 2011 HPA Colindale Botulism Duty Doctor Protocol

Requests for Botulinum Antitoxin: Guidance for Duty Doctors

1. Background These guidelines are for duty doctors dealing with duty calls and they concentrate on advice to be given to clinicians caring for a probable (based on clinical diagnosis) case of botulism. They are not intended for giving advice on the clinical management of cases ; a responsibility that rests with the attending clinician. Key duty doctor actions in response to a suspected case of botulism are summarised in figure 1.

2. Introduction Botulinum neurotoxins are produced by the anaerobic spore forming bacterium Clostridium botulinum and, rarely by Clostridium baratii and Clostridium butyricum. There are seven neurotoxins (A-G). Illness in humans is usually caused by types A or B or E, or rarely by F. Types C, D and E cause illness in mammals, birds and fish. Type G has not been shown to be associated with disease.

All toxins block the release of acetylcholine at the neuromuscular junction, which results in flaccid paralysis.

3. Clinical presentation and routes of transmission There are three naturally occurring forms of botulism:

· Food-borne botulism · Wound botulism · Infant botulism

3.1. Food-borne botulism is caused by ingestion of pre-formed toxin in food. Usually occurs after ingestion of foods stored in airtight (i.e. anaerobic) containers – such as canned foods, homemade pickles. Very rarely, intestinal colonisation similar to infant botulism occurs in adults, usually with an underlying reason, such as gastrointestinal abnormalities.

3.2. Wound botulism occurs when anaerobic conditions in a wound contaminated with Clostridium botulinum (C. botulinum) are created and toxin is formed. Injecting drug users (skin and muscle poppers) are at risk of developing wound botulism.

3.3. Infant botulism occurs following colonisation of the gut with C. botulinum with subsequent toxin production. Infants (usually <6 months old) present with very

Version 6.0 2 November 2011 HPA Colindale Botulism Duty Doctor Protocol non-specific effects, such as weakness, hypotonia, hyporeflexia, bulbar palsies, constipation, poor feeding, dehydration. Disease progression in children can be very rapid.

3.4. There is also a further possibility that botulism may occur as a result of deliberate release or as accidental exposure following miss-injection of therapeutic neurotoxin. A deliberate release may involve airborne dissemination of toxin, producing botulism through inhalation. Alternatively, it may involve contamination of food and water supplies either with toxin or with C. botulinum bacteria.

3.5. Clinical presentation In adults irrespective of the route of transmission, characteristically there is bilateral cranial nerve impairment and descending flaccid paralysis. Blurred or double vision, dysphagia and a dry mouth are often the first complaints. These symptoms may extend to a symmetrical flaccid paralysis in a paradoxically alert patient. Fever is usually absent but may occur with concurrent infection. If onset is very rapid there may be no symptoms before sudden respiratory paralysis occurs. In foodborne or intestinal colonisation, vomiting and diarrhoea, followed by constipation precede neurological symptoms.

In infants signs are often non-specific with constipation presenting first, followed by irritability, lethargy, poor feeding, drooling, hypotonia and general weakness. This may progress to descending symmetrical weakness and respiratory failure is a major complication.

4. Management of botulism · Early administration of botulinum antitoxin (if indicated) · Search for and debride any wound no matter how trivial · Give antibiotics if wound present · Supportive treatment

Administration of antitoxin in food-borne and wound botulism is life-saving and must be expedited. Botulinum toxin binds irreversibly to motor end-plates and the aim of antitoxin administration is to block that binding process.

For infant botulism a human derived Botulinum immune Globulin BabyBIG® is available from the Infant Botulism Treatment and Prevention Program (IBTPP) California USA (510) 231-7600 http://www.infantbotulism.org/. Use of BabyBIG® significantly reduces the length of hospital stay and associated hospital costs in

Version 6.0 3 November 2011 HPA Colindale Botulism Duty Doctor Protocol patients with infant botulism. Supportive treatment must also be provided and, although paralysis may persist for several weeks, recovery may be expected.

In wound botulism management of the wounds is also important. All wounds, no matter how trivial should be surgically debrided and treatment with antibiotics (penicillin and metronidazole) should be continued until the wounds have healed completely. Sometimes lesions are not apparent and the presence of deep-seated abscesses or sinuses should be considered.

5. Clinical investigations Laboratory tests include detection of neurotoxin in clinical and food samples, detection of the organism and isolation from clinical and food samples. The specimens required depend on the clinical presentation and include:

Foodborne botulism Wound botulism Infant botulism 10 ml serum taken prior to 10 ml serum taken prior to Serum is not a useful antitoxin treatment antitoxin treatment specimen for confirmation of infant botulism Faecesa (10g) or high rectal Pus from wound/abscess or Faeces a or rectal wash wash out and 1g inoculated debrided tissue in Cooked outbin Cooked Meat Broth or into Cooked Meat Broth or Meat Broth or other anaerobic other anaerobic medium other anaerobic medium medium Vomitus, gastric contents, Vomitus, gastric contents, intestinal contents if available intestinal contents if available Food items implicated (10g) Food items implicated e.g honey, formula milk afaeces may be difficult to obtain due to constipation bsee http://www.infantbotulism.org/ for information on how to perform this procedure in infants

HPA Colindale provides a rapid diagnostic PCR assay for clinical specimens inoculated into CMB

5.1 Routine laboratory tests are NOT helpful. For food and infant botulism, clinical and food specimens must be forwarded to the HPA Colindale Foodborne Pathogens Reference Unit (FBPRU), immediately. During out of office hours, senior staff of FBPRU must be informed via the out of hours Colindale duty doctor: 0208 200 4400

5.2 Wound botulism Clinical specimens may not have the same public health urgency and will usually not need to be immediately tested if arriving out of hours.

It is important to note that botulism is a clinical diagnosis, which can be confirmed, but not refuted, by laboratory investigations. In cases of food poisoning, toxin is

Version 6.0 4 November 2011 HPA Colindale Botulism Duty Doctor Protocol present in serum or faeces in >50% of cases within one day of onset, but <25% after 3 days. C. botulinum will be present in the faeces of >70% of cases within 2 days and 40% 10 days after onset of food poisoning.

For cases of wound botulism the diagnosis is confirmed by the detection of toxin and/or isolation of organism in 41% of cases.

6. Requests for antitoxin Antitoxin is held at 32 sites around the country and at the HPA Colindale. The antitoxin is accessible by contacting the duty doctor at Colindale: 0208 200 4400.

6.1 Upon receipt of a request for antitoxin the duty doctor should: · Record patient details and those of the clinician making the request, including the level of suspicion (i.e. where in the list of differential diagnoses does botulism occur). A botulism questionnaire has been developed on which these details can be recorded. If IDU related the IDU team in Colindale is responsible for sending the questionnaire to the CCDC or the clinical team and for collating the data from the returned completed questionnaires. · Ensure that the clinical team has informed the CCDC and Consultant Microbiologist locally. Food-borne botulism is a public health emergency, which must be reported to the CCDC IMMEDIATELY. · Ensure that appropriate clinical specimens have been taken prior to administration of antitoxin (detailed above in section 5).

6.2 Further advice If duty doctors need further advice on issue or administration of antitoxin, or want to refer clinicians for medical advice, they may contact Tim Brooks( Porton Down) , Barbara Bannister (Royal Free Hospital) or Nick Beeching (Royal Liverpool University Hospital). Contact details: Dr Barbara Bannister Work: 0207 941 1823/1826 Mobile: 07831 865912

Dr Tim Brooks Work: 01980 612774 Home: 01980 612032 Mobile: 07766 775149

Version 6.0 5 November 2011 HPA Colindale Botulism Duty Doctor Protocol

Dr Nicholas Beeching Work: 0151 706 3835 Mobile: 07816320607 Royal Liverpool University Hospital switch board: 0151 706 2000

6.2. Antitoxin for food and wound botulism

6.2.1 Ensure that the distribution centre includes with the antitoxin the information sheet on how to use the antitoxin. The composition of the antitoxin changes (there is worldwide shortage and companies and sourcing changes). It is important to make sure that the clinician looking after the patient follows the instructions to the letter. The clinician should contact their local infectious disease consultant for further advice.

6.2.2 Ensure that the patient has been assessed by an expert senior physician or neurologist. Botulism is a clinical diagnosis; advice on the diagnosis and clinical management of the patient. This should be obtained from the local ID physician or from an appropriate physician with expertise in botulism.

6.3. Issue of antitoxin. There are several types of Botulinum Antitoxin available which differ in presentation, composition, dosage and administration.

6.3.1 The Botulinum Antitoxin products currently available in the UK are:

(1) Novartis Behring Botulinum antitoxin (2) Scottish National Service / MoD Botulinum antitoxin

The DH sources the antitoxin. If HPA Colindale is the issuing centre the most likely Botulinum Antitoxin available will be Novartis Behring as DH is no longer sourcing the SNBTS/MoD Botulinum antitoxin, but duty doctors should note that other issuing centres may have SNBTS/MoD antitoxin still in stock.

6.3.2 BabyBIG is issued by the Infant Botulinum Treatment and Prevention Programe in California who will provide all details regarding transport and administration of infant antitoxin.

Version 6.0 6 November 2011 HPA Colindale Botulism Duty Doctor Protocol

6.4 Technical data and information on antitoxins The technical data and information of the Novartis Berhing and MoD Botulinum Antitoxin are found in the duty doctor pack for Botulinum.

Details on the presentation, composition, dosage and administration of the available antitoxins are found in table 1 but should be read in conjunction with the product information leaflet and generic advice below. Technical data and information on BabyBIG® can be found on the IBTPP website.

6.4.1 Composition

Antitoxins held in the UK for food and wound botulism are immune serum of F(ab)2 immunoglobulin fragments sourced from animals. BabyBIG® is immune globulin derived from plasma of donors immunised with pentavalent (ABCDE) botulinum toxoid.

6.4.2 Indications Botulinum Antitoxin is for Emergency Use Only for administration to individuals on the slightest suspicion of botulism. Under no circumstances should the treatment be delayed whilst waiting for the results of laboratory investigations and clinical observations.

BabyBIG® is indicated for the treatment of patients below one year of age with botulism A and B and has also been used in the USA to treat infant botulism caused by type E.

6.4.3 Administration Advice on administration is detailed in the product information leaflet which accompanies the antitoxin and this should be followed. Botulinum antitoxins are generally given parenterally by slow intravenous infusion.

Vital signs (pulse, blood pressure and temperature) should be monitored following administration and appropriate medication for the management of acute allergic reactions (hydrocortisone, antihistamine and adrenaline) should be readily available.

For infant botulism information on administration is provided on the IBTPP website.

Novartis Berhing and MoD botulinum Antitoxin preparations do not have a product licence and should only be administered on prescription of a Medical Officer.

Version 6.0 7 November 2011 HPA Colindale Botulism Duty Doctor Protocol

BabyBIG® is the only antitoxin licensed for the treatment of infant botulism. The use of Novartis Berhing equine antitoxin is not recommended for infant botulism due to the risk of adverse reaction and at present there is insufficient clinical evidence to support the use of MoD ovine antitoxin for treatment of infant botulism.

6.4.4 Repeat antitoxin administration An effective treatment will prevent further progression of symptoms, but will not reverse established paralysis. If the patient continues to deteriorate the dose may be repeated within 24 hours (within 4-6 hours for Novartis Behring antitoxin).

The chances of an allergic reaction are higher following repeated administrations of antitoxin and the patients should be closely observed for signs of anaphylaxis and serum sickness.

A separate issue of antitoxin will be necessary for additional doses of antitoxin for an individual patient

6.4.5 Special warnings and precautions Since antitoxin is derived from animal immunoglobulin it carries a risk of allergic reactions such as anaphylaxis and serum sickness. The recipient should therefore remain under medical supervision for 24 hours following administration of the immune serum.

As with all products prepared from biological sources, the transmission of infective agents, including those of unknown origin, cannot be excluded.

Pregnancy and lactation are not contraindications for treatment with Botulism Antitoxin as the indication is a life-threatening condition. However, clinicians should note that no reproduction studies have been carried out on antitoxins so they should only be used if clearly required.

Any effect on the ability to drive and use machines is not known.

Version 6.0 8 November 2011 HPA Colindale Botulism Duty Doctor Protocol

Table 1 Botulinum Antitoxin products summary (see product information leaflet for full details)

Manufacturer Preparation Composition Dosage Administration Shelf life and storage Novartis Horse plasma Per ml: Adults and children: Slow iv injection Storage at+2 Behring proteins in 750iu Anti Type A toxin 500ml as 2 x 250ml and +8oC and 250ml bottle 500iu Anti Type B toxin bottles, infused Repeat not used after 50iu Anti Type E toxin whilst observing administration 4- the expiry date circulatory effects 6 hours later on the pack. with 250ml if clinically Open contents indicated (see should be used product immediately information) Scottish Vial of 800mg Per vial: Adults and children: Reconstitute vial 5 year shelf life National freeze dried 570iu Anti Type A toxin 1 vial with 10ml of when stored Blood sheep plasma 3800iu AntiType Btoxin water for between +2 and Transfusion proteins 115iu Anti Type C toxin injections at +25 °C Service / MoD 120iu AntiType D toxin room 305iu Anti Type E toxin temperature. By Reconstituted 340iu Anti Type F toxin gentle agitation contents should 45iu Anti Type G toxin (not shaking). be used immediately Reconstituted contents of one vial should be administered as a single 10ml slow iv injection over 1-2 mins

NB im injection only for mass casualties as causes considerable pain at injection site

Repeat administration with further vials if clinically indicated (see product information)

Version 6.0 9 November 2011 HPA Colindale Botulism Duty Doctor Protocol

Figure 1 Flow chart for dealing with a case of suspect botulism

Initial notification of a case(s) of botulism Record patient details including level of suspicion

Clinical Advice If food or infant botulism suspected, report to For local clinical advice; advise local CCDC immediately contact with the local ID Physician and neurologist. Also ensure that FSA has been informed Further clinical advice can be obtained from B Bannister, N Beeching or T Brooks.

Advice on laboratory diagnosis Laboratory confirmation: Contact senior FBPRU staff – Request (prior to antitoxin administration) collection K Grant and C Amar. and transport to HPA Colindale of following: If food or infant botulism § 10ml serum (all types of botulism) suspected, inform FBPRU staff § pus or debrided tissue (wound botulism) and ensure immediate transport § faeces (food or infant botulism) of samples for testing. Wound § foods (food botulism) botulism may not have the same public health emergency.

Issue of antitoxin Establish nearest supply to patient - there are 32 Clinical advice on sites around the UK including at HPA Colindale. administration of Antitoxin Ensure that information sheet is included with the Contact: B Bannister, antitoxin issued. N Beeching or T Brooks.

Issue of antitoxin from HPA Colindale Call duty BMS to issue antitoxin as per instruction on leaflet accompanying the antitoxin in supply

Version 6.0 10 November 2011 1018

Wound Botulism in California, 1951–1998: Recent Epidemic in Heroin Injectors

S. Benson Werner, Douglas Passaro, James McGee, Division of Communicable Disease Control, California Department Robert Schechter, and Duc J. Vugia of Health Services, Berkeley

California has reported most of the world’s wound botulism (WB) cases and nearly three- fourths of the cases reported in the United States. We reviewed the clinical, epidemiologic, and laboratory features of WB. From the first case in 1951, through 1998, a total of 127 cases were identified—93 in the last 5 years. The dramatic increase has been due to an epidemic (of WB) in people who inject black tar heroin. Whereas early cases of WB occurred after gross trauma, all but 1 of the last 102 cases occurred in drug users, primarily those who inject drugs subcutaneously (“skin poppers”). Cases are occurring disproportionately in Hispanics and women. Misdiagnosis and diagnostic delays of up to 64 days have occurred. This unprece-

dented, ongoing epidemic is now being reported in other states. We discuss the clinical and Downloaded from laboratory features that distinguish botulism from conditions that can mimic it, the relative yield of various diagnostic laboratory tests for botulism, and its treatment.

Botulism was first recognized in the early 19th century and of Health Services (CDHS) for advice on the diagnosis and care http://cid.oxfordjournals.org/ was named for sausage (from the Latin botulus), which was of patients who are suspected to have botulism. Diagnostic testing often implicated in early foodborne outbreaks. In 1895, van and the provision of antitoxin are available in the United States Ermengem identified the causative bacillus, later renamed Clos- only through public health services. California is one of the few states that both tests for botulism and releases botulinum antitoxin tridium botulinum. Besides classic foodborne botulism that re- for treatment; these activities facilitated our case findings. sults from ingesting botulinum toxin in improperly preserved All cases of WB reported to the CDHS during 1951–1998 were foods, 2 other major forms of botulism are now recognized: reviewed for demographic and clinical features. Characteristics of wound botulism (WB), first reported in 1951 [1], caused by WB patients who were injection drug users (IDUs) were compared 2 tissue infection and in situ toxin production by C. botulinum; with those who were not IDUs, by use of the x test or Fisher’s by guest on October 1, 2012 and infant botulism, first reported in 1976, caused by intestinal exact test; all P values were 2-tailed. Seasonality of onset was colonization by C. botulinum with subsequent in vivo toxin assessed by a generalization of Hewitt’s test for seasonality [4]. To production [2]. assess the economic burden of WB, inpatient hospital charges were Historically, California has accounted for 175% of all US obtained for all patients identified in 1995. WB cases [3], and US cases have comprised 190% of those Laboratory review. Except for 3 early cases of WB that met a known in the world. Between 1951 and 1998, a total of 127 previous national standard of clinical criteria, all cases were con- firmed by wound culture or by the mouse neutralization bioassay cases of WB were reported from California. We reviewed Cali- of sera or of cell-free extracts of homogenated tissue. Heroin sam- fornia’s extraordinary series of cases to clarify the rapidly ples from 12 patients who were IDUs were submitted for culture changing epidemiology and clinical presentation of WB. We at the state laboratory. found a dramatic increase of WB in individuals injecting black tar heroin (BTH), named for its appearance and consistency. Results

Methods Clinical and epidemiologic. We identified 127 cases of WB in California since the first case was recognized in 1951 (figure Clinical and epidemiologic review. Botulism is a reportable con- 1). There have been 2 striking changes in recent years: a dra- dition in California. Information is collected on a standardized matically accelerating case count and a change in the types of national reporting form augmented by information from patient records and physicians who consult with the California Department wounds from unintentional gross trauma to self-injection of illicit drugs, particularly BTH. Specifically, from 1951 through 1987, California averaged 0.5 cases per year. However, since Received 24 September 1999; revised 28 March 2000; electronically pub- lished 25 October 2000. 1988 (when the first case of WB in an IDU was identified in Reprints or correspondence: Dr. S. B. Werner, Disease Investigations Sec- California), there has been a 20-fold increase, to 9.9 cases per tion, Division of Communicable Disease Control, California Dept. of Health year. In 1998 alone, there were 28 cases. All but 1 of 102 cases Services, 2151 Berkeley Way, Rm. 708, Berkeley, CA 94704 (bwerner @dhs.ca.gov). of WB that occurred since 1990 were present in IDUs. Four types of wounds preceded the onset of WB: (1) traumatic, (2) Clinical Infectious Diseases 2000;31:1018±24 operative, (3) sinusitis, and (4) injection drug use. ᭧ 2000 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2000/3104-0026$03.00 Beginning in 1951, traumatic wounds of the extremities CID 2000;31 (October) Wound Botulism 1019 Downloaded from http://cid.oxfordjournals.org/

Figure 1. Incidence of wound botulism in California by type of wound, 1951–1998. IDU, injection drug user caused 15 WB cases. These injuries included lacerations, open maxillary drainage; aspirate of the sinuses grew C. botulinum fractures, crush injuries, punctures, and gunshot wounds. There type A. by guest on October 1, 2012 was no apparent seasonality to these injuries. Of 10 patients Beginning in 1988, there have been 105 cases of WB in IDUs. for whom there was information on wound appearance, half BTH injection was reported by all but 1 injector (who used were described as infected. Of these 15 patients, 10 (67%) were methamphetamine) for whom information was available. IDUs men. The median age was 26 years (range, 6–44 years). Of the were older than other patients with WB, with a median age of 14 with known outcome, 4 (29%) died. The median interval 40 years versus 28 years (P ! .001 ; table 1). One of the oldest from injury to onset of neurologic symptoms was 6.5 days IDUs (64 years old) had been using heroin daily for 50 years. (range, 4–13 days), and the median interval from first medical Forty-seven (45%) of the IDUs were women. Ninety-nine of consultation to first clinical consideration of WB was 5.0 days the IDUs were white, 2 were black, and the race for 4 was (range, 1–9 days). unknown. In instances where ethnicity was known, 58 (57%) Beginning in 1979, there have been 5 cases of WB that oc- of 101 IDUs were Hispanic, whereas only 2 (20%) of 10 other curred as operative complications. One case occurred in a 15- patients with WB were Hispanic (relative risk, 2.9; 95% CI, year-old girl who complained of diplopia 5 days after under- 0.8–10.0;P ! .04 ). going caesarean section. She was discharged despite diplopia There was little geographic clustering of IDU patients. Of and was much weaker when readmitted 2 days later. Endo- California’s 58 counties, 24 reported at least 1 patient, and 17 tracheal intubation resulted in a mediastinal hemorrhage, which counties reported у2 patients. Instances of WB in IDUs oc- led to cardiac arrest and stroke. At follow-up 1 year later, she curred in all months of the year, but 52 (50%) of 105 occurred was institutionalized in a vegetative state. This was the first between October and December (Hewitt’sP p .026 ). Patients nosocomial case of WB identified in the United States, and the did not know one another except for 5 couples. In 1991, a BTH first patient whose wound was not in an extremity. Three other user in southern California developed type A botulism and, 11 patients developed symptoms of botulism 2–5 days after emer- months later, so did his girlfriend. In 1995, a man and his 47- gent laparotomy to resect impacted bowel. All 3 had volvulus year-old wife in central California presented with symptoms of associated with adhesions from past appendectomies. The fifth type A botulism within 4 days of each other. Both couples patient with WB presented 1 month after a bone graft and reported sharing BTH but not needles and syringes. In 1995, external fixation for a nonhealing tibiofibular fracture sustained a third couple in northern California, described in more detail 1 year earlier. below, became ill after sharing BTH. In 1997, a man and his Beginning in 1984, there have been 2 cases of WB from si- girlfriend, both from southern California, had onsets of type nusitis in young adult men after they sniffed cocaine. One had B botulism 4 days apart after sharing BTH. In 1998, a fifth 1020 Werner et al. CID 2000;31 (October)

Table 1. Selected features of 105 wound botulism (WB) cases in injection drug users (IDUs) and 22 WB cases in persons not known to have injected drugs (non-IDUs), California, 1951–1998. Risk ratio (95% CI), Variable WB in IDUs WB in Non-IDUs IDUs vs. non-IDUs P Female sex 47 (45%) of 105 9 (41%) of 22 1.1 (0.6–1.9) .9a Race White 99 (98%) of 101 7 (70%) of 10 1.4 (0.9–2.1) .005b Black 2 (2%) of 101 2 (20%) of 10 0.1 (0.02–0.6) .04b Asian 0 (0%) of 101 1 (10%) of 10 Undefined .09b Unknown 4 12 — — Ethnicity Hispanic 58 (57%) of 101 2 (20%) of 10 2.9 (0.8–10.0) .04b Unknown 4 12 — — Age in years, median (range) 40 (20–67) 28 (6–63) — !.001c Incubation period in days,d median (range) Unknown (see text) 5 (2–13) — — Diagnostic delay, de 1951–1987 f n 014 ——Downloaded from Median (range) — 4.0 (0–9) — — — — 2.7 ע SD — 4.1 ע Mean 1988–1995 n 46 4 — — Median (range) 2.0 (0–64) 5.5 (3–8) — 0.05c — — 2.1 ע 5.5 11.2 ע SD 4.9 ע Mean 1996–1998 http://cid.oxfordjournals.org/ n 59 0 — — Median (range) 1.0 (0–19) — — — — — — 2.9 ע SD 1.6 ע Mean

a P value obtained by the Yates correction of the Mantel-Haenszel x2 test. b P value obtained by Fisher’s exact test. c P value obtained by the Kruskal-Wallis 2-sample test. d Time from suspected inoculation of Clostridium botulinum spores to onset of symptoms. e Time from initial physician consultation to first clinical consideration of botulism. f Time from initial physician consultation to first clinical consideration of botulism was not known for 4 early WB cases. by guest on October 1, 2012 couple had onsets of type A botulism 4 months apart in north- identified only by close examination of the patient’s customary ern California. injection sites and not by history. One IDU developed type A WB twice. In 1995, this BTH The interval from injection to onset of neurologic symptoms injector presented with clinical WB and a fluctuant thigh abscess; could not be identified because most IDUs injected frequently, culture grew C. botulinum type A. She underwent a 5-week hos- commonly several times a day, often in the same anatomic areas pitalization requiring intubation, mechanically assisted ventila- that later became infected, and so could not identify which tion, tracheostomy, and incision and drainage of her abscess. She particular injection (and when) was the inciting event. resumed BTH injection after she was discharged from the hos- Among IDUs, the mean interval from first medical consul- pital. She presented again in 1997 with clinical botulism and tation for signs and symptoms of WB and a physician’s first multiple abscesses; type A toxin was found in pretreatment consideration of that diagnosis was 3.1 days (range, 0–64 days). serum. Extended delays resulted from initial misdiagnosis. The most The wounds of IDUs with WB appeared to be grossly in- extreme examples were delays of 45 and 64 days after clinical fected in 80 (85%) of 94 people for whom this information was presentation. One of these was a 35-year-old homeless IDU available versus 12 (71%) of 17 patients who had other types who was misdiagnosed as having myasthenia gravis by her of WB (P p .17 ). “Abscess” was the most common description, attending neurologists at a university-affiliated medical center. often without further elaboration. Occasionally, there were ad- Six weeks after admission, a new resident rotated on service ditional comments, such as fluctuation, spontaneous draining and questioned the prevailing diagnosis. He submitted a fresh of pus, cellulitis, and, in one instance, necrosis. The appearance serum sample for testing at the state laboratory, where it was of infection was corroborated by studies of aspirated pus and found to be positive for botulinum toxin; only then was anti- of incisional and excisional biopsies that were recorded in the toxin administered. The other patient who was diagnosed very laboratory results. Although unusual injection sites (e.g., the late was a 41-year-old woman who had been managed as a vulva in one woman and the base of the tongue in another) case of Guillain-Barre´syndrome and was later transferred to were infected, the overwhelming majority of infected wounds a chronic care facility; she was dependent on a ventilator. When involved the extremities or the buttocks. Some infections were her husband, with whom she had shared BTH, was hospitalized CID 2000;31 (October) Wound Botulism 1021 for clinical botulism 9 weeks later, he questioned whether WB reported outside of California in 1990–1994, but there were 2 in could also explain his wife’s protracted illness. Serum that had 1995, 3 in 1996, and 5 in 1997. been taken from her on the fifth day of hospitalization was Nearly half of California’s 105 IDUs with WB were women. then retrieved for study, tested, and found to contain type A In contrast, women comprised only 23% of those who reported botulinum toxin. Diagnostic delays declined from a mean of ever using heroin in a 1994 national survey [8] and accounted 4.9 days to 1.6 days after we publicized the epidemic in late for only 35% of enrollees in California methadone clinics in 1995 (P ! .001 , see table 1). 1991–1995 (S. Jew, personal communication). In a study of The cost for inpatient medical care was high. All 23 patients tetanus cases reported in New York City, the risk to female with WB who were identified in 1995 were hospitalized, some IDUs was estimated to be 8–12 times greater than that for male for months. Total hospital charges were $3.78 million (median, IDUs [9]. One reason that female IDUs may be more likely $154,000; range: $19,000–$448,000). These hospital charges did than male IDUs to develop WB or tetanus is their greater not include convalescent or outpatient care or other indirect tendency to inject drugs into soft tissue (“skin pop”) than into costs. Because no patient had private medical insurance, all veins [9]. The resulting abscess formation and devitalized tissue hospital charges were paid at public or hospital expense. are thought to provide a favorable anaerobic milieu for the

Laboratory findings. Of the 127 total cases of WB, 107 (84%) growth of neurotoxin-producing clostridia (C. tetani and C. Downloaded from were due to type A botulinum toxin, 15 (12%) to type B, and 1 botulinum). Many men with WB reported becoming skin pop- (1%) to both A and B. One (1%) had botulinum toxin that could pers only after they had “used up” (sclerosed) their veins by not be typed because of inadequate serum volume, and in 3 (2%) years of injection. This may explain the relatively advanced age of the earliest patients tests for toxin type were not done. Of the (median, 40 years) of IDUs with WB.

123 cases of WB that could be typed, 89% of the IDUs and 78% The drug used almost exclusively by patients with WB has http://cid.oxfordjournals.org/ of the non-IDUs were type A (P not significant). been BTH. IDUs heat BTH in water, typically in teaspoons, Serologic testing. The clinical diagnosis of WB was most primarily to dissolve the heroin. Although ordinary non- commonly confirmed by detecting toxin in sera by use of the sporeforming bacteria and botulinum toxin itself are readily standard mouse bioassay. This test was positive in 99 (95%) of destroyed by boiling, the spores of C. botulinum are not. In 104 IDUs and in 15 (83%) of 18 non-IDUs confirmed to have fact, any dormant spores that are present may be activated to WB. germinate by such heating [10]. Wound testing: culture. C. botulinum was isolated from 41 We gathered the following information about BTH from

(65%) of 63 patients whose cultures were known to have been officials of the federal Drug Enforcement Administration, by guest on October 1, 2012 taken. The frequency of recovery of this organism from the which was corroborated by state and local drug enforcement wounds of IDUs was comparable to non-IDUs with WB: 30 agencies. BTH is relatively pure heroin produced in Mexico in (61%) of 49 versus 11 (79%) of 14 patients. small, makeshift factories next to opium poppy fields. It is Wound testing: toxin in situ. Free toxin was detected in 18 brown to black in color and can be sticky (like roofing tar) or (32%) of 56 IDU patients who had abscessed wound tissue hard (like coal). Typically, it is injected. BTH was first intro- resected. duced to the United States in the 1970s as a cheaper alternative Heroin testing. We were able to obtain only 12 very small to cocaine from Colombia and “China White” heroin from ! (pea-sized) samples of BTH ( 0.5 g each) from patients with southeast Asia. It became the predominant type of heroin in WB. These were found to be negative for C. botulinum by the California in the late 1980s, and its market share increased in state laboratory. the 1990s. BTH is sold in at least 27 states, primarily west of the Mississippi River. Its potency is variable because it is com- monly contaminated with by-products of the manufacturing Discussion process; adulterated with chemicals such as lidocaine, amphet- amine, and diphenhydramine; and diluted with inert absorbent To our knowledge, this is the largest series of WB cases ever materials such as cornstarch, lactose, mannitol, instant coffee, examined, largely because of an ongoing epidemic of WB in IDUs powdered milk, clay, banana peels, and even dirt. These ma- [5, 6]. The first case of WB identified in an IDU in the United terials are added not only to increase bulk and profitability but States occurred in New York City in 1982 [7]; California’s first also to make the BTH less tacky and more manageable. case was not identified until 1988. Through 1997, the latest year The potency of heroin dictates how many times it can be for which national figures are available, California reported 99 diluted (“cut,” “hit on,” or “stepped on”). BTH cut in Cali- (87%) of the nation’s 114 patients with WB in IDUs. The re- fornia tends to remain in California. The few cases of WB maining 15 patients were reported from the states of Washington recognized elsewhere in the United States might suggest that (n p 5 ), Arizona (n p 3 ), Oregon (n p 2 ), New York (n p 2 ), most contamination of BTH with C. botulinum is occurring New Mexico (n p 2 ), and Texas (n p 1 ) (figure 2) (J. T. Brooks, within California, but we suspect that the more likely reason CDC, personal communication). There were no cases of WB is that the diagnosis is being missed elsewhere. At least one 1022 Werner et al. CID 2000;31 (October) Downloaded from http://cid.oxfordjournals.org/ by guest on October 1, 2012

Figure 2. Wound botulism (WB) in injection drug users, United States, cumulative through 1997. California’s 1998 WB cases are also shown (hatched boxes).

California patient reported purchasing BTH in Mexico. Our portant behavioral factors. Cleaning the skin or paraphernalia inability to grow C. botulinum from 12 small BTH samples was not protective [12]. studied may mean that these samples were not from the same Delays in diagnosis (up to 64 days) were not uncommon, batch of BTH that led to WB, that spores of C. botulinum are even at major medical centers. The diseases most commonly not uniformly distributed in BTH, or that the BTH sample confused with botulism were Guillain-Barre´syndrome (espe- sizes were simply too small. We have, however, cultured C. cially the Miller Fisher variant) and myasthenia gravis. These botulinum from the washings of the inside of a syringe used by conditions are much more common [13]. Botulism can be dis- a patient with WB in 1996 [11]. tinguished from these and other diseases that mimic it by the According to drug control officials, those who harvest, pro- clinical and laboratory features (including electromyography cess, distribute, and sell BTH tend to be Hispanic. These dealers and CSF studies) outlined in table 2. We emphasize that, in are reportedly reluctant to sell to blacks, whereas Asians tend botulism, the following are true: (1) the first neurologic symp- to buy their heroin from other Asians (who market “China toms are typically diplopia, blurred vision, ptosis, or some com- White”). This may explain the virtual absence of instances of bination of these; (2) there are no objective sensory abnor- WB in patients of races other than white and why most patients malities, nor altered mental status; (3) the CSF protein is not have been Hispanic. elevated at any point in the illness; (4) electromyography The fact that patients with WB in California occur sporad- (EMG) will tend to show augmentation of the muscle action ically and rarely, even among those sharing the same BTH, potential with repetitive nerve stimulation at 20–50 Hz [14]; could suggest that C. botulinum contamination in BTH is not and (5) if disease progresses, paralysis tends to occur in a de- uniformly distributed. Alternatively, the risk of WB could relate scending, generally symmetrical fashion. to behavioral factors. Indeed, our recent study of risk factors In addition to delays in diagnosis of WB, we believe cases demonstrated that the route of administration (e.g., by skin of WB are being missed altogether. The reasons include in- popping) and the amount injected subcutaneously were im- adequate volumes of sera submitted for toxin testing; sera taken CID 2000;31 (October) Wound Botulism 1023

Table 2. Paralytic diseases that may be confused with botulism. Disease Features distinguishing this disease from botulism GBS Ascending paralysis, paresthesia, pain, and increased CSF protein after 1–2 weeks; marked slowing of NCV; EMG may help distinguish: no augmentation of MAP with RNS at 20–50 Hz, whereas augmentation is seen in botulism; antiganglioside antibody may be elevated in serum or CSF in patients with GBS or Miller Fisher variant of GBS Miller Fisher variant of GBS Internal and external ophthalmoplegia along with ataxia, which is more marked than limb weakness; areflexia tends to occur also in unweakened extremities; also, paresthesia, pain, and increased CSF protein after 1–2 weeks; NCV and EMG may help distinguish, as described above for GBS; antiganglioside antibody may be elevated in serum or CSF in patients with GBS or Miller Fisher variant of GBS Myasthenia gravis Muscular fatigability, dramatic resolution of paralysis with Tensilon (but partial improvement can be seen in ∼25% of mild botulism cases); EMG may help distinguish: decrement in MAP with RNS at 3 Hz Cerebrovascular accident involving midbrain Long tract signs, sensory deficits, and asymmetric hyperactive deep tendon reflexes Polio and other encephalitides Fever, altered mental state, abnormal CSF (pleocytosis and increased protein), and asymmetric weakness Tick paralysis Presence of a tick, paresthesia of affected extremities, ascending paralysis; rarely involves

muscles served by cranial nerves Downloaded from Wernicke’s encephalopathy Altered mental status, ataxia, history of alcohol abuse Paralytic shellfish poisoning Incubation period 15–60 min, paresthesia Carbon monoxide poisoning Altered mental status, cherry-colored skin Organophosphate poisoning Fever, altered mental state, miosis, increased oral and respiratory secretions, fasciculations, muscle cramps, paresthesia, depressed serum cholinesterase Eaton-Lambert myasthenic syndrome Associated with neoplasia, especially oat cell carcinoma of lung; muscle weakness is mainly proximal; usually spares ocular and bulbar muscles; RNS shows much greater http://cid.oxfordjournals.org/ augmentation of MAP at 20–50 Hz (e.g., 1100%) than does botulism (e.g., 30%) or myasthenia gravis Electrolyte abnormalities High serum magnesium (Mgϩ2) levels Aminoglycoside paralysis (gentamicin, streptomycin, tobramycin, etc.) Secondary to antibiotics administered under general anesthesia Poisoning with belladonna-like alkaloids Fever, tachycardia, altered mental status NOTE. CSF, cerebrospinal fluid; EMG, electromyogram; GBS, Guillain-Barre´syndrome; MAP, muscle action potential; NCV, nerve conduction velocity; RNS, repetitive nerve stimulation. by guest on October 1, 2012 too late in the course of illness to detect circulating toxin; and amount of toxin ingested, whereas in WB toxin can continue lack of consideration of the diagnosis of WB, so sera are not to be produced in situ until C. botulinum infection is eliminated. even submitted for testing. For example, unattended deaths Treatment for WB often includes not only botulinum antitoxin may be ascribed to drug overdoses when tests for opiates are but also to incision or excision, surgical debridement, irrigation, positive. and appropriate antibiotics. Normally, 1 vial of equine trivalent All patients presenting with clinical botulism should be con- botulinum antitoxin is sufficient, because each vial reportedly firmed in a reference laboratory experienced in testing for bot- contains a level of antitoxin that is 1100 times that needed to ulism. The most useful diagnostic specimen is serum (15 mL), neutralize the largest amount of circulating toxin ever measured which was found to be positive for botulism toxin in 93% of at the Centers for Disease Control and Prevention (CDC). Ac- cases. Anaerobic cultures of the wound (best done by placing cordingly, the CDC has recommended that only 1 vial of an- a specimen in an anaerobic transport pack, refrigerating it titoxin be used [15]. Twice, however, we found toxin in sera promptly, and then inoculating it into cooked meat glucose taken 12–24 h after intravenous antitoxin administration that broth) were positive for 65% of patients studied. We also de- necessitated our release of another vial of antitoxin. (Besides tected free toxin directly from abscessed tissue taken from 32% obtaining serum before botulinum antitoxin is administered, of 56 patients studied, without first growing C. botulinum. To we routinely request posttreatment sera for testing of IDU pa- our knowledge, ours is the first report on the detection of free tients with WB, to assure complete toxin neutralization.) Ide- botulinum toxin in tissue. ally, any debridement or irrigation of a wound should be done At least 15% of our patients had wounds that did not appear after antitoxin is administered, so that toxin that enters the grossly infected. Therefore, WB should be considered in the bloodstream by these procedures can be neutralized. Penicillin, differential diagnosis of cranial nerve palsy and descending administered at a dose of 10–20 million units per day, is rec- paralysis, even if injection sites or other wound sites appear ommended, although 190% of 224 strains of C. botulinum tested benign. We also identified unusual and concealed sites (e.g., by Swenson et al. [16] were also susceptible to tetracycline, vulva and base of tongue) for injection. A toxicology screen metronidazole, and chloramphenicol. Aminoglycosides should for opiates may be helpful in detecting unreported use of BTH. be avoided because they can potentiate neuromuscular block- An important distinction between foodborne botulism and ade [17, 18]. WB is that in foodborne botulism exposure is limited to the When a clinical diagnosis of botulism is made, physicians 1024 Werner et al. CID 2000;31 (October) should consider immediate administration of antitoxin before 6. Centers for Disease Control and Prevention. Wound botulism—California, laboratory confirmation (which could take days) to prevent 1995. MMWR Morb Mortal Wkly Rep 1995;44:889–92. 7. MacDonald KL, Rutherford GW, Friedman SM, et al. Botulism and botu- disease progression and death. Patients receiving antitoxin on lism-like illness in chronic drug users. Ann Intern Med 1985;102:616–8. day 1 of their symptoms tend to have fewer days on a respirator 8. Department of Health and Human Services, US Public Health Service. Na- and shorter hospitalizations than those who begin treatment tional household survey on drug abuse: population estimates 1994 (DHHS later [19]. On the other hand, antitoxin may not be necessary publication SMA 95-3093). Rockville, MD: US Department of Health when a patient is clinically stable or improving. and Human Services, 1995. As with tetanus toxin, botulinum neurotoxin may cause life- 9. Cherubin CE. Urban tetanus—the epidemiologic aspects of tetanus in nar- threatening illness at doses insufficient to immunize. One Cal- cotic addicts in New York City. Arch Environ Health 1967;14:802–8. 10. Smith LDS, Sugiyama H. Botulism—the organism, its toxins, the disease. ifornia IDU had type A WB twice, 2 years apart, just as others 2d ed. Springfield, IL: Charles C. Thomas, 1988. have had 2 episodes of foodborne botulism caused by the same 11. Werner SB, Passaro DJ, McGee J, Vugia DJ. “Shooter’s botulism”: epidemic botulinum toxin type [20–22]. Protective immunization with wound botulism in California. New Orleans, LA: Interscience Conference botulinum toxoid, as offered to selected laboratory workers, on Antimicrobial Agents and Chemotherapy, September 1996. requires multiple doses over many months [23]. 12. Passaro DJ, Werner SB, McGee J, MacKenzie WR, Vugia DJ. Wound bot- ulism among injecting drug users associated with black tar heroin. JAMA

Clinically suspected cases of botulism should be reported Downloaded from 1998;279:859–63. immediately to local public health agencies to facilitate labo- 13. Ropper AH. The Guillain-Barre´syndrome. N Engl J Med 1992;326:1130–6. ratory confirmation, release of antitoxin, and prompt investi- 14. Werner SB. Food poisoning. In: Wallace RB, ed. Maxcy-Rosenau-Last Pub- gation. If local and state officials are not immediately available, lic health and preventive medicine. 14th ed. Stamford, CT: Appleton and the CDC can be contacted directly at (404) 639-2888. Lange, 1998:263–71. 15. Shapiro RL, Hatheway C, Swerdlow DL. Botulism in the United States: a

Clinicians should be on the alert for WB cases associated http://cid.oxfordjournals.org/ with injection drug use. The problem is increasing in California clinical and epidemiologic review. Ann Intern Med 1998;129:221–8. and elsewhere. IDUs need to know that in addition to over- 16. Swenson JM, Thornsberry C, McCroskey LM, Hatheway CL, Dowell VR Jr. Susceptibility of Clostridium botulinum to thirteen antimicrobial agents. doses, HIV, endocarditis, hepatitis B and C, tetanus, and soft Antimicrob Agents Chemother 1980;18:13–9. tissue infections, including necrotizing fasciitis, there is another 17. Santos JI, Swensen P, Glasgow LA. Potentiation of Clostridium botulinum potentially lethal consequence of heroin, known as “shooter’s toxin by amingoglycoside antibiotics: clinical and laboratory observations. botulism.” Pediatrics 1981;68:50–4. 18. L’Hommedieu C, Stough R, Brown L, Kettrick R, Polin R. Potentiation of neuromuscular weakness in infant botulism by aminoglycosides. J Pediatr by guest on October 1, 2012 References 1979;95:1065–70. 19. Tacket CO, Shandera WX, Mann JM, Hargrett NT, Blake PA. Equine anti- 1. Davis JB, Mattman LH, Wiley M. Clostridium botulinum in a fatal wound toxin use and other factors that predict outcome in foodborne botulism. infection. JAMA 1951;146:646–8. Am J Med 1984;76:794–8. 2. Pickett J, Berg B, Chaplin E, Brustetter-Shafer M-A. Syndrome of botulism 20. Koenig MG, Drutz DJ, Mushlin AI, Schaffner W, Rogers DE. Type B botu- in infancy: clinical and electrophysiological study. N Engl J Med 1976; lism in man. Am J Med 1967;42:208–19. 295:770–2. 21. Beller M, Middaugh JP. Repeated type E botulism in an Alaskan Eskimo. 3. Weber JT, Goodpasture HC, Alexander H, Werner SB, Hatheway CL, Tauxe N Engl J Med 1990;322:855. R. Wound botulism in a patient with a tooth abscess: case report and review. Clin Infect Dis 1993;16:635–9. 22. Schroeder K, Tollofsrud A. Botulism from fermented trout. J Norwegian 4. Rogerson PA. A generalization of Hewitt’s test for seasonality. Int J Epi- Med Assoc 1962;82:1084–6. demiol 1996;25:644–8. 23. Cardella MA. Botulinum toxoids. In: Lewis KH, Cassel K, eds. Botulism: 5. California Department of Health Services. Wound botulism—California, proceedings of a symposium. Cincinnati, OH: US Department of Health, 1995. California Morbidity. October 1995. Education, and Welfare, 1964:113–30.

Note Added in Proof. The epidemic of wound botulism is continuing in California and, in fact, is increasing. In 1999, 38 laboratory-confirmed cases were identified as compared to a previous high of 28 cases in 1998. All but 1 of the cases in 1999 were injection drug users. Once again, a case presented a second time: a 68-year-old man was a laboratory-confirmed case in 1994 and, again, in 1999. Wound Botulism Associated With Black Tar Heroin Among Injecting Drug Users

Douglas J. Passaro, MD, MPH; S. Benson Werner, MD, MPH; Jim McGee, MSPH; William R. Mac Kenzie, MD; Duc J. Vugia, MD, MPH

Context.—Wound botulism (WB) is a potentially lethal, descending, flaccid, pa- tiallylethalparalyticdiseasethatresults ralysis that results when spores of Clostridium botulinum germinate in a wound and from ingestion of preserved food con- elaborate neurotoxin. Since 1988, California has experienced a dramatic increase taining preformed botulinum toxin. in WB associated with injecting “black tar” heroin (BTH), a dark, tarry form of the Wound botulism is a clinically similar drug. syndrome of flaccid, symmetric, de- scending paralysis that results when Objective.—To identify risk factors for WB among injecting drug users (IDUs). spores of Clostridium botulinum, an ob- Design.—Case-control study based on data from in-person and telephone ligate anaerobe, are inoculated into a interviews. wound or other devitalized tissue.12 Af- Participants.—Case patients (n=26) were IDUs who developed WB from ter gaining access to this relatively an- January 1994 through February 1996. Controls (n=110) were IDUs newly enrolled aerobic environment, the spores germi- in methadone detoxification programs in 4 counties. nate and elaborate the most potent toxin Main Outcome Measures.—Factors associated with the development of WB. known.13 Historically, the implicated Results.—Among the 26 patients, the median age was 41.5 years, 15 (58%) wound has been a crush injury or other 14 were women, 14 (54%) were non-Hispanic white, 11 (42%) were Hispanic, and gross trauma to an extremity. The first none were positive for the human immunodeficiency virus. Nearly all participants reported case of WB associated with in- jecting drug use occurred in 1982 in New (96% of patients and 97% of controls) injected BTH, and the mean cumulative dose York City.15 of BTH used per month was similar for patients and controls (27 g and 31 g, All forms of botulism are reportable respectively; P=.6). Patients were more likely than controls to inject drugs subcu- diseases in California. Since 1988, the taneously or intramuscularly (92% vs 44%, PϽ.001) and used this route of drug year of California’s first reported WB administration more times per month (mean, 67 vs 24, PϽ.001), with a greater case associated with injecting drug use, cumulative monthly dose of BTH (22.3 g vs 6.3 g, PϽ.001). A dose-response re- the number of WB cases has increased lationship was observed between the monthly cumulative dose of BTH injected dramatically, totaling 49 from 1988 subcutaneously or intramuscularly and the development of WB (␹2 for linear trend, through 1995; 46 of these cases occurred 26.5; PϽ.001). In the final regression model, subcutaneous or intramuscular injec- in IDUs, nearly all of whom injected pri- tion of BTH was the only behavior associated with WB among IDUs (odds ratio, marily heroin. In each case, the same type of heroin was used: “black tar” 13.7; 95% confidence interval, 3.0-63.0). The risk for development of WB was not heroin (BTH), a black, gummy form of affected by cleaning the skin, cleaning injection paraphernalia, or sharing needles. the drug that usually is synthesized in Conclusions.—Injection of BTH intramuscularly or subcutaneously is the makeshift factories adjacent to opium primary risk factor for the development of WB. Physicians in the western United poppy fields in several Mexican states.16 States, where BTH is widely used, should be aware of the potential for WB to occur We performed a case-control study to among IDUs. compare IDUs who developed WB with JAMA. 1998;279:859-863 other heroin users to identify risk fac- tors associated with the disease. From the Division of Communicable Disease Control, AN ESTIMATED 80 000 Americans, in- California Department of Health Services, Berkeley (Drs cluding 18 000 Californians, inject illicit METHODS Passaro, Werner, and Vugia and Mr McGee), and the 1 Division of Infectious Diseases and Geographic Medi- drugs. Compared with the general popu- Case patients (patients) were IDUs cine, Stanford University Medical School, Stanford, lation, injecting drug users (IDUs) are at who developed laboratory-confirmed Calif (Dr Passaro), and the Division of Field Epidemiol- increasedriskfordiseaseanddeath2-6 and WB from January 1, 1994, through ogy, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta, Ga (Drs Passaro and use a disproportionate amount of medical March 1, 1996. All CDHS case records Mac Kenzie). Dr Passaro is now with the Division of In- resources.7,8 Soft tissue infections are a since 1994 were reviewed, and current fectious Diseases and Geographic Medicine, Stanford particular problem.9-11 telephone numbers and addresses of pa- University Medical Center. Dr Mac Kenzie is now with the Division of Parasitic Diseases, Centers for Disease In 1994, the California Department of tients were obtained, generally through Control and Prevention. Health Services (CDHS) noted an in- hospital records or by contact tracing. Reprints: Douglas J. Passaro, MD, MPH, c/o Division creasing number of cases of wound botu- Telephone or in-person interviews were of Communicable Disease Control, California Depart- ment of Health Services, Berkeley, CA 94704 (e-mail: lism (WB), an unusual soft tissue infec- conducteddirectlywitheachpatient.Pa- [email protected]). tion. Botulism is best known as a poten- tients unable to speak because of ongo-

JAMA, March 18, 1998—Vol 279, No. 11 Wound Botulism Among Injecting Drug Users—Passaro et al 859 ©1998 American Medical Association. All rights reserved. Table 1.—Baseline Characteristics of California Wound Botulism Patients, January 1994 Through February gression to all variables associated with 1996, Compared With Control and Reference Groups a significance level of PϽ.2 in the bivar- ␹2 Odds Ratio iate analysis; likelihood ratio statis- (95% Confidence tics were compared to assess the good- Case Group Control Group Reference Group Interval), Case vs ness of fit of increasingly parsimonious Characteristics (n = 26) (n = 110)* (n = 47 809)† Control Groups multivariable models.17 Because con- Median age, y (range) 41.5 (21-69) 40.5 (24-63) 38 (5-92) . . .‡ trols were clustered by county, robust Female sex, No. (%) 15/26 (58) 37/106 (35) 16 207/47 809 (34) 2.5 (1.0-6.7) (Huber/White/sandwich) estimators of Race/ethnicity, No. (%) Black 0/26 (0) 27/109 (25) 6844/46 841 (15) 0.0 (0.0-0.6) variance were used to calculate SEs in Hispanic 11/26 (42) 43/109 (39) 18 694/46 841 (40) 1.1 (0.4-2.9) the multivariable analyses (Stata 5.0, Non-Hispanic white 14/26 (54) 37/109 (34) 20 281/46 841 (43) 2.3 (0.9-5.9) Stata Corp, College Station, Tex). Analy- Other 1/26 (4) 2/109 (2) 1022/46 841 (2) 2.1 (0.0-32.0) ses involving needle-exchange pro- Body mass index, kg/m2 25 24 ...... grams were restricted to patients and controls from Alameda, Los Angeles, San *Controls were selected new enrollees of methadone detoxification programs in 4 California counties, March to Joaquin, and Santa Clara counties. All May 1996. Reference group members were all clients of methadone detoxification or maintenance programs in 17 California counties, 1994-1995. comparisons were 2-tailed. †Sex and race information was not available from all members of the control and reference groups. ‡Ellipses indicate not available. RESULTS ing mechanical ventilation provided purchasing, storing, and using practices A total of 35 cases of laboratory-con- written responses to in-person inter- in the month before developing WB; con- firmed WB associated with injection views conducted by nursing staff. Pa- trols were asked identical questions drug use occurred in California during tientswithoutaccesstoatelephonewere aboutdrug-relatedpracticesinthemonth the 26-month study period. Of these 35 interviewed in person by staff of the Lo- before starting detoxification, including patients, all but 2 required lengthy hos- cal Assistance Branch, CDHS, Sacra- frequency and quantity of all drugs in- pitalization; all but 5 required mechani- mento, Calif. jected; frequency and method of cleaning cal ventilation. Of the 34 cases in which Controls were persons newly enrolled needlesandsyringes;frequencyofneedle botulinum toxin typing was performed, in 1 of 4 methadone detoxification pro- sharing; source of water or other solvent 30(88%)werecausedbybotulinumtoxin grams in geographically and ethnically used for dissolving heroin; type of appa- type A and 4 (12%) by type B. The case distinct cities in California (Oakland, Los ratus used for heating the heroin-water participation rate was 74%; 5 (45%) of 11 Angeles, San Jose, and Stockton) during mixture; whether cotton balls or ciga- patients diagnosed in 1994 and 21 March through May 1996. Recruitment rette filters (eg, “cottons”), through (87.5%) of 24 patients diagnosed after strategies for controls varied by clinic. At which the heroin mixture is drawn into January 1, 1995, were interviewed. the Los Angeles clinic, 25 consecutive pa- the syringe, were stored and reused; fre- Seven patients could not be located de- tients were enrolled in the study by the quency and type of skin cleansing before spite repeated attempts, and 2 refused clinic intake supervisor. At the Oakland, injection; which body sites were used for to participate. Control participation Stockton, and San Jose clinics, study in- injection; and injection technique (intra- rates varied by study site between 50% vestigators visited each of these 3 on 3 venous vs intramuscular vs subcutane- and 80% but were not precisely deter- nonconsecutive mornings. Each detoxifi- ous). Participants were also asked about mined because not all potential controls cation program patient who received their recent medical history and how fre- could be enumerated at all sites. methadone was asked to participate in quently they developed soft tissue ab- No blacks and no persons infected the study and was offered a meal voucher scesses. Each participant estimated the with the human immunodeficiency virus to encourage participation. Before inter- quantity of drugs he or she used by stated (HIV) developed WB during the study views, educational flyers about the study drug weight at point of purchase. When period; the HIV status of controls was and about WB associated with injecting this was not possible, quantities were es- not documented. Patients were less drug use (ie, “shooter’s botulism”) were timated by dollar amounts, which were likely than controls and the reference distributed at the clinics; these flyers convertedtoweightsusingthecostsmost grouptobenon-Hispanicblackandmore warned users of the local WB epidemic frequently cited by study participants: likely to be non-Hispanic white and to be among IDUs and explained warning $80 per gram of heroin and $100 per gram female, although these demographic symptomsbutdidnotdiscusshypotheses of cocaine. trends did not reach statistical signifi- about the causes of shooter’s botulism. Bivariateanalyseswereperformedby cance.Baselinecharacteristicswereoth- After each interview, participants were using the Fisher exact test or the Man- erwisesimilarbetweengroups(Table1). given verbal and written information tel-Haenszel ␹2 test (with the Yates cor- Of 34 drug purchasing, storing, and about shooter’s botulism. rection) for discrete variables and the using practices analyzed, the behavior Because our selection of methadone Student t test or the Wilcoxon 2-sample most strongly associated with the devel- clinic study sites was not random and to test for continuous variables. Since 34 opment of WB was injecting BTH sub- examine whether our control group was practices and characteristics were as- cutaneously or intramuscularly (skin- representative of California methadone sessedinthebivariateanalyses,theBon- popping) rather than intravenously clinic attendees, we also compared base- ferroni correction was used to provide a (Table 2). A total of 33 of 35 patients re- line patient characteristics of our con- stringent test of significance. Therefore, ported this route of BTH administration trol group with those of a reference bivariate associations were judged sig- at least occasionally. One of the 2 excep- group. The reference group consisted of nificantonlyifPϽ.002(ie,0.05/34).Dose- tions was an intravenous amphetamine allenrolleesinmethadonedetoxification response relationships were assessed user who insisted that she had never or maintenance programs during 1994 using the ␹2 test for linear trend (Epi skin-popped, never used BTH, and or 1995 from the 17 California counties Info 6.02, Centers for Disease Control never shared paraphernalia with BTH that have reported WB among IDUs. and Prevention, Atlanta, Ga). Multivari- users. The other exception was a patient Patients were questioned about base- able analyses were performed by apply- who reported injecting BTH but only in- line personal characteristics and 34 drug ing backwards-elimination logistic re- travenously.

860 JAMA, March 18, 1998—Vol 279, No. 11 Wound Botulism Among Injecting Drug Users—Passaro et al ©1998 American Medical Association. All rights reserved. Table 2.—Drug-Using Behaviors of California Wound Botulism Patients, January 1994 Through February 1996, Compared With Controls: Bivariate Analysis*

Odds Ratio (95% Confidence Interval), Case Group Control Group Case vs Control Behaviors (n = 26) (n = 110) Groups P Value Injection of BTH, No. (%) 25 (96) 107 (97) 0.7 (0.1-38.3) .6‡ No. of injections per month Mean (SD) 82 (39) 117 (63) . . . .01¤ Median (IQR) 90 (60-150) 90 (60-300) ... Dose of BTH per injection, g Mean (SD) 0.33 (0.21) 0.28 (0.14) ... .6¤ Median (IQR) 0.25 (0.25-0.5) 0.25 (0.25-0.31) ... Dose of BTH injected per month, g Mean (SD) 27 (20) 31 (21) ... .6¤ Median (IQR) 22.5 (12.3-37.5) 28.1 (15.0-37.5) ... Skin-pops† BTH, No. (%) 24 (92) 48 (44) 15.5 (3.5-139.5) Ͻ.001࿣ No. of skin-pops per month Figure 1.—Sex- and race-adjusted odds of being a “shooter’s botulism” patient, by cumulative monthly Mean (SD) 67 (51) 24 (45) ... Ͻ .001¤ dose of black tar heroin (BTH) injected subcutane- Median (IQR) 72 (16-112) 0 (0-28) ... ously (quartiles). Participants were asked to esti- Dose of BTH per skin-pop, g mate the quantity of heroin used (injected subcuta- Mean (SD) 0.34 (0.22) 0.29 (0.14) ... .5¤ neously or intramuscularly) daily or weekly, by Median (IQR) 0.25 (0.25-0.5) 0.25 (0.13-31.3) ... weight or by dollar amount. Weights were converted BTH dose skin-popped per month, g into monthly cumulative dollar amounts using the conversion ratio of $80 per gram of BTH.The refer- Mean (SD) 22.3 (20.5) 6.30 (12.5) ... Ͻ .001¤ ence category included those who did not inject Median (IQR) 20.0 (6.5-35.0) 0 (0.8-8) ... BTH subcutaneously or intramuscularly. Bars rep- No. of persons sharing paraphernalia resent point estimates; error bars, upper 95% con- with user in most recent month fidence intervals. Mean (SD) 0.4 (. . .) 0.6 (. . .) ... .04¤ Median (IQR) 0 (0-0) 0 (0-1) ... 6-fold higher among “occasional” subcu- Always or usually cleans paraphernalia 13 (50) 43 (39) 1.6 (0.6-4.0) .4࿣ taneousorintramuscularinjectionusers between injections, No. (%) (whose dosage was in the lowest quar- Always or usually cleans skin before 4 (16) 23 (21) 0.7 (0.2-2.3) .7࿣ injections, No. (%) tile, $20-$480 of BTH per month) and 25- No. of abscesses requiring medical fold higher among “heavy” users (whose attention or antibiotics in previous year dosage of drug injected subcutaneously Mean (SD) 3.0 (5.4) 0.6 (1.6) ... Ͻ.001¤ or intramuscularly was in the highest Median (IQR) 1 (0-3.5) 0 (0-0) ... quartile,$2000-$6300ofBTHpermonth) No. of alcoholic drinks per month (Figure 1). Mean (SD) 7.3 (19.8) 14.4 (24.3) ... .03¤ Median (IQR) 0 (0-2) 2 (0-21) ... COMMENT *Controls were selected new enrollees of methadone detoxification programs in 4 California counties, March to May From 1951 through 1995, 68 cases of 1996. IQR indicates interquartile range; BTH, black tar heroin; and ellipses, not applicable. WB were reported to CDHS. An aver- †Skin-popping is the practice of injecting drugs subcutaneously or intramuscularly.Askin-pop is a single subcutaneous or intramuscular injection. age of 0.49 WB cases per year were re- ‡P value obtained by Fisher exact test. ported from 1951 through 1987; 2.25 ¤P value obtained by the Wilcoxon 2-sample test. ࿣P value obtained by ␹2 test with Yates correction. cases per year were reported in 1988 through 1991, 3 cases in 1992, 4 in 1993, The total quantity of BTH used injections did not protect against devel- 11 in 1994, and 23 in 1995.18 From 1988 monthly by patients and controls was oping WB. Using needle-exchange pro- through 1995, only 2 WB cases among similar,buttheamountthatwasinjected grams did not protect against WB, and IDUs were reported from outside Cali- subcutaneously or intramuscularly was sharing injection paraphernalia was not fornia, and they occurred in Arizona greater among patients than among con- associated with the disease. (Foodborne and Diarrheal Disease trols (mean, 22.3 vs 6.3 grams; PϽ.001). In the final (parsimonious) multivari- Branch,CentersforDiseaseControland A dose-response relationship was ob- able model, injecting BTH subcutane- Prevention, unpublished data, 1996). served between the monthly dosage of ously or intramuscularly (odds ratio This study confirms that the ongoing BTH injected subcutaneously or intra- [OR], 13.7; 95% confidence interval [CI], epidemic of WB in California is strongly muscularly by quartile and the risk of 3.0-63.0; P=.001), sex, and race were the associated with subcutaneous or intra- developingWB(␹2 forlineartrend=26.5; only factors associated with WB. His- muscularinjection(skin-popping)ofBTH PϽ.001). This dose-response relation- panic and non-Hispanic whites had an and provides evidence to suggest that C ship remained when analysis was lim- increased risk of WB (OR, undefined; botulinum contamination of BTH is not ited to persons who reported injecting PϽ.001)andwomenhadanonsignificant theresultofspecificdrugstorageorother BTHsubcutaneouslyorintramuscularly increased risk of WB (OR, 2.2; 95% CI, using behaviors among IDUs. Although (␹2 for linear trend=4.1; P=.04). 0.8-5.9; P=.13). To better quantify the intravenous injection of heroin provides Patients reported having more ab- relationship between BTH dose and risk astrongerinitial“high,”“skinpopping”is scesses that received medical treatment for illness, race- and sex-adjusted ORs favored by users reluctant to inject intra- in the previous year than controls (Table were calculated for quartiles of cumula- venously, who desire to avoid telltale 2). Although cleaning the skin before in- tive BTH dose injected subcutaneously “track marks,” or for whom venous ac- jection may have protected against de- or intramuscularly per month. When cess is difficult because of obesity or the veloping soft tissue abscesses (P=.07, compared with IDUs who denied inject- scarring of veins from repeated use. datanotshown),cleaningtheskinbefore ing BTH subcutaneously or intramuscu- Black tar heroin was the only drug that injection or cleaning syringes between larly, the odds of developing WB were patientswithWBreportedskin-popping.

JAMA, March 18, 1998—Vol 279, No. 11 Wound Botulism Among Injecting Drug Users—Passaro et al 861 ©1998 American Medical Association. All rights reserved. is unlikely to result in a contaminated finalproduct.Second,duringdetailedin- terviews, all 26 patients denied adding Cases per Million Inhabitants othersubstancesorsolventsorusingun- (1990 US Census Figures) usual water sources when preparing 0 drugs for injection (2 controls reported <2.5 occasionally using beer or wine as a sol- vent). Therefore, inadvertent contami- 2.5-4.99 nation of heroin by individual users ≥5.0 seemsunlikelytobethesourceofCbotu- linum spores. Third, frequency of skin Yolo cleansing before injection and the type of cleanser used were not associated 3 Sacramento with WB, and the quantity of BTH used 1 San Joaquin was a more important factor than the Lake 2 4 frequency of use. Therefore, skin con- Stanislaus tamination with C botulinum appears unlikely to be a major source of this epi- 2 Sonoma demic. Fourth, the frequency of using 2 San Francisco 2 3 new paraphernalia, the frequency and 2 method of cleaning old paraphernalia, Alameda 2 and the sharing of paraphernalia were Fresno not associated with WB. Therefore, Santa Clara spread of WB via fomites or blood is un- likely to be a factor in this epidemic. For 1 San these reasons, we suspect that BTH is Bernardino most likely being contaminated when di- Monterey luted (eg, possibly with soil) after manu- facture or during distribution. Fundamental questions about this 3 outbreak remain unanswered and prob- 1 ablyreflectincompleteunderstandingof 3 heroin distribution. For example, the 5 reason that the location of cases (Figure Santa Barbara 2) was spread throughout California yet Ventura 3 essentially spared the rest of the west- 7 Riverside Los Angeles ern United States (where BTH is also distributed) is unexplained, although Orange this pattern is consistent with contami- nation of BTH during in-state distribu- tion. The lack of cases of WB occurring among black IDUs is also unclear. Drug Figure 2.—Cases of wound botulism among injecting drug users in California, by county, 1988 through 1995. enforcement officials hypothesize that Boldface numbers within each county represent the number of laboratory-confirmed cases of wound botulism distrust between black IDUs and sup- associated with injecting drug use reported by county from 1988 through 1995.19 Unshaded areas indicate counties that reported no cases of wound botulism; green shading, counties that reported fewer than 2.5 cases pliers of BTH manufactured by smaller, per million inhabitants from 1988 through 1995; yellow shading, counties that reported 2.50 to 4.99 cases per lessexperiencedproducersanddistribu- million inhabitants; and red shading, counties that reported 5 cases or more per million inhabitants. tors (which might be more highly con- taminated) has minimized the use of this Black tar heroin was introduced to US (inert materials, such as dextrose, that type of BTH among black IDUs. drugusersinthe1970sandslowlygained are used to “cut” heroin, providing bulk Reported cases of WB may represent market share from traditional white and weight and increasing the distribu- only a small fraction of this epidemic. heroin because Central and South tors’ profit margin). The tarry color and Nearly all patients diagnosed as having Americansupplierswereabletodevelop consistency of BTH has led to the use of botulism in California are described, dominant distribution networks and be- unusual diluents, including ground pa- because botulinum antitoxin is available cause initially BTH was cheaper and per fiber soaked in black shoe polish and, to California physicians only through more potent (up to 50% by weight diace- according to anecdotal reports, dirt. CDHS. However, botulism is a rapidly tylmorphine). Since the late 1980s, BTH Our study was not designed to deter- progressivediseaseandpersonswithlim- has become the predominant form of mine at which step in heroin production ited access to care or who delay seeking heroin in the United States west of the anddistributioncontaminationwasmost health care may be dying outside the hos- Mississippi River. In 1993, 20 of 21 likely to occur. However, there are sev- pital. In these circumstances, the diagno- samples of California heroin purchased eral reasons to suspect that contamina- sis of WB could be missed. For example, by undercover agents of the US Drug tion occurs during “cutting.” First, the if postmortem examination revealed de- Enforcement Agency were BTH.19 At last step in the conversion of opium to tectable serum opiate levels, an IDU present,heroinishighlyimpureandcon- BTH involves boiling the product with a might be presumed to have died from an tains contaminants (by-products of the strong acid at 150°C for several hours, overdose. In addition, there have been manufacturing process), adulterants which should destroy even heat-resis- several instances in which diagnosis of (chemicals such as methamphetamine, tant C botulinum spores. Therefore, WB has been delayed despite consulta- strychnine, or xylocaine), and diluents contamination before or during this step tion with neurologists and infectious dis-

862 JAMA, March 18, 1998—Vol 279, No. 11 Wound Botulism Among Injecting Drug Users—Passaro et al ©1998 American Medical Association. All rights reserved. ease specialists; a subset of persons with the internal surface of a syringe used by methadone clinics participating in this WB may have been diagnosed as having a patient with WB, and we have also cul- study. another neuromuscular disorder.19 tured related Clostridia species from Subcutaneous or intramuscular injec- Unlike the more widely publicized in- BTH samples belonging to other pa- tion of BTH is the primary risk factor for fectious complications of drug injection tients with WB.19 shooter’s botulism. In addition to coun- eg, HIV infection and viral hepatitis B, We do not know the source of C botu- seling IDUs to stop using BTH or, at WB is not contagious. In our study, risk linumsporescontaminatingBTH.How- least, to minimize the amount of BTH of disease was not associated with mark- ever, our findings suggest that BTH is that is injected, additional efforts are ers of exposure to other IDUs (eg, fre- contaminated before sale to IDUs and needed to increase awareness of WB quency of sharing needles). Although 3 that simple measures, such as cleaning among IDUs and health care workers clusters of WB involving pairs of “shoot- the skin before injection and cleaning sy- who serve them and to increase access of ing partners” (persons that use drugs to- ringes, are unlikely to prevent WB. heroin users to methadone detoxifica- gether) have been reported, in all 3 epi- The CDHS has taken several steps to tion and maintenance programs. Physi- sodes the partners had used the same make drug users, public health officials, cians in the western United States heroin (S.B.W., unpublished data, 1996). and physicians aware of this growing should be alerted to the potential for WB The CDHS has been unable to procure problem. In October 1995, informational occuring among IDUs. BTH samples large enough to ad- packets containing both technical and equately test for C botulinum. Further- lay fact sheets were sent to every public The authors acknowledge the work of the Local Assistance Branch, California Department of more, the samples obtained have not health jurisdiction in California for dis- Health Services, Berkeley; Sally Jew, California been closely linked to BTH samples tribution to emergency departments, Department of Alcohol and Drug Programs, Sacra- thought to have caused illness. Accord- needle-exchange programs, and metha- mento; Jeffrey Klausner, MD, MPH; and the assis- ingly, our suspicion that BTH contains done clinics. A report of the outbreak tance of staff at the East 14th Street Clinic, Oak- 16 land, the San Joaquin County Medical Center Clinic, botulism spores remains unproven. How- has been published. Education of IDUs Stockton, the Los Angeles Department of Health ever, we have cultured C botulinum from and clinic staff was also provided to the Services, and the South Valley Clinic, San Jose.

References 1. US Department of Health and Human Services, 7. Baldwin WA, Rosenfeld BA, Breslow MJ, Buch- 14. California Department of Health Services. National Institute on Drug Abuse. Final Report of man TG, Deutschman CS, Moore RD. Substance- Wound botulism in California, 1951-85. California the 1994 Household Survey on Drug Abuse. Rock- abuse related admission to adult intensive care. Morb. 1986;26:1. ville, Md: National Institute on Drug Abuse; 1995. Chest. 1993;103:21-25. 15. MacDonaldKL,RutherfordGW,FriedmanSM, 2. Wyskowski DK, Schober SE, Wise RP, Kopstein 8. ZookC,MooreF.High-costusersofmedicalcare. et al. Botulism and botulism-like illness in chronic A. Mortality attributed to the misuse of psychoactive N Engl J Med. 1980;302:996-1102. drug users. Ann Intern Med. 1985;102:616-618. drugs 1979-1988. Public Health Rep. 1993;108:565-70. 9. Haverkos HW, Lange WR. Serious infections 16. Centers for Disease Control and Prevention. 3. Goldstrim A, Herrera J. Heroin addicts and other than human immunodeficiency virus among Wound botulism, California. MMWR Morb Mortal methadonetreatmentinAlbuquerque:a22-yearfol- injection drug users. J Infect Dis. 1990;161:894- Wkly Rep. 1995;44:889-892. low-up. Drug Alcohol Depend. 1995;40:139-50. 902. 17. Selvin S. Practical Biostatistical Methods. Bel- 4. Oppenheimer E, Tobutt C, Taylor C, Andrew T. 10. Stein MD. Medical complications of intravenous mont, Calif: Duxbury Press; 1995:378-379. Death and survival in a cohort of heroin addicts from drug use. J Gen Intern Med. 1990;5:249-257. 18. Werner SB, Passaro DJ, McGee J, Vugia DJ. London clinics: a 22-year follow-up study. Addic- 11. Vlahov D, Sullivan M, Astemborski J, Nelson “Shooter’s botulism”: epidemic wound botulism in tion. 1994;89:1299-1308. KE. Bacterial infections among intravenous drug California. In: Abstracts of the 36th Annual Inter- 5. Grohnbladh L, Ohlund LS, Gunne LM. Mortality users. Public Health Rep. 1992;107:595-598. science Conference on Antimicrobial Agents and in heroin addiction: impact of methadone treatment. 12. Weber JT, Goodpasture HC, Alexander H, Chemotherapy; New Orleans, La; September 15-18, Acta Psychiatr Scand. 1990;82:223-277. WernerSB,HathewayCL,TauxeRV.Woundbotu- 1996. Abstract K109:270. 6. Marx A, Schick MT, Minder CE. Drug-related lism in a patient with a tooth abscess: case report 19. DomesticUnit,StrategicIntelligenceSection,US mortality in Switzerland from 1987 to 1989 in com- and review. Clin Infect Dis. 1993;16:635-639. Drug Enforcement Administration. Domestic Moni- parison to other countries. Int J Addictions. 1994; 13. Gill DM. Bacterial toxins: a table of lethal tor Program: 1993 Annual Summary. Washington, 29:837-860. amounts. Microbiol Rev. 1982;46:86. DC: US Drug Enforcement Administration; 1994.

JAMA, March 18, 1998—Vol 279, No. 11 Wound Botulism Among Injecting Drug Users—Passaro et al 863 ©1998 American Medical Association. All rights reserved. T h e new england journal o f medicine

images in clinical medicine

Wound Botulism

A B

Amir H. Sam, M.R.C.P. 42-year-old man with a history of subcutaneous heroin use pre- AUTHOR Sithinamsuwan RETAKE 1st Hammersmith Hospital ICM sented to the hospital with slurred speech, diplopia, and dysphagia. The phys- 2nd London, United Kingdom REG FicalFIGURE examination a-c showed bilateral ptosis (Panel A), a sluggish pupillary re- CASE 3rd [email protected] A TITLE Revised sponse to light, bilateral sixth-cranial-nerve palsies, and multiple skin abscesses on EMail Line 4-C his arms and legs. Shortly after admission,SIZE the dysphagia progressed, necessitating Enon ARTIST: mst H/T H/T Huw L.C. Beynon, F.R.C.P. intubationFILL for airway protection.Combo Single-fiber39p6 electromyography suggested a disorder Royal Free Hospital of neuromuscularAUTHOR, transmission, PLEASE NOTE: and a clinical diagnosis of wound botulism was Figure has been redrawn and type has been reset. London, United Kingdom made. He receivedPlease empirical check carefully. treatment with equine serum trivalent botulism anti- toxin and antimicrobial therapy, and the abscesses were surgically débrided (Panel B). Subsequently,JOB: 35225 botulinum toxinISSUE: was detected6-23-05 in his serum, and Clostridium botulinum was cultured from abscess specimens. The patient was extubated after 2 weeks, but complete neurologic recovery took several months. Botulinum toxin irreversibly dis- rupts stimulation-induced acetylcholine release at peripheral cholinergic synapses. The resulting clinical syndrome is typically characterized by bilateral cranial neu- ropathies and descending symmetric muscle weakness. Subcutaneous or intramus- cular injection of heroin that is contaminated with spores of C. botulinum is a major risk factor for the development of wound botulism. Copyright © 2010 Massachusetts Medical Society.

IMAGES IN CLINICAL MEDICINE The Journal welcomes consideration of new submissions for Images in Clinical Medicine. Instructions for authors and procedures for submissions can be found on the Journal’s Web site at NEJM.org. At the discretion of the editor, images that are accepted for publication may appear in the print version of the Journal, the electronic version, or both.

2444 n engl j med 363;25 nejm.org december 16, 2010 The New England Journal of Medicine Downloaded from nejm.org on October 1, 2012. For personal use only. No other uses without permission. Copyright © 2010 Massachusetts Medical Society. All rights reserved. MAJOR ARTICLE

Recurrent Wound Botulism Among Injection Drug Users in California

Jean Yuan,1,2 Gregory Inami,1 Janet Mohle-Boetani,1 and Duc J. Vugia1 1Division of Communicable Disease Control, Center for Infectious Diseases, California Department of Public Health, Richmond, California; and 2Epidemic Intelligence Service, Office of Workforce and Career Development, Centers for Disease Control and Prevention, Atlanta, Georgia Background. Botulism is an acute neurologic illness characterized by cranial nerve palsies and descending flaccid paralysis. Botulism is a rare disease and recurrent botulism even more rare. We review cases of recurrent wound botulism (WB) among injection drug users (IDUs) in California from 1993 through 2006 and describe 2 case patients. Methods. From botulism surveillance data for 1993–2006, we identified patients with .1 episode of clinical Downloaded from WB, defined as acute descending paralysis with a visible wound or recent history of injection drug use. For each patient, >1 of their WB episodes was laboratory confirmed. We extracted demographic, clinical, and laboratory information from case and laboratory reports and compared clinical characteristic frequency of initial and second WB episodes. Results. During 1993–2006, 17 IDUs had recurrent WB, 14 with 1 recurrence and 3 with 2 recurrences. Of 25 http://cid.oxfordjournals.org/ laboratory-confirmed episodes, 22 were confirmed through serum testing and 3 through wound testing. Patients were 32–61 years old, and 94% were male. All patients reported heroin injections; 88% specified black tar heroin use and 76% reported subcutaneous injection. The most common presentations were having a visible wound, speech difficulty, double vision, respiratory difficulty, and trouble swallowing. There were no significant differences in clinical presentation between initial and second episodes. Conclusions. As the California epidemic of WB among IDUs continues, WB episodes are recurring. Both clinicians and IDUs should be aware of the potential for WB to recur among IDUs to enable timely diagnosis and

early botulinum antitoxin administration and supportive care. by guest on October 1, 2012

Botulism is an acute neurologic illness characterized by naturally occurring botulism are (1) food-borne bot- cranial nerve palsies and descending flaccid paralysis. ulism resulting from ingestion of foods containing In humans, botulism is caused by botulinum neuro- preformed BoNT, (2) intestinal (infant and adult in- toxin (BoNT) types A, B, E, or F produced by Clos- testinal toxemia) botulism resulting from C. botulinum tridia species (Clostridium botulinum, C. baratii F, or colonization of the intestine followed by in situ pro- C. butyricum E) [1, 2]. Early treatment with botulinum duction of toxin, and (3) wound botulism, in which antitoxin and supportive care may prevent the pro- C. botulinum contaminates a wound and also produces gression to respiratory failure. Recovery often takes BoNT in situ [1, 2]. weeks to months, although 95% of cases in the United When first reported in 1951, wound botulism (WB) States achieve full recovery. The 3 major forms of was a rare illness associated with contaminated trau- matic or surgical wounds [3]. Since 1994, an epidemic ofWBamonginjectiondrugusers(IDUs)in California has been documented and associated with Received 16 September 2010; accepted 4 January 2011. Correspondence: Duc J. Vugia, MD, Infectious Diseases Branch, California Dept ‘‘skin popping’’ black tar heroin (BTH), the pre- of Public Health, 850 Marina Bay Pkwy, Bldg P, 2nd Fl, Richmond, CA 94804 dominant form of heroin used in the western United ([email protected]). States [4, 5]. California’s WB cases make up approx- Clinical Infectious Diseases 2011;52(7):862–866 Published by Oxford University Press on behalf of the Infectious Diseases Society of imately three-quarters of reported WB cases in the America 2011. country [6]. WB among IDUs in California often re- 1058-4838/2011/527-0001$37.00 DOI: 10.1093/cid/cir005 sulted in respiratory failure requiring mechanical

862 d CID 2011:52 (1 April) d Yuan et al ventilation and prolonged hospital stay [7, 8]. Two studies correlated nature of the data. All P values were 2-tailed. The self- found that early administration of botulinum antitoxin may reported length of time from symptom onset to presentation for decrease the length of stay in the intensive care unit [8] or the initial and subsequent infections was compared using the Wil- need and duration of mechanical ventilation [7], respectively, coxon signed-rank test. although neither study achieved statistical significance be- The CDPH Microbial Diseases Laboratory conducts tests for cause of small sample sizes. the presence of BoNT with a standard mouse bioassay [9]. The We reported the first documented case of recurrent WB early patient’s serum or culture extracts, alone or mixed with in the epidemic with an IDU having had 2 separate episodes of monovalent botulinum antitoxin, are injected into the perito- wound botulism, first in 1995 then in 1997, both due to type A neum of mice. Mice are then observed for signs or symptoms of BoNT [5]. In this report, we review and summarize the epide- botulism. If botulinum toxin is present in the sample, mice miologic and clinical characteristics of 17 IDUs with recurrent receiving the appropriate monovalent botulinum antitoxin are WB in California from 1993 through 2006 and describe 2 case asymptomatic and survive. patients’ clinical presentations in detail. RESULTS METHODS During 1993–2006, we identified 17 case patients with recurrent From botulism surveillance data collected by the California WB, 14 with 1 recurrent episode and 3 with 2 recurrent episodes.

Department of Public Health (CDPH) from January 1993 Of the 17 initial episodes of WB, 13 (76%) were laboratory Downloaded from through December 2006, we identified patients with .1episode confirmed; of the 20 recurrent episodes, 12 (60%) were labo- of clinical WB, defined as acute descending paralysis starting ratory confirmed. Ten patients had 1 laboratory-confirmed with cranial nerve palsies (including diplopia, ptosis, slurred episode, 6 had 2 laboratory-confirmed episodes, and 1 patient speech, or difficulty swallowing), with a visible wound or recent had 3. The interval from the beginning of 1 episode to the history of injection drug use. For each patient, >1 of their WB beginning of the next episode ranged from 1 to 71 months http://cid.oxfordjournals.org/ episodes must have been laboratory confirmed. (median, 16 months). In the United States, botulinum antitoxin is available exclu- Of 25 laboratory-confirmed episodes, 22 were confirmed sively through the Centers for Disease Control and Prevention through serum testing and 3 through wound testing. Twenty (CDC). In California, public health officers at CDPH review the (80%) were due to BoNT type A, and 2 were due to BoNT history and clinical presentation of suspected botulism patients type B. In the remaining episodes, collected serum samples were to determine the likelihood of clinical botulism. We collect screened for toxin, but the volume was insufficient for toxin patient information on standardized case report forms, conduct typing. One patient had recurrent WB with 2 different by guest on October 1, 2012 laboratory testing for botulism, and authorize release of botu- toxin types. Of 21 episodes in which a posttreatment serum linum antitoxin from a local CDC Quarantine Station for specimen was available and adequate for testing, 20 (95%) were treatment of clinical botulism as soon as possible while labora- negative for BoNT and 1 was positive. This patient received an tory results are pending. additional vial of antitoxin. To identify recurrent cases of WB, we matched the name and These patients were 32–61 years old at the time of their first date of birth on case report forms and laboratory test records. episode (median age, 45 years). Sixteen patients (94%) were For each case patient, we extracted demographic, clinical, risk male. Eight patients (47%) were Hispanic, 7 (41%) were non- factor, and laboratory information. We reviewed the medical Hispanic white, and 2 (12%) were non-Hispanic African charts of patients with laboratory-confirmed recurrent infection American. In 27 episodes (73%), patients had visible wounds at when demographic and clinical information was not available presentation. On physical examination, clinicians judged 20 of from the case report form. Demographic information included the visible wounds (74%) as likely to be infected. All 17 patients age, sex, race, and ethnicity. Clinical information included signs were IDUs who reported heroin injections, 15 (88%) specifying and symptoms, the time from onset of illness to treatment, BTH use. Thirteen patients (76%) reported subcutaneous in- mechanical ventilation, outcome, and the interval between ep- jection of heroin (‘‘skin popping’’), sometimes accompanied by isodes. Risk factor information included drug use, presence of intravenous or intramuscular injection. Three patients reported a visible wound, and whether the wound appeared infected. only intramuscular injection of heroin. Laboratory information included the source of sample, in- The common presenting characteristics reported from WB cluding serum or wound; BoNT type, including A, B, and E; and episodes, by initial and second episodes, are listed in Table 1. posttreatment BoNT test results. Clinical characteristic fre- The most common signs were having a visible wound, speech quency of WB patients’ initial and second episodes was com- difficulty, double vision, respiratory difficulty, and trouble pared with the McNemar test, which accounted for the swallowing. Although the frequency of clinical characteristics

Recurrent Wound Botulism in Drug Users d CID 2011:52 (1 April) d 863 Table 1. Selected Clinical Characteristics of Initial and Second Episodes of Wound Botulism in 17 Patients-California, 1993–2006

No. (%) of patients Characteristic Initial episode Second episode Pa Visible wound 15 (88) 10 (59) .06 Speech difficulty 14 (82) 9 (53) .13 Double vision 11 (65) 9 (53) .59 Respiratory difficulty 8 (47) 9 (53) .74 Trouble swallowing 12 (71) 7 (41) .17 Descending paralysis 9 (53) 6 (35) .44 Ptosis 9 (53) 7 (41) .56 Dry mouth 4 (24) 4 (24) ..99 Decreased deep tendon reflexes 2 (12) 4 (24) .41 Mechanical ventilation 10 (59) 7 (41) .32 Interval before presentation, days Median 3 2 .19b Range 0–20 0–10

NOTE. Three patients had 2 recurrent episodes. Their third episodes are not included here. a P values were calculated using the McNemar test, unless otherwise indicated. Downloaded from b P value obtained by Wilcoxon signed-rank test. was not significantly different between initial and second epi- her care required a tracheostomy and gastrostomy feeding tube. sodes, several characteristics were less frequent with second She was discharged to a rehabilitation hospital for a 1-month episodes. Patients reported having symptoms for 0 to 20 days stay. http://cid.oxfordjournals.org/ (median, 3 days) before they sought care for their initial episode. Seventeen months later, the patient presented with 2 days of Nine of the 17 patients presented to care more quickly in sub- shortness of breath and 1 day of blurry vision, neck weakness, sequent infections (median, 2 days; range, 0–10 days), but the difficulty talking, and unsteadiness while walking. On admis- median length of time to presentation for subsequent infection sion, she had an infected wound and normal neurologic ex- was not significantly different from that of initial infection. In all amination, but, over the first 3 days of admission, she developed but 1 episode, antitoxin was administered within 1 day of pre- worsening respiratory distress leading to mechanical ventilation. senting for care. Where information was available, all patients During this 6-week hospitalization, her neurologic evaluation by guest on October 1, 2012 were treated with antibiotics and the infected wounds were was inconclusive. Her Tensilon test was negative; the protein in debrided. Among the 17 patients, 10 (59%) required mechanical her cerebrospinal fluid was mildly elevated; serum testing was ventilation for their first episode, and 4 patients were mechan- negative for BoNT. She received antibiotics for the wound in- ically ventilated for every episode. fection but did not receive botulinum antitoxin and stabilized with supportive care. On discharge, she was transferred to a re- Two Clinical Case Presentations habilitation hospital. Case A. Case A was a patient with 3 episodes of wound bot- Three months later, this now 38-year-old woman presented ulism, 1 of which was seronegative. This 37-year-old woman with 2 days of headache and 1 day of worsening dysphagia and presented in December 1995 with 2 days of neck weakness with diplopia. On admission, the patient had ptosis, diplopia, dys- inability to hold her head steady, sluggishness, ptosis, and pro- phagia, dysarthria, dry mouth, vomiting, constipation, and gressive respiratory distress. Her past medical history was sig- difficulty breathing. On physical examination, she was alert and nificant for a 7-year history of injecting BTH. She was brought to oriented with bilateral ptosis, diminished deep tendon reflexes, the emergency room with respiratory insufficiency and ob- and several skin abscesses. With worsening respiratory distress tunded. She was urgently intubated, and physical examination and an iatrogenic pneumothorax, the patient was intubated and revealed right lateral rectus palsy, bilateral ptosis, neck weakness, admitted to the intensive care unit. She received botulinum inability to laterally rotate her neck to the right, bilateral antitoxin, antibiotics, and wound debridement. BoNT type A shoulder shrug weakness, and a fluctuant abscess on her right was detected in a serum sample. She improved over an 18-day thigh. She was treated with botulinum antitoxin, antibiotics, and hospital stay and was transferred to a rehabilitation hospital. wound debridement. BoNT type A was detected in a serum Case B. Case B was a patient with recurrent WB due to specimen. During her 5-week hospitalization, the patient had a a different type of C. botulinum. A 51-year-old man presented in myocardial infarction with subsequent dilated cardiomyopathy; November 1999 with an 8-day history of slowly progressive

864 d CID 2011:52 (1 April) d Yuan et al generalized muscle weakness, sore throat, difficulty swallowing, Therefore, their initial C. botulinum infections had likely re- and double vision. For 3 days, his eyelids had been drooping and solved. However, case-patients continued to inject BTH and he had had increasing difficulty breathing that he had attributed then returned with recurrent WB. Case B had a different BoNT to chronic obstructive pulmonary disease. His past medical type with each episode, which could have resulted only from history was significant for injection drug use, including skin reinfection. popping BTH. On the day of admission, the patient was noted to These recurrent cases suggest that exposure to BoNT by have disconjugate gaze, ptosis, wide-based gait, absent gag reflex, means of WB does not confer protective immunity. Exposure to decreased deep tendon reflexes, symmetrical descending weak- some BoNT via the food-borne route may be similarly non- ness beginning with his cranial nerves, and multiple puncture immunogenic, as suggested by Beller and Middaugh [10] in wounds on his thighs. During his evaluation in the emergency their description of repeat type E botulism in an Alaskan room, the patient had a respiratory arrest requiring emergent Eskimo. The small dose of BoNT causing disease may be less mechanical ventilation. He was treated with botulinum anti- than the amount needed to induce an effective immune re- toxin, antibiotics, and wound debridement. BoNT type A was sponse [16]. detected in his serum sample. After a prolonged hospitalization, Even after severe WB episodes that frequently included he recovered fully and was discharged home. intubation and prolonged hospital stays [7, 8], many Cal- Four years later, he presented as a 55-year-old man to the ifornia IDUs, including our case patients, continued to inject same emergency room. He had had 5 days of difficulty swal- BTH. If these near-death experiences do not change behavior

lowing and neck weakness and 3 days of difficulty walking. On among these IDUs and if severe disease from exposure to Downloaded from the day of admission, he was noted to have weakness in cranial BoNT does not confer immunity, recurrent WB will add to nerves III, VII, IX, XI, and XII; cellulitis and fluctuant abscesses the health care burden as the WB epidemic among IDUs on both forearms; and rapidly worsening respiratory distress continues. requiring intubation. He again recovered fully after treatment Immunizing IDUs against botulism may prevent WB and with botulinum antitoxin, antibiotics, and wound debridement. recurrent episodes. An investigational pentavalent (ABCDE) http://cid.oxfordjournals.org/ BoNT was not detected in his serum sample, but BoNT type B botulinum toxoid has been distributed by the CDC for labora- was isolated from a right arm abscess. tory workers at high risk and by the military for protection of troops [2]. However, the vaccine is not a US Food and Drug DISCUSSION Administration-licensed product and requires multiple doses and boosters [17]. Although vaccinating IDUs for botulism Botulism is a rare disease and recurrent botulism even more might reduce WB in this population, WB is still rare, and IDUs rare. There have been only 6 previous publications together are at higher risk for many other infectious agents, limiting the by guest on October 1, 2012 documenting 6 cases of recurrent food-borne botulism, and potential impact of these efforts on the overall health of IDUs. there has been only 1 publication documenting a case of re- Continued efforts to reduce injection drug use and to educate current wound botulism, which is included in this series [5, 10– current users on the infectious and other risks associated with 15]. In this report, we document 17 patients with recurrent illicit drug use are crucial to improve the health of the IDU wound botulism in California, 7 of whom had laboratory con- population. firmation for both initial and recurrent episodes. As the epi- These recurrent cases are likely to represent only a portion of demic of WB among IDUs continues, recurrent WB episodes are all recurrent cases of WB. The ability to diagnose all WB cases is taking place. As with previous reports of WB in California, these limited by the number of affected IDUs who present for medical recurrent WB cases occurred among IDUs reporting BTH use. attention, the clinical experience of emergency room physicians, In our patients, although the frequency of severe respiratory and the sensitivity of laboratory testing for botulism. BoNT may difficulty, descending paralysis, and mechanical ventilation circulate only intermittently, and the overall sensitivity of lab- tended to be less frequent with recurrent episodes, they still oratory tests of clinical specimens has been reported to be as low occurred in significant numbers of recurrent cases, highlighting as 33%–44% [18]. The mouse bioassay used by our state labo- the potential for recurrent WB to still result in severe disease. ratory was able to confirm only 25 (68%) of 37 total clinical WB These recurrent WB cases resulted from reinfection through episodes in these patients, the sensitivity of the test being the continued BTH injections. Among episodes of laboratory- same as in a recent CDPH study involving 73 WB patients confirmed WB infection, serum testing after antitoxin admin- during 2005–2007 [19]. Physicians should be aware of the lim- istration did not detect toxin in 95% of infections, when ited sensitivity of the mouse bioassay and base their diagnosis on adequate posttreatment serum specimens were collected. All clinical assessment when the test is negative. were treated with antibiotics to which C. botulinum should be Although the median time for presentation of recurrent cases susceptible, and their abscesses were incised and debrided. was 2 days from symptom onset, many still waited 3–7 days,

Recurrent Wound Botulism in Drug Users d CID 2011:52 (1 April) d 865 despite experience with symptoms from prior infection. IDUs 7. Sandrock CE, Murin S. Clinical predictors of respiratory failure and might be unaware that WB could recur. Given the severity of long-term outcome in black tar heroin-associated wound botulism. Chest 2001; 120:562–6. WB and the need for early clinical diagnosis, both clinicians and 8. Offerman SR, Schaefer M, Thundiyil JG, Cook MD, Holmes JF. IDUs should be aware of the potential for recurrence of wound Wound botulism in injection drug users: time to antitoxin correlates botulism with continued injection drug use to allow for timely with intensive care unit length of stay. West J Emerg Med 2009; 10:251–6. diagnosis and early administration of botulinum antitoxin and 9. Centers for Disease Control and Prevention. Botulism in the United appropriate supportive care. States, 1899-1996: handbook for epidemiologists, clinicians, and lab- oratory workers. US Department of Health, Education, and Welfare. Acknowledgments Atlanta, GA: CDC; 1998. 10. Beller M, Middaugh JP. Repeated type E botulism in an Alaskan Es- We thank California clinicians who have cared for injection drug users kimo. N Engl J Med 1990; 322:855. with wound botulism and reported them to their local health department, 11. Shneider VP. A recurrence of botulism. Klin Med (Mosk) 1988; and local and state public health workers who reported these cases to the 66:126–7. California Department of Public Health. 12. Raszeja B. Case of recurrent botulism. Pol Tyg Lek (Wars) 1954; Potential conflicts of interest. All authors: no conflicts. 9:1204–5. 13. Schwartz J, Champy M, Muller J. Recurrence of botulism six years after References the first infection; esophageal paralysis. Strasb Med 1953; 4:355–60. 14. Bilusic M, Pattathil J, Brescia M, McHugh W, Zaboski M, Schanzer B. 1. Sobel J. Botulism. Clin Infect Dis 2005; 41:1167–73. Recurrent bulbar paralysis caused by botulinum toxin type B. Clin 2. Arnon SS, Schechter R, Inglesby TV, et al. Botulinum toxin as a bi- Infect Dis 2008; 46:e72–4. ological weapon: medical and public health management. JAMA 2001; 15. Sobel J, Malavet M, John S. Outbreak of mild botulism type E illnesses 285:1059–70. with predominantly gastrointestinal symptoms from home-salted fish. Downloaded from 3. Davis JB, Mattman LH, Wiley M. Clostridium botulinum in a fatal Clin Infect Dis 2007; 45:e14–6. wound infection. JAMA 1951; 146:646–8. 16. Brett MM, Hallas G, Mpamugo O. Wound botulism in the UK and 4. Passaro DJ, Werner SB, McGee J, Mac Kenzie WR, Vugia DJ. Wound Ireland. J Med Microbiol 2004; 53(Pt 6);555–61. botulism associated with black tar heroin among injecting drug users. 17. Smith LA. Botulism and vaccines for its prevention. Vaccine 2009; JAMA 1998; 279:859–63. 27(Suppl 4):D33–9. 5. Werner SB, Passaro D, McGee J, Schechter R, Vugia DJ. Wound bot- 18. Woodruff BA, Griffin PM, McCroskey LM, et al. Clinical and labora- http://cid.oxfordjournals.org/ ulism in California, 1951-1998: recent epidemic in heroin injectors. tory comparison of botulism from toxin types A, B, and E in the United Clin Infect Dis 2000; 31:1018–24. States, 1975-1988. J Infect Dis 1992; 166:1281–6. 6. Centers for Disease Control and Prevention. National botulism 19. Wheeler C, Inami G, Mohle-Boetani J, Vugia D. Sensitivity of mouse surveillance http://www cdc gov/nationalsurveillance/botulism_ bioassay in clinical wound botulism. Clin Infect Dis 2009; 48: surveillance html. Accessed 15 July 2009. 1669–73. by guest on October 1, 2012

866 d CID 2011:52 (1 April) d Yuan et al MAJOR ARTICLE

Long-Term Outcomes of 217 Botulism Cases in the Republic of Georgia

Sami L. Gottlieb,1,2 Katrina Kretsinger,1,2 Nato Tarkhashvili,3 Neli Chakvetadze,3 Maia Chokheli,3 Marina Chubinidze,3 R. Michael Hoekstra,2 Ekaterina Jhorjholiani,3 Merab Mirtskhulava,3 Maia Moistsrapishvili,3 Merab Sikharulidze,3 Tamar Zardiashvili,3 Paata Imnadze,3 and Jeremy Sobel2 1Epidemic Intelligence Service, Office of Workforce and Career Development, and 2Division of Foodborne, Bacterial, and Mycotic Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia; and 3National Center for Disease Control, Tbilisi, Republic of Georgia

Background. The acute paralytic syndrome of botulism has been well-described; however, little is known about

its long-term consequences. Downloaded from Methods. We conducted a case-control study in the Republic of Georgia to evaluate the health of patients у6 months after they had experienced an episode of botulism. Case patients were selected on the basis of who had had a clinical diagnosis of foodborne botulism reported to the national surveillance system from 1998 through 2003. Three control subjects were randomly selected from each patient’s community. Results. We located 217 patients who had had botulism from surveillance records, with a median time since http://cid.oxfordjournals.org/ onset of illness of 4.3 years. The median age was 37 years, and 49% of the patients were female, similar to the control subjects. Most of the patients (68%) had acquired botulism from home-conserved vegetables (probably containing toxin type B), 15% had been hospitalized for 11 month, and 25% had required mechanical ventilation. Six patients died. Of the remaining 211 patients, 68% reported having worse health at the time of the interview than 6 years before the interview, compared with 17% of 656 control subjects (matched odds ratio, 17.6; 95% confidence interval, 10.9–28.4). Overall, 49% of the patients reported their current health as “fair” or “poor,” versus 25% of the control subjects (odds ratio, 5.0; 95% confidence interval, 3.2–7.6). Patients were more likely than control subjects to report fatigue, weakness, dizziness, dry mouth, and difficulty lifting objects (P ! .05 , for

each). Patients were more likely than control subjects to report difficulty breathing caused by moderate exertion by guest on October 1, 2012 (P ! .001 ) but not by minimal exertion or at rest. Patients were also more likely to report being limited in vigorous activities, walking 3 blocks, and climbing 3 flights of stairs (P ! .05 , for each). Finally, patients reported feeling significantly worse than control subjects for 6 of 11 questions regarding psychosocial well-being (P ! .05 , for each). In a multivariable model involving patients who had had botulism, mechanical ventilation during acute illness, older age, and region of residence independently predicted worse health. Conclusions. Several years after acute botulism, patients reported significant health, functional, and psychosocial limitations that are likely to be consequences of the illness.

Foodborne botulism is a severe paralytic illness caused lism is sparse. The few studies that exist involved very by ingestion of neurotoxins of Clostridium botulinum. small numbers of cases, and no studies have reported The acute illness is well-described and involves cranial outcomes 13 years after illness [2–5]. More information nerve dysfunction, symmetric descending flaccid pa- about the long-term outcomes of botulism is needed, ralysis, and potential respiratory failure that can last especially in regions where the incidence of disease is months, until nerve endings regenerate [1]. However, high. the literature regarding long-term outcomes of botu- The Republic of Georgia is a former Soviet republic with a population of 4.4 million people and has the highest nationally reported rate of botulism in the

Received 13 November 2006; accepted 29 March 2007; electronically published world. The annual incidence of botulism in Georgia is 5 June 2007. 0.9 cases per 100,000 persons, compared with 0.01 cases Reprints or correspondence: Dr. Sami Gottlieb, Centers for Disease Control and Prevention, 1600 Clifton Rd., MS E-02, Atlanta, GA 30333 ([email protected]). per 100,000 persons in the United States [6, 7]. Most Clinical Infectious Diseases 2007;45:174–80 cases of botulism in Georgia are attributed to con- This article is in the public domain, and no copyright is claimed. sumption of home-conserved vegetables containing 1058-4838/2007/4502-0005 DOI: 10.1086/518890 botulinum toxin type B; a smaller proportion of cases

174 • CID 2007:45 (15 July) • Gottlieb et al. have been linked to ingestion of smoked fish (more likely con- functional, and psychosocial status questions were modified taining toxin type E) [7]. We used the comprehensive records from the 36-item short form questionnaire [10, 11]. Inter- of Georgia’s national botulism surveillance system to locate and viewers also assessed the ability of study participants to stand interview individuals who had botulism during 1998–2003 to from a seated position without using their arms and to walk evaluate the long-term health, functional, and psychosocial 5 m without assistance. consequences of botulism in Georgia. Data analysis was performed with SAS software, version 9.1 (SAS Institute). Analyses for all outcome variables comparing METHODS patients with control subjects were matched for neighborhood or location of residence. Patients and control subjects in the We conducted a case-control study comparing interview re- same neighborhood or location were pooled to create larger sponses of patients who had had botulism with those of control strata on the basis of an assumption of exchangeability. We did subjects from the same areas of residence. A patient was defined not match patients and control subjects on age group or sex; as a Georgia resident who had received a clinical diagnosis of rather, we planned a priori to control for these variables in all botulism that was reported to the national surveillance system analyses. We also controlled for Georgian ethnicity and car during 1998–2003 and who had lived for at least 6 months ownership (as an additional measure of socioeconomic status) after the onset of illness. Although there is no formal case because of differing distributions of these variables between definition used by all clinicians in Georgia, a previous sur-

patients and control subjects. We asked participants about un- Downloaded from veillance system evaluation, with in-depth chart review of 1700 derlying medical diagnoses but did not control for these because Georgian botulism cases, found that reported cases had char- of very small numbers of prior diagnoses among both patients acteristic presentations [8]. Diagnostic testing for botulinum and control subjects. Thus, all analyses were matched on neigh- toxin, using standard mouse bioassay, became available in borhood or location and also adjusted for age group, sex, Geor- Georgia only in late 2001 [9]. A control subject was defined as gian ethnicity, and car ownership. P values, ORs, and 95% CIs http://cid.oxfordjournals.org/ an individual aged 16 years who resided in the same community were calculated using conditional logistic regression. as a patient but had no history of botulism. For patients who had had botulism, a multivariable logistic Patient and control subject interviews were conducted during regression model was created to evaluate predictors of worse 2004. Patient households were located using national surveil- health at the time of the interview, compared with 6 years before lance records kept by the National Center for Disease Control the interview. Variables in the model included region of resi- in Tbilisi, which included names and addresses. Three control dence, sex, age group, ethnicity, car ownership, food source of subjects were selected per patient using a “random walk”

botulism, acute symptoms, length of hospital stay, need for by guest on October 1, 2012 method. From each patient’s residence, interviewers walked a ventilation when the patient had botulism, and length of time randomly generated distance and direction to arrive at a start since botulism onset. address. The first control subject household was either the The study protocol and study instruments were reviewed by household at the start address or, if that household was un- the human subjects committees at the National Center for Dis- occupied or nonresidential, the household nearest to it. Selec- ease Control (Tbilisi, Georgia) and the Centers for Disease tion of the second and third control subject households pro- Control and Prevention (Atlanta, Georgia). ceeded using the same method. Within each household, a household member (aged 16 years) was randomly selected to RESULTS be the control subject, regardless of whether that person was home at the time of initial contact. There were 270 patients who had had botulism and had locating If the target subject in a household was unable to complete information in the 1998–2003 national surveillance records, the interview because of incapacity or absence from the house- which represented 146 events of botulism: 53 outbreaks of bot- hold after 3 attempts (for both patients and control subjects) ulism involving 177 people and 93 sporadic cases. Diagnostic or because of death (for patients only), the interview was con- testing was performed for 67 cases; 34 cases were laboratory ducted by proxy (another household member). In-person in- confirmed (33 involved toxin type B, and 1 involved toxin type terviews were conducted in Georgian or Russian after written E). Overall, 217 eligible patients (81%) were located, and all informed consent (from adults) or assent with parental per- agreed to participate in the study. A total of 657 eligible control mission (for minors) was obtained. subjects were identified, and all agreed to participate. Selected The standard study instrument enabled collection of de- characteristics of patients and control subjects are shown in mographic information from all participants and characteristics table 1. of acute botulism illness from patients. Questions about current Among the 217 patients who had had botulism who were symptoms were tailored to capture the types of neuromuscular included in the study, the median time between onset of bot- dysfunction associated with botulism, and additional health, ulism illness and the date of the interview was 4.3 years (range,

Long-Term Outcomes of Botulism • CID 2007:45 (15 July) • 175 Table 1. Selected characteristics of patients who had had botulism and control subjects.

Patients Control subjects Characteristic (n p 217) (n p 656) P Residence 1.0 Rural 92 (42) 275 (42) Urban 125 (58) 381 (58) Sex .10 Male 110 (51) 284 (43) Female 107 (49) 372 (57) Age .26 Median years (range) 37 (10–72) 37 (7–85) !20 years of age 35 (17) 166 (25) 20–35 years of age 66 (31) 154 (23) 36–50 years of age 56 (27) 160 (24) 150 years of age 54 (26) 176 (27) Education .47 Elementarya 0(0) 4(1) Downloaded from Secondaryb 70 (32) 180 (27) College 48 (22) 171 (26) University 99 (46) 301 (46) Ethnicity .01

Georgian 146 (67) 477 (73) http://cid.oxfordjournals.org/ Non-Georgianc 71 (33) 179 (27) Items owned by a person in the household TV 213 (98) 650 (99) .26 Refrigerator 200 (92) 623 (95) .10 Car 92 (42) 332 (51) .02 House phone 98 (45) 330 (50) .07 Cell phone 126 (58) 380 (58) .98 by guest on October 1, 2012 Baseline medical historyd Myocardial infarction 1 (0) 3 (0) .96 Stroke 0 (0) 2 (0) .98 Congestive heart failure 2 (1) 4 (1) .61 Chronic pulmonary disease 4 (2) 2 (0) .04 Diabetes 3 (1) 8 (1) .84 Osteoarthritis 5 (2) 13 (2) .72

NOTE. Data are no. (%) of persons, unless otherwise indicated. a Completed 1–9 years of school. b Completed 10–11 years of school. c Includes Russian, Azerbaijani, Armenian, and other. d Diagnoses received у6 years before the interview.

0.6–6.4 years); 211 patients (97%) were still alive, and 6 (3%) lowing (87% of patients), difficulty walking (87% of patients), had died 16 months after the onset of botulism. Most patients difficulty speaking (86% of patients), and constipation (84% (74%) had been hospitalized for a duration of 1 week to 1 of patients). Fewer patients reported having had difficulty month; 11% had been hospitalized for ! 1 week, and 15% had breathing caused by moderate exertion (63% of patients) or been hospitalized for 11 month. Overall, 25% of patients had minimal exertion (53% of patients) or difficulty breathing when required ventilatory assistance. The majority of patients re- at rest (41% of patients) during acute illness. ported having had the following symptoms during acute bot- Of the 214 patients for whom the food source of botulism ulism illness: weakness (97% of patients), fatigue (94% of pa- illness was known, 68% had become ill from home-conserved tients), dizziness (93% of patients), blurred vision (91% of vegetables, 30% from smoked fish, and 1% from other foods. patients), diplopia (88% of patients), dry mouth (88% of pa- Patients whose food source of botulism was home-conserved tients), difficulty lifting things (88% of patients), difficulty swal- vegetables (more likely involving toxin type B) were signifi-

176 • CID 2007:45 (15 July) • Gottlieb et al. Downloaded from

Figure 1. Reported general health status of 211 patients who had had botulism and 656 control subjects 6 years before the interview and at the time of interview. http://cid.oxfordjournals.org/ cantly more likely than those with other food sources of bot- ported their current health as “fair” or “poor,” compared with ulism to have been hospitalized longer (P p .003 ) and to report 25% of control subjects (OR, 5.0; 95% CI, 3.2–7.6). having had symptoms of diplopia, dry mouth, difficulty swal- Patients were significantly more likely than control subjects lowing, difficulty speaking, and difficulty walking (P ! .03 , for to report current symptoms of fatigue, weakness, dizziness, dry each) during acute illness. Patients who had become ill from mouth, and difficulty lifting things (P ! .05 , for each) (figure home-conserved vegetables were also more likely than those 2). The current symptoms most commonly reported by patients by guest on October 1, 2012 who had become ill from other food sources to have needed mechanical ventilation (28% of patients vs. 17% of patients), but this difference was not statistically significant (P p .08 ). Of the 211 patients who were alive at the time of the inter- view, 144 (68%) reported having worse health at the time of the interview than 6 years before the interview, compared with 112 (17%) of 656 control subjects (OR, 17.6; 95% CI, 10.9– 28.4). This comparison did not substantially change after ex- cluding 20 patients whose botulism illness had occurred 16 years before the interview (OR, 18.2; 95% CI, 11.0–30.2), nor after controlling for current symptoms as a proxy for potential symptom differences 6 years before the interview (OR, 15.8; 95% CI, 9.4–26.4). Patients were more likely than control sub- jects to report worse health whether the patient had required mechanical ventilation during botulism (OR, 33.4; 95% CI, Figure 2. Selected current symptoms reported by 211 patients who 11.3–98.9) or had not required mechanical ventilation (OR, had had botulism and 656 control subjects. All symptoms shown in the 16.3; 95% CI, 9.6–28.1). There was no interaction between age figure were reported significantly more frequently by patients than by group and reported worse health (P p .10 ). Figure 1 shows the control subjects (P ! .05 , for each). All comparisons were matched on reported general health status 6 years before the interview and neighborhood or location of residence and adjusted for age group, sex, Georgian ethnicity, and car ownership (as an additional measure of so- at the time of interview. Patients were more likely than control cioeconomic status). There were no significant differences in reports of subjects to remember having had better health status 6 years blurred vision, diplopia, constipation, difficulty swallowing, difficulty walk- before the interview (P p .004 ); however, patients reported ing, difficulty speaking, or difficulty breathing caused by minimal exertion worse current health (P ! .001 ). Overall, 49% of patients re- or when at rest.

Long-Term Outcomes of Botulism • CID 2007:45 (15 July) • 177 were generalized fatigue (48% of patients vs. 28% of control subjects; OR, 3.6; 95% CI, 2.4–5.5) and weakness (43% of patients vs. 25% of control subjects; OR, 3.2; 95% CI, 2.1–4.8). There were no significant differences between patients and con- trol subjects with regard to reporting blurred vision, diplopia, constipation, and difficulty swallowing, speaking, or walking. Patients were more likely than control subjects to report dif- ficulty breathing caused by moderate exertion, such as walking or lifting heavy items (18% of patients vs. 10% of control subjects;P p .001 ), but were not more likely to report difficulty breathing caused by minimal exertion, such as bathing or get- ting dressed, or difficulty breathing when at rest. Reported health limitations in various daily activities are shown in table 2. Compared with control subjects, patients were Figure 3. Selected emotional factors reported by 211 patients who more likely to have limitations in vigorous activities (e.g., run- had had botulism and 656 control subjects. The figure shows the per- ning or participating in sports), lifting or carrying groceries, centage of respondents who reported that for “a good bit,” “most,” or у “all” of the time in the past 4 weeks, they felt the emotional factors

climbing 3 flights of stairs, and walking 3 blocks. Patients Downloaded from shown in the figure. All emotional factors in the figure were reported at did not have significant limitations in moderate activities (e.g., significantly different frequencies by patients and control subjects (P ! moving a table), climbing 1 flight of stairs, bending or kneeling, .05, for each). All comparisons were matched on neighborhood or location walking 1 block, or bathing or dressing. of residence and adjusted for age group, sex, Georgian ethnicity, and car Figure 3 depicts how much of the time during the 4 weeks ownership (as an additional measure of socioeconomic status). There preceding the interview study participants reported experienc- were no significant differences in reports of feeling “down in the dumps,” http://cid.oxfordjournals.org/ ing various emotional states (figure 3). Patients were signifi- “downhearted and blue,” worn out, or tired or of having difficulty re- membering things. cantly less likely than control subjects to report that for “a good bit,” “most,” or “all” of the time during the previous 4 weeks, they felt full of pep (OR, 0.2; 95% CI, 0.1–0.3), calm and lems had interfered with normal social activities (33% of pa- peaceful (OR, 0.6; 95% CI, 0.4–0.9), energetic (OR, 0.3; 95% tients vs. 18% of control subjects;P ! .001 ). CI, 0.2–0.5), and happy (OR, 0.3; 95% CI, 0.2–0.5). Patients There were no differences between patients and control sub-

were significantly more likely than control subjects to report jects with regard to 2 limited objective measures of function. by guest on October 1, 2012 feeling nervous (OR, 2.8; 95% CI, 1.8–4.4) and unable to sleep Patients and control subjects had similar ability to stand from (OR, 1.8; 95% CI, 1.1–3.1). Overall, patients were more likely a seated position without using their arms (94% of patients vs. than control subjects to report that physical or emotional prob- 92% of control subjects;P p .6 ) and to walk 5 m without

Table 2. Limitations in activities reported by patients who had had botulism and control subjects.

Patients, % Control subjects, % Health limitation (n p 211) (n p 656) P OR (95% CI)a Moderate activitiesb 27 21 .10 1.4 (0.9–2.2) Vigorous activitiesc 64 38 !.001 5.0 (3.2–7.7) Lifting or carrying groceries 36 23 !.001 2.3 (1.6–3.5) Climbing 1 flight of stairs 24 19 .10 1.5 (0.9–2.3) Climbing 3 flights of stairs 49 28 !.001 3.7 (2.4–5.7) Walking 1 block 21 17 .19 1.4 (0.9–2.1) Walking 3 blocks 32 21 !.001 2.3 (1.5–3.6) Walking 12 kilometers 40 26 !.001 2.4 (1.6–3.7) Bending, kneeling 23 22 .76 1.1 (0.7–1.7) Bathing or dressing 13 15 .45 0.8 (0.5–1.4)

a All ORs were matched on neighborhood or location of residence and adjusted for age group, sex, Georgian ethnicity, and car ownership (as an additional measure of socioeconomic status). b Activities such as moving a table. c Activities such as running, lifting heavy objects, and participating in sports. “Yes” response includes responses of “don’t know,” because the participant was unable to perform vigorous activities.

178 • CID 2007:45 (15 July) • Gottlieb et al. assistance (99% of patients vs. 99% of control subjects; comes among smaller cohorts of patients after outbreaks of P p .6). botulism [2–5]. In an evaluation of 13 patients after an outbreak A multivariable model of data from patients who had had of type B botulism, most patients still had symptoms of dyspnea botulism revealed 3 independent predictors of having worse and easy fatigability and had measurable exercise limitation 2 overall health at the time of the interview, compared with 6 years after the outbreak [5]. One year after an outbreak of type years previous to the interview: requirement for ventilation A botulism, all but 2 of 28 patients reported some residual during botulism, age group, and region of residence. Patients symptoms, with fatigue (68%) being most commonly reported who had required mechanical ventilation during acute botulism [4]. In previous evaluations, generalized muscular and auto- illness were 15 times more likely to report worse health (OR, nomic symptoms of botulism appeared to persist longer than 5.3; 95% CI, 1.4–20.0). Compared with patients aged !20 years, ocular and bulbar symptoms [3, 5, 12], which is an observation patients in older age groups were more likely to report worse supported by our findings. Adverse psychosocial outcomes have health (age 20–35 years, OR, 2.9 [95% CI, 0.9–9.0]; age 36–50 been described among the survivors of 2 outbreaks of botulism years, OR, 3.5 [95% CI, 1.1–11.1]; age 150 years, OR, 4.9 [95% [2, 3]. In 1 of these studies, measures of psychosocial well- CI, 1.3–18.1]). Differences in health by region, with patients being did not approximate those from before illness until 3 from a few regions being more or less likely to report worse years after onset of botulism [2]. health than those from Tbilisi, were not specific to patients The Republic of Georgia has the highest nationally reported

who had had botulism; a similar pattern was also seen among rate of botulism in the world, with an annual incidence of 0.9 Downloaded from control subjects. Additional variables that were controlled for cases per 100,000 persons, and has a centralized national sur- but did not predict worse health among patients who had had veillance system that has gathered locating information for all botulism were sex, ethnicity, car ownership, food source of patients who had had botulism over several years [7]. These botulism (i.e., home-conserved vegetables vs. smoked fish or factors, along with the excellent participation rate of patients other foods, as a proxy for toxin type), length of hospital stay, and control subjects, provided us with a unique opportunity http://cid.oxfordjournals.org/ length of time since the patients had botulism, and acute symp- to locate and evaluate a large group of patients who had had toms during botulism. botulism, even several years after illness. The median time since illness in our study was 4.3 years, well beyond the evaluation DISCUSSION period in previous studies [2–5]. Another strength of this study We found that individuals who had had botulism in the Re- was inclusion of a large control group for comparison of health public of Georgia were substantially more likely than their status; matching on location of residence likely reduced con- by guest on October 1, 2012 neighbors to report a decline in general health since the time founding by unmeasurable socioeconomic and cultural factors before their illnesses. Even several years after acute illness, pa- that may have been linked to both perceived health status and tients who had had botulism were more likely than control risk of botulism. subjects to experience fatigue, generalized weakness, dizziness, This study also had several limitations. First, health outcome dry mouth, difficulty lifting things, and difficulty breathing survey instruments, such as the 36-item short form question- caused by moderate exertion. Ocular and bulbar signs and naire, have not been validated in Georgia, and cultural differ- symptoms, such as diplopia, difficulty swallowing, and difficulty ences in health perceptions may limit generalizability to other speaking, which had been prominent during acute botulism, populations. Second, no standard case definition for botulism had largely resolved. Patients who had had botulism were more has been used in Georgia, and misclassification was possible. likely to report limitations in several measures of functional However, a previous study found that clinical diagnoses of status, such as climbing 3 flights of stairs and walking у3 botulism in Georgia were generally valid [8], most patients in blocks; less vigorous activities were not affected. In addition, our study reported having had characteristic symptoms during patients who had had botulism reported worse overall psycho- acute illness, and almost all patients reported having consumed social status than did control subjects, with patients being sig- high-risk foods before illness onset. Thus, we believe misclas- nificantly less likely to report feeling happy, calm and peaceful, sification was unlikely. Third, the majority of botulism cases or full of pep during the 4 weeks preceding the interview. in Georgia are thought to be caused by ingestion of toxin type Mechanical ventilation during acute illness, older age, and re- B in home-conserved vegetables [7]. Although type-specific gion of residence were the only predictors of worse health testing for botulinum toxin has been limited in Georgia, toxin among patients who had had botulism. type A has rarely been found [7]. This is distinct from the These data represent, to our knowledge, the largest evaluation situation in the United States, where botulism from home- to date of the long-term health, functional, and psychosocial conserved vegetables is most commonly attributable to toxin consequences of botulism. Our findings are consistent with type A [6]. Differences in the severity of illness caused by dif- those of the few previously published studies describing out- ferent types of botulinum toxin have been well described; pa-

Long-Term Outcomes of Botulism • CID 2007:45 (15 July) • 179 tients with type A botulism tend to have more severe illness Acknowledgments

[6, 13]. Interestingly, the patients in our study who had acquired Financial support. The US Department of Health and Human Ser- botulism from home-conserved vegetables (likely involving vices’ Biotechnology Engagement Program. The funding source played no toxin type B) reported having had generally more severe acute role in designing the study; collecting, analyzing, or interpreting the data; illness, compared with those who had acquired botulism from writing the report; or deciding to submit it for publication. Potential conflicts of interest. All authors: no conflicts. fish and other foods (likely involving toxin type E); however, the food source of botulism did not predict worse long-term health on multivariable analysis. Finally, we were unable to References perform objective measurements of functional status, such as 1. Shapiro RL, Hatheway C, Swerdlow DL. Botulism in the United States: electromyelographic examination or pulmonary function test- a clinical and epidemiologic review. Ann Intern Med 1998; 129:221–8. ing. The subjective perception of worse health and the symp- 2. Cohen FL, Hardin SB, Nehring W, et al. Physical and psychosocial health status 3 years after catastrophic illness—botulism. Issues Ment toms reported by patients after having had botulism are still Health Nurs 1988; 9:387–98. important and substantial, regardless of objective findings; 3. Mann JM, Martin S, Hoffman R, Marrazzo S. Patient recovery from however, objective characterization of these symptoms in future type A botulism: morbidity assessment following a large outbreak. Am studies could help to better define the pathophysiologic char- J Public Health 1981; 71:266–9. 4. Schmidt-Nowara WW, Samet JM, Rosario PA. Early and late pulmo- acteristics of the long-term consequences of botulism and, thus, nary complications of botulism. Arch Intern Med 1983; 143:451–6.

possible prevention strategies. 5. Wilcox P, Andolfatto G, Fairbarn MS, Pardy RL. Long-term follow-up Downloaded from One additional consideration is whether the adverse out- of symptoms, pulmonary function, respiratory muscle strength, and comes we delineated are truly a consequence of botulism, rather exercise performance after botulism. Am Rev Respir Dis 1989; 139: 157–63. than a consequence of critical illness in general. Persistent mus- 6. Sobel J, Tucker N, Sulka A, McLaughlin J, Maslanka S. Foodborne cle weakness and fatigue have been well documented for 1–2 botulism in the United States, 1990–2000. Emerg Infect Dis 2004; 10: years after acute respiratory distress syndrome [14, 15], and 1606–11. http://cid.oxfordjournals.org/ such symptoms are thought to occur after a wide variety of 7. Varma JK, Katsitadze G, Moiscrafishvili M, et al. Foodborne botulism in the Republic of Georgia. Emerg Infect Dis 2004; 10:1601–5. other critical illnesses and are perhaps related to a critical-illness 8. Varma JK, Katsitadze G, Moiscrafishvili M, et al. Signs and symptoms polyneuropathy or myopathy [16, 17]. Emotional sequelae after predictive of death in patients with foodborne botulism—Republic of acute respiratory distress syndrome have also been described Georgia, 1980–2002. Clin Infect Dis 2004; 39:357–62. [18]. Two features of our study suggest that the adverse long- 9. Centers for Disease Control and Prevention. Botulism in the United States, 1899–1996. Handbook for epidemiologists, clinicians, and lab- term outcomes found among patients who had had botulism oratory workers. 1998. Available at: http://www.cdc.gov/ncidod/dbmd/ are specific to botulism. First, in the studies finding adverse diseaseinfo/files/botulism.pdf. Accessed 25 May 2007. by guest on October 1, 2012 outcomes after other critical illnesses, almost all patients had 10. McHorney CA, Ware JE Jr, Raczek AE. The MOS 36-item short-form undergone mechanical ventilation, often for extended periods. health survey (SF-36). II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 1993; 31: In our study, only 25% of the patients who had had botulism 247–63. had required mechanical ventilation, and even patients who 11. Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health sur- had not required mechanical ventilation were significantly more vey (SF-36). I. Conceptual framework and item selection. Med Care likely than control subjects to report having had worse general 1992; 30:473–83. 12. Cherington M. Botulism. Ten-year experience. Arch Neurol 1974; 30: health since a period before illness. Second, as in previous 432–7. studies of botulism outcomes [3, 5, 12], symptoms of cholin- 13. Woodruff BA, Griffin PM, McCroskey LM, et al. Clinical and laboratory ergic autonomic blockade that are prominent during botulism comparison of botulism from toxin types A, B, and E in the United (e.g., dry mouth) were present at long-term follow-up, which States, 1975–1988. J Infect Dis 1992; 166:1281–6. suggests a botulism-specific effect. 14. Davidson TA, Caldwell ES, Curtis JR, Hudson LD, Steinberg KP. Re- duced quality of life in survivors of acute respiratory distress syndrome In conclusion, several years after having botulism, patients compared with critically ill control patients. JAMA 1999; 281:354–60. report significant health, functional, and psychosocial limita- 15. Herridge MS, Cheung AM, Tansey CM, et al. One-year outcomes in tions that are likely to be consequences of the illness. Future survivors of the acute respiratory distress syndrome. N Engl J Med studies should include additional objective measures of health 2003; 348:683–93. 16. Combes A, Costa MA, Trouillet JL, et al. Morbidity, mortality, and and functional limitations and should describe the links be- quality-of-life outcomes of patients requiring у14 days of mechanical tween specific elements of the acute illness (e.g., toxin type, ventilation. Crit Care Med 2003; 31:1373–81. use and timing of botulinum antitoxin treatment, and length 17. Hudson LD, Lee CM. Neuromuscular sequelae of critical illness. N of ventilation) and the presence of adverse outcomes in patients Engl J Med 2003; 348:745–7. 18. Kapfhammer HP, Rothenhausler HB, Krauseneck T, Stoll C, Schelling who survive the illness. This information could facilitate de- G. Posttraumatic stress disorder and health-related quality of life in velopment of strategies to prevent long-term sequelae of long-term survivors of acute respiratory distress syndrome. Am J Psy- botulism. chiatry 2004; 161:45–52.

180 • CID 2007:45 (15 July) • Gottlieb et al. INVITED ARTICLE FOOD SAFETY Frederick J. Angulo, Section Editor

Botulism

Jeremy Sobel Foodborne and Diarrheal Diseases Branch, Centers for Disease Control and Prevention, Atlanta, Georgia Downloaded from Botulism is a rare disease with 4 naturally occurring syndromes: foodborne botulism is caused by ingestion of foods con- taminated with botulinum toxin, wound botulism is caused by Clostridium botulinum colonization of a wound and in situ toxin production, infant botulism is caused by intestinal colonization and toxin production, and adult intestinal toxemia botulism is an even rarer form of intestinal colonization and toxin production in adults. Inhalational botulism could result from aerosolization of botulinum toxin, and iatrogenic botulism can result from injection of toxin. All forms of botulism http://cid.oxfordjournals.org/ produce the same distinct clinical syndrome of symmetrical cranial nerve palsies followed by descending, symmetric flaccid paralysis of voluntary muscles, which may progress to respiratory compromise and death. The mainstays of therapy are meticulous intensive care (including mechanical ventilation, when necessary) and timely treatment with antitoxin.

Botulism is a rare, naturally occurring disease that can also be All of the toxins are large, single polypeptides of similar struc- caused by accidental or intentional exposure to botulinum tox- ture that exert their action on the cholinergic system at the

ins. All forms of botulism manifest essentially the same distinct presynaptic motor-neuron terminal by blocking acetylcholine at St Vincents University Hospital Library on September 9, 2012 clinical syndrome of symmetrical cranial nerve palsies that may transmission across the neuromuscular junction [9–12], caus- be followed by descending, symmetric flaccid paralysis of vol- ing neuromuscular blockade, resulting in flaccid paralysis. The untary muscles, which may progress to respiratory compromise toxins also affect the adrenergic system, but this apparently and death. happens without significant consequences. Rapid diagnosis, provision of intensive care, and adminis- Although the exact lethal dose has not been quantified, bot- tration of botulinum antitoxin are the cornerstones of treat- ulinum toxins are the most potent toxins known; extrapolations ment. Antitoxin is available exclusively from public health au- can be made from primate studies. Commonly cited estimated thorities, who immediately investigate potential sources to lethal doses for purified crystalline botulinum toxin type A for prevent additional illness [1]. Botulinum toxin is a category A a 70-kg man are 0.09–0.15 mg when introduced intravenously, biological agent, and it has been extensively weaponized by 0.80–0.90 mg when introduced inhalationally, and 70 mg when governmental military programs [2–6] and was deployed by a introduced orally [2]. However, considerably lower figures have terrorist group [7, 8]. been generated by other studies [13, 14] and from estimates based on cases in humans [15]. Because mouse bioassay is the THE ORGANISM AND ITS TOXINS standard method for detection and quantification of toxin, Clostridium botulinum is ubiquitously found in soil and aquatic toxin is usually quantified in terms of biological activity in sediments. C. botulinum produces 7 immunologically distinct terms of mouse intraperitoneal lethal dose (MIPLD50). toxins, which are designated by the letters A–G. Several related Under stress, C. botulinum forms a spore that survives stan- clostridial species (e.g., Clostridium baratii and Clostridium bu- dard cooking and food-processing measures. However, the con- tyricum) can produce some botulinum toxins as well. All pro- fluence of conditions permitting spore germination—anaerobic duce a clinically similar, highly recognizable syndrome. Human milieu, nonacidic pH, and low salt and sugar content—is rarely cases are caused mostly by toxin types A, B, E, and (rarely) F. achieved in food, which explains the small number of food- borne botulism cases. The technique of modern industrial can- ning (i.e., retort canning) was developed expressly for killing Received 26 April 2005; accepted 16 June 2005; electronically published 29 August 2005. Reprints or correspondence: Dr. Jeremy Sobel, Foodborne and Diarrheal Diseases Branch, C. botulinum spores [16, 17]. In contrast with the spore, bot- CDC, MS-A38, 1600 Clifton Rd., NE, Atlanta, GA 30333 ([email protected]). ulinum toxins are temperature sensitive, and all toxins are in- Clinical Infectious Diseases 2005;41:1167–73 This article is in the public domain, and no copyright is claimed. activated by heating to 85 C for 5 min [17]. 1058-4838/2005/4108-0014 Conditions in the normal human intestine are not conducive

FOOD SAFETY • CID 2005:41 (15 October) • 1167 to germination and vegetation of C. botulinum. C. botulinum except for the absence of gastrointestinal symptoms. Often, the spores are routinely ingested and excreted by humans without abscess is a minor lesion, at times no more than a small furuncle germination, toxin production, or any harm to the person or an abscess that resembles mild cellulitis. Whenever possible, through whom they pass. The exceptions are the small number abscesses should be cleaned and debrided, tissue material of infants who develop infant botulism and the handful of should be collected in anaerobic culture tubes for testing, and adults who develop adult toxemic infectious botulism. appropriate antimicrobial therapy should be provided. For the clinician, ascertainment of a history of injection drug BOTULISM SYNDROMES AND THEIR use—particularly use of black tar heroin—is crucial. When EPIDEMIOLOGY combined with a compatible presentation, a history of such drug use is highly predictive of wound botulism. Foodborne botulism. Foodborne botulism is caused by con- Infant botulism. Infant botulism results from absorption sumption of foods contaminated with botulinum toxin. C. bot- of toxin produced in situ by C. botulinum colonization of the ulinum grows and elaborates toxin only when the food presents Downloaded from intestines of certain infants aged !1 year [25]. This is the most conditions that include an anaerobic milieu, a pH of !4.5, low common form of botulism, with ∼80–100 cases reported an- salt and sugar content, and a temperature of 4C–121C [18]. nually in the United States [19]. Colonization is believed to Home-canned foods have long constituted a major source of occur because normal bowel florae that could compete with C. intoxication in the continental United States [19, 20]. Tradi-

botulinum have not been fully established. Studies have impli- http://cid.oxfordjournals.org/ tional Alaska Native dishes, which are fermented and consumed cated honey consumption as a risk factor for illness, but honey without cooking, also pose a substantial risk [21]. consumption probably accounts for only up to 20% of cases In the United States, during 1990–2000, the median number [26]. For unknown reasons, the highest incidence is found in of foodborne botulism cases per year was 23 (range, 17–43 the vicinity of Philadelphia, Pennsylvania [27]. The clinical pre- cases) Most cases are sporadic (i.e., they are not part of out- sentation resembles that of adult forms of disease, with com- breaks); outbreaks are typically small, involving 2 or 3 persons. Nevertheless, outbreaks caused by commercial or restaurant- mon symptoms that include inability to suck and swallow, weakened voice, ptosis, and floppy neck and that may progress prepared foods do occur [22]. at St Vincents University Hospital Library on September 9, 2012 Additional evidence for the diagnosis of foodborne botulism to generalized flaccidity and respiratory compromise [28]. Spe- can be provided by obtaining a 3–5-day food history from the cific therapy for infant botulism is a newly licensed human- patient; reported consumption of home-canned food substan- source antitoxin, which halves the median hospitalization pe- tially enhances the probability of foodborne botulism. Whether riod from 6 to 3 weeks. With appropriate intensive care, the close contacts may have shared foods should also be noted. It survival rate is nearly 100% with or without antitoxin therapy is important that the clinician solicit this information early, [25]. because if the patient’s case progresses to respiratory failure Adult intestinal toxemia botulism. Adult intestinal toxe- and mechanical ventilation despite administration of suppor- mia botulism has resulted from absorption of toxin produced tive and specific therapy, the patient’s ability to communicate in situ by rarely occurring intestinal colonization in a few adults further information will be compromised. in the United States by botulinum-toxin producing Clostridia. Wound botulism. Wound botulism is caused by contam- Typically, patients have some anatomical or functional bowel ination of a wound with C. botulinum spores from the envi- abnormality or are using antimicrobials, which may permit ronment and subsequent germination of these spores and pro- protection of normally fastidious Clostridia species from com- duction of toxin in the anaerobic milieu of an abscess. The petition with normal bowel flora [29–33]. Protracted symptoms condition was exceedingly rare until the early 1990s; since that and relapse in the face of antitoxin treatment due to ongoing time, the western United States has experienced a dramatic and intraluminal production of toxin may be observed. Diagnosis continuing increase in its incidence, almost exclusively among in a patient with sporadic botulism and no known food or injection drug users [23]. Almost all persons with injection wound source rests on demonstration of prolonged excretion drug–associated cases are users of so-called “black tar heroin,” of organisms and toxin in the stool. a specific preparation of heroin, and persons who partake in Inhalational botulism. Inhalational botulism is not a nat- “skin-popping” (i.e., injection of the black tar heroin into tis- urally occurring disease. The syndrome was described once sues, as opposed to veins) [24]. The typical patient is an adult among German laboratory workers in 1962, with symptoms in the fourth or fifth decade of life, who has a long history of resembling those of foodborne botulism [3]. Deliberate dis- use of injected black tar heroin, and who resides in the western semination of botulinum toxin by aerosol could produce an United States. The incubation period is hard to establish, be- outbreak of inhalational botulism [2]. cause most patients inject drugs several times daily. The clinical Iatrogenic botulism. Iatrogenic botulism is caused by in- syndrome is indistinguishable from that of foodborne botulism, jection of botulinum toxin for cosmetic or therapeutic pur-

1168 • CID 2005:41 (15 October) • FOOD SAFETY poses. Doses recommended for cosmetic treatment are too low Table 1. Protocols for clinicians evaluating suspected cases to cause systemic disease. Higher doses injected for treatment of botulism. of muscle movement disorders have caused anecdotal cases of Clinicians evaluating a suspected case of botulism should immedi- systemic botulism-like symptoms [34]. Injection of unlicensed, ately seek clinical consultation and should administer antitoxin, highly concentrated botulinum toxin caused severe botulism as follows in 4 patients who received it for cosmetic purposes (unpub- For suspected foodborne botulism, wound botulism, or botulism lished data). of unknown source, the state health department should be contacted via its 24-h emergency telephone number; if there is no response, the Centers for Disease Control and Preven- CLINICAL PRESENTATION tion’s Director’s Emergency Operations Center should be con- tacted (at 770-488-7100) The clinical syndrome of botulism is highly distinctive, con- For suspected infant botulism occurring in any state, the Califor- sisting of symmetrical cranial nerve palsies, followed by sym- nia Department of Health Services, Infant Botulism Treatment and Prevention Program should be contacted (at 510-540- metrical descending flaccid paralysis that may progress to re- Downloaded from 2646) spiratory arrest [19, 35]. For a sporadic (isolated) case, the differential diagnosis is not extensive, and the combination of neurological findings and specific laboratory tests provide nerve VII produces expressionless facies, and dysphagia is highly sensitive clinical diagnosis pending laboratory confir- caused by cranial nerve IX paralysis, which may present as mation [17]. A cluster of у2 cases with compatible symptoms regurgitation (at times nasal) of masticated food or beverages. http://cid.oxfordjournals.org/ is essentially pathognomonic, because the illnesses that most Dysarthria is prominent. Prominent autonomic symptoms in- resemble botulism do not produce outbreaks. The diagnosis in clude anhydrosis with severe dry mouth and throat (which may sporadic cases and even in small outbreaks is frequently missed, produce pronounced mucosal erythema and pain and which however, because botulism is a rare disease with which most has been mistaken for pharyngitis) and postural hypotension. clinicians are unfamiliar [36]. Every case of botulism is a public In some cases, pharyngeal collapse secondary to cranial nerve health emergency, and immediately upon suspecting the di- paralysis may compromise the airway and require intubation agnosis, the clinician should report the suspected case to the in the absence of respiratory muscle compromise. In foodborne at St Vincents University Hospital Library on September 9, 2012 24-h emergency telephone number of the state health depart- botulism, particularly that involving toxin types B and E, the ment [1]. The state health department will initiate an epide- gastrointestinal symptoms nausea and vomiting may precede miologic investigation to determine the source of infection and neurological symptoms. It is unknown whether these symptoms to identify exposed persons, and it will also put the physician are caused by direct action of botulinum toxin, other products in contact with the Centers for Disease Control and Preven- of C. botulinum, or some other contaminant of spoiled food. tion’s (CDC’s) 24-h botulism consultancy service (table 1). The These symptoms have never been reported in cases of wound on-call CDC consultant will review the case with the clinician botulism [23], nor have corresponding signs been observed in over the telephone and, if indicated, will help arrange for lab- primate experiments in which pure toxin was administered oratory confirmation by testing appropriate specimens at a intragastrically or intravenously [13, 37–41]. Therefore, these public health laboratory; the CDC will also arrange for ship- symptoms may be absent in patients with illness resulting from ment of antitoxin, which, in the United States, is available exposure to pure toxin. exclusively from the CDC [17]. The state health departments Cranial nerve palsies may be followed by flaccid, descending, of California and Alaska maintain their own botulism clinical completely symmetric paralysis of voluntary muscles, affecting consultation services. (in order) the muscles of the neck, shoulders, the proximal and Infant botulism causes sporadic illness only and has no po- then distal upper extremities, and the proximal followed by tential to cause an epidemic [25]. Clinical consultation and distal lower extremities. Paralysis of the diaphragm and acces- human-source antitoxin licensed for treatment of infant bot- sory breathing muscles may result in respiratory compromise ulism are available on a 24-h basis from the California De- or arrest. Eventual constipation is a nearly universal symptom; partment of Health Services’ Infant Botulism Treatment and vital signs are usually normal, with preservation of normal Prevention Program (telephone, 510-540-2646; table 1). range blood pressure possibly representing equilibrium between Cranial nerve palsies are invariably presenting symptoms of vagal blockade and extensive peripheral vasodilatation, both botulism (figure 1). The absence cranial nerve palsies or their caused by the toxin. In some cases, hypotension occurs. Deep onset after other true neurological symptoms have made their tendon reflexes progressively disappear. The ultimate extent of appearance rules out the disease. Extraoccular muscle paralysis paralysis in untreated patients and the rapidity of progression is due to paralysis of cranial nerves III, IV, and VI and manifests are variable. Symptoms may be limited to a few cranial nerves as blurry vision or frank diplopia and as an inability to ac- or may progress to complete paralysis of all voluntary muscles, commodate near vision. Ptosis is prominent. Paralysis of cranial and symptoms may progress over hours to days, with the rate

FOOD SAFETY • CID 2005:41 (15 October) • 1169 volume, resulting from paralysis of diaphragmatic and acces- sory respiratory muscles. Routine laboratory tests and radiological studies are not use- ful for diagnosis of botulism. Lumbar puncture reveals normal CSF values—in particular, the protein level is normal, in con- trast to Guillain-Barre´syndrome (GBS). Brain imaging studies may help rule out rare stroke syndromes that produce non- lateralizing symptoms. The Tensilon test helps for diagnosis of myasthenia gravis. In experienced hands, electromyography can be an exceedingly helpful adjunct to diagnosis. In affected mus- cles, findings consistent with neuromuscular junction blockage, normal axonal conduction, and potentiation with rapid repet- itive stimulation are indicative of botulism [42]. Downloaded from Differential diagnosis. In the setting of an outbreak, in which several persons would present with signs and symptoms of botulism, the diagnosis readily suggests itself. The situation

for the lone or sporadic patient with botulism (who may, in http://cid.oxfordjournals.org/ fact, be the herald case in a larger outbreak) is more precarious, because there is general unfamiliarity with the syndrome. How- ever, if the diagnosis is considered, the clinician should im- mediately call the state health department emergency number, and expert consultation at no charge will be provided within minutes over the phone [1]. The differential diagnosis includes GBS, myasthenia gravis, stroke syndromes, Eaton-Lambert syn- at St Vincents University Hospital Library on September 9, 2012 drome, and tick paralysis. Less likely conditions include tetro- Figure 1. Artist’s rendition of an adult patient with mild botulism. dotoxin and shellfish poisoning, antimicrobial-associated pa- Note ptosis and facial paralysis, manifesting as youthful, unlined, and ralysis, and a host of conditions due to even rarer poisons. A seemingly inexpressive facies. The patient is fully alert. (Illustrator, James K. Archer). thorough history and meticulous physical examination can ef- fectively eliminate most competing diagnoses. GBS, a rare au- toimmune, demyelinating polyneuropathy that follows an acute apparently proportional to the dose. Toxin binding is noncom- infection (with Campylobacter jejuni in one-third of cases), petitive and irreversible. Nerve terminals do regenerate slowly, presents in 95% of cases as an ascending paralysis and never allowing for eventual full recovery in 95% of cases in the United occurs in outbreaks [43]. Five percent of GBS cases present States. Paralysis resolves in weeks to months and often requires with the Miller Fisher variant; this is characterized by the triad extended outpatient rehabilitation therapy. of ophthalmoplegia, ataxia, and areflexia, which are easily mis- The sensory system is unaffected. Intellectual function is pre- taken for descending paralysis [44, 45]. A history of recent served throughout. Patients are able to respond appropriately diarrheal or respiratory infection may be elicited, and in the to questions. Once intubated, they can continue communicat- case of the former, stool culture may yield an enteric pathogen, ing by signal using fingers or toes, so long as paralysis has not especially C. jejuni, although this is of no use in the acute affected the digits. Tragically, in some instances, the patient’s setting. CSF protein levels are elevated in persons with all forms ptosis, expressionless facies, and altered voice have been inter- of GBS. However, these elevations may be delayed until several preted as signs of mental status changes due to alcohol intox- days after symptom onset, so a negative finding must be in- ication, drug overdose, encephalitis, or meningitis, and critical terpreted in light of the duration of symptoms, and lumbar components of the history, including potential sources of toxin, puncture may need to be repeated. In experienced hands, elec- were not sought by questioning. Because of skeletal muscle tromyography may demonstrate findings consistent with GBS paralysis, patients who experience respiratory distress do not but not botulism in adult patients; electromyography should present with signs of agitation, such as restlessness, tossing, not be performed in infants. A strongly positive Tensilon test gasping, thrashing, or flailing, and they may appear to be placid result, with or without presence of autoantibodies, confirms and detached, even as they near respiratory arrest. Death in the diagnosis of myasthenia gravis. Borderline positive Tensilon patients with untreated botulism results from airway obstruc- test results have been reported for patients with botulism. In tion from pharyngeal muscle paralysis and inadequate tidal most cerebrovascular accidents, a careful physical examination

1170 • CID 2005:41 (15 October) • FOOD SAFETY will usually uncover asymmetry of paralysis and upper motor public health officials. In general, ingested toxin is not de- neuron signs; brain imaging studies can help reveal the rare monstrable in serum 11 week after exposure. Toxin can be basilar stroke that produces symmetric bulbar palsies. Eaton- isolated from stool samples farther in the course of illness, and Lambert syndrome usually manifests as proximal limb weakness the toxin is stable in many food matrices for a considerably in a patient already debilitated by cancer. Tick paralysis is a longer period. rare condition of flaccid paralysis that closely resembles bot- ulism and that is caused by neurotoxins present in certain ticks. THERAPEUTICS The ticks should be sought, especially on the scalp (but also Supportive intensive care. During the first decades of the 20th on other parts of the body), and removal of the ticks reportedly century in the United States, the mortality rate among patients results in rapid reversal of paralysis [46]. with botulism was 60%–70%, even when equine antitoxin was Laboratory diagnosis and confirmation. Confirmation of administered in heroic doses. During the late 1940s and 1950s, botulism rests on demonstration of the toxin in specimens of

the mortality rate decreased precipitously, until it reached the Downloaded from patient serum, gastric secretions, or stool or in a food sample current rate of 3%–5% [20]. The difference resulted largely [17]. Demonstration of C. botulinum in a patient’s stool sample from the development of modern intensive care techniques— or in cultures of wound material is generally satisfactory for principally, mechanical ventilation. Persons with suspected bot- diagnosis of adult botulism syndromes and is considered de- ulism should be placed immediately in an intensive care setting,

finitive for diagnosis of infant botulism. The standard test is a http://cid.oxfordjournals.org/ with frequent monitoring of vital capacity and institution of bioassay involving intraperitoneal injection of toxin into mice mechanical ventilation if required. Paralysis due to botulism is and observation of the development of botulism-specific symp- protracted, lasting weeks to months, and meticulous intensive toms. This test can detect concentrations on the order of 1 care is required during this period of debilitation. MIPLD /mL. Toxin type is determined by injecting a panel of 50 Antitoxin therapy. The only specific treatment for botu- mice with mixtures of test sample and a monoclonal type- lism is administration of botulinum antitoxin. Antitoxin can specific antitoxin (e.g., anti-A or anti-B) and by observing arrest the progression of paralysis and decrease the duration of which antitoxin confers protection on the mice. The mouse paralysis and dependence on mechanical ventilation. Antitoxin at St Vincents University Hospital Library on September 9, 2012 bioassay is performed in a limited number of public health should be given early in the course of illness, ideally !24 h after laboratories. From the time that mice are injected, final results onset of symptoms [49, 50], because antitoxin neutralizes only may not be available for 24 h or even 48 h. Accordingly, all toxin molecules that are yet unbound to nerve endings. Animal clinical management decisions and initial public health inter- experiments confirm this relationship [13, 14, 38, 41]. Use of ventions are determined solely on the basis of clinical diagnosis. antitoxin is associated with adverse effects, including anaphy- Clinical samples for suspected cases of foodborne botulism laxis, other hypersensitivity reactions, and serum sickness. Ap- include serum (10 mL), vomitus or gastric secretions, stool proximately 9% of persons treated in previous decades, when у (ideally 25 mg), and suspect foods in original containers. For the recommended antitoxin dose was 2–4-fold higher than at suspected wound botulism cases, samples include 10 mL of present, experienced hypersensitivity reactions [51]. Of patients serum and anaerobic wound material; for infant botulism, stool who were treated with 1 vial of antitoxin in the past few years, is the preferred material [17]. The overall sensitivity of labo- !1% experienced serious reactions. Before administration of ratory tests of clinical specimens has been reported to be as antitoxin, skin testing should be performed to test for sensitivity low as 33%–44% [47, 48] but varies inversely with the time to serum or antitoxin. Administration of 1 vial of botulism elapsed between symptom onset and sample collection. Ac- antitoxin produces serum levels of toxin type–specific anti- cordingly, when syndromes other than infant botulism are sus- bodies capable of neutralizing serum toxin concentrations pected, serum (at least one 10-mL red-top serum tube, spun many-fold in excess of those reported for patients with botulism and separated) should be obtained immediately, and it should [52]. always be obtained before administration of antitoxin, because Given the high predictive value of objectively noted sym- antitoxin will neutralize all circulating toxin and render the test metric cranial nerve palsies in the setting of an outbreak, all meaningless. If possible, the earliest available serum sample previously healthy patients who have this symptom during an (such as the sample obtained for hospital admission blood- outbreak should received a diagnosis of probable botulism. work) should be salvaged and preserved for testing. Vomit sam- During outbreaks, for reasons that remain uncertain, not all ples should be collected, and if a nasogastric tube is placed, exposed persons manifest symptoms of botulism [53–55]. gastric secretions should be collected immediately. Because con- Toxin may be distributed unevenly in contaminated foods. stipation almost always occurs, stool samples should be col- However, host factors may play a role, because at least 1 person lected by means of a sterile water enema. These samples should has had demonstrable levels of toxin in circulation without be refrigerated but not frozen pending shipping directions from manifesting any clinical symptoms [53]. Persons who may have

FOOD SAFETY • CID 2005:41 (15 October) • 1171 been exposed by consumption of implicated foods should be 10. Hatheway CL. Clostridium botulinum and other clostridia that produce botulinum neurotoxin. In: Hauschild AHS, Dodds KL, eds. Clostridium observed closely, and if they develop symptoms compatible with botulinum: ecology and control in foods. New York: Marcel Dekker, botulism, they should be treated with antitoxin immediately. 1992:3–20. Isolation and infection control. Standard precautions 11. Hatheway CL. Clostridium botulinum. In: Bartlett JG, Blacklow NR, should be exercised when evaluating and treating patients. Bot- eds. Infectious diseases. Orlando: Saunders Company, 1991:1583–6. 12. Sugiyama H. Clostridium botulinum neurotoxin. Microbiol Rev ulinum toxin cannot be absorbed through intact skin. Toxin 1980; 44:419–48. can be absorbed through mucosal surfaces, the eye, and non- 13. Dack GM, Wood WL. Serum therapy of botulism in monkeys. J Infect intact skin. No case of person-to-person transmission of bot- Dis 1928; 42:209–12. 14. Ono T, Karashimada T, Iida H. Studies on the serum therapy of type ulinum has ever been described, including in patient care set- E botulism (part III). Jpn J Med Sci Biol 1970; 28:177–91. tings. Nevertheless, persons exposed to bodily fluids or stool 15. Morton HE. The toxicity of Clostridium botulinum type A toxin for from patients with botulism should be advised of the early signs various species of animals, including man. Philadelphia: The Institute for Cooperative Research, University of Pennsylvania, 1961. of botulism and should report for evaluation if these are noted. 16. Meyer K. The protective measures of the state of California against At present, there is no licensed vaccine for botulinum toxins. botulism. Am J Prev Med 1931; 5:261–93. Downloaded from 17. Centers for Disease Control and Prevention (CDC). Botulism in the CONCLUSIONS United States, 1899–1996, handbook for epidemiologists, clinicians and laboratory workers. Atlanta, GA: CDC, 1998. Regardless of the mode of exposure, botulinum toxins produce 18. International Commission on Microbiological Specifications for Foods. Clostridium botulinum. In: Micro-organisms in foods 5: characteristics a distinctive syndrome of cranial nerve palsies that may be of microbial pathogens. New York: Blackie Academic & Professional, http://cid.oxfordjournals.org/ followed by descending flaccid paralysis. Effective treatment 1996:68–111. depends on provision of intensive care and rapid administration 19. Shapiro R, Hatheway CL, Swerdlow DL. Botulism in the United States: of botulinum antitoxin based on clinical presentation, because a clinical and epidemiologic review. Ann Intern Med 1998; 129. 20. Gangarosa EA. Botulism in the US, 1899–1969. Am J Epidemiol laboratory diagnosis is time-consuming. Free expert clinical 1971; 93:93–101. consultation and antitoxin can be rapidly obtained from the 21. Wainright RB, Heyward WL, Middaugh JP, Hatheway CL, Harpster CDC by contacting public health authorities (table 1). Rapid AP, Bender TR. Food-borne botulism in Alaska, 1947–1985: epide- miology and clinical findings. J Infect Dis 1988; 157:1158–62. diagnostics and more-advanced specific therapy may accelerate 22. Sobel J, Tucker N, McLaughlin J, Maslanka S. Foodborne botulism in at St Vincents University Hospital Library on September 9, 2012 diagnosis and facilitate treatment. the United States, 1990–2000. Emerg Infect Dis 2004; 10:1606–11. 23. Werner SB, Passaro DJ, McGee J, Schechter R, Vugia DJ. Wound bot- Acknowledgments ulism in California, 1951–1998: a recent epidemic in heroin injectors. Clin Infect Dis 2000; 31:1018–24. Potential conflicts of interest. J.S.: no conflicts. 24. Passaro DJ, Werner SB, McGee J, MacKenzie W, Vugia D. Wound botulism associated with black tar heroin among injecting drug users. References JAMA 1998; 279:859–63. 25. Arnon S. Infant botulism. In: Feigen RD, Cherry JD, eds. Textbook of 1. Shapiro R, Hatheway C, Becher J, Swerdlow DL. Botulism surveillance pediatric infectious diseases, 4th ed. Philadelphia: WB Saunders, 1998: and emergency response: a public health strategy for a global challenge. 1570–7. JAMA 1997; 278:433–5. 26. Spika JS, Shafer N, Hargrett-Bean N. Risk factors for infant botulism 2. Arnon SS, Schechter R, Inglesby TV, et al. Botulism toxin as a biological in the United States. Am J Dis Child 1989; 143:828–32. weapon: medical and public health management. JAMA 2001; 285: 27. Long S. Epidemiologic study of infant botulism in Pennsylvania: report 1059–70. of the infant botulism study group. Pediatrics 1985; 75:928–34. 3. Middlebrook JL, Franz DR. Botulinum toxins. In: Sidell FR, Takafuji 28. Arnon SS, Midura TF, Clay SA, Wood RM, Chin J. Infant botulism: TE, Franz DR, eds. Medical aspects of chemical and biological warfare. epidemiological, clinical, and laboratory aspects. JAMA 1977; 237: Washington, DC: Borden Institute, Walter Reed Army Medical Center, 1946–51. 1997:643–54. 29. Chia JK, Clark JB, Ryan CA, Pollack M. Botulism in an adult associated 4. Ekeus R. Report of the Secretary General on the status of the imple- with food-borne intestinal infection with Clostridium botulinum.N mentation of the Special Commission’s plan for the ongoing moni- Engl J Med 1986; 315:239–54. toring and verification of Iraq’s compliance with relevant parts of Sector 30. Arnon S. Botulism as an intestinal toxemia. In: Blaser MJ, Smith PD, C of Security Council Resolution 687. New York: United Nations Spe- Ravdin JI, Greenberg HB, Guerrant RL, ed. Infections of the gastro- cial Commission, 1991. intestinal tract. New York: Raven Press, 1995:257–71. 5. Cordesman A. Weapons of mass destruction in the Gulf and greater 31. Fenicia L, Franciosa G, Pourshaban M, Aureli P. Intestinal toxemia Middle East: force trends, strategy, tactics and damage effects. Wash- botulism in two young people, caused by Clostridium butyricum type ington DC: Center for Strategic and International Studies, 1998:18–52. E. Clin Infect Dis 1999; 29:1381–7. 6. Bermudez J. The armed forces of North Korea. London: IB Tauris, 32. Griffin PM, Hatheway CL, Rosenbaum RB, Sokolow R. Endogenous 2001. antibody production to botulinum toxin in an adult with intestinal 7. Tucker J. Toxic terror: assessing the terrorist use of chemical and bi- colonization botulism and underlying Crohn’s disease. J Infect Dis ological weapons. Cambridge: MIT Press, 2000. 1997; 175:633–7. 8. WuDunn S, Miller J, Broad WJ. How Japan germ terror alerted world. 33. McCroskey L, Hatheway CL, Woodruff BA, Greenberg JA, Jurgenson New York Times 26 May 1998:A1, A10 P. Type F botulism due to neurotoxigenic Clostridium baratii from an 9. Hauschild. Clostridium botulinum. In: Doyle M, ed. Foodborne bac- unknown source in an adult. J Clin Microbiol 1991; 29:2618–20. terial pathogens. New York: Marcel Dekker, 1989:112–89. 34. Bakheit AM, Ward CD, McLellan DL. Generalized botulism-like syn-

1172 • CID 2005:41 (15 October) • FOOD SAFETY drome after intramuscular injections of botulinum toxin type A: a 44. Asbury A. New concepts of Guillian-Barre syndrome. J Child Neurol report of two cases [letter]. J Neurol Neurosurg Psychiatry 1997; 62: 2000; 15:183–91. 198. 45. Willison HJ, O’Hanlon GM. The immunopathogenesis of Miller Fisher 35. Hughes JM, Blumenthal JR, Merson MH, Lombard GL, Dowell VR, syndrome. J Neuroimmunol 1999; 100:3–12. Gangarosa EJ. Clinical features of types A and B food-borne botulism. 46. Felz MW, Smith CD, Swift TR. A six-year-old girl with tick paralysis. Ann Intern Med 1981; 95:442–5. N Engl J Med 2000; 342:90–4. 36. St. Louis ME, Peck SHS, Bowering D, et al. Botulism from chopped 47. Woodruff BA, Griffin PM, McCroskey LM, et al. Clinical and laboratory garlic: delayed recognition of a major outbreak. Ann Intern Med comparison of botulism from toxin types A, B, and E in the United 1988; 108:363–8. States, 1975–1988. J Infect Dis 1992; 166:1281–6. 37. Franz DR, Pitt LM, Clayton MA, Hanes MA, Rose KJ. Efficacy of 48. Dowell VR Jr, McCroskey LM, Hatheway CL, Lombard GL, Hughes prophylactic and therapeutic administration of antitoxin for inhalation JM, Merson MH. Coproexamination for botulinal toxin and clostrid- botulism. In: DasGupta BR, ed. Botulinum and tetanus neurotoxins. ium botulinum: a new procedure for laboratory diagnosis of botulism. New York: Plenum Press, 1993:473–6. JAMA 1977; 238:1829–32. 38. Gunnison JB, Meyer KF. Susceptibility of monkeys, goats and small 49. Tacket CO, Shandera WX, Mann JM, Hargrett NT, Blake PA. Equine animals to oral administration of botulinum toxin types B, C, and D. antitoxin use and other factors that predict outcome in type A food- J Infect Dis 1930; 46:335–40.

borne botulism. Am J Med 1984; 76:794–8. Downloaded from 39. Herrero BA, Ecklund AE, Streett CS, Ford DF, King JK. Experimental 50. Chang GY, Ganguly G. Early antitoxin treatment in wound botulism botulism in monkeys—a clinical pathological study. Exp Mol Pathol results in better outcome. Eur Neurol 2003; 49:151–3. 1967; 6:84–95. 51. Black RE, Gunn RA. Hypersensitivity reactions associated with botu- 40. Stookey JL, Street CS, Ford DF. Preliminary studies on the disappear- linal antitoxin. Am J Med 1980; 69:567–70. ance of botulinum toxin from the circulating blood of rhesus monkeys. 52. Hatheway CL, Snyder JD, Seals JE, Edell TA, Jewis GE Jr. Antitoxin Edgewood Arsenal, MD: US Army Edgewood Arsenal Chemical Re- search and Development Laboratories, 1965. levels in botulism patients treated with trivalent equine botulism an- http://cid.oxfordjournals.org/ 41. Oberst FW, Crook JW, Cresthull P, House MJ. Evaluation of botulism titoxin to toxin types A, B, and E. J Infect Dis 1984; 150:407–12. antitoxin, supportive therapy, and artificial respiration in monkeys with 53. Kalluri P, Crowe C, Reller M, et al. An outbreak of botulism from experimental botulism. Clin Pharmacol Ther 1968; 9:209–14. food sold at a salvage store in Texas. Clin Infect Dis 2003; 37:1490–5. 42. Cherington M. Electrophysiologic methods as an aid in diagnosis of 54. Townes J, Cieslak PR, Hatheway CL, et al. An outbreak of type A botulism: a review. Muscle Nerve 1982; 5:S28-9. botulism associated with a commercial cheese sauce. Ann Intern Med 43. Pascuzzi RM, Fleck JD. Acute peripheral neuropathy in adults: Guil- 1996; 125:558–63. lain-Barre syndrome and related disorders. Neurol Clin 1997; 15: 55. Angulo FJ, Getz J, Taylor JP, et al. A large outbreak of botulism: the 529–47. hazardous baked potato. J Infect Dis 1998; 178:172–7. at St Vincents University Hospital Library on September 9, 2012

FOOD SAFETY • CID 2005:41 (15 October) • 1173 ORIGINAL PAPER

The prevalence- of beta-haemolytic streptococci in throat specimens from healthy children and adults Implications for the clinical value of throat cultures Ronny K Gunnarsson’, Stig E Holm2 and Margareta Soderstrom‘ ’ Department of Primary Health Care, Goteborg University, Department of Clinical Bacteriology, UmeA University, Sweden.

Received September 1995. Accepted December 1996.

Scand J Prim Health Care 1997;15:149-55. ISSN 0281-3432 summer season and large during the mid-winter season. Conclusion - Prevalence of BHS varies with age and season in Objective - To examine the influence of age and season of the healthy individuals and patients with throat pain. During the year on the carrier rate of beta-haemolytic streptococci (BHS) summer, it is much more difficult to interpret the result of a in healthy individuals and patients with throat pain. throat culture in individuals aged el6 years. Design - The prevalence of BHS in throat specimens from healthy individuals was compared with that from patients with Key words: streptococcal infections microbiology, streptococcal throat pain of the same age in a defined geographical area, infections diagnosis, Streptococcus pyogenes isolation and pu- collected during the same mid-winter and late summer periods. rification, child day care centres, throat culture. Results -The prevalence of BHS in healthy individuals was low before the age of 3 years (1.9-7.1%) and in adults 216 years Ronny K Gunnarsson, MD, Research and Development Unit, (2.4-3.7%) and highest in the age group 3-15years (5.0-21.2%). Primary Health Care, Boris Hospital - Adm 11, 9501 82 Borh, The difference in prevalence of BHS between healthy individ- Sweden. uals and patients with throat pain was small during the late

For personal use only. It is well known that it is difficult to use only clinical riod (15/7-15/9) in 1991. Throat specimens were col- signs to distinguish between a tonsillopharyngitis lected from patients and non infectious individuals liv- caused by beta-haemolytic streptococci (BHS) and a ing in Elfsborg County in Sweden. The specimens were non-BHS-caused tonsillopharyngitis (1,2). Thus throat collected in routine medical care by the ordinary staff, cultures or rapid tests that indicate presence of group A who were instructed to take the throat specimens ac- beta-haemolytic streptococci (GABHS) are useful. cording to the& routine method; this was to rub a swab However, not all patients with a positive microbiologi- onto the tonsils and the soft palate. The study was cal test for GABHS and symptoms from the throat have approved by the ethics committee, Goteborg University. a GABHS-caused tonsillopharyngitis. Some of these pa- tients may have a virus infection and are only carriers of Non infectious individuals

Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12 GABHS (1,3-5). and may not benefit from treatment Throat specimens were taken from 400 pre-school chil- with antibiotics at that time (4,6). It would be of interest dren, aged 16 years, visiting 12 different children’s to identify the characteristics of those patients. One welfare clinics and one paediatric outpatient clinic. The major factor that influences the clinical value of micro- children were grouped in four strata according to type of biological testing is the prevalence of asymptomatic day care, information about which was given by the carriers of GABHS. parents on direct request. The four groups were charac- The present study aimed to compare the prevalence terized as; children in full-time day care 230 hourst of BHS in healthy individuals and patients with a sore week (DCC+), in part-time day care <30 hourstweek throat by collecting throat specimens in routine medical (DCC-), in family day care (FDC), and in home care care. Furthermore, we wanted to measure the influence (HC). Throat specimens were taken from 381 school- of age, season of the year, and type of child day care on children aged 7-15 years at 11 different schools. Throat the prevalence of BHS in healthy persons and patients specimens from 516 adults aged 216 years were taken at with a sore throat. 21 different primary health care centres when they, as patients, visited for a non-infectious cause. All individ- METHODS uals lacked signs of respiratory tract infections, had not The design was a prospective study that ran during the consumed antibiotics during the previous four weeks, mid-winter period (14/1-17/2) and the late summer pe- and did not have known diabetes mellitus or any known

Scand J Prim Health Care 1997; 15 150 R.K. Gunnarsson et al.

Table I. Prevalence of BHS' in healthy individuals.

~ ~~ ~ Growth of beta-haemolytic streptococci Age Sea- Total All groups Group A Group C Group G (W son3 n n % n % n % n % 0-2 years 1.3 (0.8) s 112 8 7.1 6 5.4 0 - 2 1.8 3-6 years 4.8 (1.0) S 99 21 21.2 17 17.2 1 1.o 3 3.0 7-9 years 9.1 (0.6) S 42 4 9.5 1 2.4 1 2.4 2 4.8 10-15 years 12.7 (1.3) S 149 24 16.1 19 12.8 3 2.0 2 1.3 116 years 50.1 (16.3) S 188 7 3.7 0 - 3 1.6 4 2.1 0-2 years 1.0 (0.8) w 104 2 1.9 0 - 0 - 2 1.9 3-6 years 4.9 (1.1) W 85 11 12.9 6 7.1 1 1.2 4 4.7 7-9 years 8.8 (0.9) W 20 1 5.0 0 - 1 5.0 0 - 10-15 years 13.0 (1.5) W 170 16 9.4 11 6.4 0 - 5 2.9 116 years 47.4 (17.6) W 328 8 2.4 1 0.3 2 0.6 5 ' 1.5 Beta-haemolytic streptococci 2Mean age in the age group, years (SD) 3S= Summer (15 July - 15 Sept.) W = Winter (14 Jan. - 17. Feb.).

immunodeficiency disorder. The distribution of speci- nostic laboratory in modified Stuart medium. The swab mens is given in Table I. was gently rolled over 1/3 of the surface of a blood agar plate, followed by streaking with a sterile loop over the Patients rest of the surface. The plate was a double layered During the same periods, the results were registered selective Columbia blood agar plate with Polymyxin B, from all consecutive throat swabs that were sent to the Neomycin and Nalidixic acid. The plate was incubated micribiological laboratory in Bods with a referral stat- overnight at 37°C in 5% COZ. The swab was finally ing that the patient had throat pain. The specimens came inoculated in a tube containing broth with serum for S. For personal use only. from the same geographical area as the specimens from detection of streptolysin This tube was incubated the healthy individuals. The distribution of specimens, overnight at 30°C. If the plate showed growth of BHS 11435, is given in Table II. and the tube showed haemolysis, the sample was con- sidered to contain BHS. If both were without signs of Throat culture beta-haemolytic activity the agar plate was incubated All throat specimens were transported to the same diag- for a further 24 hours. If there was still no sign of

Table XI. Prevalence of BHS' in patients with throat pain'. Growth of beta-haemolytic streptococci Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12 Age Sea- Total All groups Group A Group C Group G (SW3 son4 n n YO n % n % n . Yo 0-2 years 1.7 (0.8) S 17 2 11.8 2 11.8 0 0 - 3-6 years 5.0 (1.1) S 21 7 33.3 7 33.3 0 0 - 7-9 years 8.5 (0.9) S 7 3 42.9 3 42.9 0 0 - 10-15 years 12.7 (1.5) S 8 1 12.5 1 12.5 0 - 0 - 116 years 33.9 (14.3) S 121 19 15.7 9 7.4 6 5.0 4 3.3

0-2 years 1.8 (0.9) W 23 6' 26.1 5 21.7 0 0 - 3-6 years 4.9 (1.2) W 25 14' 56.0 12 48.0 0 0 - 7-9 years 8.3 (0.7) W 21 6 28.6 6 28.6 0 0 - 10-15 years 13.3 (1.7) W 24 9 37.5 7 29.2 1 1 4.2 216 years 34.8 (13.6) W 168 36 21.4 32 19.0 0 3 1.8 ' Beta-haemolytic streptococci Consecutive throat specimens sent to the microbiological laboratory from routine medical care, and the referral states throat pain 3Mean age in the age group, years (SD) 4S = Summer (15 July - 15 Sept.) W = Winter (14 Jan. - 17 Feb.) 'Information about group missing in three specimens.

Scand J Prim Health Care 1997; I5 The prevalence of beta-huemolytic streptococci 15 1

a4 T RESULTS The prevalence of BHS was higher in patients with throat pain (Table 11) than in healthy individuals (Table I). The most common finding was GABHS followed by BHS of serogroup G. BHS belonging to groups other than A. C or G were not found.

Season of the year The difference in prevalence of BHS in healthy individ- uals and patients with throat pain was smaller during the late summer than during the mid-winter (Fig. 1). During Fig. 1. Seasonal variation in the prevalence of BHS. the late summer the prevalence of BHS was not statisti- cally significantly higher in patients with throat pain than in healthy individuals, except for adults 216 years (p50 colonies. Each culture showing BHS 3-15 years (5.0-21.2%). The correlation between age underwent latex agglutination (StreptexB) to determine and the proportion of BHS found in throat specimens to which serogroup (Lancefield group) it belonged. was similar among patients with throat pain (Table 11). For personal use only. Statistical analysis Day care To compare the groups, the chi-square test with Yates' Information concerning type of day care was obtained from correction was used; when the numbers were small 377 of the 400 (94%) non-infectious pre-school children Fisher's exact test was used. (Table m).The mean age in the groups DCCi, DCC-, and FDC was comparable and varied within 3.5-4.8 years. The children in the HC-group were younger, mean age of 1.9

Table 111. Prevalence of BHS' in healthy pre-school children (16 years) related to type and length of stay in day care. Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12 Growth of beta-haemolytic streptococci

~ Total All groups Group A Group C Group G Day care Season2 n n 95 n % n % n % DCC+' S 21 I 33 6 29 0 0 1 5 DCC-4 S 28 5 18 3 11 1 4 1 4 FDc5 S 39 5 13 4 10 0 0 1 3 HC6 S 111 9 8 7 6 0 0 2 2

DCC+ W 12 3 25 3 25 0 0 0 0 DCC- W 16 1 6 1 6 0 0 0 0 FDC W 38 3 8 1 3 0 0 2 5 HC W 112 6 5 1 1 I 1 4 4 ' Beta-haemolytic streptococci *S = Summer (197 - 15/9). W = Winter (14/1 -1712) 'Presence at day care centres 230 hourdweek 4Presence at day care centres <30 hourdweek 'Family day care 6Home care

Scand I Prim Healrh Care 1997; 15 152 R.K. Gunnarsson et al.

Table IV. Semiquantification of GABHS' in throat specimens. Quantification of GABHS Total Growth of Season Age n GABHS2 Abundant' Moderate4 Sparses Summer 0-6 years 211 23 7 11 3 Healthy 7-15 years 191 20 8 7 5 216 years 188 0

Winter 0-6 years 189 6 5 1 0 Healthy 7-15 years 190 11 4 6 1 216 years 328 1 1 0 0

Summer 0-6 years 38 9 4 3 2 Patients6 7-15 years 15 4 3 1 0 216 years 121 9 9 0 0

Winter 0-6 years 48 17 15 2 0 Patients 7-15 years 43 13 11 2 0 216 years 168 32 29 2 1 Beta-haemolytic stxeptococci of Lancefield group A 21f the number in this column exceeded the sum of abundant, moderate and sparse growth, then information about quantification was missing in some cultures 150 colonies 11-50 colonies 1-10 colonies Consecutive throat specimens sent to the microbiologocal laboratory from routine medical care, and the referral states throat pain.

years, hence this group is not comparable with the other during similar circumstances. By using the routine pro- For personal use only. groups. No differences in prevalence of BHS were ob- ceduF, a study may reflect the normal clinical situation. served between children in DCC+, DCC-, and FDC. Of the previous investigations presenting carrier rates of BHS in healthy individuals, we found only one (3) that Quantification of growth clearly stated that the specimens were recovered in rou- Semiquantification of GABHS (Table IV) showed that, tine medical care by the ordinary staff. if GABHS was present, abundant growth was present in Our culture technique was not optimal. Using 85% of the patients with throat pain and in 41% of the anaerobic atmosphere and non-selective culture media healthy individuals. This difference was present in mid- might yield higher numbers of BHS. However, using a winter (p4.0037) and late summer (p=O.013). better technique would change the absolute number of positive cultures but not the correlation between the

Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12 patient groups or the seasons. DISCUSSION We used throat specimens that consecutively arrived The findings that the difference in prevalence of BHS in at the microbiological laboratory in Bods as represen- healthy individuals and patients with throat pain was ting patients. We only used throat specimens for which small during the late summer and large during mid- the physicians clearly stated that the patient had throat winter, and that there was a high prevalence of BHS in pain, thus eliminating specimens collected to &ace car- healthy individuals aged 3-15 years, may have implica- riers. tions for the clinical value of a throat culture. Seasonal variation Methodological aspects It is well known that streptococcal throat infection, in a Since all the cultures in our study were collected in temperate climate, is a disease of the winter and early routine medical care, the prevalence of BHS could be spring (1,7,8). Our intention was to compare patients expected to be slightly lower than in other studies in with a sore throat and healthy individuals during periods which the cultures were collected by specially trained when we could expect maximal and minimal prevalence personnel. However, our purpose was to compare the of BHS among patients. Therefore the periods mid- prevalence of BHS in healthy individuals and patients winter and late summer were chosen. We found that the with throat pain in whom the specimens were collected prevalence of BHS in the throat among healthy individ-

Scad J Prim Health Cure 1997; 15 The prevalence of beta-haemolytic streptococci 153

uals in most’age groups was higher in the summer than care centres. It is reasonable to assume that day care in the winter, while an opposite seasonal variation was involving crowding of children increases the carrier rate found in patients with a sore throat. in healthy pre-school children, but this has not yet been The higher prevalence of BHS in healthy individuals shown. in the late summer season could reflect a changed strep- tococcal epidemiology. Thus, if a new serotype ap- Quantification of growth peared that had not been circulating for some time, it When GABHS was present, we found abundant growth could gain a higher prevalence than the strains that had of GABHS in 85% of the patients with throat pain and been present for months. If this was the case then we in 41% of the healthy individuals. In patients with an should have seen a similar increase in the prevalence of infection caused by BHS, the growth is usually abun- BHS in patients, but in fact we saw a decrease in the dant. Thus, some authors state that quantification is prevalence of BHS from the mid-winter to the late sum- useful in order to distinguish between carrier state and mer season in patients. Seasonal variations in preva- infection caused by BHS (1.15). However, as shown in lence of BHS are normally not correlated to the intro- the present study, healthy carriers often have abundant duction of a new serotype (9). Thus, in the present study growth of BHS in throat specimens (5,616). Further- the appearance of a new serotype was not explicitly more, if the growth is sparse it does not exclude symp- tested. tom-giving infection (17,18). One should also remember Since we studied only two limited periods during one that in routine medical care the sampling technique is year, we cannot conclude that the carrier rates are not controlled and this will further diminish the value of higher during the summer than during the winter season. quantification in routine diagnosis. Our conclusion, that Other seasonal factors need to be considered and quantification alone is unreliable in the single case to further evaluated. One that might reduce the prevalence distinguish carrier state from disease, is in accordance of BHS among healthy children is lower attendance at with other studies (5,16,19). The semiquantitative day care centres (10) or schools during holidays. One method used might not fully meet scientific stringency, that might increase the prevalence of BHS among but it is representative of techniques usually performed healthy children is the effect of crowding on return from in routine diagnosis. holiday. Since the carrier rate among healthy individ- uals and patients with throat pain showed opposite sea- Diflerent serotypes of BHS

For personal use only. sonal fluctuations, one may speculate that BHS are Our finding that most BHS-caused infections in the either more virulent during the winter season, or that throat were associated with GABHS is in accordance individuals are more vulnerable to a streptococcal infec- with other studies (15). BHS of group C and G may tion during the winter, due perhaps to the higher inci- occasionally cause throat infection (20). dence of viral infections (1 1). The present study confirmed previous findings (5) Some studies (3,5,12) have reported no seasonal vari- that, in samples with growth of BHS, the proportion of ation in prevalence of BHS in healthy individuals, but group C and G was higher in the healthy individuals no comparison of prevalence of BHS in age-matched than in the patients. Furthermore, in healthy individuals patients was done. the prevalence of group C and G, compared with that of group A, was higher in adults than in children (12). Influence of age Thus, growth of BHS belonging to group C and G in Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12 The strong correlation between age and the prevalence adult patients and growth of BHS of any serotype in of BHS is caused by a mechanism whereby children children with throat pain may represent a carrier state, gradually acquire immunity to the prevalent serologic and it might therefore be difficult to interpret the find- types of group A streptococci (13). Thus, we as well ings with respect to the aetiology of disease. However, as others (14) can state that a throat culture is more overtreating a few symptomatic adult patients who have useful in adults than in children in predicting disease a low incidence of throat infection, with growth of caused by BHS. group C and G, is a minor problem. It is a bigger problem to overtreat many children who have a high Influence of day care incidence of throat infection, carrying GABHS, and the Due to the strong correlation between age and type of problem of overtreating patients with no proven growth day care, no conclusions concerning the influence of the of BHS is even greater. type of day care could be drawn in this study. Some authors report a trend for children in public child day Streptococcal epidemiology care to have a higher prevalence of BHS than children The proportion of symptomatic patients with a throat being taken care of at home (5). Stromberg et al. (12) infection who harbour BHS varies with season, epi- found higher carrier rates in children aged four years not demiological situation, and age; thus, the rate may vary attending day care centres than in those attending day from 15% to 30% (21,22). Based on the preliminary

Scand J Prim Health Care 1997; 15 154 R.K. Gunnarsson et al.

clinical diagnosis, the probability of streptococcal throat beta hemolytic streptococci among patients in general infection may be judged as “unlikely”, “possible”, or practice. Acta Path01 Microbiol Immunol Scand [B] “probable”. Patients judged as having a “probable” 1985;93:347-5 1. 4. Gerber MA, Randolph MF, Mayo DR. The group A strep streptococcal infection are treated with antibiotics, and tococcal carrier state. A reexamination. Am J Dis Child those judged as having an “unlikely” infection are not. 1988;142:562-5. For patients judged as having a “possible” infection the 5. Begovac J, Bobinac E, Benic B. Desnica B. Maretic T. result of a microbiological test could decide whether Basnec A, et al. Asymptomatic pharyngeal carriage of treatment with antibiotics was necessary. However, some beta-haemolytic streptococci and streptococcal pharyngitis of these patients might be carriers, with an infection among patients at an urban hospital in Croatia. Eur J caused by other agents. It is assumed that, in patients with Epidemiol 1993;9:405- 10. 6. Breese BB, Breese HC, Le C. Carriers. In: Breese BB, tbroat pain who have an infection caused by agents other Breese HC, editors. Beta Hemolytic streptococcal dis- than BHS, the carrier rate of BHS is similar to that of eases. Boston: Houghton Mifflin Professional Publishers, healthy individuals (3-5,23). A higher proportion of 1978: 143-53. asymptomatic carriers will diminish the clinical value of a 7. Breese BB. Breese HC, Le C. Epidemiology. In: Breese positive throat culture. Using the results from the present BB. Breese HC, editors. Beta Hemolytic streptococcal diseases. Boston: Houghton Mifflin Professional Pub- use study, it could be difficult to the throat cultures during lishers, 1978:34-64. the summer season in patients judged as having a “possi- 8. Gerber MA. Comparison of throat cultures and rapid strep ble” streptococcal infection when they are between 3-15 tests for diagnosis of streptococcal pharyngitis. Pediatr years of age. The clinical value of a negative throat culture Infect Dis J 1989;8:820-4. is not affected by asymptomatic carriers, and a negative 9. Prakash K, Lakshmy A. Streptococcal throat carriage in throat culture will, with approximately 97% probability, school children with special reference to seasonal inci- rule out BHS the aetiological agent (24,25). dence. Southeast Asian J Trop Med Public Health as 1992;23:705- 10. There is a problem of overtreatment of symptomatic 10. Ro&n C, Christensen P, Hovelius B, Prellner K. A longi- patients with throat infection and no proven growth of tudinal study of the nasopharyngeal carriage of pneu- BHS. However, the problem of overtreating carriers of mococci as related to pneumococcal vaccination in chil- BHS also needs to be considered. The carrier rate of dren attending day-care centres. Acta Otolaryngol BHS among healthy individuals has been assessed to (Stockh) 1984;98:524-32. vary between 6.7% and 26% (2,26), and the prevalence 11. Hakansson A, Kidd A, Wadell G, Sabharwal H, Svanborg C. Adenovirus infection enhances in vim adherence of For personal use only. of GABHS between 0.6% and 11.3% (3,12). Our find- Streptococcus pneumoniae. Infect Immun 1994,62:2707-14. ings are in agreement with these figures, and thus there 12. Stromberg A, Schwan A, Cars 0. Throat carrier rates of seems to be a risk of overtreating carriers. beta-hemolytic streptococci among healthy adults and children. Scand J Infect Dis 1988;20411-7. Conclusions 13. Quinn RW. Hemolytic streptococci in Nashville school The prevalence of BHS varies with age. The differences children. South Med J 1980;73:288-96. 14. Stromberg A. Studies on group A streptococcal pharyn- in prevalence of BHS between patients with throat pain gotonsillitis. Diagnostic and therapeutic aspects (thesis). and age-matched healthy individuals were greater dur- Uppsala: The Faculty of Medicine Univ of Uppsala, 1987. ing the winter than in the summer. This result empha- 15. Breese BB, Breese HC, Le C. Bacteriology and serology. sizes that during the summer it is much more difficult to In: Breese BB, Breese HC, editors. Beta Hemolytic strep- interpret the result of a throat culture due to a high tococcal diseases. Boston: Houghton Mifflin Professional Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12 carrier rate among healthy individuals aged 46years. Publishers, 1978:9-33. 16. Box QT, Cleveland RT, Willard CY.Bacterial flora of the upper respiratory tract. Comparative evaluation by anter- ior nasal, oropharyngeal, and nasopharyngeal swabs. Am J ACKNOWLEDGEMENTS Dis Child 1961;102:39-47. We thank the staff at the microbiology laboratory in 17. Gerber MA, Randolph MF, Chanatry J, Wright LL, De- Bods for processing the cultures. Financial support was Meo KK, Anderson LR. Antigen detection test for strep- obtained from the Research Committee, Health Depart- tococcal pharyngitis: evaluation of sensitivity with respect ment of Elfsborg County. to true infections. J Pediatr 1986;108:654-8. 18. Huck W, Reed BD, French T, Mitchell RS. Comparison of the Directigen 1-2-3 Group A Strep Test with culture for detection of group A beta-hemolytic streptococci. J Clin REFERENCES Microbiol 1989;27: 1715-8. 1. Roos K. The diagnostic value of symptoms and signs in 19. Gerber MA. Office cultures for group A streptococci. acute tonsillitis in children over the age of 10 and in Pediatr Infect Dis J 1991;10:960-1. adults. Scand J Infect Dis 1985;17:259-67. 20. Arditi M, Shulman ST, Davis AT, Yogev R. Group C 2. Hansen JG, Schmidt H, Bitsch N. Sore throat. Principles beta-hemolytic streptococcal infections in children: nine of diagnosis and treatment. Ractitioner 1983;227:937-48. pediatric cases and review [see comments]. Rev Infect Dis 3. Hoffmann S. The throat carrier rate of group A and other 1989;11:34-45.

Scand J Prim Health Care 1997: I5 The prevalence of beta-haemolytic streptococci 155

21. Tenjarla G, Kumar A, Dyke JW. TestPack Strep A kit for Streptococcal pharyngitis. Comparison of latex agglutin- the rapid detection of group A streptococci on 11,088 ation and throat culture. Clin Pediatr (Phila) 1988;27:431- throat swabs in a clinical pathology laboratory. Am J Clin 4. Path01 1991;96:759-61. 25. Lewey S, White CB, Lieberman MM, Morales E. Evalu- 22. Hoffmann S. An algorithm for a selective use of throat ation of the throat culture as a follow-up for an initially swabs in the diagnosis of group A streptococcal pharyngo- negative enzyme immunosorbent assay rapid streptococcal tonsillitis in general practice. Scand J Prim Health Care antigen detection test. Pediatr Infect Dis J 1988;7:765-9. 1992;10:295-3OO. 26. Rudin L, Rotta J, Blomqvist C, Benslimane A, Berger 23. Centor Rh4, Meier FA, Dalton HP. Throat cultures and Jekic 0, Kereselidze T, et al. Multicentre evaluation of a rapid tests for diagnosis of group A streptococcal pharyn- direct coagglutination test for group A streptococci. Eur J gitis. Ann Intern Med 1986;105:892-9. Clin Microbiol 1987;6:303-5. 24. White CB, Hamis R, Weir MR, Gonzales I, Bass JW. For personal use only. Scand J Prim Health Care Downloaded from informahealthcare.com by 62.49.91.193 on 08/08/12

Scad J Prim Health Care 1997; 15

MAJOR ARTICLE

Intravenous Immunoglobulin G Therapy in Streptococcal Toxic Shock Syndrome: A European Randomized, Double-Blind, Placebo-Controlled Trial

Jessica Darenberg,1 Nahla Ihendyane,1 Jan Sjo¨lin,3 Ewa Aufwerber,2 Sven Haidl,4 Per Follin,5 Jan Andersson,1 Anna Norrby-Teglund,1 and the StreptIg Study Groupa

1 2 Center for Infectious Medicine, Department of Medicine, Karolinska Institutet, Huddinge University Hospital, Division of Medicine, Department Downloaded from of Infectious Diseases, Karolinska Hospital, Stockholm, 3Department of Medical Sciences, Uppsala University Hospital, Uppsala, 4Department of Infectious Diseases, Malmo¨ University Hospital, Malmo¨, and 5Department of Infectious Diseases, Linko¨ping University Hospital, Linko¨ping, Sweden

(See the Editorial Commentary by Stevens on pages 341–3) http://cid.oxfordjournals.org/

The efficacy and safety of high-dose intravenous polyspecific immunoglobulin G (IVIG) as adjunctive therapy in streptococcal toxic shock syndrome (STSS) were evaluated in a multicenter, randomized, double-blind, placebo-controlled trial. The trial was prematurely terminated because of slow patient recruitment, and results were obtained from 21 enrolled patients (10 IVIG recipients and 11 placebo recipients). The primary end point was mortality at 28 days, and a 3.6-fold higher mortality rate was found in the placebo group. A significant decrease in the sepsis-related organ failure assessment score at days 2 (P p .02 ) and 3 (P p .04 ) by guest on August 8, 2012 was noted in the IVIG group. Furthermore, a significant increase in plasma neutralizing activity against superantigens expressed by autologous isolates was noted in the IVIG group after treatment (P p .03 ). Although statistical significance was not reached in the primary end point, the trial provides further support for IVIG as an efficacious adjunctive therapy in STSS.

The 2 most severe manifestations of invasive infections secreted, which have been shown to be important in caused by group A streptococci (GAS) are necrotizing pathogenesis. A pivotal role for superantigens in me- fasciitis and streptococcal toxic shock syndrome (STSS). diating the systemic effects of STSS has been shown [2, The latter is characterized by hypotension and mul- 3]. Previous studies have shown that protective hu- tiorgan failure early in the course of infection [1]. GAS moral immunity to both cell-associated and soluble express several virulence factors, both cell bound and GAS virulence factors are important in preventing in- vasive disease [4–8]. Patients with invasive GAS disease had significantly lower serum levels of protective an- tibodies against M-protein and superantigens, com- Received 9 January 2003; accepted 20 March 2003; electronically published 17 July 2003. pared with serum samples from noninvasive cases Financial support: Baxter Healthcare; Swedish Foundation for Strategic Research [4–7]. These findings provided evidence that lack of and the Swedish Research Council (to J.A. and A.N.-T.). protective humoral immunity against the relevant GAS a Members of the study group are listed at the end of the text. virulence factors contributes to susceptibility to invasive Reprints or correspondence: Dr. Anna Norrby-Teglund, Center for Infectious Medicine F59, Dept. of Medicine, Karolinska Institutet, Huddinge University infection. Hospital, S-141 86 Stockholm, Sweden ([email protected]). There is a definite need for adjunctive therapy in Clinical Infectious Diseases 2003;37:333–40 severe invasive GAS infections because the mortality 2003 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2003/3703-0003$15.00 rate can be as high as 80% despite prompt antimicrobial

IVIG in Streptococcal Toxic Shock • CID 2003:37 (1 August) • 333 therapy in many cases [9]. Previous studies on humoral im- OBC II; Abbott Laboratories, Diagnostics Division). Exclusion munity in invasive GAS disease have highlighted the impor- criteria included known hypersensitivity to IVIG or underlying tance of antibodies in protection against these infections, and diseases that were expected to cause death within 3 months. they suggest that immunoglobulins might be a useful adjunctive Treatment assignments. Patients were randomly assigned therapy. Human polyspecific intravenous IgG (IVIG) was sug- in a 1:1 manner to receive either IVIG (Endobulin S/D; Baxter) gested as a potential adjunctive therapy for invasive GAS dis- or an equal volume of 1% albumin (5% albumin [Baxter], eases mainly because of its ability to neutralize a wide variety diluted in 0.9% normal saline). IVIG was provided intrave- of superantigens and to facilitate opsonization of streptococci nously for 3 consecutive days at 1 g/kg of body weight on day [10]. An observational cohort study of IVIG therapy involving 1 and 0.5 g/kg on days 2 and 3. Patients received clindamycin patients with STSS reported decreased mortality rates in pa- (600 mg iv t.i.d.) in combination with intravenous benzylpen- tients treated with IVIG, compared with controls [11]. Here, icillin corresponding to a maximum dosage of 12 g per day, we report the results of the first multicenter, placebo-controlled for a minimum of 3 days after inclusion. In the case of penicillin trial to evaluate the safety and efficacy of IVIG as adjunctive intolerance, cefuroxime was provided in dosages of 1.5 g 3 times therapy for patients with STSS. per day. The total period of antibiotic treatment was stan- dardized to at least 14 days. Plasmapheresis was not allowed to be performed for enrolled patients because it would likely SUBJECTS, MATERIALS, AND METHODS

interfere with the study therapy. Downloaded from Data collection. The study included 5 phases: screening, Study protocol. A multicenter, randomized, double-blind, baseline (maximum of 4 h), treatment (days 1–14), follow-up placebo-controlled clinical trial of IVIG treatment of patients (days 14–28), and long-term follow-up (days 28–180). At base- with STSS, with or without necrotizing fasciitis, was performed. line, demographic and clinical information was obtained, in- The trial was investigator-driven and involved 17 hospitals in cluding a Simplified Acute Physiology Score II to determine http://cid.oxfordjournals.org/ Sweden, Norway, Finland, and The Netherlands. The study was the severity of illness [12]. Organ failures were defined by means initiated in January 1999 and, on the basis of intent-to-treat of the criteria outlined in the consensus definition for STSS analysis, was projected to include a finite number of 120 pa- proposed by The Working Group on Severe Streptococcal In- tients evenly randomized (1:1) to the IVIG and placebo groups. fections [13]. Information concerning the infection, including However, as a result of slow patient recruitment, the trial had progression of tissue infection (i.e., time to cessation of skin to be prematurely terminated in May 2001. The study was approved by the regional ethics committee and drug agency involvement), number of tissue lesions, microbiological data, by guest on August 8, 2012 authority in each country. Written informed consent was ob- and antibiotic use, was gathered during all study phases. Other tained from all subjects or their legal guardians. The clinical clinical and laboratory data were followed as long as the patient coordinating center (Huddinge University Hospital, Stock- required intensive care. Severity of organ dysfunction was de- holm, Sweden) was available 24 h per day throughout the study termined on days 1–3 during the stay in the intensive care unit to answer investigators’ questions regarding patients’ eligibility by the Sequential Organ Failure Assessment (SOFA) score [13]. and safety issues and to handle the reporting of serious adverse Blood samples for in vitro studies were obtained before infusion events. The study was performed in accordance with the Dec- of study drug on days 1–3, as well as postinfusion day 3 and laration of Helsinki. on days 28 and 180. Bacterial isolates were recovered from all Selection criteria. Adult patients (age, у18 years) who patients with positive blood culture results. In cases in which fulfilled the clinical criteria for STSS, including hypotension blood isolates were not available, isolates recovered from throat and multiorgan failure, as described in the consensus definition or skin/tissue aspirates were obtained. that was proposed by the Working Group on Severe Strepto- Study variables. The primary objective was to determine coccal Infections [1], were eligible for enrollment. Patients whether IVIG therapy resulted in a decrease in mortality during could be enrolled before results from bacteriological cultures the first 28 days, compared with placebo. The secondary ob- were obtained if they had clinical symptoms of STSS and if a jective was to evaluate whether there were differences between streptococcal infection was suspected. Suspicion of a strepto- the 2 groups in time to resolution of shock, time to no further coccal infection included tissue involvement; suspected tissue progression of the tissue infection, and survival on day 180, or involvement as indicated by severe local pain; recent history of differences in mental, renal, respiratory, or cardiovascular prob- streptococcal infections in the family or symptoms of tonsillitis, lems at day 180. pharyngitis, or scarlet fever–like rash; or local wound with in- The safety of the study drugs was assessed by an independent flammation. To facilitate identification of GAS cases, throat data and safety monitoring committee that reviewed nonserious swabs and aspirates from patients with skin or soft-tissue in- adverse events, serious adverse events, disease-related events, fections were screened using rapid antigen tests (Strept-A Plus indices of organ dysfunction (hepatic, renal, pulmonary, ce-

334 • CID 2003:37 (1 August) • Darenberg et al. Table 1. Inclusion characteristics of the patients provided high-dose in- travenous polyspecific IgG or placebo.

IVIG group Placebo group Characteristic (n p 10) (n p 11) Age, years Mean 51.3 52.6 Median (range) 54 (28–72) 52 (35–83) Male sex 4 (40) 6 (55) SAPS II scorea Mean 53 51 Median (range) 52 (28–78) 54 (28–98) SOFA scorea Mean 11.0 11.0 Median (range) 10.7 (6–16) 11.5 (6–19) Definitive STSS 8 (80) 10 (91) Other TSS 2 (20)b 1(9)c

Deep-tissue infection 6 (60) 7 (64) Downloaded from No. of lesions Mean 1.7 1.8 Median (range) 1 (1–5) 1 (1–4) Time of hypotension before inclusion, h Mean 33.5 23.3 http://cid.oxfordjournals.org/ Median (range) 34.0 (2.5–74.8) 21.2 (2.9–59.4) Dysfunctional organs or systems Renal failure 7 (70) 8 (73) Coagulopathy 6 (60) 9 (82) Liver failure 5 (50) 6 (55) Adult respiratory distress syndrome 4 (40) 2 (18)

Erythematous rash 2 (20) 3 (27) by guest on August 8, 2012 Soft-tissue necrosis 5 (50) 5 (45) Total no. of organ failures Mean 3.1 3.0 Median 3.0 3.0 IL-6 plasma levels,a median ng/mL (range) 1.4 (0.1–29.1) 2.3 (0.1–1424) IL-8 plasma levels,a median ng/mL (range) 0.3 (0.1–1.6) 1.6 (0.1–374)

NOTE. Data are no. (%) of patients, unless otherwise indicated. IVIG, intravenous IgG; SAPS II, Simplified Acute Physiology II; SOFA, Sepsis-related Organ Failure Assessment; STSS, streptococcal toxic shock syndrome; TSS, toxic shock syndrome. a Values obtained during screening phase. b One patient with mixed infection caused by Staphylococcus aureus and group B strep- tococcus, and one patient with mixed anaerobic infection in which gram-positive cocci were identified. c Infection caused by Pseudomonas aeruginosa. rebral, cardiac, metabolic, and hematologic), and deaths due formed to detect genes encoding the streptococcal superantigen to all causes. (ssa) and the pyrogenic exotoxins speA ,speB, speF, speG, speH, Strain characterization. Clinical isolates were identified and smeZ/2, as previously described [15]. as GAS by conventional methods. T-typing of the GAS isolates Preparation of bacterial culture supernatant. Isolates was performed at the Swedish Institute for Infectious Disease were cultured overnight in 15 mL Todd-Hewitt broth (Difco) Control (Solna, Sweden), in accordance with standard agglu- supplemented with 1.5% yeast extract (Difco) at 37C. Culture tination procedures. DNA from the streptococcal isolates was supernatants were separated from the bacteria by centrifuga- prepared as described elsewhere [14]. PCR amplification that tion, and proteins in the culture supernatants were ethanol used primer pairs specific for each superantigen gene was per- precipitated overnight, as previously described in detail [16].

IVIG in Streptococcal Toxic Shock • CID 2003:37 (1 August) • 335 Table 2. Characterization of group A streptococcal isolates’ isolated from patient blood by Ficoll-Hypaque gradient cen- serotype and genotype distribution. trifugation, transferred to adhesion glass slides (Erie Scientific), and fixed with 2% formaldehyde (Fisher Scientific). The cells No. (%) of strains were stained for specific cell markers and cytokines via im- IVIG groupa Placebo groupb Strain characteristics (n p 7) (n p 10) munohistochemistry, as previously described in detail [17]. In brief, the cells were permeabilized with balanced salt solu- T serotype tion–saponin to allow intracellular staining. The following T1 1 (14) 7 (70) monoclonal antibodies were used: antihuman TNF- (cocktail T12 3 (43) 0 (0) a NT 2 (29) 1 (10) of MabI and MabII, murine IgG1; Pharmingen), antihuman T3/T3.B3264 0 (0) 2 (20) IFN-g (cocktail of 1-DIK and 7-B6-1, murine IgG1; MabTech; T4 1 (14) 0 (0) both used at a concentration of 2 mg/mL). Antihuman CD3 spe genotypec (Leu 4, murine IgG1, Becton Dickinson; used at a 1:10 dilu- speA 3 (43) 8 (80) tion), and antihuman CD68 (EMB II, murine IgG1; Dako; used speH 5 (71) 4 (40) at a 1:200 dilution). A biotinylated goat antimouse IgG1 (Cal- ssa 3 (43) 1 (10) tag) was used as secondary antibody. The immunohistochem- speA and speH 2 (29) 3 (30) ical staining was achieved using avidin-peroxidase reagent (Vec-

speH and ssa 3 (43) 1 (10) tastain Elite-ABC; Vector Laboratories) in combination with Downloaded from speA, speH, and ssa 2 (29) 1 (10) the substrate diaminobenzidine (Vector Laboratories). Dupli- Mean no. of superantigen genes 5.6 5.2 cate fields were stained for the cytokines and single fields for the cell markers. Frequencies of cytokine-producing cells and NOTE. IVIG, intravenous IgG; NT, not typable; spe, streptococcal pyro- genic exotoxin; ssa, streptococcal superantigen. specific cell types were assessed by direct microscopy, and http://cid.oxfordjournals.org/ a Seven strains available; 2 patients were infected with other bacteria, and 500–700 cells were counted per field. 1 had only a positive group A streptococcal rapid antigen test result. b Ten group A streptococcal strains available; 1 placebo patient had infec- Assessment of plasma cytokine levels. IL-6 and IL-8 con- tion caused by a non–group A streptococcal isolate. centrations in patient plasma were determined by multiplex c All strains harbored the speB, speF, speG (except 1), and smeZ (except 1) genes. cytokine analyses using Fluorokine MAP kits (R&D Systems) and the Luminex 100 instrument (Luminex), in accordance The supernatants were tested by dose-response experiments to with the manufacturers’ instructions. determine the optimal concentration to be used for the neu- Neutralization assay. PBMCs were isolated from a healthy by guest on August 8, 2012 tralizing assay, and a 1:100 dilution was found to be optimal donor by Ficoll-Hypaque gradient centrifugation. The PBMCs for all isolates. Cell-free culture supernatants were stored at (1 ϫ 10 6 cells/mL) were cultured in RPMI 1640 medium sup- Ϫ20C until used. plemented with 25 mmol/L HEPES, 4 mmol/L L-glutamine, Cytokine assessment at the single-cell level. PBMCs were and 100 U/mL penicillin-streptomycin. The cells were stimu-

Table 3. Primary and secondary end points of a study assessing the efficacy of administration of high-dose intravenous polyspecific IgG.

All included patients Patients with GAS only IVIG group Placebo group IVIG group Placebo group End point (n p 10) (n p 11) (n p 8) (n p 10) Primary: mortality day 28, no. (%) of patients 1 (10) 4 (36) 1 (12.5) 3 (30) Secondary Time to resolution of shock,a h Mean 88 122 100 122 Median (range) 96 (2–159) 108 (47–294) 108 (2–159) 108 (47–294) Time to no further progression of NF/cellulitis, h Mean 68b 36c 69c 36c Median (range) 20 (2–168)b 24 (19–72)c 20 (2–168)c 24 (19–72)c Mortality day 180, no. (%) of patients 2 (20) 4 (36) 1 (12.5) 3 (30)

NOTE. GAS, group A streptococci; IVIG, intravenous IgG; NF, necrotizing fasciitis. a In the survivors. b Seven patients. c Five patients.

336 • CID 2003:37 (1 August) • Darenberg et al. Statistical analysis. The binary data on mortality were analyzed by Fisher’s exact test. For all other comparisons, the Wilcoxon-Mann-Whitney U test was used. This test was also used for analyses of cytokines and differences in the neutralizing capacity of bacterial culture supernatants by patient plasma. The P values are based on 2-sided tests.

RESULTS

The trial was terminated prematurely as a result of slow patient recruitment. The main reason for the slow patient recruitment seemed to be a low incidence of STSS in the participating countries during the trial period. Only 21 patients, 10 of whom received IVIG and 11 of whom received placebo, were included (table 1). All surviving enrolled patients were observed for the entire study period. GAS strains were isolated from 17 of the patients. For 1 patient, the strain could not be cultured because Downloaded from of previous antibiotic therapy, but GAS was implicated as the causative agent on the basis of a positive result of a rapid antigen Figure 1. Initial Sepsis-related Organ Failure Assessment (SOFA) scores and changes (mean SE) during treatment in polyspecific intra- test performed on aspirate samples from the affected skin area. venous IgG (IVIG)– and placebo-treated patients. Differences were ana- The majority (94%) of the GAS isolates were obtained from lyzed by the Mann-Whitney U test, and P values are indicated. sterile sites—that is, blood cultures (53%) or cultures of tissue http://cid.oxfordjournals.org/ samples (41%). The nonsterile isolate was obtained from a lated with the cell-free bacterial culture supernatants (diluted throat swab. In 3 patients with suspected STSS, non-GAS strains 1:100), in the presence of 2.5% heat-inactivated plasma, sup- were isolated from blood cultures, including Pseudomonas aeru- plemented with 2.5% heat-inactivated fetal calf serum, as pre- ginosa, group B streptococcus in combination with Staphylo- viously described [16]. Patients’ plasma samples were tested for coccus aureus, and a mixed anaerobic infection in which gram- inhibitory activity against their own bacterial isolate. After 72 positive cocci had been observed by Gram stain and direct 3 h, the cells were pulsed for 6 h with 1 mCi per well of [ H]- microscopy. by guest on August 8, 2012 thymidine (specific activity, 6.7 Ci/mmol; ICN Biomedicals). At the time of inclusion in the trial, clinical and demographic Phytohemagglutinin-L (Sigma) was used as a positive control characteristics were similar in the IVIG and placebo groups at a concentration of 1 mg/mL. All samples were assayed in (table 1). Levels of IL-6 and IL-8 were found to be somewhat triplicate. The 2 IVIG batches used in the trial were tested for higher in the plasma samples obtained from the placebo group neutralizing activity against culture supernatants from all clin- before study drug infusion day 1 (table 1). There was an ov- ical isolates essentially as described above, but cells were stim- errepresentation of T1 and T3 isolates, and there was a higher ulated in the presence of 5% fetal calf serum. prevalence of the speA gene among strains isolated from the

Table 4. Clinical characteristics of patients who died who received high-dose intravenous polyspecific IgG (IVIG) or placebo.

Duration of hypotension Total no. of Age, Days to Deep-tissue before organ failures Superantigen Treatment years Sex death infection inclusion, h at inclusion Clinical isolate genotype profile IVIG 72 F 163 Yes 11 3 GBS + S. aureus NA IVIG 28 F 2 No 38 4 GAS T4a speAϪ,GϪ,HϪ, ssa+ Placebo 83 F 1 No 5 2 P. aeruginosa NA Placebo 64 M 8 No 40 3 GAS T1a speA+,G+,HϪ, ssaϪ Placebo 60 F 7 Yes 26 4 GAS T3a speA+,G+,HϪ, ssa+ Placebo 67 F 15 Yes 34 2 GAS NTa speAϪ,G+,H+, ssaϪ

NOTE. GAS, group A streptococci; GBS, group B streptococci; NA, not applicable; NT, not typable; P. aeruginosa, Pseudomonas aeruginosa; S. aureus, Staphylococcus aureus; spe, streptococcal pyrogenic exotoxin; ssa, streptococcal superantigen. a Indicates serotype of GAS strain.

IVIG in Streptococcal Toxic Shock • CID 2003:37 (1 August) • 337 respect to duration of hypotension before inclusion in the study, number of organ failures, or prevalence of deep-tissue infection. The mean age of patients who died was 62 years, compared with 48 years for survivors. The serotypes of GAS isolates varied among the patients who died and included T1, T3, T4, and a nontypeable strain. Also, the superantigen genotype profile of the strains varied. In 2 patients who died, bacteria other than GAS caused the episodes of toxic shock. Cytokine assessment at the single-cell level revealed no dif- ferences between the groups or within treatment groups over time with respect to frequencies of TNF-a– and IFN-g– producing cells, or CD3- and CD68-positive cells (data not shown). Analyses of plasma cytokine levels showed a gradual decrease in IL-6 and IL-8 levels over time in all patients; this did not differ between the IVIG and placebo groups. The results obtained from the neutralization assays are sum- marized in figure 2. In the placebo group, all plasma samples Downloaded from obtained before or after therapy lacked neutralizing activity against the tested GAS supernatants. In the IVIG group, low Figure 2. Neutralization of proliferative activity of bacterial culture supernatants by patient plasma. Plasma from patients treated with po- neutralizing activity was found in plasma before therapy. All lyspecific intravenous IgG (IVIG; open circles) or placebo (filled triangles) patients except one attained neutralizing activity in their plasma were tested for neutralizing activity against culture supernatants prepared after administration of IVIG. The IVIG patient who did not http://cid.oxfordjournals.org/ from the patients’ own isolates, as detailed in Subjects, Materials, and attain neutralizing activity against her isolate was the patient Methods. One patient was infected with Staphylococcus aureus (Staph.), who had a mixed infection caused by a group B streptococcal as indicated. Study drug was administered on days 1, 2, and 3, as indicated by arrows. Plasma samples obtained preinfusion on days 1, 2 strain and a S. aureus strain. Since the group B streptococcal and 3, on postinfusion day 3, and at follow-up on day 28 were analyzed. isolate did not induce a proliferative response when tested in Differences were analyzed by the Mann-Whitney U test, and P values vitro, the plasma neutralizing assay was performed by using are indicated. the S. aureus isolate. Significant differences between the treat-

ment groups were found at day 2 and after infusion of study by guest on August 8, 2012 drug on day 3; the IVIG group had significant higher neu- placebo group, compared with the IVIG group (table 2). How- tralizing activity (P p .003 and .04, respectively). In vitro ex- ever, the majority of strains harbored genes encoding for 5–7 periments showed that all GAS strains were inhibited 87%– different superantigens, and there was no difference between 100% by the 2 IVIG batches used in the study, whereas the S. the groups with regard to the total number of superantigen aureus strain was only inhibited to 65% (data not shown). genes. Reported were 6 severe adverse events (i.e., deaths) and 12 The primary end point was mortality over 28 days. The adverse events or disease-related events, including arterial fib- mortality rate was 3.6-fold higher in the placebo group than rillation [2], diarrhea, prostatitis, cerebral embolus or epilepsy, in the IVIG group, but statistical significance was not reached lung bleeding, candida septicemia, hypotension, adult respi- p (P .3 ), presumably because of the small sample size (table ratory distress syndrome/pneumonia, soft-tissue necrosis, fever 3). No significant differences could be seen for the secondary not responsive to antimicrobial treatment, and epileptic con- end points. However, the reduction in the SOFA score was vulsion. None of the events were reported to be related to the significantly higher in the IVIG group than in the placebo group study drug. on days 2 and 3 (P p .02 and .04, respectively; figure 1). Similar results for primary and secondary end points were obtained when only data from patients with proven STSS were analyzed DISCUSSION (table 3). A change in the SOFA score was also observed when only patients with proven STSS were included in the analysis, Several case reports have described the use of IVIG therapy in but the change in the SOFA score was significant only on day patients with severe invasive GAS infections, including STSS, 2, the mean values being Ϫ1.1, Ϫ1.6, and Ϫ2.3 on treatment necrotizing fasciitis, and necrotizing myositis [18–26]. Support days 1, 2, and 3, respectively. for a clinical efficacy of IVIG in STSS was provided by an Clinical and strain characteristics of the fatal cases are shown observational cohort study of IVIG therapy conducted in Can- in table 4. Patients who died did not differ from survivors with ada, which reported a significantly reduced mortality rate

338 • CID 2003:37 (1 August) • Darenberg et al. among IVIG-treated patients than among control subjects [11]. Earlier reports have indicated that there may be a rapid con- However, confounding factors in this study that potentially sumption of the relevant antibodies in patients with STSS and affected mortality data included study of historical controls and that multiple doses of IVIG may be required to maintain an increased use of clindamycin among IVIG-treated patients. adequate level of neutralization of superantigens [11, 16, 36]. To provide evidence for a clinical efficacy of IVIG in STSS, In this study, the patients received 3 infusions of IVIG with we conducted a multicenter placebo-controlled trial designed the highest dosage (1 g/kg body weight) provided at day 1, to evaluate the efficacy and safety of high-dose IVIG in this followed by 0.5 g/kg on days 2 and 3. Comparison of plasma disease setting. However, as a result of a low incidence of STSS samples obtained after the first dose of IVIG (sample obtained in the participating countries during the study period, patient before treatment day 2) and before the third dose on day 3 recruitment was slow, and the trial was prematurely terminated also indicated a rapid consumption of antibodies: 83% of the after enrollment of 21 patients. The IVIG and placebo groups samples showed a reduction in neutralizing activity from day were well matched with respect to clinical and demographic 2 to day 3. characteristics, but there were differences in strain character- The results of this study, even though they did not reach istics between the groups. Serotypes T1 and T3 were overre- significant differences in the primary end point, provide further presented among isolates from the placebo group, and the speA support for the use of IVIG as adjunctive therapy in patients gene, which is commonly found in these serotypes, was con- with STSS. Although it would be desirable to have a larger controlled trial to provide statistical support for a clinical ef- sequently more prevalent in strains isolated from the placebo Downloaded from group. ficacy, the low incidence of STSS is a major obstacle that caused However, and most importantly, all GAS strains harbored a premature termination of the present study and is likely to genes encoding for several different superantigens, and there discourage the conduct of further trials. Considering this as was no difference between the groups in the total number of well as the high mortality rate and the clinical and functional superantigen genes. Although the T1M1 and T3 serotypes have data published to date, it seems reasonable to recommend IVIG http://cid.oxfordjournals.org/ been the most common serotypes found in the recent outbreaks as adjunctive therapy for the treatment of STSS until further of severe streptococcal infections and, therefore, are believed data are available. to be more invasive, other serotypes are known to cause severe and even fatal infections [4, 27–35]. This point is underscored StreptIg STUDY GROUP in the current study, in which isolates recovered from the pa- Principal investigator. Jan Andersson. tients who died included serotypes of T1, T3, and T4, as well Study committee. Jan Andersson, Anna Norrby-Teglund, as a nontypeable strain. Furthermore, the superantigen gene by guest on August 8, 2012 Martin Lee, and Ragnar Norrby. profile differed in these isolates, with 2 strains lacking the speA Investigators. Sweden: Erling B. Myhre, Jonas Cronqvist, gene. Therefore, we find it unlikely that the overrepresentation Birgitta Svanteson, Sven Haidl, Inga Odenholdt, Jan Sjo¨lin, Mia of T1 and T3 strains in the placebo group would significantly Furebring, Bo So¨derquist, Per Follin, Vanda Friman, Ewa Auf- affect the outcome of the study. werber, Bengt Ga˚rdlund, Jan Ha¨ggqvist, Gunilla Sundelin, Anders Analyses of the primary end point revealed a trend toward Ha˚kansson, Bjo¨rn Eriksson, Go¨ran Gu¨nther, and Jan Smedjega˚rd. a reduced mortality rate in IVIG-treated patients as compared Norway: Claus Ola Solberg, Stein Lund-Tønnessen, Steinar with those receiving placebo (10% vs. 36%). These results were Skrede, Haakon Sjursen, Elisabeth von der Lippe, Oddbjørn Bru- supported by significantly better improvement of organ dys- bakk, Bjørn Myrvang, Kjell B. Hellum, Nils Smith-Erichssen, function during treatment days 1–3 in the IVIG group, as as- Øystein Strand, and Aira Bucher. Finland: Asko Ja¨rvinen. The sessed by the SOFA score, which has been considered especially Netherlands: Jaap van Dissel, Peter Veldkamp, Jos van der Meer, suitable for sequential analyses [13]. Bart Jan Kullberg, Jan Verhoef, and Ellen Mascini. United King- We also performed in vitro studies to detect any differences dom: Jonathan Cohen and Shiranee Sriskandan. Denmark: Gitte between the IVIG and placebo groups with respect to super- Kronborg and Christian Fischer. antigen-neutralizing activity and cytokine levels. No difference Streptococcal serotyping. Birgitta Henriques Normark. between the groups was noted in the cytokine levels before and Data and safety committee. Johan Giesecke, Go¨ran Her- after therapy. Both groups showed an equal decrease in the mere´n, and Jørgen Hilden. levels of serum IL-6 and IL-8 on day 2. A significant increase Scientific advisor. Don E. Low. in superantigen-neutralizing activity after administration of IVIG could be seen, whereas plasma neutralizing activity re- Acknowledgments mained low in the placebo group. This is in agreement with previous reports that IVIG confers neutralizing antisuperan- The gift of group A streptococcal rapid antigen tests by Ab- tigen antibodies to the patients [11, 16, 36]. bott Scandinavia AB is acknowledged. We also acknowledge the

IVIG in Streptococcal Toxic Shock • CID 2003:37 (1 August) • 339 assistance of the pharmacies and health care workers at the 17. Andersson J, Nagy S, Bjo¨rk L, Abrams J, Holm S, Andersson U. Bac- terial toxin–induced cytokine production studied at the single-cell level. participating centers. Immunol Rev 1992; 127:69–96. 18. Cawley M, Briggs M, Haith LJ, et al. Intravenous immunoglobulin as adjunctive treatment for streptococcal toxic shock syndrome associated References with necrotizing fasciitis: case report and review. Pharmacotherapy 1999; 19:1094–8. 1. Working Group on Severe Streptococcal Infections. Defining the group 19. Barry W, Hudgins L, Donta ST, Pesanti EL. Intravenous immuno- A streptococcal toxic shock syndrome: rationale and consensus defi- globulin therapy for toxic shock syndrome. JAMA 1992; 267:3315–6. nition. The Working Group on Severe Streptococcal Infections [com- 20. Chiu CH, Ou JT, Chang KS, Lin TY. Successful treatment of severe ment]. JAMA 1993; 269:390–1. streptococcal toxic shock syndrome with a combination of intravenous 2. Kotb M. Bacterial pyrogenic exotoxins as superantigens. Clin Microbiol immunoglobulin, dexamethasone and antibiotics. Infection 1997; 25: Rev 1995; 8:411–26. 47–8. 3. Norrby-Teglund A, Kotb M. Host-microbe interactions in the path- 21. Lamothe F, D’Amico P, Ghosn P, Tremblay C, Braidy J, Patenaude J. ogenesis of invasive group A streptococcal infections. J Med Microbiol Clinical usefulness of intravenous human immunoglobulins in invasive 2000; 49:849–52. group A streptococcal disease: case report and review. Clin Infect Dis 4. Holm SE, Norrby A, Bergholm A-M, Norgren M. Aspects of the path- 1995; 21:1469–70. ogenesis in serious group A streptococcal infections in Sweden 22. Nadal D, Lauener RP, Braegger CP, et al. T cell activation and cytokine 1988–1989. J Infect Dis 1992; 166:31–7. release in streptococcal toxic shock–like syndrome. J Pediatr 1993; 122: 5. Eriksson B, Andersson J, Holm S, Norgren M. Invasive group A strep- 727–9. tococcal infections: T1M1 isolates expressing pyrogenic exotoxins A 23. Mahieu L, Holm S, Goossens H, Van Acker K. Congenital streptococcal and B in combination with selective lack of toxin-neutralizing anti- toxic shock syndrome with absence of antibodies against streptococcal Downloaded from bodies are associated with increased risk of streptococcal toxic shock pyrogenic exotoxins. J Pediatr 1995; 127:987–9. syndrome. J Infect Dis 1999; 180:410–8. 24. Perez CM, Kubak BM, Cryer HG, Salemugodam S, Vespa P, Farmer 6. Norrby-Teglund A, Pauksens K, Holm SE, Norgren M. Relation be- tween low capacity of human sera to inhibit streptococcal mitogens D. Adjunctive treatment of streptococcal toxic shock syndrome with and serious manifestation of disease. J Infect Dis 1994; 170:585–91. intravenous immunoglobulin: case report and review. Am J Med 7. Basma H, Norrby-Teglund A, McGeer A, et al. Opsonic antibodies to 1997; 102:111–3. the surface M protein of group A streptococci in pooled normal im- 25. Stegmayr B, Bjorck S, Holm S, Nisell J, Rydvall A, Settergren B. Septic http://cid.oxfordjournals.org/ munoglobulins (IVIG): potential impact on the clinical efficacy of IVIG shock induced by group A streptococcal infection: clinical and ther- therapy for severe invasive group A streptococcal infections. Infect apeutic aspects. Scand J Infect Dis 1992; 24:589–97. Immun 1998; 66:2279–83. 26. Yong J. Necrotising fasciitis. Lancet 1994; 343:1427. 8. Basma H, Norrby-Teglund A, Guedez Y, et al. Risk factors in the 27. Cockerill FR 3d, MacDonald KL, Thompson RL, et al. An outbreak pathogenesis of invasive group A streptococcal infections: role of pro- of invasive group A streptococcal disease associated with high carriage tective humoral immunity. Infect Immun 1999; 67:1871–7. rates of the invasive clone among school-aged children. JAMA 1997; 9. Davies DH, McGeer A, Schwartz B, et al. Invasive group A streptococcal 277:38–43. infections in Ontario, Canada. The Ontario Group A Streptococcal 28. Eriksson BK, Andersson J, Holm SE, Norgren M. Epidemiological and

Study Group. N Engl J Med 1996; 335:547–54. clinical aspects of invasive group A streptococcal infections and the by guest on August 8, 2012 10. Norrby-Teglund A, Stevens DL. Novel therapies in streptococcal toxic streptococcal toxic shock syndrome. Clin Infect Dis 1998; 27:1428–36. shock syndrome: attenuation of virulence factor expression and mod- 29. Gaworzewska E, Colman G. Changes in the pattern of infection caused ulation of the host response. Curr Opin Infect Dis 1998; 11:285–91. by Streptococcus pyogenes. Epidemiol Infect 1988; 100:257–69. 11. Kaul R, McGeer A, Norrby-Teglund A, et al. Intravenous immuno- 30. Hoge CW, Schwartz B, Talkington DF, Breiman R, MacNeill EM, En- globulin therapy for streptococcal toxic shock syndrome—a compar- glender SJ. The changing epidemiology of invasive group A strepto- ative observational study. Canadian Streptococcal Study Group. Clin coccal infections and the emergence of streptococcal toxic shock–like Infect Dis 1999; 28:800–7. syndrome: a retrospective population-based study. JAMA 1993; 269: 12. Le Gall JR, Lemeshow S, Leleu G, et al. Customized probability models 384–9. for early severe sepsis in adult intensive care patients. Intensive Care 31. Martin PR, Høiby EA. Streptococcal serogroup A epidemic in Norway Unit Scoring Group. JAMA 1995; 273:644–50. 1987–1988. Scand J Infect Dis 1990; 22:421–9. 13. Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score 32. Nakashima K, Ichiyama S, Iinuma Y, et al. A clinical and bacteriologic to assess the incidence of organ dysfunction/failure in intensive care investigation of invasive streptococcal infections in Japan on the basis units: results of a multicenter, prospective study. Working Group on of serotypes, toxin production, and genomic DNA fingerprints. Clin “Sepsis-Related Problems” of the European Society of Intensive Care Infect Dis 1997; 25:260–6. Medicine. Crit Care Med 1998; 26:1793–800. 33. Johnson D, Stevens D, Kaplan E. Epidemiologic analysis of group A 14. Jasir A, Tanna A, Efstratiou A, Schalen C. Unusual occurrence of M streptococcal serotypes associated with severe systemic infections, rheu- type 77, antibiotic-resistant group A streptococci in southern Sweden. J Clin Microbiol 2001; 39:586–90. matic fever, or uncomplicated infections. J Infect Dis 1992; 166:374–82. 15. Chatellier S, Ihendyane N, Kansal RG, et al. Genetic relatedness and 34. Schwartz B, Facklam R, Breiman R. Changing epidemiology of group superantigen expression in group A streptococcus serotype M1 isolates A streptococcal infection in the USA. Lancet 1990; 336:1167–71. from patients with severe and nonsevere invasive diseases. Infect Im- 35. Stro¨mberg A, Romanus V, Burman L. Outbreak of group A strepto- mun 2000; 68:3523–34. coccal bacteremia in Sweden: an epidemiological and clinical study. J 16. Norrby-Teglund A, Kaul R, Low DE, et al. Plasma from patients with Infect Dis 1991; 164:595–8. severe invasive group A streptococcal infections treated with normal 36. Norrby-Teglund A, Kaul R, Low DE, et al. Evidence for the presence polyspecific IgG inhibits streptococcal superantigen–induced T cell of streptococcal superantigen neutralizing antibodies in normal po- proliferation and cytokine production. J Immunol 1996; 156:3057–64. lyspecific IgG (IVIG). Infect Immun 1996; 64:5395–8.

340 • CID 2003:37 (1 August) • Darenberg et al. Review Article Necrotizing Fasciitis

Abstract Joseph M. Bellapianta, MD Necrotizing fasciitis is a rare but life-threatening soft-tissue infection Karin Ljungquist, MD characterized by rapidly spreading inflammation and subsequent necrosis of the fascial planes and surrounding tissue. Infection Ellis Tobin, MD typically follows trauma, although the inciting insult may be as Richard Uhl, MD minor as a scrape or an insect bite. Often caused by toxin- producing, virulent bacteria such as group A streptococcus and associated with severe systemic toxicity, necrotizing fasciitis is rapidly fatal unless diagnosed promptly and treated aggressively. Necrotizing fasciitis is often initially misdiagnosed as a more benign soft-tissue infection. The single most important variable influencing mortality is time to surgical débridement. Thus, a high degree of clinical suspicion is necessary to avert potentially disastrous consequences. Orthopaedic surgeons are often the first to evaluate patients with necrotizing fasciitis and as such must be aware of the Dr. Bellapianta is Chief Orthopaedic presentation and management of this disease. Timely diagnosis, Resident, Albany Medical Center, Albany, NY. Dr. Ljungquist is broad-spectrum antibiotic therapy, and aggressive surgical Orthopaedic Resident, Ohio State débridement of affected tissue are keys to the treatment of this University Medical Center, serious, often life-threatening infection. Columbus, OH. Dr. Tobin is Clinical Associate Professor, Department of Medicine, Albany Medical College, Albany, NY. Dr. Uhl is Professor of lthough necrotizing fasciitis gar- rotizing fasciitis” in 1952 and de- Surgery, Division of Orthopaedic nered tabloid fame as a sensa- Surgery, Albany Medical College. A scribed the inflammation and necro- tionalistic “new” and dangerous sis of the subcutaneous fat and deep Dr. Uhl or a member of his “flesh-eating bacterium” in the immediate family is a member of a fascia, with sparing of the muscle, 1990s,1 it was described by Hip- speakers’ bureau or has made paid that are the cardinal features of this 2 presentations on behalf of AO and pocrates in the 5th century BCE. devastating disease.7 This character- has received research or institutional Necrotizing fasciitis has since had ization agrees with the current belief support from CONMED Linvatec, many names, including phagedena, DePuy, Smith & Nephew, Stryker, that a disease process, rather than a and Synthes. None of the following phagedena gangrenosum, progressive particular organism, is implicated in authors or a member of their bacterial synergistic gangrene, and this condition. immediate families has received nonclostridial gas gangrene.3 The anything of value from or owns first report in the United States was stock in a commercial company or institution related directly or made in 1871 by Confederate Army Etiology indirectly to the subject of this surgeon Joseph Jones, MD, who de- article: Dr. Bellapianta, scribed “hospital gangrene” with a Necrotizing fasciitis typically follows Dr. Ljungquist, and Dr. Tobin. mortality rate of almost 50%.4 In an injury to the involved site. Even a Reprint requests: Dr. Bellapianta, 1883, Fournier5 identified his ep- minor lesion may be sufficient to al- Albany Medical College, Suite 202, onymic form of gangrene and so de- low bacteria to breach the skin bar- 1367 Washington Avenue, Albany, NY 12206. scribed the pathology of necrotizing rier. Necrotizing fasciitis has been as- fasciitis affecting the perineum and sociated with minor or blunt trauma, J Am Acad Orthop Surg 2009;17: 6 8 174-182 external genitalia. Meleney identi- insect bites, surgical incisions, cuts, fied hemolytic streptococcus as the abrasions, contusions, injection sites, Copyright 2009 by the American Academy of Orthopaedic Surgeons. etiologic agent of the disease in cutaneous ulcers, perirectal ab- 1924. Wilson coined the term “nec- scesses, incarcerated hernia,7 burns,

174 Journal of the American Academy of Orthopaedic Surgeons Joseph M. Bellapianta, MD, et al splinters, childbirth, chicken pox, Figure 1 Figure 2 penetrating injury, and muscle strains.9 Although the skin is the most common portal of entry, in 45% of cases, no definitive access point can be found,10 possibly be- cause the inciting insult was so mi- nor as to be forgotten. The extremi- ties are most commonly involved, but necrotizing fasciitis can affect any body part.11 Infections of the trunk and perineal regions have a higher mortality rate than those of the extremities, presumably because amputation is not feasible as a life- saving option.7 Developing blisters on the arm in necrotizing fasciitis. Note the presence of ruptured bullae draining serosanguineous fluid in Risk Factors Necrotizing fasciitis of the arm. the lower left quadrant of the Note the ecchymosis, erythema, figure. and presence of serous drainage Any host condition that results in an from ruptured blisters. immunocompromised state is a risk quently, the infection is often incor- factor for necrotizing fasciitis. Diabe- rectly identified as cellulitis. A high in- tes mellitus, the most common co- dex of suspicion is critical to avoid disproportionate to injury seen ear- morbidity, is present in 18% to 60% morbidity and mortality because, al- lier in the process gives way to anal- of cases.8 Other risk factors include though cellulitis without deeper in- gesia as cutaneous nerves are de- obesity,12 peripheral vascular dis- volvement is usually treatable with an- stroyed. Superficial fat and fascia ease,13 intravenous drug use,13 alco- tibiotics alone, necrotizing fasciitis necrose, producing the watery, gray- hol abuse,14 malnutrition,8 smok- requires surgical débridement.9 The ish, often foul-smelling “dishwater ing,13 chronic cardiac disease,13,15 triad of swelling, erythema, and inor- pus” characteristic of the disease.20 chronic corticosteroid therapy,16 dinate pain—often disproportionate By this stage, the patient shows con- chronic immune suppression,17 can- to the observed lesion—in a patient stitutional symptoms of fever, chills, cer,17 and age.17 The continued or who is thought to have cellulitis hypotension, tachycardia, and possi- chronic use of nonsteroidal anti- should raise suspicion of necrotizing bly an altered level of consciousness, inflammatory drugs has been sug- fasciitis. Pain may precede warming and is critically ill21 (Table 1). gested as a possible risk factor7 be- and induration of skin (ie, wooden Acute renal failure is present in 35% cause it may mask initial symptoms skin) by several hours.18 Rapid mi- of patients, coagulopathy in 29%, liver and delay diagnosis.18,19 Although gration of the margins of erythema function tests are abnormal in 28%, numerous risk factors have been and skin induration (>1 cm/hr) de- acute respiratory distress syndrome is identified, half of all cases of necro- spite the use of intravenous antibiot- seen in 14%, and bacteremia is seen in tizing fasciitis occur in previously ics is another important clue in the 46%.22 The clinician must not be in- healthy individuals.8 early stages of the disease.3 appropriately reassured by the ab- Classic signs of necrotizing fasciitis sence of these findings because many Clinical Presentation develop as the disease progresses. patients may lack these signs on pre- Blisters and bullae form and drain sentation. In one series, only 47% of Necrotizing fasciitis typically begins as first serosanguineous and then hem- patients presented with classic skin a slightly inflamed area of soft tissue orrhagic fluid (Figure 1). The skin changes of bullae, vesicles, and ne- that suddenly and dramatically may also show violaceous discolora- crosis, and only 51% were febrile.23 progresses to overt fasciitis with sys- tion before turning frankly necrotic Disproportionately severe pain, the temic toxicity. Early diagnosis is made and sloughing (Figure 2). Crepitus most sensitive symptom, is noted in difficult by a paucity of skin findings in may be present if there is soft-tissue nearly 100% of patients with necro- the first stages of disease. Conse- gas. Edema develops rapidly. Pain tizing fasciitis.24

March 2009, Vol 17, No 3 175 Necrotizing Fasciitis

Diagnosis Table 1 Table 2 Physical Examination Findings in Variables and Scoring of the Diagnosis is primarily clinical, and Patients With Necrotizing Laboratory Risk Indicator for Fasciitis Necrotizing Fasciitis (LRINEC) no test is as valuable as a high degree of suspicion. Laboratory and radio- Finding Patients (%) Value Score logic evaluations can be helpful but Pain 100 C-reactive protein level are best used for confirmation of the Erythema 95 (mg/L) diagnosis. Pursuing such tests should Edema 82 <150 0 never delay surgical intervention. Av- Cellulitis 75 >150 4 erage time from onset of initial Fever (>38.5° C) 70 White blood cell count symptoms to diagnosis is 2 to 4 days, (cells/µL) Discoloration 49 although some cases may take weeks <15,000 0 Crepitus 25 to be correctly identified.25 15,000–25,000 1 Vesicles 16 >25,000 2 Disorientation 14 Laboratory Evaluation Hemoglobin level (g/dL) Initial laboratory evaluation should Adapted with permission from Childers BJ, >13.5 0 Potyondy LD, Nachreiner R, et al: 11–13.5 1 include a complete blood count, Necrotizing fasciitis: A fourteen-year comprehensive metabolic panel, and retrospective study of 163 consecutive <11 2 patients. Am Surg 2002;68:109-116. coagulation studies. Blood cultures Sodium level (mmol/L) should be taken to obtain pathologic ≥135 0 diagnosis of involved microorgan- ficity (76%) and has a poor positive <135 2 isms and antibiotic sensitivities to predictive value (26%) for necrotiz- Creatinine level (mg/dL) guide future targeted antibiotic ther- ing fasciitis, suggesting that it is use- ≤1.6 0 apy. Arterial blood gases should be ful only for ruling out the disease. >1.6 2 assessed, especially when signs of Another model for evaluating group Glucose level, mg/dL sepsis are present.7 Azotemia, hy- A streptococcus soft-tissue infections ≤180 0 ponatremia,hypoproteinemia,throm- found that C-reactive protein levels >180 1 bocytopenia, hematuria, elevated >16 mg/dL are 89% sensitive and creatine kinase and erythrocyte sedi- Adapted with permission from Anaya DA, 90% specific for necrotizing fasciitis, Dellinger EP: Necrotizing soft-tissue mentation rates, metabolic acido- and that creatine kinase levels >600 infection: Diagnosis and management. sis, hypoalbuminemia, anemia, and U/L are 58% sensitive and 95% spe- Clin Infect Dis 2007;44:705-710. hyperbilirubinemia are commonly cific for necrotizing fasciitis as op- noted. Hypocalcemia secondary to posed to cellulitis.18 tissue infections when 6 was used as 27 extensive fat necrosis may be seen. Wong et al27 developed the Labora- the cut-off score. Few metabolic abnormalities may be tory Risk Indicator for Necrotizing noted early on, but as the disease Fasciitis (LRINEC). This method dif- Radiologic Evaluation proceeds toward sepsis and organ ferentiates necrotizing fasciitis from Plain radiographs are most useful for failure, laboratory findings may be- other soft-tissue infections using the detecting gas in soft tissues. How- come extremely deranged. parameters of C-reactive protein ever, their utility is limited, as studies In addition to clinical monitoring, level, total WBC count, and hemo- are often normal until infection and routine laboratory evaluations can globin, serum sodium, creatinine, necrosis are advanced, and in many be used to distinguish necrotizing and glucose levels (Table 2). The cases, soft-tissue gas never presents. fasciitis from other soft-tissue infec- probability of necrotizing fasciitis is Computed tomography (CT) scans tions. Wall et al26 described a model <50% with a score ≤5, 50% to 75% are more beneficial than plain radio- based on the white blood cell (WBC) with a score of 6 or 7, and >75% graphs. Increased attenuation of the count and serum sodium levels. They with a score ≥8 (minimum score, 0; subcutaneous fat with stranding is reported that concurrent findings of maximum, 13). The model had a seen in 80% of cases, and fascial a WBC count >15,400 cells/µL and a 92% positive predictive value and a thickening may be present.25 Images serum sodium level <135 mmol/L are 96% negative predictive value for may show gas in soft tissues or 90% sensitive for necrotizing fascii- detecting early cases of necrotizing tracking along fascial planes.25 CT tis. However, this model lacks speci- fasciitis in patients with severe soft- scans are especially sensitive in iden-

176 Journal of the American Academy of Orthopaedic Surgeons Joseph M. Bellapianta, MD, et al tifying soft-tissue edema and often tizing fasciitis to be made within 15 suppuration of the arteries and veins can be more useful than physical ex- minutes of biopsy and reduces the (Figure 3). Obliterative vasculitis of amination in defining the margins of time between onset of symptoms and the subcutaneous vessels is seen early infection. However, abnormal CT diagnosis. Although mortality rates in the disease.13 There is expansion findings are not universal, and cases are lower with this technique, it is of fibrous septa by edema, mixed in- have been reported of necrotizing somewhat cumbersome and requires flammatory cell infiltrate, and early fasciitis with negative CT findings.24 the immediate availability of an ex- fibroblastic proliferation.29 As the 28 Magnetic resonance imagining perienced pathologist. Frozen- disease progresses, lesions develop studies have demonstrated a high section biopsy is best used in cases in liquefactive necrosis at all involved sensitivity (93% to 100%) for diag- which the diagnosis is unclear and tissue levels. A dense neutrophil- the patient is stable. Surgical inter- nosing necrotizing fasciitis.24 Lique- predominant inflammatory infiltrate vention should never be postponed factive tissue necrosis and inflamma- is present, and Gram stain of the af- to perform such a biopsy in cases in tory edema create fascial fluid that is fected tissues is usually positive.24 which necrotizing fasciitis is strongly detected as abnormally increased sig- suspected. nal intensity on T2-weighted images. On T1-weighted images, necrosis Microbiology and edema present as variably in- Mortality Aerobic, anaerobic, gram-positive creased signal intensity along thick- and -negative organisms, and even 29 There are an estimated 500 to 1,000 ened deep fascial planes. Although fungi have been implicated in necro- cases of necrotizing fasciitis in the magnetic resonance imaging is very tizing fasciitis (Table 3). The infec- United States annually,8 with mortal- useful in delineating the extent of in- tion can be classified into three ity ranging from 6% to 76%.27 A re- fection, it is of lesser priority than groups based on Gram stain and cent review of 147 patients revealed surgical débridement of necrotic tis- bacterial culture results. Type 1 con- a mortality rate of 9.3%, suggesting sue when a patient is unstable with stitutes 80% to 90% of all cases and that outcomes are improving, pre- worsening signs of sepsis. is a polymicrobial infection involving sumably because of more effective non–group A Streptococcus along patient management.30 Patients aged Biopsy with anaerobes and/or facultative <1 year or >60 years have the highest The diagnostic benchmark for necro- anaerobes. It often involves entero- mortality rates.13 Thrombocytopenia, tizing fasciitis is the finding of fascial bacteriaceae as well. Typically, four abnormal liver function tests, low se- necrosis during surgery. Outside the to five species will be cultured from rum albumin level,31 acute renal fail- operating room, tissue-based diag- the wound. This type is associated ure, and elevated blood lactate level nostic procedures may be of use. The with postoperative abdominal and are significantly associated with mor- “finger test” is a bedside procedure perineal infections; it is most com- tality,7 while advanced age, strepto- that can confirm necrotizing fasciitis. mon in immunosuppressed patients. coccal toxic shock syndrome, and Under local anesthesia, a 2-cm inci- Virulence is not well understood but immunocompromised status are in- sion is made down to the deep fascia, is speculated to be the result of syn- dependent predictors of death.32 A and a gloved finger is inserted to its ergy between bacterial species. study of 99 patients with necrotizing base. The presence of dishwater pus, Type 2 necrotizing fasciitis is de- fasciitis found that the risk of death lack of bleeding on incision, and lack fined by the presence of group A β- increased by 4% for every year of of tissue resistance to blunt finger hemolytic streptococci, typically as a life.32 However, the most important dissection define a positive test. single agent,20 although Staphylococ- clinical variable affecting mortality is Frozen-section biopsy can also be cus or other organisms may be the time from admission to initial dé- done at bedside. A small elliptical present. This is the “flesh-eating” bridement, making rapid diagnosis section of skin, deep tissue, and fas- presentation of necrotizing fasciitis.13 of the utmost importance.7 cia is taken from the suspected area, Infection usually occurs in an ex- along with another from the leading tremity and can develop in healthy edge of any erythema, induration, or Histopathology individuals. necrosis. The specimens are then Type 3 necrotizing fasciitis is caused submitted for Gram stain, frozen sec- Necrotizing fasciitis is characterized by marine vibrios (gram-negative rods). tion, and culture. This method al- by necrosis of the superficial fascia, Vibrio vulnificus is thought to be the lows an accurate diagnosis of necro- fat, and nerves, with thrombosis and most virulent of these agents. Infection

March 2009, Vol 17, No 3 177 Necrotizing Fasciitis

Figure 3 Table 3 Organisms Identified in Necrotizing Fasciitis

Gram-positive aerobic bacteria Group A β-hemolytic streptococci Group B streptococci Enterococci Coagulase-negative staphylococci Staphylococcus aureus Bacillus spp Gram-negative aerobic bacteria A, Low-magnification photomicrograph demonstrating necrotizing fasciitis. Escherichia coli Note the skeletal muscle tissue (upper right) encircled on the left and below Pseudomonas aeruginosa by infected fascial tissues featuring extensive acute suppurative inflammation Enterobacter cloacae (hematoxylin-eosin, original magnification ×20). B, Higher-magnification photomicrograph demonstrating the intense infiltrate of neutrophils in the Klebsiella spp fascial septum, with extension into the adjacent skeletal muscle fibers Proteus spp (hematoxylin-eosin, original magnification ×100). Serratia spp Acinetobacter calcoaceticus Citrobacter freundii usually begins with a puncture wound ders group A Streptococcus one of Pasteurella multocida caused by a fish, or a cut or insect bite the most common human pathogens, Anaerobic bacteria that is then exposed to seawater or ma- and it is capable of producing a vari- Bacteroides spp rine animals. Soft-tissue damage is ety of clinical disorders, including Clostridium spp thought to be mediated by an extracel- pharyngitis, cellulitis, impetigo, and Peptostreptococcus spp lular toxin synthesized by the vibrio scarlet fever. It is among the few bac- Marine Vibrio spp organism.20 Klebsiella, Escherichia teria that can cause a wound infec- Vibrio vulnificus coli, and other uncommon organ- tion, cellulitis, or necrotizing fasciitis Vibrio parahaemolyticus isms have also been reported as caus- within 24 hours following surgery.37 Vibrio damsela ative agents of necrotizing fasci- Vibrio alginolyticus itis.31,33 Recently, cases of necrotizing Fungi fasciitis resulting from community- Treatment Candida spp acquired methicillin-resistant Staphy- There are five parameters of therapy for Aspergillus spp lococcus aureus (MRSA) have been necrotizing fasciitis: early diagnosis and reported.34,35 Rhizopus débridement, broad-spectrum antibiot- The microbiology of type 2 necro- ics, aggressive resuscitation, frequent re- Adapted with permission from Green RJ, tizing fasciitis is the best character- Dafoe DC, Raffin TA: Necrotizing fasciitis. evaluation, and comprehensive nutri- ized of the three groups. The natural Chest 1996;110:219-229. tional support (Figure 4). habitat of group A Streptococcus is human skin and mucosal surfaces. It Surgical Débridement The initial incision should be made has coevolved with humans and does directly over the involved skin, paral- not survive outside a human host. Surgical débridement is the founda- lel to the neurovascular bundles and The bacterium boasts an enormous tion of management because it is the down to the deep fascia. The surgeon and evolving molecular diversity, fastest and most effective way to re- may encounter dull, gray, avascular ne- driven by horizontal transmission duce the bacterial load and halt the crotic tissue demonstrating a lack of re- among group A streptococci strains necrotic process. This is the only in- sistance to blunt dissection, lack of and also between group A and other tervention proven to increase the rate bleeding of the fascia, and the presence streptococci. Acquisition of pro- of survival.8 The goal of surgery is to of foul-smelling dishwater pus.24 The phages confers virulence through remove at the first débridement all margins of débridement should be phage-associated factors and in- necrotic tissue, including muscle and advanced until skin and fascia nor- creases bacterial survival against skin if necessary, in addition to the mally adherent to the deep fascia are host defenses.36 This diversity ren- involved fascia.11 encountered and must terminate in

178 Journal of the American Academy of Orthopaedic Surgeons Joseph M. Bellapianta, MD, et al

Figure 4

Treatment algorithm for necrotizing fasciitis. IVIG = intravenous immunoglobulin G, MRI = magnetic resonance imaging *Amputation may be indicated at the first surgical intervention if the infection is quickly progressing, necrosis has consumed most of the involved limb, and limb salvage is deemed impossible. viable, vascularized tissue. Poorly for either adults or children.39 formation of , and perfused tissue will act as a nidus for Careful management of the extensive may reduce wound pain. continued bacterial proliferation.13 wounds that often result from this rad- Once the infection has resolved and Although amputation does not im- ical form of surgery is critical. The a bed of healthy granulation tissue is prove mortality in certain studies,38 it wound must be reevaluated daily be- present, the wound may be closed with may be necessary with some very ag- cause repeated débridement is com- skin flaps or split-thickness grafts.7 gressive infections as a first-line sur- monly required. The wound should be Many cases will require reconstruc- gery to avoid death. Some authors kept covered to protect against second- tive surgery with possible free-tissue have described conservative manage- ary infection, encourage the formation transfer.13 Wound management tech- ment in pediatric patients, allowing of granulation tissue, and absorb in- niques developed for burn victims tissue demarcation to occur over a flammatory exudates. Both alginate and may be applicable, and the successful period of several days before surgical hydrogel dressings have been used suc- use of the skin substitute Integra (In- intervention, but this approach has cessfully. Topical negative pressure ther- tegra LifeSciences, Plainsboro, NJ) not been adopted at our institution apy can reduce edema and stimulate the following surgical débridement for

March 2009, Vol 17, No 3 179 Necrotizing Fasciitis

Table 4 Initial Antibiotic Management for Necrotizing Fasciitis

Gram Stain Result

Polymicrobial With Gram- Gram-positive Cocci Gram-positive Cocci in positive Cocci and Gram- Initial Empiric Therapy* in Clusters Pairs or Chains negative Bacilli

Clindamycin plus any of the Clindamycin plus vancomy- Clindamycin plus any of the follow- Clindamycin plus any of the following: imipenem, mero- cin, or monotherapy with ing: piperacillin/tazobactam, following: imipenem, mero- penem, ampicillin/sulbactam, linezolid ampicillin/sulbactam, or high- penem, ampicillin/sulbactam, piperacillin/tazobactam dose penicillin piperacillin/tazobactam

* Before results of Gram stain are available necrotizing fasciitis has been re- cocci in clusters seen on Gram stain is among patients with necrotizing fas- ported. By providing early wound adequate to support the use of vanco- ciitis.42 coverage, grafting can be delayed un- mycin or linezolid because of the in- til the patient has recovered from the creased prevalence of MRSA.34,35,41 In Supportive Care 40 initial injury. contrast, Gram stain revealing gram- Adequate fluid resuscitation and positive cocci in pairs or chains is blood pressure support are essential Antibiotic Therapy treated with clindamycin and a because patients with necrotizing fas- β-lactam antibiotic. When the Gram Antibiotics are an important comple- ciitis often are septic at presentation ment to surgical therapy but cannot stain reveals a polymicrobial flora, or become so very quickly. Nutri- be the sole treatment because fascia initial empiric therapy should be tional support is necessary because is poorly vascularized, and the dis- continued. As culture results and loss of fluid, protein, and electrolytes ease process further reduces its blood sensitivities become available, antibi- from the large surgical wounds are supply. This results in poor antibiotic otic therapy should be further ad- comparable to those observed in delivery to the site of infection. De- justed if necessary. burn victims. In the acute phase, in- spite having little effect on the Clindamycin is advised at the earli- take of twice the basal caloric re- wound itself, antibiotics can reduce est suspicion of group A streptococ- quirement is appropriate. A patient the bacterial load, terminate toxin cal or Staphylococcus aureus infec- who is unable to tolerate enteral production, and help prevent organ tion because it shuts down bacterial feedings should receive total failure. Intravenous antibiotic ther- ribosome function, which inhibits parenteral nutrition.13 Postoperative apy should be instituted at presenta- both M protein and exotoxin pro- care must also address adequate pain tion after microbial cultures are ob- duction. Linezolid also possesses this management, such as use of patient- tained. ability. This inhibition facilitates controlled analgesia.43 Initially, the infection should be phagocytosis and suppresses the syn- treated empirically with broad- thesis of tumor necrosis factor-α, Adjunctive Therapies spectrum agents with activity against thereby reducing the overzealous im- gram-positive, gram-negative, and mune response.9 The Eagle effect de- Adjunctive therapies include intrave- anaerobic organisms until antimicro- scribes the limits of the efficacy of nous immunoglobulin G (IVIG), hy- bial sensitivities are obtained (Ta- penicillin after a streptococcal infec- perbaric oxygen (HBO), and recom- ble 4). tion has reached steady state,37 but binant human-activated protein C. Acceptable regimens include imi- the efficacy of clindamycin is not af- The latter has been cited in a case re- penem, meropenem, ampicillin/sul- fected by the size of the bacterial in- port as a promising new therapy for bactam, and piperacillin/tazobactam, oculum or the stage of growth. Clin- sepsis that may reduce mortality in all in combination with clindamycin. damycin has been shown to be more patients with group A streptococcus Gram stain of pus or deep wound tis- effective than penicillin against strep- necrotizing fasciitis.44 sue acquired during surgery should tococcus species.24 One study found In both in vivo and in vitro studies, guide initial adjustments to antibiotic that use of clindamycin reduced the IVIG has been found to inhibit the ac- therapy. The presence of gram-positive risk of hospital mortality by 89% tivation of T-cells by superantigens and

180 Journal of the American Academy of Orthopaedic Surgeons Joseph M. Bellapianta, MD, et al to inhibit the activity of streptococcal ture, seizures, and central nervous sys- references 3, 7, 9, 11-15, 18-22, 24, 26, antigens to elicit cytokine production. tem oxygen toxicity.7 Because of the 28, 30-33, 36-39, 41, 42, and 47. Level Thus, it has been speculated to be of po- lack of randomized, prospective data III/IV case-control or cohort studies in- tential use in necrotizing fasciitis caused regarding the utility of HBO therapy, clude references 8, 23, 25, 27, 29, 34, by group A streptococcus. The typical currently it is best characterized as 35, 46, and 48-51. Reference 10 is a dosage is 1 to 2 g/kg of body weight per an adjunct therapy of potential bene- level V (expert opinion) report. The re- 45 day for 2 days. Some studies have fit that should never delay or inter- mainder are review articles, case re- reported decreased mortality with fere with surgical intervention. ports, and textbook chapters. the use of IVIG in patients with Citation numbers printed in bold type streptococcal toxic shock syndrome, indicate references published within the but a recent review found no evi- Summary dence that it improves clinical out- past 5 years. 46 comes in necrotizing fasciitis. Necrotizing fasciitis is an uncommon 1. Stevens DL: The flesh-eating bacterium: HBO therapy has been proved to infection of the superficial fascia and What’s next? J Infect Dis 1999; 179(suppl 2):S366-S374. be efficacious for clostridial gan- surrounding tissue. It is typically the 47 result of a mixed infection with viru- 2. Hippocrates: On Regimen in Acute grene, but its use in necrotizing fas- Diseases. Adams F, trans. Available at: ciitis is supported only by anecdotal lent, toxin-producing bacteria, or http://greektexts.com/library/ and retrospective reports. Bacterial group A Streptococcus alone. MRSA Hippocrates/On_Regimen_In_ Acute_Diseases/eng/index.html. infection is associated with reduced appears to be emerging as an addi- Accessed December 8, 2008. tissue oxygen tension, and HBO may tional microbiologic factor. Early di- 3. Wong CH, Chang HW, Pasupathy S, be able to reverse several of the agnosis is essential because the dis- Khin LW, Tan JL, Low CO: Necrotizing pathophysiologic mechanisms poten- ease can rapidly progress to severe fasciitis: Clinical presentation, microbiology, and determinants of tiating the necrotizing process. It is systemic toxicity and shock, organ mortality. J Bone Joint Surg Am 2003; thought that reestablishment of a failure, or acute respiratory distress 85:1454-1460. normal or elevated partial oxygen syndrome. The single most effective 4. Ashurst J: The Principles and Practice of pressure by administration of HBO management tool is a high degree Surgery. Philadelphia, PA: HC Lea Publishing, 1871, p 390. can terminate the vicious cycle of in- of clinical suspicion followed by fection, ischemia, and reduced host prompt surgical débridement of all 5. Fournier JA: Gangrene foudroyante de la verge. Semaine Medicale 1883;3:345-348. defense mechanisms. HBO therapy necrotic tissue, with repeat débride- ment as needed. Broad-spectrum an- 6. Meleney FL: Hemolytic streptococcus may also be helpful in identifying in- gangrene. Arch Surg 1924;9:317-364. fection boundaries, potentially re- tibiotic therapy accompanies surgical 48 7. Carter PS, Banwell PE: Necrotising ducing unnecessary débridement. management, and clindamycin is sig- fasciitis: A new management algorithm Standard therapy is 2.0 to 2.5 atm nificantly beneficial when strepto- based on clinical classification. Int for 90 to 120 minutes twice daily un- coccal or staphylococcal species are Wound J 2004;1:189-198. til infection progression is halted. involved. Patients with necrotizing 8. Dufel S, Martino M: Simple cellulitis or a more serious infection? J Fam Pract Reported outcomes vary greatly. Es- fasciitis are often critically ill, and in- 2006;55:396-400. cobar et al49 found an 11-fold greater tensive supportive care is necessary. 9. Bisno AL, Stevens DL: Streptococcal chance of survival with HBO treat- Adjuvant therapies may be benefi- infections of skin and soft tissues. N ment and a decrease in morbidity cial, but further study is needed to Engl J Med 1996;334:240-245. with fewer amputations in a retro- prove their efficacy. 10. Mulla ZD: Treatment options in the spective study of 42 patients. Rise- management of necrotising fasciitis caused by group A streptococcus. Expert man et al50 reported that HBO use Acknowledgment Opin Pharmacother 2004;5:1695-1700. reduced mortality from 66% to 23% 11. Levine EG, Manders SM: Life- in a retrospective study of 29 pa- The authors would like to thank threatening necrotizing fasciitis. Clin 51 Dermatol 2005;23:144-147. tients. Conversely, Brown et al con- Jeffrey S. Ross, MD, for his generous cluded that there was no significant 12. Sudarsky LA, Laschinger JC, Coppa GF, help preparing the pathology slides. Spencer FC: Improved results from a reduction in mortality rate associated standardized approach in treating with HBO therapy among 54 pa- patients with necrotizing fasciitis. Ann tients with necrotizing fasciitis. References Surg 1987;206:661-665. HBO therapy carries risks. Among 13. Childers BJ, Potyondy LD, Nachreiner R, et al: Necrotizing fasciitis: A the potential complications are claus- Evidence-based Medicine: Level I/II ran- fourteen-year retrospective study of 163 trophobia, tympanic membrane rup- domized, prospective studies include consecutive patients. Am Surg

March 2009, Vol 17, No 3 181 Necrotizing Fasciitis

2002;68:109-116. 2000;191:227-231. 40. Akhtar S, Hasham S, Abela C, Phipps A: The use of Integra in necrotizing fasciitis. 14. Lamagni TL, Neal S, Keshishian C, et al: 27. Wong CH, Khin LW, Heng KS, Tan KC, Burns 2006;32:251-254. Severe Streptococcus pyogenes Low CO: The LRINEC (Laboratory Risk infections, United Kingdom, 2003-2004. Indicator for Necrotizing Fasciitis) score: 41. Young LM, Price CS: Community- Emerg Infect Dis 2008;14:202-209. A tool for distinguishing necrotizing acquired methicillin-resistant 15. Sharkawy A, Low D, Saginur R, et al: fasciitis from other soft tissue infections. Staphylococcus aureus emerging as Severe group A streptococcal soft-tissue Crit Care Med 2004;32:1535-1541. important cause of necrotizing fasciitis. Surg Infect (Larchmt) 2008;9:469-474. infections in Ontario: 1992-1996. Clin 28. Anaya DA, Dellinger EP: Necrotizing Infect Dis 2002;34:454-460. soft-tissue infection: Diagnosis and 42. Mulla ZD, Leaverton PE, Wiersma ST: 16. Geusens E, Pans S, Van Breuseghem I, management. Clin Infect Dis 2007;44: Invasive group A streptococcal infections Knockaert D: Necrotizing fasciitis of the 705-710. in Florida. South Med J 2003;96:968- leg presenting with chest wall 973. emphysema. Eur J Emerg Med 2004;11: 29. Fugitt JB, Puckett ML, Quigley MM, 49-51. Kerr SM: Necrotizing fasciitis. 43. Schroeder JL, Steinke EE: Necrotizing Radiographics 2004;24:1472-1476. fasciitis: The importance of early 17. Cox NH: Streptococcal necrotizing diagnosis and debridement. AORN J fasciitis and the dermatologist. Br J 30. Ogilvie CM, Miclau T: Necrotizing soft 2005;82:1031-1040. Dermatol 1999;141:613-614. tissue infections of the extremities and back. Clin Orthop Relat Res 2006;447: 44. Purnell D, Hazlett T, Alexander SL: A 18. Simonart T: Group a beta-hemolytic 179-186. new weapon against severe sepsis related streptococcal necrotising fasciitis: Early to necrotizing fasciitis. Dimens Crit Care diagnosis and clinical features. 31. Liu YM, Chi CY, Ho MW, et al: Nurs 2004;23:18-23. Dermatology 2004;208:5-9. Microbiology and factors affecting mortality in necrotizing fasciitis. 45. Fontes RA Jr, Ogilvie CM, Miclau T: 19. Aronoff DM, Bloch KC: Assessing the J Microbiol Immunol Infect 2005;38: Necrotizing soft-tissue infections. JAm relationship between the use of 430-435. Acad Orthop Surg 2000;8:151-158. nonsteroidal anti-inflammatory drugs and necrotizing fasciitis caused by group 32. Golger A, Ching S, Goldsmith CH, 46. Darabi K, Abdel-Wahab O, Dzik WH: A streptococcus. Medicine (Baltimore) Pennie RA, Bain JR: Mortality in Current usage of intravenous immune 2003;82:225-235. patients with necrotizing fasciitis. Plast globulin and the rationale behind it: The Reconstr Surg 2007;119:1803-1807. 20. Green RJ, Dafoe DC, Raffin TA: Massachusetts General Hospital data Necrotizing fasciitis. Chest 1996;110: 33. Muqim R: Necrotizing fasciitis: and a review of the literature. 219-229. Management and outcome. J Coll Transfusion 2006;46:741-753. Physicians Surg Pak 2003;13:711-714. 21. Kihiczak GG, Schwartz RA, Kapila R: 47. Majeski J, Majeski E: Necrotizing Necrotizing fasciitis: A deadly infection. 34. Miller LG, Perdreau-Remington F, Rieg fasciitis: Improved survival with early J Eur Acad Dermatol Venereol 2006;20: G, et al: Necrotizing fasciitis caused by recognition by tissue biopsy and 365-369. community-associated methicillin- aggressive surgical treatment. South Med resistant Staphylococcus aureus in Los J 1997;90:1065-1068. 22. Kaul R, McGeer A, Low DE, Green K, Angeles. N Engl J Med 2005;352:1445- Schwartz B: Population-based 1453. 48. Shupak A, Shoshani O, Goldenberg I, surveillance for group A streptococcal Barzilai A, Moskuna R, Bursztein S: necrotizing fasciitis: Clinical features, 35. Kalorin CM, Tobin EH: Community Necrotizing fasciitis: An indication for prognostic indicators, and associated methicillin resistant hyperbaric oxygenation therapy? Surgery microbiological analysis of seventy-seven Staphylococcus aureus causing Fournier’s 1995;118:873-878. cases. Ontario Group A Streptococcal gangrene and genital infections. J Urol Study. Am J Med 1997;103:18-24. 2007;177:967-971. 49. Escobar SJ, Slade JB Jr, Hunt TK, Cianci P: Adjuvant hyperbaric oxygen therapy 23. Rodriguez RM, Abdullah R, Miller R, 36. Currie BJ: Group A streptococcal et al: A pilot study of cytokine levels and infections of the skin: Molecular (HBO2) for treatment of necrotizing white blood cell counts in the diagnosis advances but limited therapeutic fasciitis reduces mortality and of necrotizing fasciitis. Am J Emerg Med progress. Curr Opin Infect Dis 2006;19: amputation rate. Undersea Hyperb Med 2006;24:58-61. 132-138. 2005;32:437-443.

24. Young MH, Aronoff DM, Engleberg 37. Everest E: Group A streptococcal 50. Riseman JA, Zamboni WA, Curtis A, NC: Necrotizing fasciitis: Pathogenesis fasciitis. Crit Care Resusc 1999;1:63-68. Graham DR, Konrad HR, Ross DS: and treatment. Expert Rev Anti Infect Hyperbaric oxygen therapy for Ther 2005;3:279-294. 38. Ozalay M, Ozkoc G, Akpinar S, Hersekli necrotizing fasciitis reduces mortality MA, Tandogan RN: Necrotizing soft- and the need for debridements. Surgery 25. Wysoki MG, Santora TA, Shah RM, tissue infection of a limb: Clinical 1990;108:847-850. Friedman AC: Necrotizing fasciitis: CT presentation and factors related to characteristics. Radiology 1997;203:859- mortality. Foot Ankle Int 2006;27:598- 51. Brown DR, Davis NL, Lepawsky M, 863. 605. Cunningham J, Kortbeek J: A multicenter review of the treatment of 26. Wall DB, Klein SR, Black S, de Virgilio 39. Bingöl-Kolog˘lu M, Yildiz RV, Alper B, major truncal necrotizing infections with C: A simple model to help distinguish et al: Necrotizing fasciitis in children: and without hyperbaric oxygen therapy. necrotizing fasciitis from nonnecrotizing Diagnostic and therapeutic aspects. Am J Surg 1994;167:485-489. soft tissue infection. J Am Coll Surg J Pediatr Surg 2007;42:1892-1897.

182 Journal of the American Academy of Orthopaedic Surgeons Review

Global emm type distribution of group A streptococci: systematic review and implications for vaccine development

Andrew C Steer, Irwin Law, Laisiana Matatolu, Bernard W Beall, Jonathan R Carapetis emm sequence typing is the most widely used method for defi ning group A streptococcal (GAS) strains, and has been Lancet Infect Dis 2009; 9: 611–16 applied to isolates in all regions of the world. We did a systematic review of the global distribution of GAS emm types. Centre for International Child 102 articles and reports were included (38 081 isolates). Epidemiological data from high-income countries were Health, University of predominant, with sparse data from low-income countries. The epidemiology of GAS disease in Africa and the Pacifi c Melbourne, Australia (A C Steer FRACP, I Law MAE); region seems to be diff erent from that in other regions, particularly high-income countries. In Africa and the Pacifi c, there Fiji Group A Streptococcal were no dominant emm types, a higher diversity of emm types, and many of the common emm types in other parts of the Project, Ministry of Health, world were less common (including emm1, 4, 6, and 12). Our data have implications for the development of GAS vaccines. Suva, Fiji (L Matatolu BSc); On the basis of the available data, the current formulation of the experimental multivalent emm vaccine would provide Streptococcus Laboratory, Respiratory Diseases Branch, good coverage in high-income countries, particularly USA, Canada, and Europe, but poor coverage in Africa and the Centers for Disease Control and Pacifi c, and only average coverage in Asia and the Middle East. Prevention, Atlanta, GA, USA (B W Beall PhD); and Menzies School of Health Research and Introduction if they were associated with rheumatic fever in classic Charles Darwin University, 13 Group A streptococcal (GAS) infections are a major cause studies from the USA in the mid-20th century. Recent Darwin, Australia of morbidity and mortality worldwide.1 Streptococcus studies of GAS disease in North America found that emm (J R Carapetis PhD) pyogenes causes a wide range of clinical disease. In high- types in the 26-valent vaccine accounted for 79% of all Correspondence to: income countries, pharyngitis and invasive disease are the invasive isolates from ten sites in the USA between 2000 Andrew Steer, Centre for GAS diseases of greatest public health importance, whereas and 2004, and 85% of pharyngitis isolates from 13 sites in International Child Health, University of Melbourne, 7,14 in low-income countries, acute rheumatic fever, rheumatic the USA and Canada between 2000 and 2007. Department of Paediatrics, heart disease, invasive disease, and acute post-streptococcal Diff erences in the distribution of emm sequence types Flemington Road, glomerulonephritis are the major severe diseases, with between global regions have been noted previously,7,15 but a Parkville, VIC 3052, Australia endemic streptococcal impetigo also leading to very high thorough review of the available global data has not been [email protected] morbidity. On a global scale, the overwhelming burden of undertaken, as it has for other bacteria, including GAS disease is found in low-income countries, where more Streptococcus pneumoniae.16,17 In addition, a review of the than 95% of the estimated 663 000 cases of invasive GAS coverage and potential impact of the experimental disease and more than 95% of the estimated 294 000 deaths multivalent GAS vaccine on a global scale, particularly in due to rheumatic heart disease occur. However, accurate low-income countries where the burden of disease is data are not available from most low-income countries, and greatest, has not been undertaken. Therefore, we did a published summary data are likely to be under estimates.1 global review of the distribution of emm types of GAS and Due to the size and severity of the burden of GAS disease, assessed the implications of our fi ndings for the epidemiological surveillance has been crucial to detect development of GAS vaccines. changes in disease distribution in various populations. An important part of epidemiological surveillance for GAS Methods disease has been the typing of collected bacterial isolates. Data sources Several diff erent methods are available to type GAS.2 Typing We searched for studies that described the epidemiology of based on the M protein, a cell-surface protein that is the GAS based on emm or M typing by use of a systematic major virulence and immunological determinant of GAS, approach that complied with the QUORUM guidelines.18 has been the most widely used method.3–5 Classic M-protein Figure 1 summarises our approach. Searches were done in serological typing was largely replaced by sequence typing Medline and EmBase from the start of 1990 to the end of of the 5´ end of the M protein (emm) gene in the late 1990s.6 March, 2009, by use of the search term “Streptococcus Large epidemiological studies of pharyngitis and invasive pyogenes” combined with the search terms “epidemiology”, disease have been done using emm sequence typing, “emm”, and “streptococcal M protein”. Relevant abstracts particularly in the USA, Canada, and Europe.7–10 Population- from the Lancefi eld Symposia on Streptococci and based studies using emm typing have also been done in Streptococcal Infections held between 1990 and 2008 were many other countries. also reviewed. No language restrictions were used in the Available molecular epidemiological data have informed initial search. the development of GAS vaccine candidates. Several vaccine candidates have shown promise; however, only one Study eligibility and quality assessment vaccine, a 26-valent M-protein-based vaccine, has recently All abstracts from the initial search were checked for reached clinical trials.11,12 Serotypes for this vaccine were relevance (by ACS). Studies that were clearly not population chosen if they were known to be common causes of invasive based or studies that did not report emm or M typing were GAS disease or uncomplicated pharyngitis in the USA, or excluded. Review articles and studies not dealing with www.thelancet.com/infection Vol 9 October 2009 611 Review

“other” in referenced articles were removed from all 2828 potentially relevant articles identified analyses (n=1663). from abstracts 1938 Medline 609 EMBASE Classifi cation by region, clinical specimen type, and time 281 conference proceedings period Studies were categorised by region based on United Nations Populations Prospects classifi cations,21 with some 2486 excluded 892 not population based modifi cations. The fi rst modifi cation was that the defi nition 870 not emm or M typing of high-income nations (Europe, North America, Australia, 345 not human studies 204 duplicate papers New Zealand, and Japan) was expanded to include Hong 175 review articles Kong.1 The second was that Indigenous Australians were included in the Pacifi c region because of documented epidemiological similarities between this group and Pacifi c 342 potentially relevant articles selected (full 1 articles reviewed or authors contacted) Islanders, particularly related to GAS infections. Isolates were categorised into one of four clinical disease sites/states: invasive, pharyngeal, skin, and other. “Invasive” 245 excluded 119 not population based refers to all isolates that caused invasive disease in the 71 duplicates referenced article. “Pharyngeal” refers to all isolates 30 not emm or M typing specifi ed as causing pharyngitis or throat carriage, as well 20 full data not available 5 review articles as pharyngeal isolates that were not clearly designated. “Skin” refers to all isolates that caused impetigo or other skin infection not otherwise specifi ed in the referenced 97 articles included in Review article. “Other” refers to isolates taken from patients with acute rheumatic fever, acute post-streptococcal glomer- 5 additional datasets included ulonephritis, scarlet fever, and tic disorders. Isolates of a 2 articles had 2 datasets clinical type not clearly specifi ed in the referenced article 3 additional datasets provided from personal communication were classifi ed as “not diff erentiated”. These “other” and “not diff erentiated” isolates were included in the overall and regional analyses. For disease-specifi c analyses, only 102 datasets included in Review isolates that were clearly designated as being invasive, pharyngeal, or skin isolates were included. Figure 1: Summary of selection process and reasons for exclusion of studies Studies were also categorised into two time periods (1990–99 and 2000–09) for analysis over time. Studies that human beings were excluded. After abstracts had been spanned both these time periods were not included in this screened, full papers were retrieved for all remaining analysis. articles. Studies could be of retrospective or prospective design, but had to be representative of the population in Statistical analysis which they were based. Therefore, articles that dealt with We compared the distribution of emm types between global outbreaks of disease and articles that reported emm typing region by ranking them as a percentage of the total number of antibiotic-resistant isolates only were excluded. Articles of isolates for the specifi ed region. We also compared the that contained overlapping data already contained in other distribution of emm types by clinical specimen site and articles or reports were excluded. In the event of incomplete time period within regions by use of the same technique. data in any of the fi nal set of articles, we contacted the Simpson’s index of diversity was used to measure the investigators directly to obtain a complete dataset if possible; variation of the number of emm types within a region.22 The the dataset was only included if it was complete. higher the index, the greater the probability that any two randomly selected isolates from the same population will Sequence typing be of diff erent emm types (ie, the greater the diversity of We updated emm sequence type information provided emm types). 95% CIs for the index of diversity were in some of the older studies by use of the Centers for calculated as previously described.23 Disease Control and Prevention emm database, because Vaccine coverage was defi ned as the proportion of all designation of emm sequence types has been continually isolates in the region or the clinical disease state that were updated over time (eg, the designation st3365 has been covered by the 26-valent M-protein-based GAS vaccine replaced by the designation emm119).19 We used currently under clinical investigation (emm types 1, 2, 3, 5, published data to infer emm types from M types.20 6, 11, 12, 14, 18, 19, 22, 24, 28, 29, 33, 43, 59, 75, 76, 77, 89, 92, Isolates that were not able to be emm typed in the 94, 101, 114).12 95% CI estimation of vaccine coverage was referenced articles were included in our study and determined by use of robust SEs that assumed individual classifi ed as “nontypable”, whereas isolates reported as observations were independent between studies. Data were

612 www.thelancet.com/infection Vol 9 October 2009 Review

entered into Excel (Microsoft, Redmond, WA, USA) and subsequent analysis was done with Stata version 10.1 A (StataCorp, College Station, TX, USA). 25 Included in experimental vaccine 100 Not included in vaccine Results Cumulative proportion The fi nal database contained 102 datasets, contributing 20 80 38 081 isolates (webappendix). These isolates were not equally distributed between regions, with high-income Cumulative (%) 15 60 countries contributing 32 143 isolates (84·4%), Asia contributing 2248 isolates (5·9%), Pacifi c Island countries and Indigenous Australians contributing 1383 isolates 10 40 (3·6%), the Middle East contributing 1219 isolates (3·2%), Proportion of isolates (%) Proportion Latin America contributing 757 isolates (2·0%), and Africa contributing 331 isolates (0·9%). 31 334 isolates were 5 20 designated as being invasive, pharyngeal, or skin isolates, and of these, 17 173 (54·8%) were invasive isolates, 11 953 0 0 (38·2%) were pharyngeal isolates, and 2208 (7·0%) were emmemm1 12 emm3emm4 emmemm6 75emm77emmemm2 11 emm5 emm73 skin isolates. Of the 11 953 pharyngeal isolates, 9258 (77·5%) emm28 emm89 emm22emm81emm87 emm49emm53emm58emm44emm82 emm83emm18emm92 Nontypeable were collected from cases of pharyngitis. B A total of 205 emm types were recorded, including the 25 100 category of nontypable isolates. The most common emm type was emm1, which accounted for 18·3% of all isolates in the study, followed by emm12 (11·1%), emm28 (8·5%), 20 80 emm3 (6·9%), and emm4 (6·9%). However, there were signifi cant diff erences in emm type distribution by region Cumulative (%) 15 60 and by clinical disease state.

Distribution by region 10 40 There were obvious similarities in emm type distribution between the high-income countries, Asia, the middle of isolates (%) Proportion east, and Latin America, by contrast with the distribution 5 20 in Africa and the Pacifi c region. Figure 2 compares the proportions of the 25 most common emm types in high- 0 0 income countries, Africa, and the Pacifi c (further data are st62 st463 emm12emm75 emm74emm1emm3 emm9emm5st1731emm71st3757 emm77 emm8 available from the authors).24 In high-income countries, emm25emm18emm28 emm22 emm99 emm81emmemm76 103emmemm44 114emm43 25 emm types accounted for 90·3% of all isolates, and 146 types contributed the remaining 9·7%. In Africa, C 26 emm types accounted for 62·5% of all isolates, and 25 100 65 contributed the remaining 37·5% of isolates. 26 emm types accounted for 61·8% of all isolates from the Pacifi c 20 80 region, and 74 types contributed the remaining 38·2% of

isolates. emm1 and emm12 were the two most common Cumulative (%) emm types in high-income countries, Asia, and Latin 15 60 America, and the second and third most common emm types in the Middle East, accounting for between 26·1% 10 40 and 40·0% of all isolates in these regions. By contrast, emm1 was ranked fi fth in Africa and 13th in the Pacifi c of isolates (%) Proportion region, accounting for 3·6% and 2·0% of isolates, 5 20 respectively. emm12 did not appear in the 25 most common sequence types in the Pacifi c region. Other than 0 0 emm1 and emm12, there were several other common emm emm1 emm55 emm11emm70emm33emm25emm44M53/80 emm76emm89 emm71emm22emm93emm52emm91emm73emm69 emm56emm58emm78emm14 types found in high-income countries, Asia, the middle emm103 emm100 stNS1033 east, and Latin America; these included emm4 and emm6, Nontypeable emm types which were ranked among the eight most common emm types in all four regions. By contrast, emm4 was ranked Figure 2: 25 most common emm types as proportions of all isolates in high-income countries (A), Africa (B), 32nd in the Pacifi c region and was not reported in any of and the Pacifi c region (C) the African studies included in the database, whereas In Africa emm112 and in the Pacifi c region emm74 were equal 25th, but are not included. www.thelancet.com/infection Vol 9 October 2009 613 Review

See Online for webappendix Distribution over time Simpson’s index of diversity22 (% [95% CI]) The period 1990–99 included 40 unique studies, as did Africa 98·1% (97·7–98·5) 2000–09. 22 unique studies were excluded from the analysis Asia 88·7% (88·0–89·4) because they overlapped the time periods. Seven of the ten Latin America 93·2% (92·4–94·1) most common emm types found in high-income countries Middle East 93·0% (92·4–93·5) during 1990–99 were also among the ten most common Pacifi c region 97·9% (97·7–98·1) during 2000–09, with remarkably few changes in the order High-income countries 92·1% (92·0–92·3) of prevalence. There was less overlap in Asia (four in the Combined 92·8% (92·7–92·9) ten most common in both periods), Latin America (four), Table 1: Diversity of emm types by global region Africa (three), and the Pacifi c (one). No data were available for the middle east in the period 2000–09.

Invasive Pharyngeal Skin All Diversity Africa ·· 43·7% (33·2–54·1) 21·3%* 39·0% (27·2–50·7) There was greater diversity of emm types in Africa and the Asia 65·8% (47·4–84·1) 52·5% (39·3–65·7) 41·1% (37·2–44·9) 60·5% (50·8–70·1) Pacifi c than in other regions, as indicated by the smaller Latin America 77·7% (62·6–92·8) 73·7% (49·6–97·8) 27·0% (19·6–34·4) 71·9% (46·8–97·0) proportions of isolates accounting for the fi rst 25 emm types Middle East 67·5% (57·3–77·7) 54·7% (26·6–82·8) ·· 63·2% (50·5–76·0) in these regions, and by the straighter cumulative curve Pacifi c 32·1% (14·4–49·8) 30·8% (26·9–34·6) 19·3% (10·4–28·2) 23·9% (17·8–29·9) (fi gure 2). A few emm types accounted for more than 50% High-income countries 74·9% (71·1–78·7) 77·8% (68·1–87·5) 36·8% (14·5–59·0) 72·8% (66·4–79·2) of isolates in Asia, Latin America, the Middle East, and Combined 74·2% (70·5–77·9) 73·5% (61·4–85·5) 30·6% (18·3–42·8) 69·7% (63·1–76·2) high-income countries: in Asia, three emm types accounted for 54·0% of isolates; in Latin America, six emm types Data are vaccine coverage [% (95% CI)]. *95% CI not calculated because data are from one study only. ··=Data not available. accounted for 54·2% of isolates; in the Middle East, fi ve Table 2: Vaccine coverage of isolates by region and disease emm types accounted for 53·8% of isolates; and in high- income countries, fi ve emm types accounted for 54·5% of the isolates. By contrast, 18 emm types accounted for 50·4% emm6 was ranked 29th in Africa and was not reported in of isolates in Africa and 19 emm types accounted for 51·3% any of the Pacifi c studies. of isolates in the Pacifi c region. These diff erences in diversity were also indicated by calculation of Simpson’s Distribution by clinical specimen site index (table 1): the index was higher in Africa and the No data were available for invasive disease in Africa nor Pacifi c compared with the other regions, and this diff erence for skin disease in the Middle East. There was was signifi cant on the basis of the 95% CIs.23 substantial overlap in common emm types found in invasive and pharyngeal isolates both within and Vaccine coverage between the Asian, Latin American, Middle-East, and Table 2 summarises theoretical coverage of isolates by high-income countries (further data are available from region and disease by the experimental multivalent the authors).24 However, there was less overlap when vaccine.12 Although there was some heterogeneity between common emm types found in skin isolates were studies within regions, there were clear regional patterns of compared with common emm types found in pharyngeal vaccine coverage. Overall, the 26 emm types in the and invasive isolates in these regions. For example, in experimental multivalent vaccine accounted for less than high-income countries, among the ten most common 65% of all isolates in four out of the six regions (Africa, emm types found in pharyngeal, invasive, and skin Asia, Middle East, Pacifi c region), and was particularly low isolates, eight emm types were shared between invasive in Africa and the Pacifi c region. Of the disease states, and pharyngeal isolates, whereas fi ve emm types were coverage by the vaccine was best for invasive disease and shared between skin and pharyngeal isolates, and fi ve worst for skin disease. between skin and pharyngeal isolates. In Asia, there were fi ve emm types shared between the ten most Discussion common emm types found in invasive and pharyngeal Our study has revealed diff erences in the emm type isolates, but only one type shared between skin and distribution of GAS across global regions, and in particular invasive isolates, and three types shared between skin has revealed marked diff erences in the molecular and pharyngeal isolates. In the Pacifi c region, there epidemiology in Africa and the Pacifi c region compared were four emm types shared among the ten most with high-income countries. A distinct profi le of emm types common emm types found in invasive and pharyngeal exists in Africa and the Pacifi c, with an apparent lack of isolates, two types shared between skin and invasive dominant emm types and a greater molecular diversity. isolates, and two types shared between skin and The reasons for the contrasting molecular epidemiology pharyngeal isolates. In Africa, two emm types were in Africa and the Pacifi c are not clear. It might be that shared among the ten most common types found in diff ering clinical presentations of GAS infection in these skin and pharyngeal isolates. regions contribute to diff ering emm type profi les. GAS

614 www.thelancet.com/infection Vol 9 October 2009 Review

impetigo is endemic in many parts of Africa and the Pacifi c, investigation, including C5a peptidase, GAS carbohydrate, but is much less common in high-income countries.25–28 and a peptide within the conserved region of the M protein Search strategy and The high burden of impetigo in tropical countries is known as J8.34–36 selection criteria accompanied by large numbers of circulating GAS of The main limitations to our study were the degree of These are described multiple emm types that are readily transmitted.25,29,30 Studies heterogeneity within some of the regions, the predominance in detail in the of site tropism for GAS in tropical countries have found of data from high-income countries, and the potential for Methods section. that most circulating emm types of GAS are of emm pattern selection bias related to time periods in which data were D (skin tropism) or E (both skin and pharyngeal tropism), collected. Caution needs to be exercised in interpreting the as opposed to temperate regions where there are more quantitative comparisons presented in our Review. strains of emm pattern A–C (pharyngeal tropism).15 Rapid Although we attempted to choose population-representative transmission, negative selection pressure due to a studies, there were variations in the methods used by the seemingly weak immune response to GAS in the skin,31 studies included. Therefore, formal statistical tests of and a lack of eff ective public-health control measures all between-study diff erences were avoided. An indication of support the notion that skin emm types dominate the study heterogeneity can be found by comparing vaccine epidemiological scene in many tropical settings. coverage proportions and 95% CIs within regions Our results have implications for vaccine development. (webappendix). Few data are available from regions that The M protein is highly immunopotent, and sera containing carry the highest burden of GAS disease (ie, Africa and the serotype-specifi c antibodies were shown to be protective Pacifi c).1 The relatively small number of studies from some against re-infection with the same serotype in studies done regions, particularly Africa, is a potential source of bias and in the 1950s.4,32 A 26-valent M protein vaccine has been our fi ndings might not be generalisable to all countries in found to be safe and immunogenic in human beings.12 The these regions. However, the data reviewed are further available data suggest that the current formulation would evidence of the need for better research into GAS disease in cover most disease-causing serotypes in high-income low-income countries, on which we have previously countries, particularly in USA, Canada, and Europe. reported.37,38 Another limitation of our data is that emm However, our results suggest the current formulation of types are not always adequate strain markers, because they this vaccine would provide limited coverage of disease- can be shared by unrelated clonal types.39 Assessment of causing GAS emm types in Africa and the Pacifi c, and only M-type-specifi c vaccines against the known array of clones mid-level coverage in Asia and the Middle East. Alternate that exist within given emm types is thus potentially formulations with diff erent emm type profi les would be important, particularly because changes can also occur needed for diff erent regions, and because of the even within a single clone that is represented by a single emm distribution of emm types in Africa and the Pacifi c region, a type. Such a scenario is underscored by relatively subtle larger number of emm types would need to be included in chromosomal changes that have occurred within the highly the vaccine in these regions to obtain coverage against a prevalent emm1 clone.40,41 Finally, a further limitation of our similar proportion of isolates to that in other regions. A study was that the use of only one reviewer to screen and further barrier to multivalent M-protein vaccines is the select the eligible studies might have been a source of potential for rapid emergence of new emm types. Our study selection bias. found apparently diff erent emm profi les over time in most In conclusion, our study underscores the need for further regions, although with a more stable profi le in high-income molecular epidemiological data from regions where the countries. Changes in the predominant circulating emm burden of GAS disease is greatest. However, on the basis of types within relatively short time periods has been shown to the available data from such regions, the molecular occur in individual regions in the USA,33 although this was epidemiology of GAS infections in Africa and the Pacifi c explained by the replacement of common emm types with seems to be diff erent from that in other regions, such as other common types from within other regions in the USA, high-income countries. This might be related to the high all of which are included in the 26-valent vaccine. Therefore, prevalence of GAS impetigo in regions with large numbers introduction of a multivalent M-specifi c GAS vaccine in the of circulating strains. Although the current formulation of USA would seem to be a sound strategy, although close the experimental multivalent vaccine provides good surveillance of prevalent GAS emm types within a population coverage of disease-causing emm types in most high- would be warranted for an extended period. However, there income countries, this vaccine would provide poor coverage is potential for rapid and dramatic turnover in regions in Africa and the Pacifi c, and only average coverage in Asia where there are high numbers of circulating GAS strains and the Middle East. 25 (ie, in areas where pyoderma caused by GAS is endemic). Contributors Diff erent approaches to the selection of vaccine antigens, ACS was the primary coordinator of data collection, analysis, and writing. including investigation of vaccines containing conserved IL was primarily involved in data analysis and writing. LM was primarily epitopes, might be more appropriate in low-income involved in data collection and writing. BB and JRC supervised data collection, analysis, and writing. All authors contributed substantially to the countries, particularly in Africa and the Pacifi c region, preparation of the paper. possibly in combination with selected serotype-specifi c Confl icts of interest 31,34–36 antigens. Potential conserved antigens are under There are no relevant confl icts of interest to disclose for any author. www.thelancet.com/infection Vol 9 October 2009 615 Review

Acknowledgments 20 Facklam RF, Martin DR, Lovgren M, et al. Extension of the Lancefi eld The following investigators provided additional emm typing data: classifi cation for group A streptococci by addition of 22 new M protein Robert Tanz, Northwestern University, Chicago, IL, USA; gene sequence types from clinical isolates: emm103 to emm124. Chris Van Beneden, Centers for Disease Control and Prevention, Atlanta, Clin Infect Dis 2002; 34: 28–38. GA, USA; Marc Levy, Centre Hospitalier de Polynésie Française, Papeete, 21 United Nations Population Division. World Population Prospects: The Tahiti; Rajesh Kumar, Post Graduate Institute of Medical Education and 2008 revision population database. Defi nition of major areas and Research, Chandigarh, India; Tuula Siljander, National Public Health regions. http://esa.un.org/unpp/index.asp?panel=5 Institute, Mannerheimintie, Finland; Tadayoshi Ikebe, National Institute of (accessed Aug 12, 2009). Infectious Diseases, Tokyo, Japan; Guliz Erdem, John A Burns School of 22 Simpson E. Measurement of diversity. Nature 1949; 163: 688. Medicine, University of Hawaii, Manoa, Hawaii, USA; Graham Magor, 23 Grundmann H, Hori S, Tanner G. Determining confi dence intervals Queensland Institute of Medical Research, Brisbane, Australia; and when measuring genetic diversity and the discriminatory abilities of Kerry-Ann O’Grady and Leisha Richardson, Menzies School of Health typing methods for microorganisms. J Clin Microbiol 2001; 39: 4190–92. Research, Darwin, NT, Australia. 24 Centers for Disease Control and Prevention Streptococcus Laboratory. References emm types as proportions of total disease isolates in six global regions. 1 Carapetis JR, Steer AC, Mulholland EK, Weber M. The global burden http://www.cdc.gov/ncidod/biotech/strep/emmtype_proportions.htm of group A streptococcal diseases. Lancet Infect Dis 2005; 5: 685–94. (accessed Aug 11, 2009). 2 Johnson DR, Kaplan EL, Sramek J, Bicova R, Havlicek J, Havlickova H. 25 Carapetis JR, Currie BJ, Hibble M, Sriprakash KS, Bessen DE, Laboratory diagnosis of group A streptococcal infections. Geneva: Mathews JD. Rapid turnover of multiple strains of group A WHO, 1996. streptococcus in an Australian Aboriginal community. In: Martin DR, 3 Lancefi eld RC. The antigenic complex of Streptococcus hemolyticus, I: Tagg JR, eds. Streptococci and streptococcal diseases: entering the new demonstration of a type-specifi c substance in extracts of Streptococcus millennium. Auckland: SecuraCopy, 2000: 155–58. hemolyticus. J Exp Med 1928; 47: 9–10. 26 Carapetis JR, Currie BJ, Matthews JD. Multiple strains of Streptococcus 4 Lancefi eld RC. Current knowledge of the type specifi c M antigens of pyogenes in skin sores of Australian Aborigines. J Clin Microbiol 1995; group A streptococci. J Immunol 1962; 89: 307–13. 33: 1471–72. 5 Beall B, Facklam R, Thompson T. Sequencing emm-specifi c PCR 27 Kristensen JK. Scabies and pyoderma in Lilongwe, Malawi. Prevalence products for routine and accurate typing of group A streptococci. and seasonal fl uctuation. Int J Dermatol 1991; 30: 699–702. J Clin Microbiol 1996; 34: 953–58. 28 Masawe AE, Nsanzumuhire H, Mhalu F. Bacterial skin infections in 6 Facklam R, Beall B, Efstratiou A, et al. emm typing and validation of preschool and school children in coastal Tanzania. Arch Dermatol 1975; provisional M types for group A streptococci. Emerg Infect Dis 1999; 111: 1312–16. 5: 247–53. 29 Ferrieri P, Dajani AS, Wannamaker LW, Chapman SS. Natural history 7 O’Loughlin RE, Roberson A, Cieslak PR, et al. The epidemiology of of impetigo. I. Site sequence of acquisition and familial patterns of invasive group A streptococcal infection and potential vaccine spread of cutaneous streptococci. J Clin Invest 1972; 51: 2851–62. implications: United States, 2000–2004. Clin Infect Dis 2007; 30 Reinstein CR. Epidemic nephritis at Red Lake, Minnesota. J Pediatr 45: 853–62. 1955; 47: 25–34. 8 Lamagni TL, Efstratiou A, Vuopio-Varkila J, Jasir A, Schalen C. The 31 Bessen DE, McGregor KF, Whatmore AM. Relationships between epidemiology of severe Streptococcus pyogenes associated disease in emm and multilocus sequence types within a global collection of Europe. Eurosurveillance 2005; 10: 179–84. Streptococcus pyogenes. BMC Microbiol 2008; 8: 59–71. 9 Shulman ST, Tanz RR, Kabat W, et al. Group A streptococcal 32 Lancefi eld RC. Persistence of type-specifi c antibodies in man pharyngitis serotype surveillance in North America, 2000–2002. following infection with group A streptococci. J Exp Med 1959; Clin Infect Dis 2004; 39: 325–32. 110: 271–92. 10 Tanz RR, Shulman ST, Kabat W, et al. Five-year group A streptococcal 33 Shulman ST, Stollerman G, Beall B, Dale JB, Tanz RR. Temporal pharyngitis serotype surveillance in North America, 2000–2005. changes in streptococcal M protein types and the near-disappearance Proceedings of the XVIth Lancefi eld International Symposium on of acute rheumatic fever in the United States. Clin Infect Dis 2006; Streptococci and Streptococcal Diseases, Palm Cove, Australia; 42: 441–47. Sept 25–29, 2005: 30–33. 34 Sabharwal H, Michon F, Nelson D, et al. Group A streptococcus (GAS) 11 Kotloff KL, Dale JB. Progress in group A streptococcal vaccine carbohydrate as an immunogen for protection against GAS infection. development. Pediatr Infect Dis J 2004; 23: 765–66. J Infect Dis 2006; 193: 129–35. 12 McNeil SA, Halperin SA, Langley JM, et al. Safety and 35 Shet A, Kaplan EL, Johnson DR, Cleary PP. Immune response to immunogenicity of 26-valent group A streptococcus vaccine in healthy group A streptococcal C5a peptidase in children: implications for adult volunteers. Clin Infect Dis 2005; 41: 1114–22. vaccine development. J Infect Dis 2003; 188: 809–17. 13 Hu MC, Walls MA, Stroop SD, Reddish MA, Beall B, Dale JB. 36 Batzloff MR, Hayman WA, Davies MR, Zeng M, Pruksakorn S, Immunogenicity of a 26-valent group A streptococcal vaccine. Brandt ER. Protection against group A streptococcus by immunization Infect Immun 2002; 70: 2171–77. with J8-diptheria toxoid: contribution of J8-and diptheria toxoid- 14 Shulman ST, Tanz RR, Kabat W, et al. Seven year surveillance of specifi c antibodies to protection. J Infect Dis 2003; 187: 1598–608. streptococcal pharyngitis emm types in North America. Proceedings of 37 Carapetis JR. Rheumatic heart disease in developing countries. the XVII Lancefi eld Symposium on Streptococci and Streptococcal N Engl J Med 2007; 357: 439–41. Diseases, Porto Heli, Greece; June 22–26, 2008. Abstract O1.3. 38 Tibazarwa KB, Volmink JA, Bongani MM. The incidence of acute 15 Bessen DE, Carapetis JR, Beall B, et al. Contrasting molecular rheumatic fever in the world: a systematic review of population-based epidemiology of group A streptococci causing tropical and nontropical studies. Heart 2008; 94: 1534–40. infections of the skin and throat. J Infect Dis 2000; 182: 1109–16. 39 Beall B, Gherardi G, Lovgren M, Forwick B, Facklam R, Tyrrell G. emm 16 Hausdorff WP, Bryant J, Kloek C, Paradiso PR, Siber GR. The and sof gene sequence variation in relation to serological typing of contribution of specifi c pneumococcal serogroups to diff erent disease opacity factor positive group A streptococci. Microbiology 2000; manifestations: implications for conjugate vaccine formulation and 146: 1195–209. use, part II. Clin Infect Dis 2000; 30: 122–40. 40 Cleary PP, LaPenta D, Vessela R, Lam H, Cue D. The present day 17 Hausdorff WP, Bryant J, Paradiso PR, Siber GR. Which pneumococcal globally disseminated M1 subclone of group A streptococci diff ers serogroups cause the most invasive disease: implications for conjugate from other subclones by 70 kilobases of prophage DNA and capacity vaccine formulation and use, part I. Clin Infect Dis 2000; 30: 100–21. for high-frequency intracellular invasion. Infect Immun 1998; 18 Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. 66: 5592–97. Improving the quality of reports of meta-analyses of randomised 41 Sumby P, Porcella SF, Madrigal AG, et al. Evolutionary origin and controlled trials: the QUOROM statement. Lancet 1999; 354: 1896–900. emergence of a highly successful clone of serotype M1 group A 19 Centers for Disease Control and Prevention. Streptococcus Laboratory. Streptococcus involved multiple horizontal gene transfer events. Protocol for emm typing. http://www.cdc.gov/ncidod/ J Infect Dis 2005; 192: 771–82. biotech/strep/protocol_emm-type.htm (accessed Aug 11, 2009).

616 www.thelancet.com/infection Vol 9 October 2009 MECHANISMS OF DISEASE

Mechanisms of disease

Streptolysin S and necrotising infections produced by group G streptococcus

Deepali Humar, Vivekananda Datta, Darrin J Bast, Bernard Beall, Joyce C S De Azavedo, Victor Nizet

Summary Introduction Group G streptococci are commonly part of the normal Background We encountered three patients with severe flora of human skin, pharynx, and gastrointestinal tract.1 necrotising soft tissue infections due to -haemolytic group G Human group G streptococcus isolates are subdivided on streptococcus. Due to strong clinical similarities with invasive the basis of colony size and haemolytic phenotype on infections produced by group A streptococcus, we investigated sheep blood agar. Small colony group G streptococcus a potential link of shared -haemolytic phenotype to disease exhibit variable haemolytic reactions and are classified pathogenesis. within the Streptococcus milleri group. Large colony group G streptococcus isolates, now classified as Methods Hybridisation, DNA sequencing, targeted muta- S dysgalactiae subspecies S equisimilis,2 produce robust genesis, and complementation studies were used to establish -haemolysis and are morphologically very similar to the the genetic basis for group G streptococcus -haemolytic prominent pathogen group A streptococcus. activity. The requirement of group G streptococcus Since the mid-1980s, an increase in life-threatening -haemolysin in producing necrotising infection was examined invasive infections produced by group A streptococcus in mice. has been well documented.3,4 Prominent among these syndromes is necrotising fasciitis, a destructive Findings Each patient had an underlying medical condition. -haemolytic group G streptococcus was the sole microbial Panel 1 isolate from debrided necrotic tissue. The group G streptococcus chromosome contained a homologue of the Case 1: A 52-year-old man with type II diabetes mellitus was nine-gene group A streptococcus sag operon encoding the admitted after 6 days of fever and right leg swelling. Oedema -haemolysin streptolysin S (SLS). Targeted mutagenesis of and erythema extended from knee to ankle with tender right the putative SLS structural gene sagA in group G streptococcus inguinal adenopathy. The patient received intravenous eliminated -haemolytic activity. Mice injected subcutaneously cefazolin and clindamycin but remained febrile. He developed with wild-type group A streptococcus or group G streptococcus lymphangitic streaking of his thigh and bullae on his leg. developed an inflammatory lesion with high bacterial counts, Cultures of blood and bullous fluid were negative. On day 3 the marked neutrophil infiltration, and histopathological evidence patient underwent surgical debridement. The fascia of his right of diffuse tissue necrosis. These changes were not found in medial calf was grossly thickened and necrotic, and mice injected with the isogenic group A streptococcus or group histopathology revealed extensive acute necrotising G streptococcus SLS-negative mutants. inflammation and intravascular thrombosis (figure 1). Cultures from the fascial tissue grew group G streptococcus. He was Interpretation In patients with underlying medical conditions, changed to intravenous benzylpenicillin for a 4 week course. -haemolytic group G streptococcus can produce necrotising The patient required two additional surgical debridements but soft tissue infections resembling those produced by group A improved and was discharged to a rehabilitation facility. streptococcus. The -haemolytic phenotype of group G streptococcus is produced by the exotoxin SLS, encoded by a functional homologue of the nine-gene group A streptococcus Panel 2 sag operon. SLS expression contributes to the pathogenesis of streptococcal necrotising soft tissue infection. Case 2: A 59-year-old man with hairy-cell leukaemia and neutropenia presented with fever and left calf pain. On Lancet 2002; 359: 124–29 examination the calf was mildly oedematous but tender to palpation. Full blood count showed white blood cells See Commentary page 93 0·5109/L with absolute neutrophil count 0·04109/L, Department of Medicine, Division of Infectious Diseases, University haemoglobin 125 g/L, and platelets 74109/L. He was of California, San Diego and VA Medical Center, San Diego, CA, admitted and treated with piperacillin and tobramycin. Many USA (D Humar MD); Department of Pediatrics, Division of Infectious blood cultures grew group G streptococcus and antibiotic Diseases, University of California, San Diego, CA, USA (V Datta MD, therapy was changed to benzylpenicillin. Due to continuing V Nizet MD); Toronto Medical Laboratories and the Department of pain, calf tenderness, and erythema magnetic resonance Microbiology, Mount Sinai Hospital and University of Toronto, imaging was done on day 7, showing a 107 cm fluid Toronto, Ontario, Canada (D J Bast PhD, J C S De Azavedo PhD); and Centers for Disease Control and Prevention, National Center for collection within the soleus muscle. Irrigation and debridement Infectious Diseases, Respiratory Diseases Branch, Atlanta, GA, of the abscess and surrounding muscle tissue was done. USA (B Beall PhD) Intraoperative cultures grew group G streptococcus and histology was consistent with myonecrosis. He was treated Correspondence to: Dr Victor Nizet, Division of Infectious Diseases, with intravenous penicillin for 4 weeks and granulocyte-colony University of California, San Diego, 9500 Gilman Drive, Mail Code stimulating factor. He improved on this regimen with full 0672, La Jolla, CA 92093, USA (e-mail: [email protected]) recovery after rehabilitation.

124 THE LANCET • Vol 359 • January 12, 2002 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet Publishing Group. MECHANISMS OF DISEASE

GLOSSARY

MUTAGENESIS Creation of a heritable change in a specific DNA sequence.

COMPLEMENTATION Restoration of phenotype to a bacterial mutant by reintroduction of an intact copy of the mutated gene on a plasmid vector.

OPERON A set of genes which are grouped together and transcribed on the same messenger RNA.

HOMOLOGOUS RECOMBINATION Substitution of a segment of DNA by another that is identical (homologous) or nearly so. Occurs naturally during meiotic recombination (crossing over); also used experimentally to modify the sequence of a target gene.

TRANSFORMATION A process by which the genetic material carried by an individual bacterial cell is altered by incorporation of exogenous DNA on a plasmid vector or into its genome.

ISOGENIC MUTANT A bacterial mutant differing from its wild-type parent strain by only a single genetic modification.

BACTERIOCIN A small, naturally occurring protein produced by one species of bacterium that possesses antimicrobial activity against other bacteria. Some bacteriocins exhibit toxicity to eukaryotic cells.

TRANSPOSON A relatively small DNA segment that has the ability to move from one chromosomal position to another, used experimentally for bacterial mutagenesis studies.

PULSED-FIELD GEL ELECTROPHORESIS (PFGE) Figure 1: Histopathology of debrided tissue from patient with A technique used to separate very large (megabase) DNA fragments group G streptococcal necrotising fasciitis of the calf for genotype analysis. Extensive tissue necrosis, vascular thrombosis (arrow), and neutrophilic infiltration can be seen on (A) low power (40x) and (B) high power (100x) views of the haematoxylin and eosin stained sections. infection of the subdermal soft tissues frequently complicated by toxic shock syndrome.5 By contrast, published case each of necrotising fasciitis or myositis serious group G streptococcus infections occur only caused by group G streptococcus.10,11 rarely, including endocarditis,6 septic arthritis,7 bacter- Here we report three cases of severe necrotising aemia,8 and septic shock.9 We identified only one infections due to -haemolytic group G streptococcus (panels 1, 2, and 3). Because of similar clinical Panel 3 presentations to group A streptococcus infections, we investigated a link between bacterial -haemolysin Case 3: A 58-year-old homeless man with ethanol-induced phenotype and disease pathogenesis. We used molecular cirrhosis and chronic lower extremity lymphoedema was techniques and a murine infection model to identify the admitted with a 3 day history of left thigh pain. Temperature -haemolysin of pathogenic human group G streptococcus, was 35ºC and blood pressure 70/50 mm Hg. His thigh was and assess its contribution to disease pathogenesis.12 tensely swollen and erythematous from knee to groin with bullae formation. He was treated with cefazolin and Methods clindamycin. He developed progressive acidaemia, Group G streptococcus isolates were identified by the hypotension, and coagulopathy. Intravenous immunoglobulin API 20 Strep identification system (bioMérieux, St Louis, was administered. On day 3 he developed swelling and MO, USA). Published methods were used for M-protein erythema of the left knee and left wrist, and numerous focal (emm) genotyping,13 T-antigen typing,14 opacity factor necrotic skin lesions of his digits. Gross examination of the testing,15 and PULSED-FIELD GEL ELECTROPHORESIS (PFGE) left thigh fascia showed extensive liquefaction necrosis. analysis.16 Haemolytic titres were determined in a liquid- Biopsy samples showed nectrotising fasciitis with contiguous phase assay17 in aerobic growth conditions. We used culture myonecrosis. Septic arthritis of the left knee and left wrist and TRANSFORMATION conditions as previously described.12 were present, with fascial necrosis extending into the left The group G streptococcus isolate from Case 1 (VASD1) extensor forearm. Extensive drainage and debridement were was selected for genetic and animal virulence studies. We done. Blood, knee, wrist, thigh fascia, forearm fascia, and skin did dot-blot hydridisation analysis with digoxigenin- lesion cultures all grew group G streptococcus. labelled group A streptococcus sag gene probes from the Echocardiogram was normal. He developed adult respiratory nine-gene OPERON encoding the -haemolysin streptolysin 5 distress syndrome, pneumonia, and candidaemia. He died on (SLS). A 2·4 kb HindIII fragment of group G streptococcus day 12 despite aggressive supportive care. chromosomal DNA probe that was positive for sagA by

THE LANCET • Vol 359 • January 12, 2002 • www.thelancet.com 125

For personal use. Only reproduce with permission from The Lancet Publishing Group. MECHANISMS OF DISEASE

GGS: MLQFTSNILATSVAETTQVAPGG CCCCCTTCCFSINVGGGSAQGGSGSYTPGK locally from epithelial sites (147 HU), and greater than a panel of 13 group A GAS: MLKFTSNILATSVAETTQVAPGG CCCCCTTCCFSIATGSGNSQGGSGSYTPGK streptococcal strains of differing M-protein serotype provided by the Centers for Leader peptide Propeptide Disease Control (23 HU). Prepropeptide We examined chromosomal DNA from strain VASD1 (from patient 1) and four Figure 2: Comparison of the deduced amino-acid sequences of SagA between other -haemolytic group G streptococcus group G streptococcus (GGS) and group A streptococcus (GAS) strains for homologies to the nine genes (sagA–sagI) in the group A streptococcus operon encoding SLS activity. By means of Southern blot analysis was cloned in Escherichia coli and dot-blot analysis, strong hybridisation signals were sequenced directly. obtained from group G streptococcal DNA with probes For MUTAGENESIS studies, we amplified an intragenic for each of the nine group A streptococcal genes (data fragment from the group G streptococcus sagA gene not shown). DNA sequencing showed that the sag using PCR and cloned in temperature-sensitive vector operon promoter and a terminator motif responsible for pHY304.18 This knockout plasmid was introduced into differential transcription of sagA versus sagB-I were group G streptococcus by electroporation, and trans- highly conserved. Comparison of predicted protein formants selected at 30oC. HOMOLOGOUS RECOMBINATION sequences for the group G streptococcus and group A events in the group G streptococcus chromosome were streptococcus gene products showed 89% identity and identified by shifting to the non-permissive temperature 94% similarity for SagA (figure 2), 81% identity and (37oC) while maintaining antibiotic selection. Fidelity of 92% similarity for SagB, and 74% identity and 81% site-directed recombination event was confirmed by similarity for the N-terminus of SagC, respectively. PCR. COMPLEMENTATION was done as follows: the group Confirmed plasmid integrational mutagenesis of sagA G streptococcus sagA knockout mutant was rendered yielded a group G streptococcus with no detectable - competent and transformed with vector pSagLocus haemolytic activity on SBA (figure 3). -haemolytic containing the SLS operon genes of group A strepto- activity could be partially restored to the group G coccus.12 streptococcus sagA mutant by introduction of the intact We tested group G streptococcus virulence in a group A streptococcus sag locus on a plasmid vector mouse model of streptococcal necrotising fasciitis.19,20 (figure 3). These studies show that a functional Briefly, 106 cfu of log-phase bacteria were mixed with homologue of the group A streptococcus SLS (sag) Cytodex beads (Sigma) and injected subcutaneously operon is present in group G streptococcus. The new into the right flank of hairless 4-week-old male group G streptococcal sequence information has been crl:SKH1(hrhr)Br mice (Charles River, Wilmington, submitted to the DDBJ/EMBL/GenBank databases MA, USA) Six animals were tested for the group G under the accession number AY033399. We did streptococcus parent strain and its ISOGENIC sagA additional sequencing of the sagA gene on a group C MUTANT; group A streptococcus strain NZ131 (M49) streptococcal bloodstream isolate (T107) from the and its corresponding sagA mutant were tested for Toronto Invasive Bacterial Diseases Network (TIBDN) comparison. Animals were monitored for development collection. This strain had a gene 100% homologous to of necrotic ulcers, and killed at 24 h or 48 h for sagA from group G streptococcus patient isolate quantitative culture and histopathological assessment. VASD1. Like group G streptococcus, large-colony Data were compared using the exact Wicoxon rank-sum group G streptococcus exhibiting robust -haemolysis test. are classified within S dysgalactiae subspecies equisimilis.2

Role of funding source The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing the report. Results Each group G streptococcus isolate from the patients was S dysgalactiae, subspecies Equisimilis, with T-antigen type 25 and negative opacity factor. Group G streptococcus isolates from Cases 1 and 2 possessed M- protein (emm) gene stg480 (GenBank Accession number X79520) previously identified in group G streptococcus isolates from bloodstream and wound infections. Group G streptococcus isolates from Case 3 possessed the emm gene stc74a (GenBank Accession number X60097) which has been found in various -haemolytic human group C streptococcus and group G streptococcus isolates. PFGE analysis showed that the group G streptococcus from Cases 1 and 2 were clonal, whereas the isolate from Case 3 was genetically distinct. No direct or common-source epidemiological contacts among the three patients were identified. The in-vitro Figure 3: Elimination of group G streptococcus -haemolysis by haemolytic activities of the three invasive group G mutation of sagA and partial complementation of the mutant streptococcal isolates (mean titre 149 HU) were similar phenotype upon transformation with the homologous group A to those of six group G streptococcal strains isolated streptococcus sag genes

126 THE LANCET • Vol 359 • January 12, 2002 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet Publishing Group. MECHANISMS OF DISEASE

Bacterial Necrotic Wound Wound culture Neutro- M49 strain and sagA mutant used as a control and to our strain tested ulcer culture (cfu/gm) range philic previously reported observations in M1 strains and SLS- formation (cfu/gm) (24–48 h) infiltrate negative TRANSPOSON mutants.20 Two mice infected with the (24 h) mean group G streptococcus sagA mutant appeared to have log value (24–48 h) cleared the infection by 48 h (<10 cfu/gm tissue). In the four mice without necrotic ulcers but with persistence of 6 6 Group G wild-type 100% 6·45 1·910 –4·110 ++++ bacteria at the inoculation site, up to 20% of the recovered Group G sagA 0 3·87† 1·0100–5·9105*+ mutant colonies had reverted to the wild-type -haemolytic Group A wild-type 100% 7·34 1·3105–8·7107 ++++ phenotype. The latter finding suggests an in vivo selective Group A sagA 0 4·93‡ 7·7102–1·2105 + pressure toward excision of the integrative plasmid through mutant reverse homologous recombination. Cytodex alone 0 NA NA + Six mice were tested in each group. *2/6 mice had <10 cfu/gm recovered. Discussion †p=0·0022 vs wild-type. ‡p=0·005 vs wild-type. Exact Wicoxon rank-sum test. We report three patients with necrotising soft tissue Effect of streptolysin S gene mutation on group G infections resembling group A streptococcus disease in streptococcus and group A streptococcus virulence in the which the sole microbial isolate was -haemolytic group G murine model of necrotising fasciitis streptococcus. A severe underlying medical condition was The contribution of group G streptococcus SLS present in each case. Diabetes mellitus, malignancy, and expression in the pathogenesis of necrotising fasciitis was cirrhosis are commonly reported risk factors for tested in mice (table). Within 24–48 h, animals injected development of other types of invasive group G subcutaneously with the group G streptococcal clinical streptococcus infection.6,21,22 Despite initial therapy with isolate developed necrotic ulcers at the site of inoculation, intravenous antibiotics active against group G had high bacterial counts on lesion culture, and showed streptococcus, all three patients had clinical deterioration histopathological evidence of diffuse skin and soft tissue and grew viable organisms from the necrotic tissues necrosis with substantial neutrophil infiltration. By when surgery was done. This observation reinforces the contrast, mice injected with the group G streptococcus importance of prompt and thorough surgical debridement sagA mutant did not develop necrotic ulcers, had ten-fold for the successful therapy of streptococcal nectrotic lower bacterial counts on lesion culture (p=0·0022), and fasciitis.4,5 showed minimal degrees of tissue injury or neutrophil The -haemolysin of human pathogenic group infiltration. Representative gross and microscopic G streptococcus and group C streptococcus is SLS, pathological findings are shown in figure 4. The results encoded by a nine-gene operon highly similar to that were similar to those seen with the group A streptococcus recently discovered in group A streptococcus.12 The

Figure 4: Representative gross and microscopic histopathological findings in mice infected subcutaneously with a group G streptococcal necrotising fasciitis clinical isolate (A) versus an isogenic streptolysin S-deficient mutant (B) Ulcer formation with necrotic tissue destruction, vascular thrombosis, and diffuse neutrophilic infiltrate are noted with the wild-type strain, whereas only minimal inflammatory changes are seen with the mutant.

THE LANCET • Vol 359 • January 12, 2002 • www.thelancet.com 127

For personal use. Only reproduce with permission from The Lancet Publishing Group. MECHANISMS OF DISEASE

group G streptococcus SLS precursor, SagA, retains key and IgG binding proteins, SLO, C5a peptidase, NADase, features of this BACTERIOCIN-type toxin, including a and possibly a capsule.1,35,36 Of notable predicted Gly-Gly cleavage site to yield a propeptide exception are the absence in group G streptococcus of the matching the calculated size (2·8 kD) of mature SLS.23 group A streptococcus pyrogenic exotoxins SPE-A—the Dowstream genes, including the putative modifing scarlet fever toxin A, and SPE-B, a chromosomally- enzyme sagB and ATP-binding cassette exporter sagG-I encoded cysteine protease. SPE-A production is strongly are also conserved.12 Targeted mutagenesis of the group G linked epidemiologically with strains identified in the streptococcus sagA gene abolishes -haemolytic activity, present resurgence of invasive group A streptococcus and this phenotype is partially restored upon infections.3,4 As we show with SLS, production of SPE-B, transformation of the mutant with the group A M-protein, and hyaluronic acid capsule are known to streptococcus homologue. These data confirm that genes contribute to development of group A streptococcus of the sag operon are both necessary and sufficient for SLS necrotising fasciitis in the murine model.37,38 Absence of production. pyrogenic exotoxins or differences in the coordinate The bacteriocin-like SLS precursor SagA shares no regulation of virulence factor expression39,40 may account homology whatsoever with streptolysin O (SLO), a 57-kD for the apparent inability of group G streptococcus to thiol-activated cytolysin expressed by group A, C, and G produce necrotising fasciitis in the non-compromised streptococci, for which the gene has been identified.24 By host. contrast with SLS, SLO is inactivated by oxygen, Strategies aimed at neutralisation of SLS activity could inhibited by cholesterol, produces little to no detectable be of therapeutic benefit as adjuncts to definitive surgical haemolysis on blood agar plates, is immunogenic, and and antibiotic management of streptococcal necrotising oligomerises in the red-blood-cell membrane to form a fasciitis. pore. The -haemolysin and virulence factor of the important human pathogen group B streptococcus is Contributors unrelated to SLS or SLO, but rather seems to be encoded D Humar and V Nizet reported the clinical cases. V Datta, D Humar, by a new 78·3 kD gene, cylE.18,25 An interesting DJ Bast, JCS De Azavedo, and V Nizet designed and carried out the molecular genetic and in vivo experiments. B Beall performed the emm- comparison for SLS may be the plasmid-encoded genotyping and PFGE analysis. D Humar, V Dutta, and V Nizet prepared haemolysin/cytolysin expressed by 45–60% of Enterococcus the original manuscript. All authors contributed to the revised manuscript. faecalis (formerly Group D streptococcus) isolates from D Humar and V Datta contributed equally to this work. patients. Although sharing little primary sequence similarity to SagA, this small protein toxin is composed of Conflict of interest statement two structural subunits that belong to the antibiotic class None declared. of bacteriocin peptides.26 In contrast to the wild-type parent strains, SLS-negative Acknowledgments sagA mutants of group G streptococcus and group A This work was supported by National Institute of Health/NIAID grant streptococcus are not virulent in a murine model of AI01451 and a Charles E Culpeper Foundation Biomedical Project Initiative Grant from the Rockefeller Brothers Fund. We thank streptococcal necrotising fasciitis. These findings suggest Nissi Varki for her assistance with histopathological analysis. that SLS expression is required for the pathogenesis of this destructive infection. SLS is one of the most potent cytotoxins known,27 capable of injuring a wide array of References membranes including those of lymphocytes, neutrophils, 1 Gaviria JM, Bisno AL. Group C and G streptococci. In: Stevens DL, and certain tissue culture cell lines.28-30 The precise Kaplan EL, eds. Streptococcal infections: clinical aspects, microbiology, and molecular biology. New York: Oxford University mechanisms of SLS membrane toxicity are not known. Press, 2000: 180–221. SLS does not seem to possess phospholipase action, and 2 Vandamme P, Pot B, Falsen E, Kersters K, Devriese LA. Taxonomic electron microscopic examinations of erythrocyte study of lancefield streptococcal groups C, G, and L (Streptococcus membranes damaged by SLS do not show large pores dysgalactiae) and proposal of S dysgalactiae subsp equisimilis subsp. such as those induced by SLO.31,32 We hypothesise that Int J Syst Bacteriol 1996; 46: 774–81. 3 Stevens DL, Tanner MH, Winship J, et al. Severe group A SLS contributes to the development of streptococcal streptococcal infections associated with a toxic shock- like syndrome necrotising fasciitis via direct toxicity to cells of the deep and scarlet fever toxin A. N Engl J Med 1989; 321: 1–7. soft tissues and feeding vessels, leading to cell death and 4 Bisno AL, Stevens DL. Streptococcal infections of skin and soft provoking neutrophil influx. In vitro and primate model tissues. N Engl J Med 1996; 334: 240–45. studies with group A streptococcus suggest that 5 Stevens DL. Streptococcal toxic shock syndrome associated with neutrophil-derived reactive oxide metabolites and necrotizing fasciitis. Annu Rev Med 2000; 51: 271–88. 6 Lam K, Bayer AS. Serious infections due to group G streptoccocci: proteases may act together with bacterial cytotoxins to report of 15 cases with in vitro-in vivo correlations. Am J Med 1983; accelerate necrotic fascial injury.33,34 SLS-mediated 75: 561–70. neutrophil lysis could cause release of such factors and 7 Nakata MM, Silvers JH, George WL. Group G streptococcal arthritis. prevent phagocytosis, explaining in part how wild-type Arch Intern Med 1983; 143: 1328–30. group A streptococcus and group G streptococcus persist 8 Auckenthaler R, Hermans PE, Washington JA. Group G streptococcal bacteremia: clinical study and review of the literature. Rev Infect Dis in the infection model despite intense neutrophil 1983; 5: 196–204. recruitment. 9 Honore PM, Lozana A, Defalque D, Estratiou A. Fatal septic shock SLS toxin is clearly not sufficient to trigger necrotising due to Lancefield group G streptococci. Acta Clin Belg 1996; 51: fasciitis. SLS is produced by virtually all group G 57–60. streptococcal and group A streptococcus isolates, even 10 Gaunt N, Rogers K, Seal D, Denham M, Lewis J. Necrotising fasciitis due to group C and G haemolytic streptococcus after chiropody. those isolated from asymptomatic individuals. Moreover, Lancet 1984; 1: 516. we found SLS haemolytic activity levels to be greater in 11 Wagner JG, Schlievert PM, Assimacopoulos AP, et al. Acute group G streptococcus than the more virulent group A group G streptococcal myositis associated with streptococcal toxic streptococcus. Human -haemolytic group G strepto- shock syndrome: case report and review. Clin Infect Dis 1996; 23: coccus share many other important virulence factors with 1159–61. 12 Nizet V, Beall B, Bast DJ, et al. Genetic locus for streptolysin S group A streptococcus including the antiphagocytic production by group A Streptococcus. Infect Immun 2000; 68: surface M and M-like proteins, streptokinase, fibronectin, 4245–54.

128 THE LANCET • Vol 359 • January 12, 2002 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet Publishing Group. MECHANISMS OF DISEASE

13 Beall B, Facklam R, Thompson T. Sequencing emm-specific PCR Structural analysis and proteolytic activation of Enterococcus faecalis products for routine and accurate typing of group A streptococci. cytolysin, a novel lantibiotic. Mol Microbiol 1996; 21: 1175–84. J Clin Microbiol 1996; 34: 953–58. 27 Koyama J. Biochemical studies on streptolysin S, II. Properties of a 14 Johnson DR, Kaplan EL. A review of the correlation of T-agglutination polypeptide component and its role in the toxin activity. J Biochem patterns and M-protein typing and opacity factor production in the 1963; 54: 146–51. identification of group A streptococci. 28 Ginsburg I. Mechanisms of cell and tissue injury induced by group A J Med Microbiol 1993; 38: 311–15. streptococci: relation to poststreptococcal sequelae. J Infect Dis 1972; 15 Widdowson JP, Maxted WR, Grant DL, Pinney AM. The relationship 126: 294–340. between M-antigen and opacity factor in group A streptococci. 29 Hryniewicz W, Pryjma J. Effect of streptolysin S on human and mouse J Gen Microbiol 1971; 65: 69–80. T and B lymphocytes. Infect Immun 1977; 16: 730–33. 16 From the Centers for Disease Control and Prevention. Use of pulsed- 30 Taketo Y, Taketo A. Cytolytic effect of streptolysin S complex on field gel electrophoresis for investigation of a cluster of invasive group A Ehrlich ascites tumor cells. J Biochem (Tokyo) 1966; 60: 357–62. streptococcal illness—Spokane, Washington, 1999. JAMA 1999; 282: 31 Elias N, Heller M, Ginsburg I. Binding of streptolysin S to red blood 934–35. cell ghosts and ghost lipids. Isr J Med Sci 1966; 2: 302–09. 17 Nizet V, Gibson RL, Chi EY, Framson PE, Hulse M, Rubens CE. 32 Dourmashkin RR, Rosse WF. Morphologic changes in the membranes Group B streptococcal beta-hemolysin expression is associated with of red blood cells undergoing hemolysis. Am J Med 1966; 41: injury of lung epithelial cells. Infect Immun 1996; 64: 3818–26. 699–710. 18 Pritzlaff CA, Chang JC, Kuo SP, et al. Genetic basis for the beta- 33 Ginsburg I. Is streptolysin S of group A streptococci a virulence haemolytic/cytolytic activity of group B Streptococcus. Mol Microbiol factor? APMIS 1999; 107: 1051–59. 2001; 39: 236–47. 34 Taylor FB, Jr., Bryant AE, Blick KE, et al. Staging of the baboon 19 Bisno AL, Gaviria JM. Murine model of recurrent group G response to group A streptococci administered intramuscularly: a streptococcal cellulitis: no evidence of protective immunity. descriptive study of the clinical symptoms and clinical chemical Infect Immun 1997; 65: 4926–30. response patterns. Clin Infect Dis 1999; 29: 167–77. 20 Betschel SD, Borgia SM, Barg NL, Low DE, De Azavedo JC. Reduced 35 Malke H. Genetics and pathogenicity factors of group C and G virulence of group A streptococcal Tn916 mutants that do not produce streptococci. In: Fischetti VA, Novick RP, Ferretti JJ, Portnoy DA, streptolysin S. Infect Immun 1998; 66: 1671–79. Rood JI, eds. The Gram-Positive Pathogens. Washington, DC: ASM 21 Liu CE, Jang TN, Wang FD, Wang LS, Liu CY. Invasive group G Press, 2000: 163–76. streptococcal infections: a review of 37 cases. Chung Hua I Hsueh Tsa 36 Chhatwal GS, Talay SR. Pathogenicity factors in group C and G Chih (Taipei) 1995; 56: 173–78. streptococci. In: Fischetti VA, Novick RP, Ferretti JJ, Portnoy DA, 22 Watsky KL, Kollisch N, Densen P. Group G streptococcal bacteremia. Rood JI, eds. The Gram-Positive Pathogens. Washington DC: ASM The clinical experience at Boston University Medical Center and a Press, 2000: 177–83. critical review of the literature. Arch Intern Med 1985; 145: 58–61. 37 Lukomski S, Montgomery CA, Rurangirwa J, et al. Extracellular 23 Bernheimer AW. Physical behavior of streptolysin S. J Bacteriol 1967; cysteine protease produced by Streptococcus pyogenes participates in the 93: 2024–25. pathogenesis of invasive skin infection and dissemination in mice. 24 Kehoe M, Timmis KN. Cloning and expression in Escherichia coli of Infect Immun 1999; 67: 1779–88. the streptolysin O determinant from Streptococcus pyogenes: 38 Ashbaugh CD, Warren HB, Carey VJ, Wessels MR. Molecular characterization of the cloned streptolysin O determinant and analysis of the role of the group A streptococcal cysteine protease, demonstration of the absence of substantial homology with hyaluronic acid capsule, and M protein in a murine model of human determinants of other thiol-activated toxins. Infect Immun 1984; 43: invasive soft-tissue infection. J Clin Invest 1998; 102: 550–60. 804–10. 39 Wessels MR. Regulation of virulence factor expression in group A 25 Spellerberg B, Pohl B, Haase G, Martin S, Weber-Heynemann J, streptococcus. Trends Microbiol 1999; 7: 428–30. Lutticken R. Identification of genetic determinants for the hemolytic 40 Geyer A, Schmidt KH. Genetic organisation of the M protein region activity of Streptococcus agalactiae by ISS1 transposition. J Bacteriol in human isolates of group C and G streptococci: two types of 1999; 181: 3212–19. multigene regulator-like (mgrC) regions. Mol Gen Genet 2000; 262: 26 Booth MC, Bogie CP, Sahl HG, Siezen RJ, Hatter KL, Gilmore MS. 965–76.

THE LANCET • Vol 359 • January 12, 2002 • www.thelancet.com 129

For personal use. Only reproduce with permission from The Lancet Publishing Group. ORIGINAL ARTICLE 10.1111/j.1469-0691.2005.01169.x

Invasive group A, B, C and G streptococcal infections in Denmark 1999–2002: epidemiological and clinical aspects K. Ekelund1, P. Skinhøj2, J. Madsen3 and H. B. Konradsen1

1Department of Bacteriology, Mycology and Parasitology, Statens Serum Institut, 2Department of Infectious Diseases, Rigshospitalet, Copenhagen University Hospital and 3Biostatistics Unit, Statens Serum Institut, Copenhagen, Denmark

ABSTRACT Group A streptococci (GAS) have been described frequently as an emerging cause of severe invasive infections in population-based surveillance studies, whereas the descriptions of group B, C and G streptococci (GBS, GCS and GGS) have been less frequent. Enhanced surveillance for invasive GAS, GBS, GCS and GGS was performed in Denmark in 1999–2002. A detailed questionnaire was completed for 1237 (98%) of 1260 invasive infections. GAS infections dominated (40%), followed by GGS (32%), GBS (23%) and GCS (6%). Most (74%) patients had predisposing factors, and there were no significant differences between the four serogroups when comparing the prevalence of cancer, diabetes mellitus, chronic heart or lung diseases, immunodeficiency or alcohol abuse. The overall case fatality rate at day 30 was 21%, increasing significantly to 59% for patients with streptococcal toxic shock syndrome (STSS). STSS was significantly more frequent in GAS patients (10%) than in GCS (4%), GBS (2%) and GGS (2%) patients. Regression analyses showed that, despite a younger median age among GAS patients, the probability of developing septic shock and mortality was significantly higher among GAS patients than among GBS and GGS patients. These analyses showed no significant differences between GAS and GCS infections. Invasive infections caused by GAS, GBS, GCS and GGS are still a major challenge for clinicians. Continued epidemiological and microbiological surveillance is important to assess the development of these infections and to improve preventative strategies. Keywords Case fatality rates, b-haemolytic streptococci, Denmark, invasive streptococcal infections, surveillance Original Submission: 3 December 2004; Revised Submission: 2 February 2005; Accepted: 3 February 2005 Clin Microbiol Infect 2005; 11: 569–576

before, during and after delivery [8]. GBS, together INTRODUCTION with GCS and GGS, have also been described as b-Haemolytic streptococci (BHS) groups A, B, C frequent invasive pathogens in elderly patients, and G (GAS, GBS, GCS and GGS) are a transient often in association with alcohol abuse, diabetes part of the normal flora of the pharynx, skin, mellitus, malignant diseases, and cardiac and intestinal tract and vagina, and can cause a variety peripheral vascular diseases [6,7,9]. In comparison, of invasive and non-invasive infections [1–7]. It is GAS have been described since the mid-1980s as an not possible to differentiate between invasive emerging cause of potentially fatal infections, such infections caused by GAS, GBS, GCS or GGS solely as necrotising fasciitis (NF), myositis and strepto- on the clinical presentation [1–7]. However, GBS coccal toxic shock syndrome (STSS), described remain one of the most significant causes of originally in healthy young individuals [10]. The invasive neonatal infections during the period same severe clinical presentations have been des- cribed subsequently for GBS, GCS and GGS in Corresponding author and reprint requests: K. Ekelund, several case series, but more often in elderly Streptococcus Unit, Department of Bacteriology, Mycology patients with predisposing factors. and Parasitology, Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen S., Denmark The emergence of severe GAS infections has E-mail: [email protected] resulted in enhanced surveillance of invasive

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases 570 Clinical Microbiology and Infection, Volume 11 Number 7, July 2005

GAS infections in Europe, the USA and Canada In order to validate the survey, 41 (3%) of the patients’ since 1986 [2,11–15]. In the present study, a files were read and the corresponding questionnaires were completed by one of the authors; only insignificant differ- nationwide prospective surveillance, designed ences between the two questionnaires were found. If a originally for invasive GAS infections, was used patient had more than one episode of invasive GAS, GBS, to monitor invasive infections caused by GAS, GCS or GGS infection during the period 1999–2002, the GBS, GCS and GGS. The aim was to estimate the second episode was only registered in the database if the incidence of these infections and to determine dates of the cultures were separated by > 30 days, or if the second episode was caused by an isolate belonging to a the frequency of symptoms, predisposing factors, different serogroup. Date of death, or a confirmation that an clinical presentation, treatment, complications individual was alive on 1 March 2004, was obtained from and outcome, in order to delineate any differ- the Central Office of Civil Registration for all except nine ences between invasive infections caused by patients; five patients did not have a personal identification number (i.e., tourists or recent immigrants) and were GBS, GCS, GGS and GAS in the same area excluded from the analyses of the CFRs, while four patients during the same period. were registered as missing or as having left Denmark before 1 March 2004. They were registered as alive, but their details were noted when their registration at the Central Office of MATERIALS AND METHODS Civil Registration ended. Surveillance and specimens Statistical analysis The Streptococcus Unit at the Statens Serum Institut (Copen- hagen, Denmark) serves as the National Streptococcus Refer- Data from the survey were analysed by the SAS System v. ence Centre. During the study period, invasive BHS (groups A, 8.2 (SAS Institute, Cary, NC, USA). Spearman rank correla- B, C and G) isolates were submitted from patients admitted to tion tests were performed to assess the correlation between all hospitals in Denmark (population 5.3 million). The isolates the annual incidence rates of GAS, GBS, GCS and GGS, were received as pure isolates, on a voluntary basis, from all respectively, and the time (in years). For comparisons of Danish clinical microbiological departments. All invasive BHS proportions between serogroups in relation to outcome of cultures from normally sterile sites (e.g., blood, cerebrospinal the variables described in the questionnaire, chi-square tests, fluid, synovial fluid, pleural fluid, ascites and tissue obtained adjusted for stratification with relation to age (grouped as during surgery) between 1 January 1999 and 31 December 2002 £ 40, 40–65 and > 65 years) and gender, were used. However, were collected. in situations where outcomes of the response variable were Based on information concerning the number of BHS- rare, an analogue exact test was conducted. Stratification positive blood cultures from ten of the 15 clinical microbiolo- according to age and gender was used primarily because of gical departments, the Streptococcus Unit received a constant a significant difference in the age and gender distribution proportion of these blood isolates, which on average was 79% between the serogroups. Furthermore, only women aged for GAS, 58% for GBS, 56% for GCS and 66% for GGS (overall < 40 years were included in the statistical analysis regarding average 72%). fever in relation to delivery, puerperal sepsis and recent delivery. For the comparison of age distributions between sero- Questionnaire groups and other sub-populations of the data, Kruskal– A detailed questionnaire (describing the characteristics of the Wallis or Mann–Whitney rank sum tests were used. Two patient, the infection and the outcome) was completed by the logistic regression analyses were performed to determine physician responsible for the treatment of each patient. factors which were associated significantly with STSS and Standard information (age, gender, specimen, serogroup, septic shock without fulfilling the criteria of STSS [16]. serotype, date of culture) was included in the statistical Factors included in the analysis were: gender, age (grouped analysis even if the questionnaire was not returned. Invasive as £ 40, 40–65 and > 65 years), serogroup, cancer, diabetes, GBS infections in infants aged < 90 days were considered to be chronic heart or lung diseases, skin lesions, alcohol abuse a distinct clinical entity and were not included in this study. and immunodeficiency. In order to compare the survival Bacteraemia was defined as a clinical entity associated functions between the four serogroups £ 30 days after the with identification of BHS from blood culture without any culture was obtained, a Cox proportional-hazards regression specified focus of the infection or complications. NF was analysis was performed. The following independent varia- defined as necrosis of the fascia and tissue (excluding bles were included in the model: gender, age (no grouping), muscle) and was included as a clinical diagnosis only. A serogroups, cancer, diabetes, chronic heart or lung diseases, patient with septic shock was defined as a patient with skin lesions, alcohol abuse and immunodeficiency. The invasive BHS infection and a systolic blood pressure of estimated survival function can be considered as a prognosis < 90 mm Hg (in adults). The definition of STSS was based on of mortality when previous knowledge of the patient is a consensus definition [16], including identification of BHS available. If a parameter was recorded as ‘unknown’, or the from a normally sterile site, septic shock and multi-organ registration was missing, data were not included in the failure. An overall case fatality rate (CFR) was assessed at statistical analyses. A p value of £ 0.05 was considered to be day 1 after the specimen was obtained (1-day CFR; death significant. Odds ratios and 95% CIs were calculated. within day 0 and day 1 after specimen collection), as well as Incidence rates were calculated in terms of the inci- a 30-day CFR. dence ⁄ 100 000 inhabitants ⁄ year.

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576 Ekelund et al. Invasive GAS, GBS, GCS and GGS infections in Denmark 571

RESULTS Table 1. Invasive group A, B, C and G streptococcal infections in Denmark (1999–2002) Study period No (% of isolates) Fig. 1 shows the incidence of invasive GAS Isolates from GAS GBS GCS GGS infections, together with the incidence curves of GBS, GCS and GGS, for 1988–2002. The annual Blood 454 (91.5) 273 (95.8) 70 (92.1) 382 (94.1) CSF 10 (2.0) 3 (1.1) 1 (1.3) 4 (1.0) incidence of invasive GAS infections showed Other specimens 32 (6.5) 9 (3.2) 5 (6.6) 20 (4.9) considerable variation, but statistical analysis Total 496 (100) 285 (100) 76 (100) 406 (100) did not reveal a significant trend. The incidence GAS, group A streptococci; GBS, group B streptococci; GCS, group C streptococci; of GBS and GGS showed a constant increase in GGS, group G streptococci; CSF, cerebrospinal fluid. the study period between 1999 and 2002, but this almost paralleled the increase in the number of between the four serogroups, as invasive episodes blood cultures in the same period (data not involving GCS and GGS were significantly more shown). prevalent in the older age groups compared to During the study period, the Streptococcus episodes involving both GAS and GBS (p 0.001). Unit received 1263 BHS isolates. In total, 1260 Females constituted 268 (54.0%) of 491 patients episodes of invasive BHS infections from 1231 with GAS infections, compared to 141 (49.6%)of different patients were included in the study. 274, 33 (44.0%) of 73 and 176 (43.5%) of 393 Twenty-seven patients had more than one BHS patients with GBS, GCS and GGS infections, isolate registered at the Streptococcus Unit (n =59 respectively. In the group of invasive GBS infec- isolates), with the second isolate from eight tions, there were significantly more female than patients belonging to a different serogroup than male patients in the younger age groups the first isolate. The same serogroup was isolated (p £ 0.001), a tendency that was not found with twice from 15 patients, on three occasions from the other serogroups. The overall gender distri- three patients, and on four occasions from one bution did not differ significantly between the patient. Three isolates (one GBS, one GCS and one four serogroups. GGS) were discarded because there were Fig. 2 shows that the incidence of streptococcal < 30 days between the dates of the successive infections increased with increasing age, partic- cultures. Questionnaires were received in respect ularly after the age of 55 years for GAS, after of 1237 (98.2%) of the 1260 episodes. 65 years for GBS and GGS infections, and a The distribution of invasive episodes among decade later for GCS infections. the four serogroups is shown in Table 1. The median age of the patients with invasive GAS Predisposing factors infections was 59.8 (range 0.4–97.4) years, com- pared to 64.7 (range 0.4–95.8) years, 72.8 (range The questionnaire revealed that most patients 1.4–94.6) years and 71.7 (range 0.3–99.6) years with invasive BHS infection had underlying among patients with GBS, GCS and GGS infec- predisposing factors, and that the frequencies tions, respectively. The age distribution differed differed significantly between the serogroups

16 GAS GBS GCS GGS 4.0 14 GAS GBS GCS GGS Study period 12 3.5 10 3.0 8 2.5 6 2.0 4 1.5 000 inhabitants per year 000 inhabitants per year 2 1.0 100

/100 000 inhabitants/ year 0

0.5 GGS GCS, GBS, Incidence of GAS,

Incidence of GAS. GBS. GCS. GGS GCS. GBS. Incidence of GAS. –2 0.0 0–4 5–9

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85–89 90–94 95–99 Year Age (years)

Fig. 1. Incidence ⁄ 100 000 inhabitants ⁄ year of invasive Fig. 2. Incidence ⁄ 100 000 inhabitants ⁄ year by age of inva- infections caused by group A, B, C and G streptococci in sive infections caused by group A, B, C and G streptococci Denmark 1988–2002. in Denmark 1999–2002.

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576 572 Clinical Microbiology and Infection, Volume 11 Number 7, July 2005

(p 0.014; Table 2). Skin lesions were the most Table 3. Clinical manifestations and complications of frequent predisposing factor and were found invasive infection with group A, B, C and G streptococci in Denmark (1999–2002) most often among patients with GAS infections (p £ 0.001). Even excluding Caesarian sections, GAS GBS GCS GGS n = 487 n = 278 n =75 n = 397 Overall GBS dominated BHS infections because of peri- or % % % % p value post-operative procedures (p 0.010). Cancer was Septic shock 26.9 12.1a 17.8b 10.7a £ 0.001 the second most frequent predisposing factor, but STSS 10.3 2.2 4.0 2.3a £ 0.001 NF 7.0 1.8a 2.7 1.0a £ 0.001 there was no significant difference between the Myositis 3.3 0.4a 00.5a 0.036 serogroups (p 0.129). Likewise, the data showed Pneumonia 10.3 7.6 4.0 4.8a 0.002 Meningitis 4.5 2.9 1.3 3.3 0.542 no significant differences between the serogroups Arthritis 4.1 6.1 5.3 10.3a 0.002 in the prevalence of diabetes mellitus, chronic Erysipelas 20.1 8.3a 34.7b 33.8a £ 0.001 Bacteraemia only 27.1 41.7a 32.0 27.5 £ 0.001 heart or lung diseases, human immunodeficiency Puerperal feverc 35.9 33.3 0 16.7 0.352 ICU 24.7 14.6a 15.1b 10.0a £ 0.001 virus or alcohol abuse. Mechanical ventilation 14.2 6.3a 6.9 4.6a £ 0.001 Surgery 28.9 19.6a 13.7b 17.2a £ 0.001 Renal impairment 30.7 25.2b 27.0 22.6a £ 0.001 Hepatic impairment 30.2 21.6a 25.0 21.2a 0.006 Clinical aspects Coagulator dysfunction 20.7 13.2a 18.5 10.2a £ 0.001 The variety of infections associated with the four GAS, group A streptococci; GBS, group B streptococci; GCS, group C streptococci; GGS, group G streptococci; STSS, streptococcal toxic shock syndrome; NF, serogroups is summarised in Table 3. necrotising fasciitis; ICU, intensive care unit. % ap £ 0.01 relative to serogroup A. In total, 208 (17.8 ) of 1167 episodes were bp £ 0.05 relative to serogroup A. registered as septic shock; these occurred in 124 cComparison performed among females aged < 40 years (GAS, n = 78; GBS, n =42; (26.9%) patients with GAS, 32 (12.1%) patients GCS, n = 3; GGS, n = 12). with GBS, 12 (17.8%) patients with GCS and 40 (10.7%) patients with GGS (p £ 0.001). All four Case fatality rates serogroups caused STSS, which was identified in 68 (5.5%) of 1237 patients. Most (73.5%) of the STSS During the first 30 days following collection of the patients were infected with GAS. Furthermore, specimen, 21% of patients with invasive BHS GAS was identified in 34 (75.6%) of the 45 patients infections died (23% of patients infected with with NF. Nineteen (1.5%) patients had myositis, GAS, 21% with GCS, 19% with GBS, and 18% again with a significant difference between the with GGS). Patients infected with GAS had a serogroups (p 0.036), since 16 (84.2%) patients significantly higher 1-day CFR (11%) compared were infected with GAS. For 381 (30.8%) patients, to patients infected with GBS or GGS (both 5%; clinicians recorded bacteraemia as the only diag- p £ 0.001); 9% of the GCS patients died within nosis, and this occurred more frequently with GBS 1 day, with no significant difference between GCS infections (p £ 0.001). and GAS patients (p 0.296). The hazard functions for 30-day case fatality a Table 2. Predisposing factors for invasive infection with (or, equivalently, the 30-day survival functions), group A, B, C and G streptococci in Denmark (1999–2002) as estimated by a Cox proportional-hazards GAS GBS GCS GGS regression, were significantly lower for GBS n = 472 n = 271 n =72 n = 386 Overall % % % % p value (p 0.007) and GGS (p £ 0.001) than for GAS (Table 4). No significant difference was observed Predisposing factor (‡ 1) 73.7 79.0 75.0 83.9b 0.014 Skin lesions 31.6 17.7b 33.3 36.5 £ 0.001 between the 30-day hazard functions for GAS and Peri- ⁄ post-operative 3.4 9.2b 2.8 5.7 0.010 IDU 4.9 1.1c 1.4 3.6 0.007 GCS. Table 4 also presents the significant 30-day Cancer 13.1 18.5 15.3 19.7 0.129 hazard functions for chronic heart or lung disease, Diabetes mellitus 11.0 14.0 13.9 15.8 0.904 Chronic heart or lung disease 5.5 6.3 8.3 7.5 0.925 alcohol abuse, immunodeficiency and skin le- Alcohol abuse 9.3 12.2 6.9 11.4 0.573 Immune deficiencyd 8.3 12.6 13.9 10.1 0.168 sions. NSAID 6.1 6.3 0.0 6.7 0.146

GAS, group A streptococci; GBS, group B streptococci, GCS, group C streptococci; GGS, group G streptococci; IDU, intravenous drug use; NSAID, non-steroid anti- Septic shock and STSS inflammatory drugs. aA patient can have more than one predisposing factor, and a patient can have more Overall, 59% of patients with STSS died within than one episode of invasive streptococcal infection. bp £ 0.01 relative to serogroup A. the 30-day period following the collection of cp £ 0.05 relative to serogroup A. cultures, compared to 57% of patients with septic dIncludes patients positive for human immunodeficiency virus and those receiving immunosuppressive treatment. shock without STSS, and 13% of the non-septic

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576 Ekelund et al. Invasive GAS, GBS, GCS and GGS infections in Denmark 573

Table 4. Cox proportional-hazards regression analysis of No increase in invasive GAS infections was factors associated with fatal outcome within 30 days after identified during the study period (1999–2002). invasive infection caused by group A, B, C and G streptococci in Denmark (1999–2002) The average annual incidence and fluctuating pattern of invasive GAS infections were both Cox proportional-hazards regression similar to those in previous reports [11,12]. There p value Hazard ratio (95% CI) was a three-fold increase in invasive GBS and GGS infections in Denmark during 1988–2002, Age £ 0.001 – GBS vs. GAS 0.007 0.63 (0.45–0.88) and the incidence curves showed similarities with GCS vs. GAS 0.247 0.73 (0.43–1.24) GGS vs. GAS £ 0.001 0.52 (0.38–0.70) the incidence curve of GAS until 1996. These CHLD £ 0.001 2.54 (1.53–3.32) similarities may reflect improvements in meth- Alcohol abuse £ 0.001 2.38 (1.64–3.45) Immune incompetencea 0.003 1.58 (1.17–2.14) odology or an overall increase in the number of Skin lesions 0.019 0.70 (0.52–0.94) blood cultures analysed. The incidence curves for GAS, group A streptococci; GBS, group B streptococci; GCS, group C streptococci; invasive GBS and GGS were almost parallel GGS, group G streptococci; CHLD, chronic heart or lung diseases. aIncludes patients positive for human immunodeficiency virus and those receiving in 1988–2002. The reason for the continuous immunosuppressive treatment. increases since 1998 remains unclear, although increased awareness or prolonged survival of Table 5. Results of logistic regression analyses of predis- adults with underlying chronic diseases (diabetes posing factors associated with septic shock, with or without streptococcal toxic shock syndrome, after invasive mellitus, cancer and heart disease) may be a infection caused by group A, B, C and G streptococci in contributing factor [4–9,18–20]. The incidence of Denmark (1999–2002) GGS and the emerging trend of invasive GGS

Septic shock including Septic shock excluding infections, approaching and potentially surpass- STSS STSS ing the incidence of invasive GAS infections, were

p value OR (95% CI) p value OR (95% CI) similar to those described elsewhere [20–22]. The incidence of invasive GBS infections was lower GBS vs. GAS £ 0.001 0.19 (0.08–0.45) 0.005 0.50 (0.31–0.81) GCS vs. GAS 0.101 0.37 (0.11–1.22) 0.530 0.80 (0.39–1.62) than that described in Europe and North America GGS vs. GAS £ 0.001 0.20 (0.09–0.42) £ 0.001 0.35 (0.22–0.55) [4,5,23], but was similar to the average annual Age – – £ 0.001 – CHLD – – 0.005 2.42 (1.30–4.48) incidence in a recent surveillance study from Alcohol abuse – – £ 0.001 2.29 (1.35–3.90) Finland [8]. This may be a consequence of the GAS, group A streptococci; GBS, group B streptococci; GCS, group C streptococci; moderate submission rate or may be caused by GGS, group G streptococci; CHLD, chronic heart or lung disease; STSS, strepto- coccal toxic shock syndrome. variations in bacterial virulence, host susceptibil- ity, socio-economic factors or demographic vari- shock patients. Table 5 presents the results of a ations. Comparisons of the incidence of GCS logistic regression analysis of factors associated infection are almost impossible, since only two significantly with STSS, and a similar analysis of studies have presented epidemiological data factors associated significantly with septic shock [24,25]. without fulfilling the STSS definition. The present study confirmed that GAS patients are generally younger and associated less fre- quently with predisposing factors when com- DISCUSSION pared to GBS, GCS and GGS patients [4–7,17– This study provides an overview of invasive GAS, 20,25–27]. However, no significant differences in GBS, GCS and GGS infections in Denmark during the frequencies of cancer, diabetes mellitus, chro- 1999–2002, based on nationwide enhanced pros- nic heart or lung diseases or alcohol abuse were pective surveillance. The incidence rates obtained identified in the comparisons between the four reflect the minimum estimates of invasive BHS serogroups in the present study. As described disease, since national streptococcal surveillance previously [2,4–7,9,17–20,22,25], skin was the in Denmark is based on voluntary submission of most common portal of entry, while skin and soft invasive isolates from the local clinical microbio- tissue infections, together with bacteraemia with- logy departments. However, c. 72% of invasive out focal symptoms, were the most prevalent BHS isolates were received and 98% of question- clinical presentations of invasive BHS infections. naires were submitted, and incidence rates from Patients with invasive GBS infections were infec- the county of Northern Jutland in Denmark [17] ted more often in association with surgical pro- were similar to those in the present study. cedures, and were less likely than GAS patients to

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576 574 Clinical Microbiology and Infection, Volume 11 Number 7, July 2005 have an impaired skin barrier (i.e., skin lesions or septic shock without STSS was associated with intravenous drug users). In the present study, increased age and predisposing factors such as most GBS infections were diagnosed as bacterae- chronic heart or lung diseases and alcohol abuse, mia with an unknown focus of infection, and as well as GAS. Moreover, GAS STSS patients more frequently than in previous reports [4,19]. In were younger than GAS patients with septic contrast to GBS, GGS were associated particularly shock without STSS, perhaps because a certain with skin infections. Erysipelas was recorded level of immune response is required for progres- more frequently for patients with GCS and GGS sion of multi-organ failure involved in STSS. This than for patients with GAS and GBS. Similarly, may be seen as analogous to the association cellulitis occurred in 60% of patients with inva- between the HLA class II allotype of infected sive GGS infections in a study from Israel [21]. patients and disease progression; that is, the This may reflect a new aspect of this clinical allotype influences the in-vitro immune response entity, since erysipelas has been regarded tradi- induced by streptococcal superantigens, and tionally as a specific kind of superficial cellulitis thereby identifies high and low responders to caused predominantly by GAS [28,29]. superantigens [35]. The elderly, with potential Puerperal sepsis is regarded currently as a immune senescence, or patients with immune minor problem, but the present results indicate deficiency, are associated with poor prognosis for that clinicians should still suspect streptococcal other more obvious reasons [7]. Both the fre- infections in relation to delivery, occurring either quency of GAS STSS and the 30-day CFR were sporadically or in clusters [30]. Endocarditis and comparable with those in other studies [11,12]. osteomyelitis have also been described among the The occurrence of STSS caused by GBS, GCS or clinical manifestations of invasive streptococcal GGS has been reported primarily in case series infections [3,6,7,12,18,19,31], but these entities studies [36–42], but data from population-based were not included in the present prospective studies has not been presented previously in survey. Neither was pneumonia, although it was comparison with GAS STSS. recorded by clinicians in 7% of cases, which NF was predominantly a disease caused by resulted in a frequency slightly lower than that GAS, and occurred at a frequency similar to that reported previously [5,6,20,25,26,32,33]. Pneu- in previous reports [12,31]. Since the present monia is not a diagnosis associated normally surveillance study included primarily blood iso- with BHS, as it was in the pre-antibiotic era, but lates, with only a few isolates from surgical tissue, there may be re-emergence of this entity [33]. the real prevalence will probably be higher However, there are obvious limitations in the [11,43]. A large surveillance study of invasive diagnosis, since no radiological confirmation was GBS infections [9] reported a GBS NF prevalence obtained, and objective distinctions between BHS of 0.3%, compared with 2% of patients with GBS pneumonia and invasive BHS disease complica- in the present study and elsewhere [19]. In the ted by adult respiratory distress syndrome are present study, NF occurred in 3% and 1% of the lacking. patients with GCS and GGS infections, respect- From the clinicians’ point of view, it may be ively. NF caused by GBS, GCS and GGS has been difficult to distinguish between STSS and a described previously in several case series straightforward case of septic shock, as similar [36,41,44–47]. As the clinical presentations of NF CFRs were found, and these were comparable to and STSS caused by GBS, GCS and GGS are CFRs for septic shock caused by other pathogens indistinguishable from those caused by GAS, the [34]. The case definition of STSS may identify underlying mechanisms are likely to be related. individuals with toxic septic shock, as well as Some researchers consider that the possible acqui- those who develop multi-organ failure when sition of GAS-related virulence genes, especially subjected to the stress of sepsis without toxicity, by GGS, is consistent with the hypothesis although this was not the intention of the con- that GGS may be an emerging pathogen with sensus definition [16]. Severe infections such as increased virulence potential [48,49]. septic shock (with or without STSS) and NF were The overall 21% CFR for invasive BHS infec- more likely to be caused by GAS than by GBS, tions was similar to that found previously in GCS and GGS. GAS was the only independent population-based studies [2,7,9,17,18,20,26], predictor for the development of STSS, whereas although other studies have reported lower fig-

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576 Ekelund et al. Invasive GAS, GBS, GCS and GGS infections in Denmark 575 ures [4,19]. Because information regarding patient 4. Farley MM, Harvey RC, Stull T et al. A population-based mortality was obtained from the Central Office of assessment of invasive disease due to group B Streptococ- cus in nonpregnant adults. N Engl J Med 1993; 328: 1807– Civil Registration, the cause of fatalities, and the 1811. link with BHS infection, was not investigated. 5. Tyrrell GJ, Senzilet LD, Spika JS et al. Invasive disease However, it has been estimated that c. 90% of due to group B streptococcal infection in adults: results deaths which occur within 30 days are related from a Canadian, population-based, active laboratory directly to the infection [22]. In addition to the surveillance study—1996. Sentinel Health Unit Surveil- lance System Site Coordinators. J Infect Dis 2000; 182: serotype involved, the prognosis of invasive BHS 168–173. infection was associated independently with 6. Carmeli Y, Schapiro JM, Neeman D, Yinnon AM, Alkan M. increased age and the precise predisposing factor Streptococcal group C bacteremia. Survey in Israel and (i.e., chronic heart or lung diseases, alcohol abuse analytic review. Arch Intern Med 1995; 155: 1170–1176. 7. Lewthwaite P, Parsons HK, Bates CJ, McKendrick MW, or immunodeficiency). However, patients with Dockrell DH. Group G streptococcal bacteraemia: an skin lesions were associated negatively with opportunistic infection associated with immune senes- death compared to patients without skin lesions. cence. Scand J Infect Dis 2002; 34: 83–87. This was interpreted as a difference between 8. Lyytikainen O, Nuorti JP, Halmesmaki E et al. Invasive patients with primary bacteraemia (no skin group B streptococcal infections in Finland: a population- based study. Emerg Infect Dis 2003; 9: 469–473. lesions and without any obvious focus of the 9. Schrag SJ, Zywicki S, Farley MM et al. Group B strepto- initial infection) and secondary bacteraemia (with coccal disease in the era of intrapartum antibiotic pro- skin lesions and therefore an obvious site of initial phylaxis. N Engl J Med 2000; 342: 15–20. infection). 10. Stevens DL, Tanner MH, Winship J et al. Severe group A streptococcal infections associated with a toxic shock-like In conclusion, invasive infections caused by syndrome and scarlet fever toxin A. N Engl J Med 1989; BHS remain a major challenge for clinicians. The 321: 1–7. clinical manifestations vary, and the CFR of 21% 11. Davies HD, McGeer A, Schwartz B et al. Invasive group A increases to c. 60% in patients with septic shock. streptococcal infections in Ontario, Canada. Ontario An ageing population, and progress in the treat- Group A Streptococcal Study Group. N Engl J Med 1996; 335: 547–554. ment of chronic medical conditions, provides an 12. O’Brien KL, Beall B, Barrett NL et al. Epidemiology of expanding number of patients at risk of invasive invasive group A streptococcus disease in the United BHS. In order to improve preventive strategies States, 1995–1999. Clin Infect Dis 2002; 35: 268–276. and therapeutic procedures, further investigation 13. Eriksson BK, Andersson J, Holm SE, Norgren M. Epide- miological and clinical aspects of invasive group A of the interactions between host and pathogen is streptococcal infections and the streptococcal toxic shock needed. Continued epidemiological and micro- syndrome. Clin Infect Dis 1998; 27: 1428–1436. biological surveillance is important to assess the 14. Kaul R, McGeer A, Low DE, Green K, Schwartz B. Popu- future development of invasive BHS infections. lation-based surveillance for group A streptococcal nec- rotizing fasciitis: clinical features, prognostic indicators, and microbiologic analysis of seventy-seven cases. Ontario ACKNOWLEDGEMENTS Group A Streptococcal Study. Am J Med 1997; 103: 18–24. 15. Vlaminckx B, van Pelt W, Schouls L et al. Epidemiological We thank A. Jørgensen and A. Mortensen for valuable help in features of invasive and noninvasive group A streptococ- preparing the manuscript and compiling the data, respect- cal disease in the Netherlands, 1992–1996. Eur J Clin ively. We also thank the clinical microbiological departments Microbiol Infect Dis 2004; 23: 434–444. who submitted isolates to the Statens Serum Institut and the 16. Working Group on Severe Streptococcal Infections. Defi- physicians who provided clinical information. The study was ning the group A streptococcal toxic shock syndrome. approved by the Scientific-Ethical Committees for Copenhagen Rationale and consensus definition. JAMA 1993; 269: 390– and Frederiksberg (no. (KF) 11-156 ⁄ 00) and by the Danish Data 391. Protection Agency (no. 2001-41-0807). 17. Kristensen B, Schonheyder HC. A 13-year survey of bac- teraemia due to beta-haemolytic streptococci in a Danish county. J Med Microbiol 1995; 43: 63–67. REFERENCES 18. Bradley SF, Gordon JJ, Baumgartner DD, Marasco WA, Kauffman CA. Group C streptococcal bacteremia: analysis 1. Cartwright K. Group A streptococcal infections in of 88 cases. Rev Infect Dis 1991; 13: 270–280. humans. Soc Appl Bacteriol Symp Ser 1997; 26(suppl): 52–61. 19. Blancas D, Santin M, Olmo M, Alcaide F, Carratala J, 2. Efstratiou A. Group A streptococci in the 1990s. J Anti- Gudiol F. Group B streptococcal disease in nonpregnant microb Chemother 2000; 45(suppl): 3–12. adults: incidence, clinical characteristics, and outcome. Eur 3. Salata RA, Lerner PI, Shlaes DM, Gopalakrishna KV, J Clin Microbiol Infect Dis 2004; 23: 168–173. Wolinsky E. Infections due to Lancefield group C strep- 20. Sylvetsky N, Raveh D, Schlesinger Y, Rudensky B, Yinnon tococci. Medicine (Baltimore) 1989; 68: 225–239. AM. Bacteremia due to beta-hemolytic Streptococcus group

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576 576 Clinical Microbiology and Infection, Volume 11 Number 7, July 2005

G: increasing incidence and clinical characteristics of 36. Jankowski S, MacLachlan K, Stephenson J, Thomas P. patients. Am J Med 2002; 112: 622–626. Necrotising infections and group G streptococcus. Lancet 21. Cohen-Poradosu R, Jaffe J, Lavi D et al. Group G strepto- 2002; 359: 2277. coccal bacteremia in Jerusalem. Emerg Infect Dis 2004; 10: 37. Natoli S, Fimiani C, Faglieri N et al. Toxic shock syndrome 1455–1460. due to group C streptococci. A case report. Intens Care Med 22. Hindsholm M, Schonheyder HC. Clinical presentation and 1996; 22: 985–989. outcome of bacteraemia caused by beta-haemolytic strep- 38. Bateman AC, Ramsay AD, Pallett AP. Fatal infection tococci serogroup G. APMIS 2002; 110: 554–558. associated with group C streptococci. J Clin Pathol 1993; 46: 23. Persson E, Berg S, Trollfors B et al. Serotypes and clinical 965–967. manifestations of invasive group B streptococcal infections 39. Ojukwu IC, Newton DW, Luque AE, Kotb MY, Menegus in western Sweden 1998–2001. Clin Microbiol Infect 2004; M. Invasive Group C Streptococcus infection associated 10: 791–796. with rhabdomyolysis and disseminated intravascular 24. Barnham MR, Weightman NC. Changing incidence of coagulation in a previously healthy adult. Scand J Infect Dis detected streptococcal bacteraemia in North Yorkshire, 2001; 33: 227–229. England. Ind J Med Res 2004; 119(suppl): 160–163. 40. Thomas S, Cunha BA. Group B streptococcal toxic shock- 25. Schugk J, Harjola VP, Sivonen A, Vuopio-Varkila J, Val- like syndrome with fulminant cellulitis. Heart Lung 1996; tonen M. A clinical study of beta-haemolytic groups A, B, 25: 497–499. C and G streptococcal bacteremia in adults over an 8-year 41. Tang WM, Ho PL, Yau P, Wong WK, Yip DKP. Report of 2 period. Scand J Infect Dis 1997; 29: 233–238. fatal cases of adult necrotizing fasciitis and toxic shock 26. Dahl MS, Tessin I, Trollfors B. Invasive group B strepto- syndrome caused by S. agalactiae. Clin Infect Dis 2000; 31: coccal infections in Sweden: incidence, predisposing fac- e5–e17. tors and prognosis. Int J Infect Dis 2003; 7: 113–119. 42. Schlievert PM, Gocke JE, Deringer JR. Group B strepto- 27. Kiska DL, Thiede B, Caracciolo J et al. Invasive group A coccal toxic shock-like syndrome: report of a case and streptococcal infections in North Carolina: epidemiology, purification of an associated pyrogenic toxin. Clin Infect clinical features, and genetic and serotype analysis of Dis 1993; 17: 26–31. causative organisms. J Infect Dis 1997; 176: 992–1000. 43. Sharkawy A, Low DE, Saginur R et al. Severe group A 28. Eriksson BK, Jorup-Ronstrom C, Karkkonen K, Sjoblom streptococcal soft-tissue infections in Ontario: 1992–1996. AC, Holm SE. Erysipelas: clinical and bacteriologic spec- Clin Infect Dis 2002; 34: 454–460. trum and serological aspects. Clin Infect Dis 1996; 23: 1091– 44. Holmstrom B, Grimsley EW. Necrotizing fasciitis and toxic 1098. shock-like syndrome caused by group B Streptococcus. 29. Swartz MN. Clinical practice. Cellulitis. N Engl J Med 2004; South Med J 2000; 93: 1096–1098. 350: 904–912. 45. Sing A, Trebesius K, Heesemann J. Diagnosis of Strepto- 30. Ejlertsen T, Prag J, Pettersson E, Holmskov A. A 7-month coccus dysgalactiae subspecies equisimilis (group C strepto- outbreak of relapsing postpartum group A streptococcal cocci) associated with deep soft tissue infections using infections linked to a nurse with atopic dermatitis. Scand J fluorescent in situ hybridization. Eur J Clin Microbiol Infect Infect Dis 2001; 33: 734–737. Dis 2001; 20: 146–149. 31. Zurawski CA, Bardsley M, Beall B et al. Invasive group A 46. Sharma M, Khatib R, Fakih M. Clinical characteristics of streptococcal disease in metropolitan Atlanta: a popula- necrotizing fasciitis caused by group G Streptococcus: case tion-based assessment. Clin Infect Dis 1998; 27: 150–157. report and review of the literature. Scand J Infect Dis 2002; 32. Stromberg A, Romanus V, Burman LG. Outbreak of group 34: 468–471. A streptococcal bacteremia in Sweden: an epidemiologic 47. Korman TM, Boers A, Gooding TM, Curtis N, Visvanathan and clinical study. J Infect Dis 1991; 164: 595–598. K. Fatal case of toxic shock-like syndrome due to group C 33. Muller MP, Low DE, Green KA et al. Clinical and epi- Streptococcus associated with superantigen exotoxin. J Clin demiologic features of group A streptococcal pneumonia Microbiol 2004; 42: 2866–2869. in Ontario, Canada. Arch Intern Med 2003; 163: 467– 48. Kalia A, Bessen DE. Presence of streptococcal pyrogenic 472. exotoxin A and C genes in human isolates of group 34. Matot I, Sprung CL. Definition of sepsis. Intens Care Med G streptococci. FEMS Microbiol Lett 2003; 219: 291–295. 2001; 27(suppl 1): S3–S9. 49. Kalia A, Enright MC, Spratt BG, Bessen DE. Directional 35. Kotb M, Norrby-Teglund A, McGeer A et al. An immuno- gene movement from human-pathogenic to commensal- genetic and molecular basis for differences in outcomes of like streptococci. Infect Immun 2001; 69: 4858–4869. invasive group A streptococcal infections. Nat Med 2002; 8: 1398–1404.

2005 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 569–576