Fever in the ICU*

Paul E. Marik, MD, FCCP

Fever is a common problem in ICU patients. The presence of fever frequently results in the performance of diagnostic tests and procedures that significantly increase medical costs and expose the patient to unnecessary invasive diagnostic procedures and the inappropriate use of antibiotics. ICU patients frequently have multiple infectious and noninfectious causes of fever, necessitating a systematic and comprehensive diagnostic approach. , , and blood stream are the most common infectious causes of fever. The urinary tract is unimportant in most ICU patients as a primary source of infection. Fever is a basic evolutionary response to infection, is an important host defense mechanism and, in the majority of patients, does not require treatment in itself. This article reviews the common infectious and noninfectious causes of fever in ICU patients and outlines a rational approach to the management of this problem. (CHEST 2000; 117:855–869)

Key words: ; fever; ICU; sinusitis; ; ventilator-associated pneumonia

Abbreviations: CDC ϭ Centers for Control and Prevention; CFU ϭ colony-forming units; ELISA ϭ - linked immunosorbent assay; IL ϭ ; TNF ϭ ; UTI ϭ urinary tract infection; VAP ϭ ventilator-associated pneumonia

ever is a common problem in ICU patients. The primarily involved in the development of fever in- F presence of fever frequently results in the per- clude interleukin (IL) 1, IL-6, and tumor necrosis formance of diagnostic tests and procedures that factor (TNF)-␣.2–13 The interaction between these significantly increase medical costs and expose the cytokines is complex, with each being able to up- patient to unnecessary invasive diagnostic proce- regulate and down-regulate their own expression as dures and the inappropriate use of antibiotics. The well as that of the other cytokines. These cytokines main diagnostic dilemma is to exclude noninfectious bind to their own specific receptors located in close causes of fever and then to determine the site and proximity to the preoptic region of the anterior .2,3 Here, the receptor inter-

For editorial comment see page 627 action activates phospholipase A2, resulting in the liberation of plasma membrane as likely of those with . ICU pa- substrate for the cyclo-oxygenase pathway. Some tients frequently have multiple infectious and non- cytokines appear to increase cyclo-oxygenase expres- infectious causes of fever,1 necessitating a systematic sion directly, leading to liberation of and comprehensive diagnostic approach. This article E2. This small lipid mediator diffuses across the reviews the common infectious and noninfectious blood barrier, where it acts to decrease the rate causes of fever in ICU patients and outlines a of firing of preoptic warm-sensitive neurons, leading rational approach to the management of these pa- to activation of responses designed to decrease heat tients. loss and increase heat production.2,14 In a small proportion of hospitalized patients, may result from increased sympathetic activity with Pathogenesis of Fever increased heat production. Cytokines released by monocytic cells play a cen- tral role in the genesis of fever. The cytokines Significance of Fever Fever appears to be a preserved evolutionary * From the Department of Internal Medicine, Section of Critical 15–20 Care, Washington Hospital Center, Washington, DC. response within the animal kingdom. With few Manuscript received May 11, 1999; revision accepted October exceptions, reptiles, amphibians, and fish, as well as 25, 1999. several invertebrate species, have been shown to Correspondence to: Paul E. Marik, MD, Department of Internal Medicine, Washington Hospital Center, 110 Irving St NW, manifest fever in response to challenge with micro- Washington, DC 20010-2975; e-mail: [email protected] organism.15–19 Increased body temperature has been

CHEST / 117/3/MARCH, 2000 855 shown to enhance the resistance of animals to infec- nary artery is considered the optimal site for core tion.21,22 Although fever has some harmful effects, temperature measurement; however, this method fever appears to be an adaptive response that has requires placement of a pulmonary artery cathe- evolved to help rid the host of invading pathogens. ter.40–42 Infrared thermometry has been demon- Temperature elevation has been shown to enhance strated to provide values that are a few tenths of a several parameters of immune function, including degree below temperatures in the pulmonary artery antibody production, T-cell activation, production of and brain.43–46 Rectal temperatures obtained with a cytokines, and enhanced and mercury or electronic probe are often function.23–26 Furthermore, some pathogens such as a few tenths of a degree higher than core tempera- Streptococcus pneumoniae are inhibited by febrile ture.40–42 Rectal temperatures are perceived by pa- temperatures.27 tients as unpleasant and intrusive. Furthermore, It has long been known that increasing body access to the rectum may be limited by patient temperature is associated with improved outcome position, with an associated risk of rectal trauma. from infectious . The preantibiotic era pro- Oral measurements are influenced by events such as vides abundant, although uncontrolled data, on the eating and drinking and the presence of respiratory deliberate use of elevated body temperature to treat devices delivering warmed gases.43 Axillary measure- infections. The beneficial effects of hot baths and ments substantially underestimate core temperature malarial in were noted as early as the and lack reproducibility.43 Body temperature is 15th century.28 In mammalian models, increasing therefore most accurately measured by an intravas- body temperature results in enhanced resistance to cular thermistor, but measurement by infrared ear infection.29–32 In a retrospective analysis of 218 thermometry or with an electronic probe in the patients with Gram-negative bacteremia, Bryant and rectum is an acceptable alternative.47 Normal body colleagues33 reported a positive correlation between temperature is generally considered to be 37.0°C maximum temperature on the day of bacteremia and (98.6°F) with a circadian variation of between 0.5 to survival. Similarly, Weinstein and colleagues34 re- 1.0°C.2,14 The definition of fever is arbitrary and ported that a temperature Ͼ 38°C increased survival depends on the purpose for which it is defined. The in patients with spontaneous bacterial . Society of Critical Care Medicine practice parame- Dorn and colleagues35 reported that children with ters define fever in the ICU as a temperature who were treated with acetaminophen Ͼ 38.3°C (Ն 101°F).47 Unless the patient has other had a longer time to crusting of lesions than when features of an infectious process, only a temperature treated with placebo. Ͼ 38.3°C (Ն 101°F) warrants further investigation. An elevated body temperature may, however, also be associated with a number of deleterious effects, Fever Patterns most notably an increase in cardiac output, oxygen consumption, carbon dioxide production, and energy Attempts to derive reliable and consistent clues expenditure.36 Oxygen consumption increases by from evaluation of a patient’s fever pattern is fraught approximately 10% per degree Celsius.36 These with uncertainly and not likely to be helpful diagnos- changes may be poorly tolerated in patients with tically.2,14,48 Most patients have remittent or inter- limited cardiorespiratory reserve. In patients who mittent fever that, when due to infection, usually have suffered a cerebrovascular accident or trau- follow a diurnal variation.48 Sustained fevers have matic head injury, moderate elevations of brain been reported in patients with Gram-negative pneu- temperature may markedly worsen the resulting monia or CNS damage.48 The appearance of fever at injury.37 Maternal fever has been suggested to be a different time points in the course of a patient’s cause of fetal malformations or spontaneous abor- illness may however provide some diagnostic clues. tions.38,39 However, this association has not been Fevers that arise Ͼ 48 h after institution of mechan- rigorously tested. ical ventilation may be secondary to a developing pneumonia.49,50 Fevers that arise 5 to 7 days postop- eratively may be related to formation.51 Definitions and Measurement of Fever Fevers that arise 10 to 14 days postinstitution anti- biotics for intra-abdominal abscess may be due to Accurate and reproducible measurement of body fungal infections.52–54 temperature is important in detecting disease and in monitoring patients with an elevated temperature. A Causes of Fever in the ICU variety of methods are used to measure body tem- perature, combining different sites, instruments, and As outlined above, any disease process that results techniques. The mixed venous blood in the pulmo- in the release of the proinflammatory cytokines IL-1,

856 Reviews IL-6, and TNF-␣ will result in the development of be considered in ICU patients are listed in Table fever. While infections are the commonest cause of 1.1,55,66–68 For reasons that are not entirely clear, fever in ICU patients, many noninfectious inflamma- most noninfectious disorders usually do not lead to a tory conditions cause the release of the proinflam- fever Ͼ 38.9°C (102°F); therefore, if the tempera- matory cytokines with a febrile response.55–61 Simi- ture increases above this threshold, the patient larly, it is important to appreciate that not all patients should be considered to have an infectious etiology with infections are febrile. Approximately 10% of as the cause of the fever.67 However, patients with septic patients are hypothermic and 35% are normo- fever may have a temperature Ͼ 102°F.69–71 thermic at presentation. Septic patients who fail to Similarly, fever secondary to blood transfusion may develop a temperature have a significantly higher be Ͼ 102°F.72,73 mortality than febrile septic patients.62–64 The reason Most of those clinical conditions listed in Table 1 that patients with established infections fail to de- are clinically obvious and do not require additional velop a febrile response is unclear; however, prelim- diagnostic tests to confirm their presence. However, inary evidence suggests that this aberrant response is a few of these disorders require special consider- not due to diminished cytokine production.65 ation. Although drug-induced fever is commonly The presence of fever in an ICU patient fre- cited as a cause of fever,74 Ͻ 300 cases of this quently triggers a battery of diagnostic tests that are condition have been reported in the literature.70 costly, expose the patient to unnecessary risks, and Furthermore, only a single case of drug fever has often produce misleading or inconclusive results. It been reported in an ICU patient population.1 How- is therefore important that fever in ICU patient be ever, on the basis of the number of evaluated in a systematic, prudent, clinically appro- administered to patients in the ICU, one would priate, and cost-effective manner. expect drug fever to be a relatively common event. Although the true incidence of this disorder is unknown, drug fever should be considered in pa- Noninfectious Causes of Fever in the ICU tients with an otherwise unexplained fever, particu- larly if they are receiving ␤-lactam antibiotics, pro- A large number of noninfectious disorders result cainamide, or diphenylhydantoin.70 Drug fever is in tissue injury with and a febrile usually characterized by high spiking temperatures reaction. Those noninfectious disorders that should and shaking .70 It may be associated with a with and eosinophilia. Relative bradycardia, although commonly cited, is uncommon.67,70,74 Table 1—Noninfectious Causes of Fever in the ICU Atelectasis is commonly implicated as a cause of Noninfectious Causes fever. Standard ICU texts list atelectasis as a cause of fever, although they provide no primary source.51,75 /drug withdrawal Indeed a major text states that “fever is (48 h postoperative) Posttransfusion fever almost always present [in patients with atelecta- 51 76 Drug fever sis].” However, Engeron studied 100 postopera- Cerebral /hemorrhage tive patients and was unable to Adrenal insufficiency demonstrate a relationship between atelectasis and Myocardial infarction fever. Furthermore, when atelectasis is induced in Acalculous experimental animals by ligation of a mainstem 77,78 Ischemic bowel bronchus, fever does not occur. However, Kisala Aspiration pneumonitis and coworkers79 demonstrated that IL-1 and TNF-␣ ARDS (both acute and late fibroproliferative phase) levels of macrophage cultures from atelectatic lungs were significantly increased compared with the con- Fat emboli Transplant rejection trol lungs. The role of atelectasis as a cause of fever Deep venous thrombosis is unclear; however, atelectasis probably does not Pulmonary emboli cause fever in the absence of pulmonary infection. /pseudogout Febrile reactions complicate about 0.5% of blood Hematoma transfusions, but may be more common following Cirrhosis (without primary peritonitis) 72,80,81 GI bleed platelet transfusion. Antibodies against mem- Phlebitis/thrombophlebitis brane of transfused leukocytes and/or plate- Adrenal insufficiency lets are responsible for most febrile reactions to IV contrast reaction cellular blood components.72 Febrile reactions usu- Neoplastic fevers ally begin within 30 min to 2 h after a blood-product Decubitus ulcers transfusion is begun. The fever generally lasts be-

CHEST / 117/3/MARCH, 2000 857 tween 2 h and 24 h and may be preceded by chills.73 feature on ultrasound examination, with a specificity An acute leucocytosis lasting up to 12 h commonly of 90% and a sensitivity of 100%.93,94 In ICU occurs following a blood transfusion.82 patients, hepatobiliary scintigraphy has a high false- Patients with the ARDS may progress to a “chron- positive rate (Ͼ 50%), limiting the value of this ic” stage characterized by pulmonary fibroprolifera- test.95 However, a normal scan virtually excluded tion and fevers. Meduri and coworkers1,83 have acalculous cholecystitis. CT scanning has been re- demonstrated that fever and leukocytosis may result ported to have a high sensitivity and specificity; from the inflammatory-fibrotic process present in however, no prospective studies have been per- the airspace of patients with late ARDS in the formed comparing ultrasonography with CT scan- absence of pulmonary infection. Corticosteroids ap- ning in the diagnosis of acalculous cholecystitis.96 pear to be associated with an improvement in lung The management of acalculous cholecystitis is injury and reduced mortality.83,84 Some authors rec- somewhat controversial.85,89,97 However, with the ommend an open lung biopsy prior to commencing development of more advanced radiologic imaging corticosteroid therapy, in order to obtain histologic techniques, percutaneous cholecystostomy may be evidence of the fibroproliferative phase of ARDS the procedure of choice. Kiviniemi and coworker98 and to exclude infection. demonstrated diminution of in 94% of patients, Acalculous cholecystis occurs in approximately with normalization of fever in 90% and leukocyte 1.5% of critically ill patients.85,86 While relatively count in 84% of patients treated by percutaneous uncommon, acalculous cholecystitis is an important cholecystostomy. The procedure is associated with “noninfectious” cause of fever in critically ill patients, few complications and is the definitive therapy in as it is frequently unrecognized and therefore poten- most patients.99 Open is, however, tially life threatening.85,86 The pathophysiology of recommended should the abdominal signs, fever, acalculous cholecystitis is related to the complex and leucocytosis not improve within 48 h of percu- interplay of a number of pathogenetic mechanisms, taneous cholecystostomy.85,89,97 including gallbladder ischemia, bile stasis with inpis- While fever may occur in patients with deep sation in the absence of stimuli for emptying of the venous thrombosis, in patients suspected of deep gallbladder, positive-end expiratory pressure, and venous thrombosis, the predictive value of fever is parenteral nutrition.87–92 Bacterial of the poor.100 Furthermore, in critically ill ICU patients, gallbladder appears to be a secondary phenome- fever without other features of ileofemoral thrombo- non.89 sis is uncommon and does not warrant routine The diagnosis of acalculous cholecystitis is often venography as part of the initial diagnostic workup of exceedingly difficult and requires a high index of pyrexia in ICU patients.1,101 suspicion. Pain in the right upper quadrant is the finding that most often leads the clinician to the correct diagnosis, but it may frequently be ab- Infectious Causes of Fever sent.85,86,89 Nausea, vomiting, and fever are other associated clinical features. The clinical findings and The prevalence of nosocomial infection in ICUs laboratory workup in patients with acalculous chole- has been reported to vary from 3 to 31%.102–108 Data cystitis are, however, often nonspecific. The most from the National Nosocomial Infection Surveillance difficult patients are those recovering from abdomi- system database from 1986 to 1990 documented nal who deteriorate again, misleadingly sug- nosocomial infection in 10% of the 164,034 patients, gesting a flare-up of the original infection. Rapid with a strong correlation between ICU length of stay diagnosis is essential because ischemia may progress and the development of infection.103 In a point rapidly to and perforation, with attendant prevalence study conducted in 1992, The EPIC increase in the already high morbidity and mortali- Study Investigators104 reported on the prevalence of ty.89 The diagnosis should therefore be considered in nosocomial infections in 10,038 patients hospitalized every critically ill patient who has clinical findings of in 1,417 European ICUs. In this study, 20.6% of sepsis with no obvious source. patients had an ICU-acquired infection, with pneu- Radiologic investigations are required for a pre- monia being the most common (46.9%), followed by sumptive diagnosis of acalculous cholecystitis. Ultra- urinary tract infection (17.6%) and blood stream sound is the most common radiologic investigation infection (12%). This data must, however, be inter- used in the diagnosis of acalculous cholecystitis; preted with some caution. The presence and type of features include increased wall thickness, intramural infection in these studies was documented according lucencies, gallbladder distension, pericholecystic to the “standard definitions” of the Centers for fluid, and intramural sludge.93,94 Wall thickness Ն 3 Disease Control and Prevention (CDC).109,110 The mm is reported to be the most important diagnostic definitions of nosocomial infection published by the

858 Reviews CDC may, however, not be applicable to ICU The optimal technique(s) for diagnosis of VAP re- patients.109,110 For example, according to the most mains unclear as a uniformly agreed on “gold stan- recent definitions published in 1988, the presence of dard,” for the diagnosis is lacking.111,118,124,133–135 rales and purulent sputum or the presence of new The impact that diagnostic tests for VAP have on chest radiographic findings and change in sputum patient outcome is controversial. Using a decision character were used to diagnose pneumonia.110 In analysis method, Sterling and coauthors136 demon- patients receiving mechanical ventilation, less than a strated that invasive or semi-invasive microbiological third of patients with these features would be con- diagnostic techniques improved the outcome of pa- sidered to have pneumonia using invasive diagnostic tients with suspected VAP. However, Luna and methods.111–114 Similarly, fever and a urine culture colleagues137 and Rello and coworkers138 have dem- of Ն 105 colony-forming units (CFU)/mL was con- onstrated that the most important factor affecting sidered diagnostic of urinary tract infection. As is outcome in patients with VAP is the early initiation discussed below, the presence of these two finding in of appropriate antibiotic therapy. In the study by catheterized critically ill ICU patients does not rep- Luna et al,137 the mortality of patients who were resent infection of the urinary tract. changed from inadequate antibiotic therapy to ap- The most common infections reported in ICU pa- propriate therapy based on the results of the BAL tients are pneumonia, followed by sinusitis, blood was comparable to the mortality of those patients stream infection, and catheter-related infection.1,102–108 who continued to receive inadequate therapy. Kollef Table 2 lists the most important sites of infection in and Ward,139using noninvasive mini-BAL to diag- ICU patients. As is discussed below, urinary tract nose VAP, confirmed these findings. It should how- infection is probably unimportant in most ICU pa- ever be noted that patients who have clinical features tients. of VAP and in whom VAP is “excluded” based on quantitative culture of lower respiratory tract secre- tions and in whom antibiotics are stopped have a Ventilator-Associated Pneumonia significantly lower mortality than those patient who are culture positive.121,139 Invasive or noninvasive Ventilator-associated pneumonia (VAP) occurs in sampling of lower respiratory tract sections with approximately 25% of patients undergoing mechan- quantitative culture therefore allows for the safe 49,115–118 ical ventilation. The impact of VAP on pa- discontinuation of antibiotics in the “culture nega- 117,119,120 tient outcome has been much debated ; tive” patients.123,125,140–145 Furthermore, as the ini- 121 however, Fagon and colleagues reported an attrib- tial empiric antibiotic regimen must be broad and utable mortality of 27%. The optimal management of cover both Gram-positive and negative organisms, patients with suspected VAP requires confirmation these techniques allow for narrowing of the spec- of the diagnosis and identification of the responsible trum once a has been isolated in those pathogen(s) in order to provide appropriate antimi- patients with confirmed pneumonia. This approach crobial therapy. The diagnosis of VAP remains one of to suspected VAP will result in significant cost savings the most difficult clinical dilemmas in critically ill and reduce the selection of resistant organisms.113 patients receiving mechanical ventilation.49 Clinical criteria alone have been shown to be unreliable in the diagnosis of this condition.113,115,122 A number of invasive and minimally invasive techniques have Sinusitis been reported to aid in the diagnosis of VAP. The Because paranasal sinusitis is usually clinically number of methods currently available attest to the silent in intubated patients, it is not widely appreci- 49,112,120,123–132 fact that no single method is ideal. ated that nosocomial sinusitis is an important source of infection and fever in critically ill patients. Fur- thermore, many ear, nose, and throat surgeons are of Table 2—Common Infectious Causes of Fever in the the belief that paranasal sinusitis in intubated pa- ICU tients receiving mechanical ventilation does not Infectious Causes cause fever or systemic signs of infection. Nosoco- mial sinusitis is particularly common following nasal VAP Sinusitis intubation, with an incidence of up to 85% after a 146–151 Catheter-related sepsis week of intubation. The incidence of nosoco- Primary Gram-negative septicemia mial sinusitis appears to be lower in patients in C difficile whom both the endotracheal and gastric tubes are Abdominal sepsis placed orally.146–151 The diagnosis of sinusitis re- Complicated wound infections quires a CT scan and cannot be accurately assessed

CHEST / 117/3/MARCH, 2000 859 using standard radiography or echography.152 Sinus- silver-impregnated subcutaneous cuffs.156,159–162 itis is diagnosed by total opacification or the presence These studies have generally shown poor or incon- of an air fluid level within any of the paranasal sistent results. It has been suggested that antimicro- sinuses. The maxillary sinus is most commonly in- bial bonding of central venous catheters may be the volved; however, most patients with radiologic max- most effective method of reducing the rate of cath- illary sinusitis have abnormalities of the ethmoid and eter colonization and catheter-related sepsis.163,164 sphenoid sinuses.148 Since radiologic abnormalities Several types of antiseptic or antimicrobial coatings of the paranasal sinuses do not necessarily imply have been developed, including catheters coated infection, diagnosis of infectious maxillary sinusitis with chlorhexidine gluconate and silver sulfadiazine, requires transnasal puncture following appropriate as well as with and rifampin. While a disinfection of the nares.146,148,150,153 When the eth- number of studies have demonstrated the incidence moid or sphenoid sinuses only are involved, bacteri- of catheter-related sepsis to be lower with chlorhexi- ologic specimens can be obtained by an open eth- dine/sulfadiazine-coated catheters,165–167 not all moidectomy/sphenoidotomy.146 Sinus infection is studies have duplicated these findings.168–170 Fur- diagnosed by the presence of pus associated with thermore, Darouiche and colleagues154 have demon- high quantitative cultures of implicated pathogens. strated that central venous catheters impregnated Rouby and colleagues148 reported that only 38% of with minocycline and rifampin are associated with a patients with radiologic maxillary sinusitis had true significantly lower rate of catheter colonization and infectious sinusitis. In the series reported by Rouby blood stream infection than catheters coated with et al,148 there was normalization of the core temper- chlorhexidine and silver sulfadiazine. ature and WBC count following removal of all nasal Central venous catheterization via the femoral and tubes, followed by transnasal puncture and drainage internal jugular veins are reported to have a similar in the patients with infectious maxillary sinusitis. infection rates, which are higher than that for cath- These authors did not use IV antibiotics. Similarly, in eters inserted via the subclavian approach.154,163,165,171 the series reported by Grindlinger and colleagues146 Replacement of a colonized catheter over a guide- and by Deutschman and coworkers,147 resolution of wire is associated with rapid recolonization of the sinusitis was associated with normalization of the replacement catheter.172 If catheter sepsis is sus- temperature and WBC count. Paranasal sinusitis is pected, the catheter should be changed to a new site, best treated by removal of all nasal tubes together with culture (quantitative or semiquantitative) of the with drainage of the maxillary sinuses. Broad-spec- catheter tip.154,172–176 In patients with limited venous trum antibiotics are generally recommended.146,147 access or in patients in whom catheter sepsis is less likely, the catheter can be changed over a guidewire; however, withdrawal blood cultures and culture of Catheter-Associated Sepsis the catheter tip should be performed and the cath- eter removed if the cultures are positive. Catheter-associated sepsis is defined as blood stream infection due to an organism that has colo- nized a vascular catheter. Approximately 5% of Urinary Tract Infection patients with indwelling vascular catheters (uncoat- ed) will develop blood stream infection (Ϸ 10 infec- Urinary tract infections (UTIs) have been reported tions/1,000 catheter days).154–158 The incidence of to be common in ICU patients, where they are catheter-associated sepsis increases with the length reported to account for between 25 to 50% of all of time the catheter is in situ, the number of ports, infections.102–108 However, it is likely that most of and increases with the number of manipulations. these patients had “asymptomatic bacteriuria” rather Approximately 25% of central venous catheters be- than true infections of the urinary tract. The use of come colonized (Ͼ 15 CFU), and approximately 20 antibiotics in patients with asymptomatic bacteriuria to 30% of colonized catheters will result in catheter is based on a single study performed in the early sepsis.154–158 Staphylocuccus aureus and coagulase- 1980s that may not be applicable today.177 Platt and negative staphylococci are the most common infect- colleagues177 demonstrated that in hospitalized pa- ing (and colonizing) organisms, followed by entero- tients bacteriuria with Ն 105 CFUs of per cocci, Gram-negative bacteria, and Candida milliliter of urine during bladder catheterization was species.154–158 associated with a 2.8-fold increase in mortality. A number of methods of reducing catheter colo- Based on this study, thousands of ICU patients with nization and blood stream infection have been stud- urinary tract colonization have been treated with ied, including topical antibiotics, antimicrobial flush antibiotics. solutions, subcutaneous tunneling of catheters, and Most ICU patients require an indwelling urinary

860 Reviews catheter for monitoring fluid balance and renal nous is the most dramatic manifestation of C function. The patients’ colonic flora rapidly colonizes difficile infection; these patients have marked ab- the urinary tract in these patients.178 Stark and dominal and systemic and may Maki179 have demonstrated that in catheterized pa- develop a fulminant and life-threatening colitis. tients, bacteria in the urinary system rapidly prolif- Stool assay for toxins A or B are the main clinical 5 erate to exceed 10 CFU/mL over a short period of tests used to diagnose C difficile infection.190–192 The time. Bacteriuria, defined as a quantitative culture of “gold standard” test is the tissue culture cytotoxicity Ն 5 10 CFU/mL, has been reported in up to 30% assay. This test has a high sensitivity (94 to 100%) 180 of catheterized hospitalized patients. The terms and specificity (99%). The major disadvantages of “bacteriuria” and “UTI” are generally although in- this test are its high expense and the time needed to correctly used as synonyms. Indeed, most studies in complete the assay (2 to 3 days). For these reasons, ICU patients have used bacteriuria to diagnose a this test is no longer routinely performed. UTI. Bacteriuria implies colonization of the urinary enzyme-linked immunosorbent assay (ELISA) tests tract without bacterial invasion and an acute inflam- are less sensitive (70 to 90%) than the cytotoxicity matory response.181 UTI implies an infection of the urinary tract.181 Criteria have not been developed for test, but demonstrate excellent specificity (99%) and differentiating asymptomatic colonization of the uri- can be rapidly processed, and have largely replaced 190–192 nary tract from symptomatic infection. Furthermore, the cytotoxicity assay. It is suggested that two the presence of white cells in the urine is not useful stool specimens be examined for leukocytes and for differentiating colonization from infection, as toxin ELISA test.190 Should the ELISA be negative most catheter-associated bacteriurias have accompa- and a high index of suspicion for C difficile exist, the nying pyuria.182 It is therefore unclear how many following are recommended: (1) sigmoidoscopy, catheterized patients with Ͼ 105 CFU/mL actually and/or (2) cytotoxicity assay, and/or (3) CT scan of have UTI. abdomen looking for thickened colonic wall. While catheter-associated bacteruria is common in ICU patients, data for the early 1980s indicates that Ͻ 3% of catheter-associated bacteriuric patients will Candida Infections develop bacteremia caused by organisms in the urine.183 Therefore, the surveillance for and treat- Candida species are important opportunistic ment of isolated bacteruria in most ICU patients is pathogens in the ICU. The CDC National Nosoco- currently not recommended.184 Bacteriuria should, mial Infection Study reported that 7% of all nosoco- however, be treated following urinary tract manipu- mial infections were due to candidal species.193 In lation or surgery, in patients with kidney stones, and the EPIC study,104 17% of nosocomial ICU infec- in patients with urinary tract obstruction. tions were due to fungi. Candida infections should be considered in febrile ICU patients who have been in the ICU for Ͼ 10 days and have received multiple courses of antibiotics.53 Candida species are partic- CLOSTRIDIA DIFFICILE Colitis ularly important pathogens in patients with ongoing C difficile, the agent that causes pseudomembra- peritonitis.52–54 It is important to realize that Can- nous colitis and antibiotic-associated diarrhea, has dida species are constituents of the normal flora in become a common nosocomial pathogen.185–187 Ap- about 30% of all healthy people. Antibiotic therapy proximately 20% of all hospitalized patients become increases the incidence of colonization by up to “infected” with C difficile, of whom only about a 70%.53 It is probable that most ICU patients become third develop diarrhea.185–187 The majority of hospi- colonized with Candida species soon after admission. tal inpatients infected with C difficile are asymptom- Not all patients colonized with Candida will become atic.188,189 C difficile infection commonly presents infected with Candida. Nonneutropenic patients with mild to moderate diarrhea, sometimes accom- with isolation of Candida species from pulmonary panied by lower abdominal cramping. Symptoms samples (tracheal aspirates, bronchoscopic or blind usually begin during or shortly after antibiotic ther- sampling methods), even in high concentrations, are apy but are occasionally delayed for several weeks. unlikely to have invasive .194,195 Indication Severe colitis without pseudomembrane formation for initiation of antifungal therapy in these patients may occur with profuse, debilitating diarrhea, ab- should be based on histologic evidence or identifica- dominal pain, and distension. Common systemic tion from sterile specimens. Similarly, isolation of manifestations include fever, nausea, , and Candida species from the urine in ICU patients with . A neutrophilia and increased numbers of indwelling catheters usually represents colonization fecal leukocytes are common.188,189 Pseudomembra- rather than infection. Although candiduria may be

CHEST / 117/3/MARCH, 2000 861 observed in up to 80% of patients with systemic blood cultures should not be obtained through intra- candidiasis, candidemia from a urinary tract source is vascular catheters unless the catheter has been re- extremely rare.54 cently placed.207 The volume of blood drawn in adult patients is the single most important factor governing the sensitivity Other Infections of blood cultures.180,206,208,209 Therefore, it is recom- mended that a minimum of 10 mL and preferably 20 Nosocomial is exceedingly uncommon mL of blood be removed per draw divided among in hospitalized patients who have not undergone a the minimum number of containers 196,197 neurosurgical procedure. Lumbar puncture, as recommended by the manufacturer.180,206,208,209 therefore, need not be performed routinely in ICU Resin-containing medium offers little clinical benefit patients (nonneurosurgical) who develop a fever to the majority of ICU patients.210 Once bloodstream unless they have meningeal signs or contiguous infection is identified, repeated or follow-up cultures 196,197 infection. In patients who have undergone are not necessary in most cases. Subsequent blood abdominal surgery and develop a fever, intra-abdom- cultures may be justified in patients who deteriorate inal infection must always be excluded. CT scanning clinically or those who fail to improve despite ther- of the abdomen is indicated in these patients. Simi- apy. However in some cases bacteremia may be larly, in patients who have undergone other opera- prolonged, necessitating further blood cultures dur- tive procedures, wound infection must be excluded. ing treatment (eg, staphylococcal bacteremia).

Diagnostic Evaluation Scintigraphy, CT Scanning, and Ultrasound It is important that blood cultures as well as other Examinations appropriate cultures be performed before the initi- ation of antibiotic therapy. The impact of antibiotic Scintigraphic scanning techniques have a low sen- therapy on culture positivity is illustrated in patients sitivity and specificity in ICU patients and are there- 1,211,212 with suspected VAP, where a number of studies have fore not recommended. The advantages of demonstrated that both prior and current antibiotic CT scanning and/or ultrasound over scintigraphy is therapy reduces the predictive accuracy of invasive that the results of the test can be obtained immedi- diagnostic testing. 198,199 ately with superior anatomic resolution, which can be used to guide drainage procedures.

Blood Cultures An Approach to the Critically Ill Patient Bacteremia and candidemia have been docu- With Fever mented in up to 10% of ICU patients and are an important cause of morbidity and mortality in the From the forgoing information, the following ap- ICU.200–203 Blood cultures are therefore indicated in proach is suggested in ICU patients who develop a all febrile patients. Surveillance blood cultures, how- fever (see Fig 1). Due to the frequency and excess ever, are expensive and add very little to the man- morbidity and mortality associated with bacteremia, agement of patients in the ICU.204 blood cultures are recommenced in all ICU patients Bennett and Beeson205 reported that the presence who develop a fever. A comprehensive physical of microorganisms in the blood is the initiating event examination and review of the chest radiograph is leading to fever and chills 1 to 2 h later, and that essential. Noninfectious causes of fever should be blood cultures are frequently negative at the time of excluded. In patients with an obvious focus of infec- the temperature spike. Thus blood cultures are tions (eg, purulent nasal discharge, abdominal ten- ideally drawn prior to the onset of a temperature derness, profuse green diarrhea), a focused diagnos- spike. In reality, this is not possible; therefore, tic workup is required. If there is no clinically spreading out the collection of blood cultures in- obvious source of infection and unless the patient is creases the likelihood of blood collection during clinically deteriorating (falling BP, decreased urine bacteremia. It is therefore recommended that at output, increasing , rising serum lactate least two and no more than three sets of blood concentration, falling platelet count, or worsening cultures should be obtained by separate needle sticks coagulopathy), or the temperature is Ͼ 39°C from different venipuncture sites.206 Colonization of (102°F), it may be prudent to perform blood cultures the lumen of central venous catheters occurs within and then observe the patient before embarking on a short period of time after placement. Therefore, further diagnostic tests and commencing empiric

862 Reviews Figure 1. Fever diagnostic algorithm. Dx ϭ diagnostic; ABx ϭ antibiotics; Rx ϭ therapy. antibiotics. However, all neutropenic patients with loose stools, and CT scan of the sinuses with removal fever and patients with severe (as outlined above) or of all nasal tubes. Urine culture is indicated only in progressive signs of sepsis should be started on patients with abnormalities of the renal system or broad-spectrum antimicrobial therapy immediately following urinary tract manipulation. If the patient is after obtaining appropriate cultures. at risk of abdominal sepsis or has any abdominal In patients whose clinical picture is consistent with signs (, distension, unable to tolerate en- infection and in whom no clinically obvious source teral feeds) CT scan of abdomen is indicated. Pa- has been documented, removal of all central lines tients with right upper quadrant tenderness require Ͼ 48 h old (with semiquantitative or quantitative an abdominal ultrasound. culture) is recommended as well as stool for WBC Reevaluation of the patient’s status after 48 h count and C difficile toxin in those patients with using all available results and the evolution of the

CHEST / 117/3/MARCH, 2000 863 patients clinical condition is essential. If fever per- References sists despite empiric antibiotics and no source of 1 Meduri GU, Mauldin GL, Wunderink RG, et al. Causes of infection has been identified, empiric antifungal fever and pulmonary densities in patients with clinical therapy may be indicated if the patient has risk manifestations of ventilator-associated pneumonia. Chest factors for candidal infection. Additional diagnostic 1994; 106:221–235 tests may be appropriate at this time, including 2 Mackowiak PA. Concepts of fever. Arch Intern Med 1998; 158:1870–1881 venography, a differential blood count for eosino- 3 Saper CB, Breder CD. The neurologic basis of fever. N Engl phils (diagnosis of drug fever), and abdominal J Med 1994; 330:1880–1886 imaging. 4 Dinarello CA, Cannon JG, Mancilla J. Interleukin-6 as an endogenous pyrogen: induction of in brain but not peripheral blood mononuclear cells. Brain Res 1991; 562:199–206 Treatment of Fever in the ICU 5 Dinarello CA, Wolff SM. The role of interleukin-1 in disease. N Engl J Med 1993; 328:106–113 Almost all febrile ICU patients are treated with 6 Dinarello CA, Cannon JG, Mier JW, et al. Multiple biolog- acetaminophen and external cooling methods to ical activities of human recombinant interleukin 1. J Clin Invest 1986; 77:1734–1739 render the patients “afebrile.” However, fever is a 7 Dinarello CA. Interleukin-1 and the pathogenesis of the basic evolutionary response to infection and may be acute-phase response. N Engl J Med 1984; 311:1413–1418 an important host defense mechanism. The prepon- 8 Fontana A, Weber E, Dayer JM. Synthesis of interleukin derance of evidence suggests that temperature in the 1/endogenous pyrogen in the brain of endotoxin-treated mice: a step in fever induction? J Immunol 1984; 133:1696– range of the usual fever renders host defenses more 1698 active and many pathogens more susceptible to these 9 Gourine AV, Rudolph K, Tesfaigzi J, et al. Role of hypotha- defenses. Therefore, it seems illogical to treat fever lamic interleukin-1 beta in fever induced by cecal ligation per se. In addition, temperature is an important and puncture in rats. Am J Physiol 1998; 275(3 Pt 2):R754– R761 physical sign, allowing the physician to monitor the 10 Leon LR, White AA, Kluger MJ. Role of IL-6 and TNF in response to treatment. Furthermore, acute hepatitis and survival during sepsis in mice. Am J may occur in ICU patients with reduced glutathione Physiol 1998; 275(1 Pt 2):R269–R277 reserves (alcoholics, malnourished, etc.) who have 11 Kluger MJ, Kozak W, Leon LR, et al. The use of knockout mice to understand the role of cytokines in fever. Clin Exp received regular therapeutic doses of acetamino- Pharmacol Physiol 1998; 25:141–144 phen. Based on this data, it is recommenced that 12 Klir JJ, McClellan JL, Kluger MJ. Interleukin-1 beta causes febrile episodes not be routinely treated with anti- the increase in anterior hypothalamic interleukin-6 during pyretic therapy; an evaluation of the relative benefits LPS-induced fever in rats. Am J Physiol 1994; 266(6 Pt and risks of treatment should be evalu- 2):R1845–R1848 13 Klir JJ, Roth J, Szelenyi Z, et al. Role of hypothalamic ated in each individual case. Fever should, however, interleukin-6 and tumor necrosis factor-alpha in LPS fever be treated in patients with acute brain insults, pa- in rat. Am J Physiol 1993; 265(3 Pt 2):R512–R517 tients with limited cardiorespiratory reserve (ie, isch- 14 Mackowiak PA, Bartlett JG, Borden EC, et al. Concepts of emic heart disease), and in patients in whom the fever: recent advances and lingering dogma. Clin Infect Dis temperature increases above 40°C (104°F).2,14,37,213–217 1997; 25:119–138 15 Kluger MJ, Kozak W, Conn CA, et al. The adaptive value of blankets are frequently used in ICU fever. Infect Dis Clin N Am 1996; 10:1–20 218 patients with febrile episodes. However, studies 16 Kluger MJ, Vaughn LK. Fever and survival in rabbits have demonstrated that hypothermia blankets are no infected with Pasteurella multocida. J Physiol 1978; 282: more effective in cooling patients than are antipy- 243–251 retic agents.218 Furthermore, the use of hypothermia 17 Bernheim HA, Kluger MJ. Fever and antipyresis in the lizard Dipsosaurus dorsalis. Am J Physiol 1976; 231:198–203 blankets is associated with large temperature fluctu- 18 D’Alecy LG, Kluger MJ. Avian febrile response. J Physiol 218 ations and rebound hyperthermia. In addition, 1975; 253:223–232 there is a fundamental illogic to the use of external 19 Vaughn LK, Bernheim HA, Kluger MJ. Fever in the lizard application of to lower temperature in a patient Dipsosaurus dorsalis. Nature 1974; 252:473–474 with true fever. Because of the altered hypothalamic 20 Covert JB, Reynolds WW. Survival value of fever in fish. Nature 1977; 267:43–45 set point, the patient is already responding as if to a 21 Bernheim HA, Kluger MJ. Fever: effect of drug-induced cold environment. External cooling may result in antipyresis on survival. Science 1976; 193:237–239 augmented hypermetabolism and a persistent fever. 22 Kluger MJ, Ringler DH, Anver MR. Fever and survival. Indeed, Lenhardt and colleagues219 demonstrated Science 1975; 188:166–168 that active external cooling in volunteers with in- 23 Jampel HD, Duff GW, Gershon RK, et al. Fever and immunoregulation: III. Hyperthermia augments the pri- duced fever increased oxygen consumption by 35 to mary in vitro humoral immune response. J Exp Med 1983; 40% and was associated with a significant increase in 157:1229–1238 epinephrine and levels. 24 van Oss CJ, Absolom DR, Moore LL, et al. Effect of

864 Reviews temperature on the chemotaxis, phagocytic engulfment, 47 O’Grady NP, Barie PS, Bartlett J, et al. Practice parameters digestion and O2 consumption of human polymorphonu- for evaluating new fever in critically ill adult patients. Crit clear leukocytes. J Reticuloendothel Soc 1980; 27:561–565 Care Med 1998; 26:392–408 25 Biggar WD, Bohn DJ, Kent G, et al. Neutrophil migration in 48 Musher DM, Fainstein V, Young EJ, et al. Fever patterns: vitro and in vivo during hypothermia. Infect Immunol 1984; their lack of clinical significance. Arch Intern Med 1979; 46:857–859 139:1225–1228 26 Azocar J, Yunis EJ, Essex M. Sensitivity of human natural 49 Chastre J, Fagon JY, Trouillet JL. Diagnosis and treatment killer cells to hyperthermia. Lancet 1982; 1:16–17 of nosocomial pneumonia in patients in intensive care units. 27 Styrt B, Sugarman B. Antipyresis and fever. Arch Intern Clin Infect Dis 1995; 21(Suppl 3):S226–S237 Med 1990; 150:1589–1597 50 Kollef MH. Ventilator-associated pneumonia: a multivariate 28 Dennie CC. A history of syphilis. Springfield, IL: Charles C. analysis. JAMA 1993; 270:1965–1970 Thomas, 1962 51 Hiyama DT, Zinner MJ. Surgical complications. In: 29 Small PM, Tauber MG, Hackbarth CJ, et al. Influence of Schwartz SI, Shires GT, Spencer FC et al, eds. Principles of body temperature on bacterial growth rates in experimental surgery. New York, NY: McGraw-Hill, 1994; 455–487 pneumococcal meningitis in rabbits. Infect Immunol 1986; 52 Calandra T, Bille J, Schneider R, et al. Clinical significance 52:484–487 of Candida isolated from peritoneum in surgical patients. 30 Sande MA, Sande ER, Woolwine JD, et al. The influence of Lancet 1989; 2:1437–1440 fever on the development of experimental Streptococcus 53 Petri MG, Konig J, Moecke HP, et al. Epidemiology of pneumoniae meningitis. J Infect Dis 1987; 156:849–850 invasive in ICU patients: a prospective multicenter 31 Anderson KJ, Kuhn RE. Elevated environmental tempera- study in 435 non-neutropenic patients. Paul-Ehrlich Society ture enhances immunity in experimental Chagas’ disease. for , Divisions of Mycology and Pneumonia Infect Immunol 1989; 57:13–17 Research. Intensive Care Med 1997; 23:317–325 32 Carmichael LE, Barnes FD, Percy DH. Temperature as a 54 Nolla-Salas J, Sitges-Serra A, Leon-Gil C, et al. Candidemia factor in resistance of young puppies to canine herpesvirus. in non-neutropenic critically ill patients: analysis of prognos- J Infect Dis 1969; 120:669–678 tic factors and assessment of systemic antifungal therapy. 33 Bryant RE, Hood AF, Hood CE, et al. Factors affecting Study Group of Fungal Infection in the ICU. Intensive Care mortality of gram-negative rod bacteremia. Arch Intern Med Med 1997; 23:23–30 1971; 127:120–128 55 Ferrara JL. The febrile platelet transfusion reaction: a 34 Weinstein MR, Iannini PB, Stratton CW, et al. Spontaneous cytokine shower. Transfusion 1995; 35:89–90 bacterial peritonitis: a review of 28 cases with emphasis on 56 Chen CC, Wang SS, Lee FY, et al. Proinflammatory cyto- improved survival and factors influencing prognosis. Am J kines in early assessment of the prognosis of acute pancre- Med 1978; 64:592–598 atitis. Am J Gastroenterol 1999; 94:213–218 35 Dorn TF, DeAngelis C, Baumgardner RA, et al. Acetamin- 57 Osman MO, El-Sefi T, Lausten SB, et al. Sodium fusidate ophen: more harm than good for chickenpox? J Pediatr and the cytokine response in an experimental model of acute 1989; 114:1045–1048 pancreatitis. Br J Surg 1998; 85:1487–1492 36 Manthous CA, Hall JB, Olson D, et al. Effect of cooling on 58 Feuerstein GZ, Wang X, Barone FC. The role of cytokines oxygen consumption in febrile critically ill patients. Am J in the neuropathology of stroke and neurotrauma. Neuro- Respir Crit Care Med 1995; 151:10–14 immunomodulation 1998; 5:143–159 37 Ginsberg MD, Busto R. Combating hyperthermia in acute 59 DeGraba TJ. The role of inflammation after acute stroke: stroke: a significant clinical concern. Stroke 1998; 29:529– utility of pursuing anti-adhesion molecule therapy. Neurol- 534 ogy 1998; 51(3 Suppl 3):S62–S68 38 Badawi N, Kurinczuk JJ, Keogh JM, et al. Intrapartum risk 60 Carlstedt F, Lind L, Lindahl B. Proinflammatory cytokines, factors for newborn encephalopathy: the Western Australian measured in a mixed population on arrival in the emergency case-control study. BMJ 1998; 317:1554–1558 department, are related to mortality and severity of disease. 39 Quinn M, Matthews TG. Does maternal pyrexia embarrass J Intern Med 1997; 242:361–365 the fetus? Irish Med J 1984; 77:218–219 61 Marx N, Neumann FJ, Ott I, et al. Induction of cytokine 40 Schmitz T, Bair N, Falk M, et al. A comparison of five expression in leukocytes in acute myocardial infarction. J Am methods of temperature measurement in febrile intensive Coll Cardiol 1997; 30:165–170 care patients. Am J Crit Care 1995; 4:286–292 62 Clemmer TP, Fisher CJ, Jr, Bone RC, et al. Hypothermia in 41 Milewski A, Ferguson KL, Terndrup TE. Comparison of the sepsis syndrome and clinical outcome. The Methylpred- pulmonary artery, rectal, and tympanic membrane temper- nisolone Severe Sepsis Study Group. Crit Care Med 1992; atures in adult intensive care unit patients. Clin Pediatr 20:1395–1401 1991; 30(4 Suppl):13–16 63 Sprung CL, Peduzzi PN, Shatney CH, et al. Impact of 42 Nierman DM. Core temperature measurement in the inten- encephalopathy on mortality in the sepsis syndrome. The sive care unit. Crit Care Med 1991; 19:818–823 Veterans Administration Systemic Sepsis Cooperative Study 43 Erickson RS, Kirklin SK. Comparison of ear-based, bladder, Group. Crit Care Med 1990; 18:801–806 oral, and axillary methods for core temperature measure- 64 Arons MM, Wheeler AP, Bernard GR, et al. Effects of ment. Crit Care Med 1993; 21:1528–1534 on the physiology and survival of hypothermic 44 Hayward JS, Eckerson JD, Kemna D. Thermal and cardio- sepsis. Crit Care Med 1999; 27:699–707 vascular changes during three methods of resuscitation from 65 Marik PE, Zaloga GP. Hypothermia and cytokines in septic mild hypothermia. Resuscitation 1984; 11:21–33 [abstract]. Crit Care Med 1999; 27(Suppl):A136 45 Shiraki K, Konda N, Sagawa S. Esophageal and tympanic 66 Clarke DE, Kimelman J, Raffin TA. The evaluation of fever temperature responses to core blood temperature changes in the intensive care unit. Chest 1991; 100:213–220 during hyperthermia. J Appl Physiol 1986; 61:98–102 67 Cunha BA. Fever in the critical care unit. Crit Care Clin 46 Shiraki K, Sagawa S, Tajima F, et al. Independence of brain 1998; 14:1–14 and tympanic temperatures in an unanesthetized human. 68 Chambers LA, Kruskall MS, Pacini DG, et al. Febrile J Appl Physiol 1988; 65:482–486 reactions after platelet transfusion: the effect of single versus

CHEST / 117/3/MARCH, 2000 865 multiple donors. Transfusion 1990; 30:219–221 sonic diagnosis. Am J Surg 1981; 142:290–292 69 Hanson MA. Drug fever: remember to consider it in 94 Deitch EA. Utility and accuracy of ultrasonically measured diagnosis. Postgrad Med 1991; 89:167–170 gallbladder wall as a diagnostic criteria in biliary tract 70 Mackowiak PA, LeMaistre CF. Drug fever: a critical ap- disease. Dig Dis Sci 1981; 26:686–693 praisal of conventional concepts; an analysis of 51 episodes 95 Kalff V, Froelich JW, Lloyd R, et al. Predictive value of an in two Dallas hospitals and 97 episodes reported in the abnormal hepatobiliary scan in patients with severe inter- English literature. Ann Intern Med 1987; 106:728–733 current illness. Radiology 1983; 146:191–194 71 Mackowiak PA. Drug fever: mechanisms, maxims and mis- 96 Blankenberg F, Wirth R, Jeffrey RBJ, et al. Computed conceptions. Am J Med Sci 1987; 294:275–286 tomography as an adjunct to ultrasound in the diagnosis of 72 Barton JC. Nonhemolytic, noninfectious transfusion reac- acute acalculous cholecystitis. Gastrointest Radiol 1991; tions. Semin Hematol 1981; 18:95–121 16:149–153 73 Rutledge R, Sheldon GF, Collins ML. Massive transfusion. 97 Roslyn JJ, Zinner MJ. Gallbladder and extrahepatic biliary Crit Care Clin 1986; 2:791–805 system. In: Schwartz SI, Shires GT, Spencer FC et al, ed. 74 Wood AJJ. Adverse drug reactions. In: Fauci AS, Braunwald Principles of surgery. New York, NY: McGraw-Hill, 1994; E, Isselbacher KJ et al, eds. Harrison’s principles of internal 1367–1399 medicine. New York, NY: McGraw-Hill, 1998; 422–430 98 Kiviniemi H, Makela JT, Autio R, et al. Percutaneous 75 Fry DE. Postoperative fever. In: Mackowiak PA, ed. Fever: cholecystostomy in acute cholecystitis in high-risk patients: basic mechanisms and management. New York, NY: Raven an analysis of 69 patients. Int Surg 1998; 83:299–302 Press, 1991; 243–254 99 van Overhagen H, Meyers H, Tilanus HW, et al. Percuta- 76 Engoren M. Lack of association between atelectasis and neous cholecystectomy for patients with acute cholecystitis fever. Chest 1995; 107:81–84 and an increased surgical risk. Cardiovasc Intervent Radiol 77 Shields RT. Pathogenesis of postoperative pulmonary atel- 1996; 19:72–76 ectasis an experimental study. Arch Surg 1949; 48:489–503 100 Diamond PT, Macciocchi SN. Predictive power of clinical 78 Lansing AM. Mechanism of fever in pulmonary atelectasis. symptoms in patients with presumptive deep venous throm- Arch Surg 1963; 87:168–174 bosis. Am J Phys Med Rehabil 1997; 76:49–51 79 Kisala JM, Ayala A, Stephan RN, et al. A model of pulmo- 101 Marik PE, Andrews L, Maini B. The incidence of deep nary atelectasis in rats: activation of alveolar macrophage and venous thrombosis in ICU patients. Chest 1997; 111:661– cytokine release. Am J Physiol 1993; 264(3 Pt 2):R610–R614 664 80 Menitove JE, McElligott MC, Aster RH. Febrile transfusion 102 Brown RB, Hosmer D, Chen HC, et al. A comparison of reaction: what blood component should be given next? Vox infections in different ICU’s within the same hospital. Crit Sanguinis 1982; 42:318–321 Care Med 1985; 13:472–476 81 Snyder EL, Stack G. Febrile and nonimmune transfusions 103 Jarvis WR, Edwards JR, Culver DH, et al. Nosocomial reactions. In: Rossi EC, Simon TL, Moss GS, eds. Principles infection rates in adult and pediatric intensive care units in of transfusion medicine. Baltimore, MD: Williams and the United States. National Nosocomial Infections Surveil- Wilkins, 1991 lance System. Am J Med 1991; 91(3B):185S–191S 82 Fenwick JC, Cameron M, Naiman SC, et al. Blood transfu- 104 Vincent JL, Bihari DJ, Suter PM, et al. The prevalence of sion as a cause of leucocytosis in critically ill patients. Lancet nosocomial infection in intensive care units in Europe: 1994; 344:855–856 results of the European Prevalence of Infection in Intensive 83 Meduri GU, Belenchia JM, Estes RJ, et al. Fibroprolifera- Care (EPIC) Study. EPIC International Advisory Commit- tive phase of ARDS: clinical findings and effects of cortico- tee. JAMA 1995; 274:639–644 steroids. Chest 1991; 100:943–952 105 Daschner FD, Frey P, Wolff G, et al. Nosocomial infections 84 Meduri GU, Headley S, Golden E, et al. Effect of prolonged in intensive care wards: a multicenter prospective study. methylprednisolone therapy in unresolving acute respiratory Intensive Care Med 1982; 8:5–9 distress syndrome: a randomized controlled trial. JAMA 106 Bueno-Cavanillas A, Delgado-Rodriguez M, Lopez-Luque 1999; 280:159–165 A, et al. Influence of nosocomial infection on mortality rate 85 Orlando R, Gleason E, Drezner AD. Acute acalculous in an intensive care unit. Crit Care Med 1994; 22:55–60 cholecystitis in the critically ill patient. Am J Surg 1983; 107 Bjerke HS, Leyerle B, Shabot MM. Impact of ICU nosoco- 145:472–476 mial infections on outcome from surgical care. Am Surg 86 Long TN, Heimbach DM, Carrico CJ. Acalculous cholecys- 1991; 57:798–802 titis in critically ill patients. Am J Surg 1978; 136:31–36 108 Potgieter PD, Linton DM, Oliver S, et al. Nosocomial 87 Johnson EE, Hedley-White J. Continuous positive-pressure infections in a respiratory intensive care unit. Crit Care Med ventilation and portal flow in dogs with pulmonary edema. 1987; 15:495–498 J Appl Physiol 1972; 33:385–389 109 Haley RW, Quade D, Freeman HE, et al. Study on the 88 Johnson EE, Hedley-White J. Continuous positive pressure efficacy of nosocomial infection control (SENIC Project): ventilation and choledochoduodenal flow resistance. J Appl summary of study design. Am J Epidemiol 1980; 111:472– Physiol 1975; 39:937–942 485 89 Barie PS, Fischer E. Acute acalculous cholecystitis. J Am 110 Garner JS, Jarvis WR, Emorl TG, et al. CDC definitions for Coll Surg 1995; 180:232–244 nosocomial infections. Am J Infect Control 1988; 16:128– 90 Malet PF. Acalculous cholecystitis: a new perspective. Gas- 140 troenterology 1990; 99:1529–1530 111 Baker AM, Bowton DL, Haponik EF. Decision making in 91 Roslyn JJ, Pitt HA, Mann L, et al. Parenteral nutrition- nosocomial pneumonia: an analytic approach to the inter- induced gallbladder disease: a reason for early cholecystec- pretation of quantitative bronchoscopic cultures. Chest tomy. Am J Surg 1984; 148:58–63 1995; 107:85–95 92 Petersen SR, Sheldon GF. Acute acalculous cholecystitis: a 112 Chastre J, Trouillet JL, Fagon JY. Diagnosis of pulmonary of hyperalimentation. Am J Surg 1979; 138: infections in mechanically ventilated patients. Semin Respir 814–817 Infect 1996; 11:65–76 93 Deitch EA, Engel JM. Acute acalculous cholecystitis: ultra- 113 Fagon JY, Chastre J, Domart Y, et al. Nosocomial pneumo-

866 Reviews nia in patients receiving continuous mechanical ventilation: cultures of endotracheal aspirates for the diagnosis of ven- prospective analysis of 52 episodes with use of a protected tilator-associated pneumonia. Am Rev Respir Dis 1993; specimen brush and quantitative culture techniques. Am 148:1552–1557 Rev Respir Dis 1989; 139:877–884 133 Marquette CH, Copin MC, Wallet F, et al. Diagnostic tests 114 Meduri GU. Diagnosis and of ventila- for pneumonia in ventilated patients: prospective evaluation tor-associated pneumonia. Clin Chest Med 1995; 16:61–93 of diagnostic accuracy using histology as a diagnostic gold 115 Fagon JY, Chastre J, Hance AJ, et al. Detection of nosoco- standard. Am J Respir Crit Care Med 1995; 151:1878–1888 mial lung infection in ventilated patients: use of a protected 134 Kirtland SH, Corely DE, Winterbauer RH, et al. The specimen brush and quantitative culture in 147 patients. Am diagnosis of ventilator-associated pneumonia: a comparison Rev Respir Dis 1988; 138:110–116 of histologic, microbiologic, and clinical criteria. Chest 1997; 116 Torres A, Aznar R, Gatell JM, et al. Incidence, risk, and 112:445–457 prognosis factors of nosocomial pneumonia in mechanically 135 Corely DE, Kirtland SH, Winterbauer RH, et al. Reproduc- ventilated patients. Am Rev Respir Dis 1990; 142:523–528 ibility of the histologic diagnosis of pneumonia among a 117 Timsit JF, Chevret S, Valcke J, et al. Mortality of nosocomial panel of four pathologists: analysis of a gold standard. Chest pneumonia in ventilated patients: influence of diagnostic 1997; 112:458–465 tools. Am J Respir Crit Care Med 1996; 154:116–123 136 Sterling TR, Ho EJ, Brehm WT, et al. Diagnosis and 118 Timsit JF, Misset B, Goldstein FW, et al. Reappraisal of treatment of ventilator-associated pneumonia: impact on distal diagnostic testing in the diagnosis of ICU-acquired survival; a decision analysis. Chest 1996; 110:1025–1034 pneumonia. Chest 1995; 108:1632–1639 137 Luna CM, Vujacich P, Niederman MS, et al. Impact of BAL 119 Craig CP, Connelley S. Effect of intensive care unit noso- data on the therapy and outcome of ventilator-associated comial pneumonia on duration and mortality. Am J Infect pneumonia. Chest 1997; 111:676–685 Control 1984; 12:233–238 138 Rello J, Gallego M, Mariscal D, et al. The value of routine 120 Bregeon F, Papazian L, Visconti A, et al. Relationship of microbial investigation in ventilator-associated pneumonia. microbiologic diagnostic criteria to morbidity and mortality Am J Respir Crit Care Med 1997; 156:196–200 in patients with ventilator-associated pneumonia. JAMA 139 Kollef MH, Ward S. The influence of mini-BAL cultures on 1997; 277:655–662 patients outcomes: implications for the antibiotic manage- 121 Fagon JY, Chastre J, Hance AJ, et al. Nosocomial pneumo- ment of ventilator-associated pneumonia. Chest 1998; 113: nia in ventilated patients: a cohort study evaluating attribut- 412–420 able mortality and hospital stay. Am J Med 1993; 94:281–288 140 Wearden PD, Chendrasekhar A, Timberlake GA. Compar- 122 Fagon JY, Chastre J, Hance AJ, et al. Evaluation of clinical ison of nonbronchoscopic techniques with bronchoscopic judgment in the identification and treatment of nosocomial brushing in the diagnosis of ventilator-associated pneumo- pneumonia in ventilated patients. Chest 1993; 103:547–553 nia. J Trauma 1996; 41:703–707 123 Marik PE, Brown WJ. A comparison of bronchoscopic vs 141 Bello S, Tajada A, Chacon E, et al. “Blind” protected blind protected specimen brush sampling in patients with specimen brushing versus bronchoscopic techniques in the suspected ventilator-associated pneumonia. Chest 1995; aetiolological diagnosis of ventilator- associated pneumonia. 108:203–207 Eur Respir J 1996; 9:1494–1499 124 Chastre J, Fagon JY, Bornet-Lecso M, et al. Evaluation of 142 Papazian L, Thomas P, Garbe L, et al. Bronchoscopic or bronchoscopic techniques for the diagnosis of nosocomial blind sampling techniques for the diagnosis of ventilator- pneumonia. Am J Respir Crit Care Med 1995; 152:231–240 associated pneumonia. Am J Respir Crit Care Med 1995; 125 Kollef MH, Bock KR, Richards RD, et al. The safety and 152:1982–1991 diagnostic accuracy of minibronchoalveolar lavage in pa- 143 Leal-Noval SR, Alfaro-Rodriguez E, Murillo-Cabeza F, et al. tients with suspected ventilator-associated pneumonia. Ann Diagnostic value of the blind brush in mechanically-venti- Intern Med 1995; 122:743–748 lated patients with nosocomial pneumonia. Intensive Care 126 Allaouchiche B, Jaumain H, Dumontet C, et al. Early Med 1992; 18:410–414 diagnosis of ventilator-associated pneumonia: is it possible to 144 Rouby JJ, De Lassale EM, Poete P, et al. Nosocomial define a cutoff value of infected cells in BAL fluid? Chest bronchopneumonia in the critically ill: histologic and bacte- 1996; 110:1558–1565 riologic aspects. Am Rev Respir Dis 1992; 146:1059–1066 127 Speich R, Wust J, Hess T, et al. Prospective evaluation of a 145 Marik PE, Careau P. A comparison of mini-bronchoalveolar semiquantitative dip slide method compared with quantita- lavage and blind-protected specimen brush sampling in tive bacterial cultures of BAL fluid. Chest 1996; 109:1423– ventilated patients with suspected pneumonia. J Crit Care 1429 1998; 13:67–72 128 Kollef MH, Eisenberg PR, Ohlendorf MF, et al. The 146 Grindlinger GA, Niehoff J, Hughes L, et al. Acute paranasal accuracy of elevated concentrations of endotoxin in bron- sinusitis related to nasotracheal intubation of head-injured choalveolar lavage fluid for the rapid diagnosis of gram- patients. Crit Care Med 1987; 15:214–217 negative pneumonia. Am J Respir Crit Care Med 1996; 147 Deutschman CS, Wilton P, Sinow J, et al. Paranasal sinusitis 154:1020–1028 associated with nasotracheal intubation: a frequently unrec- 129 Timset JF, Misset B, Azoulay E, et al. Usefulness of airway ognized and treatable source of sepsis. Crit Care Med 1986; visualization in the diagnosis of nosocomial pneumonia in 14:111–114 ventilated patients. Chest 1996; 110:172–179 148 Rouby JJ, Laurent P, Gosnach M, et al. Risk factors and 130 Jourdain B, Joly-Guillou ML, Dombret MC, et al. Useful- clinical relevance of nosocomial maxillary sinusitis in the ness of quantitative cultures of BAL fluid for diagnosing critically ill. Am J Respir Crit Care Med 1994; 150:776–783 nosocomial pneumonia in ventilated patients. Chest 1997; 149 Fassoulaki A, Pamouktsoglou P. Prolonged nasotracheal 111:411–418 intubation and its association with inflammation of paranasal 131 Torres A, el-Ebiary M, Fabregas N, et al. Value of intracel- sinuses. Anesth Analg 1989; 69:50–52 lular bacteria detection in the diagnosis of ventilator associ- 150 Holzapfel L, Chastang C, Demingeon G, et al. Randomized ated pneumonia. Thorax 1996; 51:378–384 study assessing the systematic search for maxillary sinusitis 132 el-Ebiary M, Torres A, Gonzalez J, et al. Quantitative in nasotracheally mechanically ventilated patients. Am J

CHEST / 117/3/MARCH, 2000 867 Respir Crit Care Med 1999; 159:695–701 169 Pemberton LB, Ross V, Cuddy P, et al. No difference in 151 Hansen M, Poulsen MR, Bendixen DK, et al. Incidence of catheter sepsis between standard and antiseptic central sinusitis in patients with nasotracheal intubation. Br J An- venous catheters: a prospective randomized trial. Arch Surg aesth 1988; 61:231–232 1996; 131:986–989 152 Zinreich SJ. Paranasal sinus imaging. Otolaryngol Head 170 Ciresi DL, Albrecht RM, Volkers PA, et al. Failure of Neck Surg 1990; 103:863–869 antiseptic bonding to prevent central venous catheter-re- 153 Holzapfel L, Chevret S, Madinier G, et al. Influence of lated infection and sepsis. Am Surg 1996; 62:641–646 long-term oro- or nasotracheal intubation on nosocomial 171 Williams JF, Seneff MG, Friedman BC, et al. Use of femoral maxillary sinusitis and pneumonia: results of a prospective, venous catheters in critically ill adults: prospective study. randomized, clinical trial. Crit Care Med 1993; 21:1132– Crit Care Med 1991; 19:550–553 1138 172 Cook D, Randolph A, Kernerman P, et al. Central venous 154 Darouiche R, Raad I, Heard SO, et al. A comparison of two catheter replacement strategies: a systemic review of the antimicrobial-impregnated central venous catheters. N Engl literature. Crit Care Med 1997; 25:1417–1424 J Med 1999; 340:1–8 173 Cobb DK, High KP, Sawyer WT, et al. A controlled trial of 155 Maki DG. Pathogenesis, prevention, and management of scheduled replacement of central venous and pulmonary infections due to intravascular devices used for infusion artery catheters. N Engl J Med 1992; 327:1062–1068 therapy. In: Bisno A, Waldvogel F, eds. Infections associated 174 Maki DG, Weise CE, Sarafin HW. A semiquantitative with indwelling medical devices. Washington, DC: Ameri- culture method for identifying intravenous-catheter related can Society for Microbiology, 1989; 161–177 infection. N Engl J Med 1977; 296:1305–1309 156 Mimoz O, Pieroni L, Lawrence C, et al. Prospective, 175 Raad I, Sabbagh MF, Rand KH, et al. Quantitative tip randomized trial of two antiseptic solutions for prevention of culture methods and the diagnosis of central venous cathe- central venous or arterial catheter colonization and infection ter-related infections. Diagn Microbiol Infect Dis 1991; in intensive care unit patients. Crit Care Med 1996; 24: 15:13–20 1818–1823 176 Sherertz RJ, Raad I, Belani A, et al. Three-year experience 157 Tacconelli E, Tumbarello M, Pittiruti M, et al. Central with sonicated vascular catheter cultures in a clinical micro- venous catheter-related sepsis in a cohort of 366 hospitalized biology laboratory. J Clin Microbiol 1990; 28:76–82 patients. Eur J Clin Microbiol Infect Dis 1997; 16:203–209 177 Platt R, Polk BF, Murdock B, et al. Mortality associated with 158 Wenzel RP, Edmond MB. The evolving technology of nosocomial urinary tract infection. N Engl J Med 1982; venous access. N Engl J Med 1991; 340:48–49 307:637–642 159 Maki DG, Brand JD. A comparative study of polyantibiotic 178 Daifuku R, Stamm W. Association of rectal and urethral and iodophor ointments in prevention of vascular catheter- colonization with urinary tract infection in patients with related infection. Am J Med 1981; 70:739–744 indwelling catheters. JAMA 1984; 252:2028–2030 160 Timsit JF, Sebille V, Farkas JC, et al. Effect of subcutaneous 179 Stark RP, Maki DG. Bacteriuria in the catheterized patient: tunneling on internal jugular catheter-related sepsis in crit- what quantitative level of bacteriuria is relevant? N Engl ically ill patients: a prospective randomized multicenter J Med 1984; 311:560–564 study. JAMA 1996; 276:1416–1420 180 Brown DF, Warren RE. effect of sample volume on yield of 161 Flowers RI, Schwenzer J, Kopel RF. Efficacy of an attach- positive blood cultures from adult patients with hematolog- able subcutaneous cuff for the prevention of intravascular ical malignancy. J Clin Pathol 1990; 43:777–779 catheter-related infection: a randomized controlled trial. 181 Paradisi F, Corti G, Mangani V. Urosepsis in the critical care JAMA 1989; 261:878–883 unit. Crit Care Clin 1998; 114:165–180 162 Hasaniya NWMA, Angelis M, Brown MR, et al. Efficacy of 182 Warren JW. Catheter-associated urinary tract infections. subcutaneous silver-impregnated cuffs in preventing central Infect Dis Clin North Am 1997; 11:609–619 venous catheter infections. Chest 1996; 109:1030–1032 183 Krieger JN, Kaiser DL, Wenzel RP. Urinary tract etiology of 163 Raad I, Darouiche R. Prevention of infections associated in hospitalized patients. J Infect Dis with intravascular devices. Curr Opin Crit Care 1996; 1983; 148:57–62 2:361–365 184 Garibaldi RA, Mooney BR, Epstein BJ, et al. An evaluation 164 Raad I, Darouiche RO, Hachem R, et al. Antimicrobial of daily bacteriologic monitoring to identify preventable durability and rare ultrastructural colonization of indwelling episodes of catheter-associated urinary tract infection. Infect central catheters coated with minocycline and rifampin. Crit Control 1982; 3:466–470 Care Med 1998; 26:219–224 185 Wilcox MH, Smyth ET. Incidence and impact of Clostrid- 165 Raad I, Darouiche R, Dupuis J, et al. Central venous ium difficile infection in the UK, 1993–1996. J Hosp Infect catheters coated with minocycline and rifampin for the 1998; 39:181–187 prevention of catheter-related colonization and blood 186 Brazier JS. The epidemiology and typing of Clostridium stream infections: a randomized, double-blind trial. Ann difficile. J Antimicrob Chemother 1998; 41(Suppl C):47–57 Intern Med 1997; 127:267–274 187 Lai KK, Melvin ZS, Menard MJ, et al. Clostridium difficile- 166 Maki DG, Stolz SM, Wheeler S, et al. Prevention of central associated diarrhea: epidemiology, risk factors, and infection venous catheter-related bloodstream infection by use of an control. Infect Control Hosp Epidemiol 1997; 18:628–632 antiseptic-impregnated catheter: a randomized, controlled 188 Adams HP, Brott TG, Crowell RM, et al. Guidelines for the trial. Ann Intern Med 1997; 127:257–266 management of patients with acute ischemic stroke: a 167 Civatta JM, Hudson-Civatta J, Ball S. Decreasing catheter- statement for the healthcare professionals from a special related infection and hospital costs by continuous quality writing group of the stroke council, American Heart Asso- improvement. Crit Care Med 1996; 24:1660–1665 ciation. Circulation 1994; 90:1588–1601 168 Sherertz RJ, Hard SO, Raad II, et al. Gamma radiation- 189 Borriello SP. Pathogenesis of Clostridium difficile infection. sterilized, triple-lumen catheters coated with a low concen- J Antimicrob Chemother 1998; 41(Suppl C):13–19 tration of chlorhexidine were not efficacious at preventing 190 Manabe YC, Vinetz JM, Moore RD, et al. Clostridium catheter infections in intensive care unit patients. Antimi- difficile colitis: an efficient clinical approach to diagnosis. crob Agents Chemother 1996; 40:1995–1997 Ann Intern Med 1995; 123:835–840

868 Reviews 191 Brazier JS. The diagnosis of Clostridium difficile-associated units. French ICU Group for Severe Sepsis. JAMA 1995; disease. J Antimicrob Chemother 1998; 41(Suppl C):29–40 274:968–974 192 Kelly CP, Pouthoulakis C, LaMont JT Clostridia Difficile 204 Levin PD, Hersch M, Rudensky B, et al. Routine surveil- Colitis. N Engl J Med 1994; 330:257–262 lance blood cultures: their place in the management of 193 Beck-Sague C, Jarvis WR. Secular trends in the epidemiol- critically ill patients. J Infect 1997; 35:125–128 ogy of nosocomial fungal infections in the United States, 205 Bennett IL, Beeson RB. Bacteremia: a consideration of 1980–1990. National Nosocomial Infections Surveillance some experimental and clinical aspects. Yale J Biol Med System. J Infect Dis 1993; 167:1247–1251 1954; 262:241–262 194 Rello J, Esandi ME, Diaz E, et al. The role of Candida sp 206 Chandrasekar PH, Brown WJ. Clinical issues of blood isolated from bronchoscopic samples in nonneutropenic cultures. Arch Intern Med 1994; 154:841–849 patients. Chest 1998; 114:146–149 207 Wormser GP, Onorato IM, Preminger TJ, et al. Sensitivity 195 el Ebiary M, Torres A, Fabregas N, et al. Significance of the and specificity of blood cultures obtained through intravas- isolation of Candida species from respiratory samples in critically ill, non-neutropenic patients: an immediate post- cular catheters. Crit Care Med 1990; 18:152–156 mortem histologic study. Am J Respir Crit Care Med 1997; 208 Ilstrup DM, Washington JA. The importance of volume of 156(2 Pt 1):583–590 blood culture in the detection of bacteremia and fungemia. 196 Adelson-Mitty J, Fink MP, Lisbon A. The value of lumbar Diagn Microbiol Infect Dis 1983; 1:107–110 puncture in the evaluation of critically ill, non-immunosup- 209 Mermel LA, Maki DG. Detection of bacteremia in adults: pressed, surgical patients: a retrospective analysis of 70 consequences of culturing an inadequate volume of blood. cases. Intensive Care Med 1997; 23:749–752 Ann Intern Med 1993; 119:270–272 197 Metersky ML, Williams A, Rafanan AL. Retrospective anal- 210 Levin PD, Yinnon AM, Hersch M, et al. Impact of the resin ysis: are fever and mental status indications for lumbar blood culture medium on the treatment of critically ill puncture in a hospitalized patient who has not undergone patients. Crit Care Med 1996; 24:797–801 neurosurgery. Clin Infect Dis 1997; 25:285–288 211 Davis LP, Fink-Bennett D. Nuclear medicine in the acutely 198 Dotson RG, Pingleton SK. The effect of antibiotic therapy ill patient: II. Crit Care Clin 1994; 10:383–400 on recovery of intracellular bacteria from bronchoalveolar 212 Meduri GU, Belenchia JM, Massie JD, et al. The role of lavage in suspected ventilator- associated nosocomial pneu- gallium-67 scintigraphy in diagnosing sources of fever in monia. Chest 1993; 103:541–546 ventilated patients. Intensive Care Med 1996; 22:395–403 199 Souweine B, Veber B, Bedos JP, et al. Diagnostic accuracy 213 Berrouschot J, Sterker M, Bettin S, et al. Mortality of of protected specimen brush and bronchoalveolar lavage in space-occupying (’malignant’) middle cerebral artery infarc- nosocomial pneumonia: impact of previous antimicrobial tion under conservative intensive care. Intensive Care Med treatments. Crit Care Med 1998; 26:236–244 1998; 24:620–623 200 Crowe M, Ispahani P, Humphreys H, et al. Bacteraemia in 214 Clifton GL, Allen S, Barrodale P, et al. A phase II study of the adult intensive care unit of a teaching hospital in moderate hypothermia in severe brain injury. J Neuro- Nottingham, UK, 1985–1996. Eur J Clin Microbiol Infect trauma 1993; 10:263–271 Dis 1998; 17:377–384 215 Marion DW, Obrist WD, Carlier PM, et al. The use of 201 Valles J, Leon C, Alvarez-Lerma F. Nosocomial bacteremia moderate therapeutic hypothermia for patients with severe in critically ill patients: a multicenter study evaluating head injuries: a preliminary report. J Neurosurg 1993; epidemiology and prognosis. Spanish Collaborative Group 79:354–362 for Infections in Intensive Care Units of Sociedad Espanola 216 Marion DW, Penrod LE, Kelsey SF, et al. Treatment of de Medicina Intensiva y Unidades Coronarias (SEMIUC). traumatic brain injury with moderate hypothermia. N Engl Clin Infect Dis 1997; 24:387–395 J Med 1997; 336:540–546 202 Brun-Buisson C, Doyon F, Carlet J. Bacteremia and severe 217 Mackowiak PA. Fever: blessing or curse? A unifying hypoth- sepsis in adults: a multicenter prospective survey in ICUs esis. Ann Intern Med 1994; 120:1037–1040 and wards of 24 hospitals. French Bacteremia-Sepsis Study 218 O’Donnel J, Axelrod P, Fisher C, et al. Use of and effec- Group. Am J Respir Crit Care Med 1996; 154(3 Pt 1):617– tiveness of hypothermia blankets for febrile patients in the 624 intensive care unit. Clin Infect Dis 1997; 24:1208–1213 203 Brun-Buisson C, Doyon F, Carlet J, et al. Incidence, risk 219 Lenhardt R, Negishi C, Sessler DI, et al. The effects of factors, and outcome of severe sepsis and in physical treatment on induced fever in humans. Am J Med adults: a multicenter prospective study in intensive care 1999; 106:550–555

CHEST / 117/3/MARCH, 2000 869