Neurosurg Focus 16 (6):e2, 2004, Click here to return to Table of Contents

Infectious origins of

DAVID B. COHEN, M.D., M.P.H. Department of Orthopaedic Surgery, Johns Hopkins University, Baltimore, Maryland

Infections in the spine can lead to a wide range of problems for both the patient and physician. There is perhaps no more devastating complication than the neurological consequences of a cauda equina syndrome due to spinal infec- tion. A variety of organisms and origins can make diagnosis and treatment of spinal infection a difficult task. Both pyo- genic and nonpyogenic organisms can cause vertebral involvement and can result in an epidural abscess with neuro- logical compromise. Nonetheless, these two types of infections vary greatly in terms of associated patient demographics, clinical course, and treatments. The purpose of this paper was to review these types of infections and summarize treatment recommendations for this difficult condition.

KEY WORDS • cauda equina syndrome • infection • epidural abscess • tuberculosis

Infections in the spine can lead to a wide range of prob- Neurological deficits ranging from paraplegia to para- lems for both the patient and physician. There is perhaps paresis are present in 70 to 80% of patients. The reported no more devastating complication than the neurological incidence of fever varies greatly in the literature, ranging consequences of a cauda equina syndrome caused by spi- from 35 to 70%.31,35 Approximately one third of patients nal infection. A variety of organisms and origins can make will present with loss of bowel or bladder function. diagnosis and treatment of spinal infection a difficult task. Both pyogenic and nonpyogenic organisms can cause ver- Diagnostic Evaluation tebral involvement and can result in an epidural abscess with neurological compromise. Regardless, these two Typically, laboratory studies can be helpful in confirm- types of infections vary greatly in terms of associated ing the diagnosis of a spinal epidural abscess, but results patient demographics, clinical course, and treatments. of such tests are usually nonspecific. An elevated periph- eral white blood cell count is present in 50 to 60% of pa- tients,7,15,20,31,35 whereas an ESR and/or C-reactive protein PYOGENIC SPINAL EPIDURAL ABSCESS level is present in more than 90%.30,31,35 Both ESR and C- Patient Demographics reactive protein level are typically elevated following any spinal procedure (including lumbar puncture).36 Thus, Pyogenic spinal epidural abscesses are a rare but clini- although they can be useful markers of the effectiveness cally important entity with an overall incidence of be- of treatment of a spinal epidural abscess, their diagnostic tween 0.2 and 1.2 cases/10,000 hospital admissions.3,13,31,33 value is limited. The mean age at presentation ranges from 50 to 60 years, Obtaining tissue for culture and peripheral blood cul- and men are twice as likely to harbor the disease than tures prior to instituting antibiotic therapy is optimal. women (~2:1 male/female ratio).7,9,27,31,35 Most patients Peripheral blood cultures can demonstrate the offending who develop a spinal epidural abscess are immu- organism 50 to 70% of the time, whereas direct tissue cul- nocompromised due to chronic disease (50–60% of pa- tures demonstrate yields of 90%.5,6,8,18 tients), have a history of intravenous drug abuse (~35% of patients), or have undergone spinal surgery (10–20% of patients).31,35 Neuroimaging Studies Plain radiographs of the spine are the easiest imaging Clinical Features studies to obtain, but will typically remain normal for 2 to The initial diagnosis of a spinal epidural abscess is often 3 weeks after the onset of infection. Initial radiographic difficult because the classic clinical triad of fever, back findings, which include loss of disc space height, loss of 3 soft tissue planes, and subchondral endplate haziness lag pain, and progressive neurological deficit is not always 5,6 present. On clinical presentation, is the most behind the patient’s clinical course. Radionuclide imag- common physical symptom (85–90% of patients).14,31,35 ing can often demonstrate changes prior to radiography. Indium-111, 67Ga, and 99Tc scans all play a role in the diag- nosis of an associated with an epidural Abbreviations used in this paper: CT = computerized tomography; abscess, but are not specific for the presence of a spinal ESR = erythrocyte sedimentation rate; MR = magnetic resonance. epidural abscess.28,37

Neurosurg. Focus / Volume 16 / June, 2004 5

Unauthenticated | Downloaded 09/25/21 12:59 PM UTC D. B. Cohen

Given the need to obtain accurate images of both soft py.7,18,30 The surgical approach for decompression should and osseus tissues in the context of a spinal infection, MR be modified based on particular characteristics of the case imaging has become the modality of choice in evaluat- including the location of the infection (anterior compared ing spinal infections. In the presence of an infection, T2- with posterior), the region of the spine involved (cervical, weighted signal intensity will increase due to the asso- thoracic, or lumbar), the presence of a significant bone or ciated edema, whereas T1-weighted signal intensity will paraspinal abscess, and the degree of bone destruction decrease due to replacement of the marrow fat.1,2 Ga- revealed on CT scanning. dolinium enhancement can help differentiate infection Many cervical abscesses are treated using anterior de- from postsurgical change and pus from granulation tis- compression with corpectomy and fusion with an auto- sue.1,26,29,32 This combination allows MR imaging studies graft.31 In the thoracic spine the approach can vary de- to be 95% accurate in the diagnosis of spinal infections pending on the location of compression, but generally a and epidural abscesses (Fig. 1).1,26,29,32 Sagittal MR images posterior decompression of an isolated epidural abscess of the entire spine are recommended to identify the span can be accomplished by performing a laminectomy. The of the abscess and possible skip lesions. Because of its use of a costotransversectomy can facilitate a moderate accuracy, MR imaging has supplanted CT scanning as the degree of anterior decompression through the same inci- modality of choice in detecting spinal infections and sion.7,10 In the lumbar spine posterior decompression is epidural abscesses. Nonetheless, CT scanning still plays often adequate, unless there is a large abscess in the ante- an important role in guiding a biopsy as well as in assess- rior spine as well. Either a transforaminal or an anterior ing the degree of bone destruction. approach can be used to drain the abscess, but a more ade- quate debridement can be achieved through an anterior Microbiological Characteristics procedure.7,10 If there is a significant amount of anterior bone de- Staphylococcus aureus remains the most common path- struction or a significant paraspinal/bone abscess, then ogen in pyogenic spinal epidural abscesses, occurring in 7,18,27,31 formal anterior spinal debridement is recommended. between 25 and 60% of patients. During the past Through various thoracotomies, any level of the thoracic decade, however, antibiotic-resistant organisms as well as spine from T-2 to T-11 can be adequately accessed. From an increase in intravenous drug abuse have revealed many the level of T11–L1, a 10th rib thoracotomy can be ex- new pathogens. Methicillin-resistant S. aureus now ac- tended into a transthoracic transdiaphragmatic retroperi- counts for between 15 and 20% of pathogens in spinal epi- toneal approach to access the spine. The lumbar spine dural abscess, with gram-negative pathogens such as from L-2 to the sacrum can be easily accessed via a stan- Pseudomonas and Escherichia coli becoming more com- 31 dard retroperitoneal approach. Although significant focal mon as well. scarring and inflammation near the locus of infection can make these approaches technically challenging, excellent Methods of Treatment exposure can normally be achieved. The mainstay of treatment for pyogenic spinal epidural Once exposure is achieved, the infection should be abscess is surgical decompression of the thecal sac with widely debrided by excision of infected and necrotic bone, drainage of the abscess and long-term antibiotic thera- disc, and paraspinal tissues. The thecal sac is normally de-

Fig. 1. Left: Axial T1-weighted Gd-enhanced MR image revealing an epidural abscess with rim enhancement.

Center: Sagittal T2-weighted MR image demonstrating increased signal in vertebral bodies and discs. Right: Sagittal

T1-weighted Gd-enhanced MR image demonstrating epidural granulation tissue.

6 Neurosurg. Focus / Volume 16 / June, 2004

Unauthenticated | Downloaded 09/25/21 12:59 PM UTC Infectious origins of cauda equina syndrome compressed on debridement. Reconstruction of the local and the immunodeficient population, granulomatous in- defect with the aid of a structural autologous bone graft fections are on the rise. is then performed. Spinal position is then maintained Infection with Mycobacterium tuberculosis is relatively through bracing, or a second-stage posterior spinal stabi- uncommon in most of the technologically developed na- lization procedure is performed either on the same day or tions; however, it is estimated that more than 1.7 billion at a later date. Appropriate antibiotic therapy is then con- people worldwide have been infected during their life- tinued for 6 weeks or longer to eradicate infection. time.17,24 Although most primary cases of tuberculosis are Nonsurgical methods that include percutaneous biopsy pulmonary in nature, during reactivation of the disease, of the abscess followed by intravenous antibiotic therapy musculoskeletal involvement can frequently occur. This can play a role in the management of spinal epidural ab- results in approximately 10% of all patients with tubercu- scess; however, these methods should only be undertaken losis demonstrating musculoskeletal involvement together in very limited instances. Some authors have asserted that with Pott disease (tuberculous spondylodiscitis) and ac- patients with complete paraplegia lasting longer than 48 to counts for half of all musculoskeletal cases. 72 hours can be treated nonsurgically because they have little chance of neurological recovery.30,31 Patients with Clinical Presentation pan-spinal involvement as well as those medically unable 31,35 Rather than experiencing an acute onset of symptoms, to undergo surgery can also be treated nonsurgically. patients with Pott disease typically have a more insidious The most controversial group to treat nonsurgically con- onset of symptoms, especially compared with patients sists of those patients with a lumbosacral epidural abscess suffering from pyogenic infections. Back pain is an almost and a nondiagnostic neurological examination. Some au- universal symptom in patients with spinal tuberculosis,16, thors indicate that appropriate culture methods, intraven- 21,24 but neurological signs vary widely, although 10 to ous antibiotic agents, and careful monitoring of the neuro- 80% of patients present with deficits.16,21,24 Fever is not logical factors can be an acceptable treatment regimen in 22,23,35,39 typical even though it occurs in approximately 30 to 50% this group. Any decline in the patient’s neurological of patients.16,21,24 Other constitutional symptoms such as condition, however, mandates the need for rapid surgical 4,7,15,25,38 malaise and weight loss are far more common as a result intervention. of the chronic nature and insidious progression of the dis- ease.16,17,21,24 These disease characteristics can often lead to Patient Outcomes significant deformity at the time of diagnosis, which is Patient outcome following treatment of a spinal epidur- uncommon in pyogenic infections. al abscess varies depending on a number of factors. Those with multiple medical conditions or a history of spinal Diagnostic Evaluation surgery appear to have worse outcomes.19,31,35 Some au- thors assert that patients with methicillin-resistant S. aur- As in cases of pyogenic infections, laboratory studies can be helpful in confirming a diagnosis of nonpyogenic eus infections also have worse outcomes,31 but this may be due to a delay in obtaining appropriate antibiotic coverage infection, but results are frequently inconclusive. An ele- for the infection. Patients with thoracic infections demon- vated peripheral white blood cell count does not typically strate worse outcomes than those with infections at either occur in spinal tuberculosis. In contrast, the chronic nature of the disease or an associated immunosuppressed state cervical or lumbar sites,30,31 perhaps because of the vascu- lar watershed area of the in the thoracic spine. will often cause a lower than normal count. Typically, the ESR is moderately elevated.16,17,21,24 A purified protein de- Both the degree of neurological dysfunction and the rivative test is often positive, but can result in a false-neg- duration of time between the onset of symptoms and sur- ative finding in a chronically debilitated or immunosup- gical treatment can affect patient outcomes. In one series pressed patient.17 Sputum cultures and acid-fast staining 45% of those presenting with paraparesis regained ambu- may be helpful if active pulmonary disease is present, but latory status, and every patient who presented with ambu- 31 require many weeks to grow. Definitive diagnosis is usu- latory status was able to maintain it. Furthermore, these ally made based on CT-guided biopsy and cultures of the authors indicated that in patients who present with para- lesion, but again this can require long periods for culture paresis, treatment administered within 24 hours can result growth. More recently, subjecting the biopsy specimen to in better outcomes, with up to 90% of patients regaining a polymerase chain reaction study has been shown to be some neurological function and 45% becoming ambulato- more than 93% accurate in making a diagnosis and can ry compared with 45% regaining no function when treat- help detect drug-resistant strains before initiating antibiot- ment was delayed for more than 24 hours. ic therapy.17,24

NONPYOGENIC EPIDURAL ABSCESS Radiological Imaging Unlike pyogenic infections, plain radiographs will often Patient Demographics demonstrate many bone changes due to the insidious na- Although most spinal infections in the technologic- ture of the nonpyogenic infection. A varying degree of ally developed regions are the result of pyogenic organ- deformity from bone erosion is typically seen at the time isms, in areas that are less technologically developed and of patient presentation.21 Nuclear imaging modalities are in the immunocompromised population, nonpyogenic or- less useful and reliable than in pyogenic infections, with ganisms are responsible for most epidural abscesses. In false-negative rates approaching 40 and 70% for 99mTc and the inner cities of the US among both recent immigrants 67Ga scans, respectively.17

Neurosurg. Focus / Volume 16 / June, 2004 7

Unauthenticated | Downloaded 09/25/21 12:59 PM UTC D. B. Cohen

Magnetic resonance imaging remains the best single progressive neurological deficit (unless it is complete modality for the evaluation of extent of disease and the paraplegia lasting 3 days), severe spinal instability and detection of epidural compression and granulomas. deformity, or failure of medical management to control Gadolinium enhancement will often allow for differentia- disease.24,34 tion of abscess and granulomas.5,16 In nonpyogenic infec- Nonsurgical Management. The nonsurgical manage- tions, the disc spaces are relatively resistant to invasion ment of nonpyogenic epidural abscesses can be stated in compared with disc spaces in pyogenic disease. This will three simple concepts: nutritional support, spinal immobi- often result in an image of multiple-level vertebral in- lization, and appropriate antibiotic therapy. Due to the volvement, with relative sparing of the discs (Fig. 2). chronic and often indolent nature of these infections, most patients present with protein-calorie malnutrition. This Microbiological Characteristics type of malnutrition causes impaired cell-mediated immu- In less technologically developed nations, tuberculosis nity, and impaired pulmonary immune defenses can have is endemic and M. tuberculosis accounts for the majority a direct impact on the success of any therapy. Thus nutri- of all organisms causing nonpyogenic epidural abscess- tional supplementation must be a priority in these cases.17 es.24,34 In technologically developed countries, tuberculo- Spinal immobilization through either bedrest or orthosis sis is not generally endemic, but M. tuberculosis still caus- to control pain and deformity is the standard of care. Or- es most of the nonpyogenic infections that occur. Note, thoses ranging from halo vests to cervicothoracolumbar however, that other mycobacteria, fungi, and spirochetes devices may be needed, depending on the location and cause granulomatous infections in the immunocompro- extent of spinal involvement. A minimum of 3 months of mised populations. Mycobacteri aceae, Actinomyceta- brace therapy is required, but the degree of deformity and ceae, Nocardia spp., Candida spp., and Aspergillus spp. osseous destruction and the disease response may length- have all been seen as pathogens in nonpyogenic spinal en the period of this treatment.21,24,34 Antibiotic therapy infections and epidural abscesses.5,6 should be initiated once it is confirmed that diagnostic tis- sue has been obtained. Triple antibiotic therapy involving Treatment Methods a 12-month administration of isoniazid, rifampin, and pyr- Nonpyogenic epidural abscesses are amenable to azinamide is standard, but the regimen may need to be both nonsurgical and surgical management, depending on modified if there is a failure to respond or if drug resis- the individual case. In general, the goals of treatment of tance is present, which can be determined based on poly- tuberculous epidural infections include disease eradica- merase chain reaction testing.24,34 For nontuberculous or- tion, pain relief, preservation of neurological function, and ganisms, different drug regimens are needed; hence, it is spinal stability.24,34 Because most patients with these infec- paramount to have an infectious disease specialist control tions are debilitated due to the chronic nature of the dis- the antibiotic care. ease, nonsurgical management is often preferred. Thus, Surgical Management. The surgical management of indications for surgical care are limited and include failure nonpyogenic epidural infection is normally limited to cas- to obtain diagnostic tissue by other means, presence of a es of unstable deformity or progressive neurological defi-

Fig. 2. Left: Sagittal T1-weighted MR image obtained in a patient with lumbar kyphosis and reactivation of Pott dis-

ease. Note the primary involvement of multiple vertebral bodies. Center: Sagittal T2-weighted MR image exhibiting

relative sparing of disc spaces and epidural compression. Right: Sagittal T1-weighted Gd-enhanced image displaying relative sparing of disc spaces and epidural involvement.

8 Neurosurg. Focus / Volume 16 / June, 2004

Unauthenticated | Downloaded 09/25/21 12:59 PM UTC Infectious origins of cauda equina syndrome cit. Given that isolated posterior infections are extremely tion, in Frymoyer JW, Ducker TB, Hadler NM, et al (eds): The rare, the mainstay of surgical treatment includes a com- Adult Spine: Principles and Practice. New York: Raven plete and thorough debridement of all necrotic bone, soft Press, 1991, Vol 2, pp 1873–1890 tissues, discs, and exudates back to health-viable margins. 11. Govender S, Parbhoo AH: Support for the anterior column with allografts in tuberculosis of the spine. J Bone Joint Surg Br With debridement, the spinal canal will be decompressed. 81:106–109, 1999 Following copious irrigation, the spine is reconstructed 12. Guven O, Kumano K, Yalcin S, et al: A single stage posterior using structural bone grafts anteriorly and often posterior approach and rigid fixation for preventing kyphosis in the treat- spinal instrumentation.11,12,16,21,40 Traditionally, an autograft ment of spinal tuberculosis. Spine 19:1039–1043, 1994 has been the bone graft of choice. In recent years, howev- 13. Hakin RN, Burt AA, Cook JB: Acute spinal epidural abscess. er, structural allograft has been used very successfully.11,40 Paraplegia 17:330–336, 1979 Surgical management still necessitates the 12-month 14. Hancock DO: A study of 49 patients with acute spinal extradur- administration of antibiotic therapy to prevent disease re- al abscess. Paraplegia 10:285–288, 1973 currence. 15. Hlavin ML, Kaminski HJ, Ross JS, et al: Spinal epidural ab- scess: a ten-year perspective. Neurosurgery 27:177–184, 1990 Patient Outcomes 16. Hsu LC, Cheng CL, Leong JC: Pott’s paraplegia of late onset. The cause of compression and results after anterior decompres- The result of treating of nonpyogenic epidural infec- sion. J Bone Joint Surg Br 70:534–538, 1988 tions can be quite remarkable. With nonsurgical manage- 17. Jacofsky D, Currier BL: Infections of the spine, in Fardon DF, ment, if the spinal deformity can be controlled with brace Garfin SR, Abitbol JJ, et al (eds): Orthopaedic Knowledge therapy, most areas of destruction will undergo some de- Update: Spine 2. Rosemont, IL: American Academy Ortho- gree of autofusion. A significant proportion of patients paedic Surgery, 2002, pp 431–442 with neurological deficits will demonstrate a progressive 18. Kaufman DM, Kaplan JG Litman N: Infectious agents in spinal return of neurological function.11,12,16,24,34,40 epidural abscesses. Neurology 30:844–850, 1980 19. Khanna PK, Malik GM, Rock JP, et al: Spinal epidural abscess: evaluation of factors influencing outcome. Neurosurgery 39: CONCLUSIONS 958–964, 1996 Cauda equina syndrome associated with spinal infec- 20. Lifeso RM: Pyogenic spinal sepsis in adults. Spine 15: tions can result in devastating outcomes in patients. Based 1265–1271, 1990 21. Lifeso RM, Weaver P, Harder EH: Tuberculous spondylitis in on the nature of the infection (pyogenic compared with adults. J Bone Joint Surg Am 67:1405–1413, 1985 nonpyogenic) and the duration of symptomatology, either 22. Mampalam TJ, Rosegay H, Andrews BT, et al: Nonoperative surgical or nonsurgical care can play a pivotal role in the treatment of spinal epidural infections. J Neurosurg 71: long-term outcome in a patient. In general, recently pro- 208–210, 1989 gressive neurological deficits (that is, those occurring 23. Messer HD, Lenchner GS, Brust JC, et al: Lumbar spinal ab- within 48 hours) require surgical intervention in almost all scess managed conservatively. Case report. J Neurosurg 46: patients regardless of the nature of the infection. In pa- 825–829, 1977 tients with pyogenic abscesses without neurological def- 24. Moon MS: Tuberculosis of the spine. Controversies and a new icits, medical management can be initiated, but any sign of challenge. Spine 22:1791–1797, 1997 decline in neurological function mandates surgical inter- 25. Mooney RP, Hockberger RS: Spinal epidural abscess: a rapidly progressive disease. Ann Emerg Med 16:1168–1170, 1987 vention. 26. Numaguchi Y, Rigamonti D, Rothman MI, et al: Spinal epidur- References al abscess: evaluation of gadolinium-enhanced MR imaging. Radiographics 13:545–560, 1993 1. An HS, Vaccaro AR, Dolinskas CA, et al: Differentiation be- 27. Nussbaum ES, Rigamonti D, Standiford H, et al: Spinal epidur- tween spinal tumors and infections with magnetic resonance al abscess: a report of 40 cases and review. Surg Neurol 38: imaging. Spine 16 (Suppl 8):S334–S338, 1991 225–231, 1992 2. Antguaco EJ, McConnell JR, Chadduck WM, et al: MR Imag- 28. Ozuna RM, Delamarter RB: Pyogenic vertebral osteomyelitis ing of spinal epidural sepsis. AJR 149:1249–1253, 1987 and postsurgical disc space infections. Orthop Clin North Am 3. Baker AS, Ojemann RG, Swartz MN, et al: Spinal epidural ab- 27:87–94, 1996 scess. N Engl J Med 293:463–468, 1975 29. Post MJ, Sze G, Quencer RM, et al: Gadolinium-enhanced MR 4. Bouchez B, Arnott G, Delfosse JM: Acute spinal epidural ab- in spinal infection. J Comput Assist Tomogr 14:721–729, scess. J Neurol 231:343–344, 1985 1990 5. Currier BL, Eismont FJ: Infections of the spine, in Herkowitz 30. Reihsaus E, Waldbaur H, Seeling W: Spinal epidural abscess: a HN, Eismont FJ, Garfin SR, et al (eds): The Spine, ed 4. Phila- meta-analysis of 915 patients. Neurosurg Rev 23:175–205, delphia: WB Saunders, 1999, Vol II, pp 1207–1258 2000 6. Currier BL, Heller JG, Eismont FJ: Cervical spinal infections, 31. Rigamonti D, Liem L, Sampath P, et al: Spinal epidural abscess: in Clark CR, Ducker TB, Dvorak J, et al (eds): The Cervical contemporary trends in etiology, evaluation, and management. Spine, ed 3. Philadelphia: Lippincott-Raven, 1998, pp 659–690 Surg Neurol 52:189–197, 1999 7. Danner RL, Hartman BJ: Update of spinal epidural abscess: 35 32. Sandhu FS, Dillon WP: Spinal epidural abscess: evaluation with cases and review of the literature. Rev Infect Dis 9:265–274, contrast-enhanced MR imaging. AJNR 12:1087–1093, 1991 1987 33. Schlossberg D, Shulman JA: Spinal epidural abscess. South 8. Eismont FJ, Bohlman HH, Soni PL, et al: Pyogenic and fungal Med J 70:669–673, 1977 vertebral osteomyelitis with paralysis. J Bone Joint Surg Am 34. Sixth report of the Medical Research Council Working Party on 65:19–29, 1983 Tuberculosis of the Spine: Five-year assessments of controlled 9. Ericsson M, Algers G, Schliamser SE: Spinal epidural abscess- trials of ambulatory treatment, debridement and anterior spinal es in adults: review and report of iatrogenic cases. Scand J In- fusion in the management of tuberculosis of the spine: studies fect Dis 22:249–257, 1990 in Bulawayo (Rhodesia) and in Hong Kong. J Bone Joint Surg 10. Frymoyer JW: Segmental instability: overview and classifica- Br 60:163–177, 1978

Neurosurg. Focus / Volume 16 / June, 2004 9

Unauthenticated | Downloaded 09/25/21 12:59 PM UTC D. B. Cohen

35. Tang HJ, Lin HJ, Liu YC, et al: Spinal epidural abscess—expe- of spinal epidural abscesses: case report and review. Clin In- rience with 46 patients and evaluation of prognostic factors. J fect Dis 15:22–27, 1992 Infect 45:76–81, 2002 40. Yilmaz C, Selek HY, Gurkan I, et al: Anterior instrumentation 36. Thelander U, Larsson S: Quantitation of C-reactive protein for the treatment of spinal tuberculosis. J Bone Joint Surg Am levels and erythrocyte sedimentation rate after spinal surgery. 81:1261–1267, 1999 Spine 17:400–404, 1992 37. Tzen KY, Yen TC, Yang RS, et al: The role of 67Ga in the early Manuscript received April 23, 2004. detection of spinal epidural abscesses. Nucl Med Commun 21: Accepted in final form May 2, 2004. 165–170, 2000 Address reprint requests to: David B. Cohen, M.D., M.P.H., 38. Verner EF, Musher DM: Spinal epidural abscess. Med Clin Department of Orthopaedic Surgery, Johns Hopkins University, North Am 69:375–384, 1985 601 North Caroline Street, Baltimore, Maryland 21287. email: 39. Wheeler D, Keiser P, Rigamonti D, et al: Medical management [email protected].

10 Neurosurg. Focus / Volume 16 / June, 2004

Unauthenticated | Downloaded 09/25/21 12:59 PM UTC