Article Text

Download PDFPDF

Spinal epidural abscess: the importance of early diagnosis and treatment
  1. A R Mackenziea,
  2. R B S Lainga,
  3. C C Smitha,
  4. G F Kaarb,
  5. F W Smithc
  1. aThe Infection Unit, Aberdeen Royal Infirmary, Aberdeen, bDepartment of Neurosurgery, Aberdeen Royal Infirmary, Aberdeen, cDepartment of Radiology, Woodend Hospital, Aberdeen
  1. A R Mackenzie, The Infection Unit, Aberdeen Royal Infirmary, Aberdeen AB25 2ZN, UK. Telephone 0044 1224 681818; fax 0044 1224 685307.


OBJECTIVES To remind clinicians of the dangers of delayed diagnosis and the importance of early treatment of spinal epidural abscess.

METHODS A review of the literature on spinal epidural abscess and a comparison of the published literature with local experience.

RESULTS Imaging with MRI or CT enables early diagnosis of spinal epidural abcess and optimal therapy is surgical evacuation combined with 6–12 weeks (median 8 weeks) of antimicrobial chemotherapy. Clinical features are fever, pain, and focal neurological signs and may be associated with preceding and pre-existing bone or joint disease. The commonest aetiological organism is S aureus.

CONCLUSION Early diagnosis and appropriate early antimicrobial chemotherapy with surgery is associated with an excellent prognosis.

  • spinal epidural abscess
  • diagnosis
  • treatment
  • magnetic resonance imaging

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Spinal epidural abscess is a rare but potentially devastating condition. When recognised and treated early the outcome is excellent.1 Late diagnosis or inadequate treatment may leave the patient with long term severe or disabling incapacity. The need for early diagnosis and the benefits of appropriate early antimicrobial chemotherapy with surgery are discussed.


Spinal epidural abscess is an uncommon condition with an estimated incidence of 0.2–2.0/10 000 hospital admissions,2 3 and a peak incidence in the sixth and seventh decades of life.3 4 At this hospital over a five year period we have seen eight patients admitted from the community with spinal epidural abcess, giving an estimated incidence of 0.25/10 000 admissions. Their median age was 61.5 (range 48–75) The disease is rare in children, with fewer than 90 cases reported.5 Conditions commonly associated with spinal epidural abcess include diabetes mellitus, intravenous drug misuse, chronic renal failure, alcoholism, and cancer.2-4


Most spinal epidural abcesses are thought to result from the haematogenous spread of bacteria usually from a cutaneous or mucosal source and this occurred in five of our eight patients. In the series by Baker et al, 2 haematogenous spread of infection was identified in 26% of cases, the sources of bacteraemia including furuncles, pharyngitis, and dental abscesses. Haematogenous spinal epidural abcess are primarily located in the posterior aspect of the spinal canal.6 The direct spread of infection into the epidural space from a source adjacent to the spine—for example, spondylitis or discitis—is also well described.7 8 In such cases the abscesses are often located in the anterior aspect of the spinal canal and are thought to originate from the adjacent vertebral body. Direct spread from a paraspinal abscess may also occur and was the infective origin in one of our patients. In a recent review of 43 cases of spinal epidural abcess, 44% had coexistent vertebral osteomyelitis.9Vertebral osteomyelitis was present in one of our eight patients (12.5%).

Postoperative abscesses account for 16% of all spinal epidural abcesses2 and epidural catheter insertion is another recognised predisposing factor.10 11 None of our patients had postoperative abscesses.

Blunt trauma is reported to precede the symptoms of spinal epidural abcess in 15%-35% of cases7 and it is postulated that trauma may result in the formation of an epidural haematoma which subsequently becomes infected.12 A history of recent trauma was recorded in two (25%) of our patients although the source of bacteraemia, as is often the case, was neither obvious nor identified in either patient.

Clinical features

The typical features of spinal epidural abcess are recognisable clinically—notably fever, spinal pain and tenderness, and radiating root pain followed by limb weakness. Pain is the most consistent symptom and occurs in virtually all patients at some time during their illness. Spinal pain and fever are usually the only symptoms present before a precipitous neurological deterioration occurs13and were the only features in two of our patients Most patients have major neurological signs before surgery5 which are thought to result from a compromised spinal cord vasculature rather than direct cord compression.14 When septicaemia dominates the clinical picture, as in two of our patients, whose presenting features included jaundice, fever, confusion, and rais- ed fibrinogen degradation products, the neurological symptoms may go unnoticed. This may also be the case in patients confined to bed. In patients with chronic infection,constitutional symptoms of fever, weight loss, and systemic upset may predominate over the neurological syndrome10 and lead to late diagnosis and treatment.

The superior quality of MRI over CT is demonstrated in this figure. Sagittal T1 weighted MRI of lumbosacral spine showing osteomyelitis of the L5 vertebral body (large arrow) with spinal epidural abcess L4-S1 anteriorly displacing the meninges (small arrows); (A) before contrast and (B) after contrast. (C) Axial MRI of L5 vertebra showing anterior displacement of the thecal sac by a spinal epidural abcess (arrowed).


The described sites of spinal epidural abcess are variable. Only 20% occur anterior to the spinal cord.4 Our experience suggests that abscesses in the cervical or upper thoracic region are not uncommon—three of our eight patients had cervical or thoracic spinal epidural abcesses. Larger studies, however, describe a preponderance of lower thoracic and lumbar abscesses.15The patients with lumbar abscesses may be misdiagnosed as having a herniated lumbar disc.16


Staphylococcus aureus was isolated in six of our eight patients (75%). This is the most commonly recognised causative organism and accounts for 57%-73% of reported abscesses.2 10 14 17 In a review of five studies between 1930 and 1982,4 Danner and Hartman found thatS aureus was isolated in 62%, other gram positive cocci cultured in 10%, gram negative organisms in 18%, and anaerobes in 2%. After S aureus,Mycobacterium tuberculosis is the most frequent cause and accounted for 25% of infections in the series reported by Kaufman et al.18 Streptococcus milleri has previously been reported as a cause of spinal epidural abscess19 and is recognised as an organism with a tendency to abscess formation.20 Haemophilus parainfluenzae,21 Brucella species,22 andActinomyces israelii 23 are amongst the many other isolates described. Disseminated fungal infections such as cryptococcosis, aspergillosis, and blastomycosis are rare causes and usually arise in immunocompromised patients.18 Aspergillus species are known to cause spinal epidural abcess in patients with AIDS.24


The extent and location of an abscess is best visualised using MRI. Multiplanar imaging and visualisation of the soft tissues—including the spinal cord, paraspinal musculature, and the disc—confers a distinct advantage of MRI over CT and myelography25 and where available, should always be utilised (figure A, B, C).

Antimicrobial therapy

Empirical antimicrobial treatment of spinal epidural abcess should be bactericidal, start early, be delivered intravenously in high doses, and be continued well into the postoperative period.7Initial treatment should be broad spectrum and include a combination of drugs with bactericidal activity against staphylococci, anaerobes, and gram negative organisms. If a methicillin sensitiveS aureus is isolated, successful treatment is possible with a first generation cephalosporin,4 a penicillinase resistant penicillin,26 combination vancomycin plus aminoglycoside,27 or trimethoprimsulphamethoxazole.28 Parenteral treatment should be continued for at least 4 weeks and may be prolonged for 8 weeks or longer if vertebral osteomyelitis is suspected.4 7 When no pathogen is isolated, broad “best guess” bactericidal cover is safest and best.


Surgical treatment entails emergency evacuation of the pus with decompression of the spinal cord and nerve roots. In the presence of paraparesis, the outcome depends on the time that elapses from its onset to surgery.18 The longer the lapse the greater is the usual long term legacy.


The prognosis is dependent on the clinical and neurological condition of the patient at presentation and any delay in the diagnosis4 29 or instigation of appropriate antimicrobial treatment. Patients who present with overt sepsis do poorly. All such patients in one series died.30 Patients who are plegic preoperatively also do poorly, although this is time dependent.29 In the series of Danner and Hartman,4 patients with preoperative neurological defects present for less than 36 hours showed some improvement, whereas nine of 11 patients with deficits present for more than 36 hours before surgery did not. In the review by Maslen et al,29 patients paralysed preoperatively for greater than 12 hours did not recover neurological function and those paralysed for greater than 36 hours often died. In our series, the six patients who had developed neurological signs underwent surgery within 8 hours of their onset and had a good outcome. Some authors have found that the outcome of cervical abscesses is worse than those arising in thoracic or lumbar sites.14 In our series, cervical cases had good outcomes compared with those with thoracolumbar abscesses. Two cervical cases had MRI as the primary spinal investigation and at relatively early stages.

Although the duration of neurological deficit before surgery determines, in part, the neurological outcome, it is not the sole factor. It has been suggested that permanent axonal or cell body damage, for which we do not have a clinical indicator, has prognostic relevance. In our experience one patient who was paraparetic for only three hours before surgery remained paraparetic afterwards. By contrast, another patient who was paraplegic for seven hours made a virtually complete recovery after surgery.


Although the mortality from spinal epidural abcess has improved from that described in the early reports, it has remained surprisingly consistent over the past several decades, at about 14%.4 29 Mortality is clearly related to delay in instituting appropriate therapy, emphasising the need for early diagnosis and treatment. In addition, these patients should be followed up for several months—even after completion of antibiotic therapy, because of the spinal instability which may result from the infection or the necessary surgical intervention.1 There were no deaths in our series.


Patients undergoing these procedures often become malnourished and poor nutritional status is associated with increased postoperative complications. Close attention should therefore be paid to the perioperative nutrition and consideration given to preoperative and postoperative nutritional supplementation.31


The appreciable neurological recovery seen in some patients reflects the value of prompt diagnosis and early treatment. The key to successful management is early diagnosis, which requires that involved clinicians consider the diagnosis. Repeated spinal and neurological examinations are essential in any patient with an unknown focus of infection and when there is spinal pain or tenderness full investigation is warranted. The increasing availability of MRI may mean that earlier diagnosis and lower morbidity from this condition should become a reality in the near future.