9.1.1 Introduction and Epidemiology

A spinal epidural abscess (SEA) is a collection of pus in the epidural space of the spinal canal. While these can present atypically and have a relatively benign course, they can also lead to neurologic compromise or death. The incidence of SEAs has risen in recent years, from 0.2 to 1 cases per 10,000 hospital admissions to 5.1 cases per 10,000 hospital admissions, which is thought to be due to the increasing aging population in the United States, improved diagnostic modalities, and increasing prevalence of risk factors such as intravenous drug use (IVDU), alcoholism, renal insufficiency, immunosuppression, and diabetes mellitus. ^{1 , 2 , 3 , 4} Other risk factors for SEA previously identified include human immunodeficiency virus (HIV) infection, dental abscesses, hemodialysis, tattoos, and acupuncture. ^{5 , 6} More broadly, SEA can result from any condition that causes bacteremia. ³ Diabetes mellitus remains the strongest risk factor for SEA, although SEAs resulting from IVDU and epidural catheterization or pain pump placement are becoming increasingly prevalent. ⁷ SEAs have a median age of onset of 50 years of age, although the highest prevalence occurs in the sixth and seventh decades of life. ³ SEAs tend to occur more commonly in males, with a ratio of men to women of 1 to 0.56 in one meta-analysis. ⁸

9.1.2 Anatomy and Pathogenesis

SEA is thought to arise through several different means, including: (1) direct inoculation from trauma, spinal surgery, or procedures such as pain catheter placement or epidural anesthesia, (2) hematogenous seeding, or (3) contiguous spread from adjacent soft tissues or bone (e.g., psoas abscess or vertebral osteomyelitis). Despite this, approximately one-third of cases have no identifiable source of the infection. ^{9 , 10 , 11}

Approximately 86% of cases arise in the thoracic or lumbar spine, while 14% arise in the cervical spine. ¹² This is in part due to the larger epidural space in the thoracolumbar spine, as well as the presence of more fatty tissue that may create a favorable environment for infection persistence. ^{9 , 13 , 14} SEAs are also found to occur more frequently in the posterior spinal column, with anterior SEAs comprising only 20% of all cases; this is thought to be due to the contiguous spread of infection from adjacent vertebral bodies in the setting of vertebral osteomyelitis or from the intervertebral disk in the setting of pyogenic infectious diskitis. ^{9 , 15} As the epidural space is a vertical sheath, epidural abscesses typically involve multiple levels of the spinal cord, averaging three to five levels in one study. ¹³

Noncontiguous SEAs (skip abscesses) may also occur, although these are much more uncommon (9% of all SEA cases), with one prior study noting delay in presentation by >7 days, serum erythrocyte sedimentation rate (ESR) >95 mm/hour at presentation, and a concomitant area of infection outside of the spine as three significant risk factors of noncontiguous SEA. ¹⁶ In this study, the probability of a skip lesion was 73% in patients with all three predictors, 13% in patients with two, 2% in patients with one predictor, and 0% in patients with none of the above predictors. ¹⁶ Damage to the spinal cord from SEA is thought to result from multiple mechanisms, including direct compression of the spinal cord, compression of local venous or arterial blood supply, and indirectly via the presence of bacterial toxins or production of inflammatory mediators during the immune response. ⁷

9.1.3 Presentation and Diagnosis

Although classically described as presenting with the triad of fever, spinal pain, and neurological deficit, SEAs often present with nonspecific initial manifestations at onset leading to delayed or missed diagnosis. ¹⁷ Prior literature has suggested that only a small proportion of patients have all three components of the classically described triad at presentation (13% in one study), highlighting the importance of accurate history and physical examination and proper diagnostic and laboratory testing. ⁶ Indeed, one recent retrospective study of 250 cases of SEA demonstrated that up to 55% of SEA cases had a missed or incorrect diagnosis during the course of initial workup with up to a 12-day delay in diagnosis of these cases. ¹⁸ Fever was also only found to be present in 48% of patients in another case series from a single tertiary care center, and another study demonstrated that the median number of visits to the emergency department was two before admission and treatment. ^{19 , 20} Additionally, 98% of patients in this case had one or more of the features missing from the classic triad of SEA. ²⁰ As such, although the diagnosis of SEA is overall uncommon, it is important to consider SEA and rule out the diagnosis before attributing presenting complaints to other etiologic sources.

Since few patients display the classic triad of symptomatology, SEAs may follow the sequence of focal and severe back pain, followed by radicular pain, motor weakness, and finally bowel or bladder incontinence and paralysis as late-stage symptoms that rapidly become permanent. ¹³ With regard to the prevalence of symptoms, back or neck pain is most common (88%), followed by fever (61%), paresis (54%), bladder/bowel dysfunction (37%), sepsis (17%), and radiculopathy (12%). ²¹ It is imperative that the diagnosis of SEA be considered in any febrile patient with spinal pain, particularly in those who have recent or known bacteremia, risk factors for SEA, or neurologic symptoms such as radicular pain or motor weakness.

Laboratory studies that may be helpful in the diagnosis of SEA include ESR (normal ranges 0–20 mm/hour in females and 0–15 mm/hour in males) and C-reactive protein (CRP) (normal range typically <10 mg/L). ESR has been shown to be of more important utility over CRP, with ESR elevated in 94% of patients in one study, while CRP was elevated in 87%. ⁸ Davis et al demonstrated ESR elevation in 100% of patients with SEA, while only 33% of patients without SEA had elevated ESR; the average elevated ESR in this study was 76.5 mm/hour. ²⁰ Serum white blood cell (WBC) count is typically less useful, with this study showing that only 60% of patients who had SEA presented with leukocytosis to the emergency department. ²⁰

Imaging is crucial for diagnosing SEA. MRI with gadolinium contrast is the imaging modality of choice, as it is highly sensitive for SEA early in the course of infection, and allows for the highest imaging resolution for localization and extent of SEA involvement (▶ Fig. 9.1). Imaging of the entire spinal cord should be considered in patients with risk factors of noncontiguous SEA as noted previously, or in those with symptoms that do not localize well to only one region of the spine. In patients in whom gadolinium contrast is contraindicated, MRI without contrast is often still sufficient for diagnosis, with the most sensitive feature being paraspinal edema. ²² In patients who cannot undergo MRI, computerized tomography (CT) scanning with intravenous (IV) contrast is the next option of choice. X-rays of the spine are typically insufficient for diagnosing SEA and may only demonstrate the longer term sequelae of changes from concomitant or causative osteomyelitis or diskitis, and myelography is typically not recommended due to its invasiveness and potential for iatrogenic contamination of the subarachnoid space. ^{2 , 3 , 13 , 23}

Once SEA has been identified by imaging, two sets of blood cultures should be obtained to try to isolate the responsible organism. Aspiration of the SEA itself can also be considered, as this is more likely to be positive than blood cultures, although the location of the SEA may make interventional radiologic aspiration difficult, especially in anterior SEA. ⁵ A lumbar puncture is typically not recommended due to the low diagnostic yield and the risk of iatrogenic inoculation of the subarachnoid space if the diagnostic needle traverses the SEA. One study demonstrated the rate of positive cultures from the SEA itself to be 90%, while blood cultures were only positive in 62% of SEA cases, with cerebrospinal fluid (CSF) being the least diagnostic at 19%. ^{5 , 13} A summary of initial presentation and workup are stated in ▶ Table 9.1.

Staphylococcus aureus is the most common organism identified as causing SEA, accounting for approximately two-thirds of all cases. ^{5 , 9 , 24 , 25} This is followed by gram-negative bacilli (16%), streptococci (9%), and coagulase-negative staphylococci (3%), which are often observed in patients with spinal instrumentation. ^{5 , 25} Within S. aureus infections, the presence of methicillin resistance ranges from 25 to 68% and varies based on institution. ^{4 , 5 , 17} An important cause of SEA in the developing world is also Mycobacterium tuberculosis, which may be in the setting of tuberculous spondylitis (Pott’s disease). ^{26 , 27} Pseudomonas aeruginosa is also an important organism to consider in SEA, particularly in patients with history of prior or current IVDU. ²⁸

Differential diagnoses for SEA include vertebral diskitis and osteomyelitis, meningitis, herpes zoster (prior to presence of vesicular lesions), degenerative disk disease or disk herniations, and metastatic tumors of the spine.

9.1.4 Management and Surgical Decision-Making

Antibiotics should be started rapidly for empiric coverage of SEA after the collection of blood cultures once the diagnosis of SEA is suspected. The duration of antibiotic therapy is typically 4 to 8 weeks of IV antibiotics dictated by culture data. This is often dictated by infectious disease consultation, and longer courses may be selected in patients in whom hardware or bone graft (in particular, allograft) must be retained in order to preserve the mechanical stability of the spine. ^{29 , 30} Frequently, empiric regimens are vancomycin (15 to 20 mg/kg every 8 to 12 hours with a goal serum trough concentration of 15 to 20 mcg/mL) and either a third- or fourth-generation cephalosporin, such as ceftriaxone (2 g IV every 12 hours) or cefepime (2 g IV every 8 hours). This allows for Staphylococcus (including methicillin-resistant S. aureus [MRSA]) coverage as well as Streptococcus, and aerobic gram-negative pathogens.

Medical management alone may be considered in patients who have no neurologic deficit and in those who have significant medical comorbidities in whom surgery may not be tolerated. ^{3 , 31} In addition, the location of the epidural abscess may influence the decision for medical management, as lumbar epidural abscesses may be better tolerated than thoracic epidural abscesses given the presence of the spinal cord at the thoracic levels as opposed to only the cauda equina in the lumbar spine below the level of the conus, which may be more tolerant of space occupying lesions such as SEA. ³² One retrospective study of 52 patients demonstrated that 83% of patients with SEA that were medically treated had good or excellent early neurologic outcomes at a median follow-up of 2 months, with excellent outcome defined as complete recovery, and good outcome defined as ability to walk without aids but with residual pain. ³³ If medical management is selected, serial neurologic examination is of paramount importance, as any deterioration in neurologic status is an indication for surgical treatment.

Based on the location of the SEA, percutaneous drainage via interventional radiologic approaches may also be considered in select cases, particularly in patients who have adjoining abscesses such as psoas or paravertebral abscesses, with one recent study demonstrating effective treatment in 69% of cases of SEA. ³⁴

With regards to surgical management, this is typically determined on a case-by-case basis, with indications for surgery being symptomatic spinal cord compression, progressive neurologic deficit, spinal instability, or persistence or progression of disease despite appropriate antibiotic therapy. In these patient groups, surgical intervention within 24 hours after the onset of neurologic symptoms leads to improved outcomes. ³⁵ Prior studies have demonstrated predictive factors for failure of medical management, including age greater than 65, impaired neurological status, diabetes, and MRSA infection. ¹ When surgical intervention is indicated, the standard of care is typically laminectomy and debridement in the case of posterior SEAs, while anterior SEAs may require anterior decompression in order to adequately decompress the lesion and subsequent stabilization via anterior and/or posterior approaches. ^{32 , 36} Cultures should be taken prior to initial irrigation. There is variability amongst surgeons with regard to preferences for volume and type of irrigation, but in our experience, this is typically 9 L of saline via gentle gravity. After initial irrigation, any hardware or graft that can be removed without compromising mechanical stability should also be removed, although typically instrumentation in nonfused spines is initially retained. At this point, any compromised bone should also be removed after appropriate preoperative planning that accounts for any mechanical instability that may be introduced. Indeed, cages, allograft, or autograft may also be beneficial or mechanically necessary in the case of larger defects in the anterior spine, with fibular allograft and humeral allograft historically used successfully for cervical and thoracolumbar spine reconstruction. ³² In these cases, surgical judgment must be used to adequately debride infected tissue while balancing the consideration of maintaining spinal stability and the introduction of hardware or graft into an infected site. In these patients, infectious disease consultation becomes particularly useful postoperatively in determination of postoperative antibiosis, which may require long-term suppression or eventual need for planned hardware removal, such as in the case of spinal fusion pending bony fusion. ^{21 , 29} Prior studies have also shown that surgical debridement alone versus debridement with instrumentation had similar rates of failure and recurrence, suggesting instrumented therapy can be safely performed in conjunction with appropriate parenteral antibiotic therapy when indicated. ³⁷

Then, irrigation with another 9 to 12 L of saline should be repeated, with cultures obtained post irrigation. Closure should typically be with nonbraided monofilament suture and vancomycin powder may be considered, although its effect is still a topic of debate and has not been specifically studied in the context of SEA. ³⁸ Negative pressure wound therapy may also be considered for its effect in improving local wound environment. ³⁹ Postoperative care should involve empiric IV antibiosis pending infectious disease consultation, with postoperative activity dictated based on the specific type of surgery performed and the relative stability of the spine pending surgical intervention, typically 6 to 8 weeks after surgery.

Follow-up MRIs may be obtained in cases of persistent infection or deterioration of symptoms. In general, shorter treatment courses using IV antibiotics are implemented in patients in whom either surgical drainage has occurred or who do not have retained hardware or concomitant osteomyelitis, while longer courses may be provided to medically managed patients, those responding poorly to antibiotic therapy, or those with retained hardware or graft as a necessity to preserve mechanical spinal column stability. ³⁷