Evidence Based Diagnosis of Spinal Epidural Abscess
Search Strategy: You first search in vain for a JAMA Rational Clinical Exam article to review this topic. As a fairly rare diagnosis, there is not a large quantity of high-quality (i.e. non-case series based) research on this topic. You focus your search on the diagnosis question by conducting a “clinical prediction guide” PUBMED Clinical Query for “epidural abscess” narrow/specific search revealing 142 articles (see http://tinyurl.com/4xdorzp). Next, you conduct a search using the second PICO question and the “prognosis” PUBMED Clinical Query for “epidural abscess” narrow/specific and identify 106 citations but none are unique from the first strategy (see http://tinyurl.com/3gwkvva). Since all but two of the articles you identify using these strategies are narratives reviews, you scan the bibliographies of several manuscripts until you find original research data to analyze. You select these articles for closer review.
A 45 year old male with a long-history of intravenous drug abuse presents to your ED complaining of lower back pain. Your review of the electronic medical record reveals that he has had 14 prior ED evaluations for back pain over the last 2-years. Several of the prior visits’ physicians’ notes express concern for drug-seeking behavior. His past medical history also includes multiple superficial cutaneous abscesses from skin-popping, hypertension, and a car-versus-pedestrian accident at age 25. He denies any fevers, chills, urinary symptoms, back trauma, or neurological symptoms.
Your physical exam shows a disheveled gentleman appearing much older than his stated age, in moderate distress holding his back. He denies any current alcohol or intravenous drug use which you clarify to mean that he has not used any substances today. He last drank alcohol 12-hours ago and last used heroin intravenously 2-days ago. His vital signs are BP 150/70, P 110, R 20, T 37.9° C, oxygen saturation (room air) 96%. His skin exam reveals multiple skin-popping scars on his anterior arms and legs, as well as an old surgical scar in the middle of his back which he tells you is from spine surgery after his pedestrian accident 20 years ago. He doesn’t know the details of his surgery or whether he has indwelling hardware. Your review of the electronic medical records did not reveal any prior spine imaging (x-rays or MRI). He has no heart murmur, midline spine tenderness, muscle spasms, palpable abdominal mass, or appreciable neurological deficits.
Cognizant of his substance abuse history, but wary of recent research suggesting the dangers of non-steroidal anti-inflammatory agents in a variety of settings, you order oxycodone and plan to re-assess him in an hour. Your astute nurse subsequently re-checks his vital signs and notes a temperature of 38.9°C at which point you decide to explore the possibility of an epidural abscess, but you are uncertain about the role of labs and imaging to rule-in or rule-out this diagnosis.
PICO Question #1 (Diagnostics)
Population: ED patients with back pain suspicious for spinal epidural abscess
Intervention: Clinical gestalt, labs, x-rays, MRI
Outcome: Prevalence, sensitivity, specificity
PICO Question #2 (Prognostics)
Population: ED patients with back pain suspicious for spinal epidural abscess
Intervention: Predictors of adverse outcomes
Outcome: Mortality, neurological morbidity, hospital length of stay
Third years: Prospective evaluation of a clinical decision guideline to diagnose spinal epidural abscess in patients who present to the emergency department with spine pain, J Neurosurg Spine 2011; 14: 765-770. (http://pmid.us/21417700)
Article 1: The clinical presentation and impact of diagnostic delays on emergency department patients with spinal epidural abscess, J Emerg Med 2004; 26:285-291
Article 2: Spinal epidural abscess – experience with 46 patients and evaluation of prognostic factors, J Infection 2002; 45:76-81
Article 3: Prospective evaluation of a clinical decision guideline to diagnose spinal epidural abscess in patients who present to the emergency department with spine pain, J Neurosurg Spine 2011; 14:765-770
Article 4: Spinal epidural abscess: a meta-analysis of 915 patients, Neurosurg Rev 2000; 232:175-204
The lifetime incidence of back pain is 90% and it accounted for over 2 million emergency department visits in 2007. In fact, back pain is the leading musculoskeletal etiology prompting patients to present to the ED. Low back pain may originate from the lumbar spine, vertebral ligaments, annulus fibrosus, vertebral periosteum, facet joints, paravertebral musculature, blood vessels, or spinal nerve roots. In addition, low back pain may be a presenting symptom for many systemic disease processes. Clinicians should seek to answer 3 key questions in the evaluation of acute back pain:
- Is a serious systemic disease causing the pain?
- Is there neurological compromise which might require surgical evaluation?
- Do social or psychological situations exist that could amplify or prolong pain?
In the current diagnostic evaluation of acute low back pain, 30% have an x-ray and 9.6% have either an MRI or a CT based upon National Hospital Ambulatory Medical Care Survey (NHAMCS) emergency department data. The use of MRI or CT increased from 3.2% to 9.6% between 2002 and 2006. Although no emergency medicine guidelines exist for the diagnostic evaluation of acute atraumatic back pain, multiple specialty societies have published guidelines upon which an evidence-based approach can be formulated. Even though back pain has a self-limited natural history in over 90% of patients, clinicians must consider more serious etiologies for each case within the context of the patient’s age, co-morbid illnesses, psychosocial stressors, potentially occult mechanisms of injury, neurological findings, and symptom duration. Emergency medicine clinical decision aids for the more serious causes of back pain have yet to be developed and validated, so costly imaging procedures that delay emergency department thoroughfare and are most often inconsequential to patient-centric outcomes must be carefully weighed against initial clinical concern and each test’s diagnostic accuracy.
Spinal epidural abscesses are pyogenic bacterial infections that can have an insidious onset before rapidly progressing to complete, irreversible paralysis. The consensus among our EM group is that misdiagnosis of spinal epidural abscesses at the first emergency department evaluation is the norm rather than the exception. A review of the malpractice claims literature over the period of 2000-2011 revealed 24 verdict and settlement reports from across the country dealing with failure to diagnose spinal epidural abscesses. About 67% resulted in defense verdicts; the other cases had verdicts ranging from $1.2 million to $19.5 million. From the perspective of lawyers across the country (who have written on these cases for legal journals) consider the following: (a) these cases typically involve patients who use the ED for their primary medical care, who tend to be thought of as “drug-seeking”, and who go to multiple providers and fail to provide consistent information or accurate/reliable historical information; (b) imaging is important – MRI is really the imaging test of choice for cases where the clinician suspects spinal epidural abscess; and (c) explicit documentation of follow up instructions is essential.
Based upon this review of the evidence, the best-estimate pre-test probability for spinal epidural abscess in ED patients with concerning back pain is 0.4%. Unfortunately, we still do not understand the accuracy of history, physical exam, and bedside lab tests to diagnosis spinal epidural abscess. Although the absence of evidence is not evidence of absence, astute clinicians must have a low threshold to order MRI to confirm or exclude the diagnosis of spinal epidural abscess understanding the dire consequences of a delayed diagnosis.
What do we know about diagnostic tests for spinal epidural abscess? In the available literature, most cases occur in adults over the age of 30-years with a male predominance. Almost all cases (98%) will have at least one risk factor identified (diabetes, intravenous drug abuse, liver disease, renal failure, indwelling catheter, recent invasive spinal procedure, vertebral fracture, immunocompromised, or distal sites of infection). The classic triad of spine pain, fever, and neurologic abnormality has a sensitivity of 2.0-7.9% for spinal epidural abscess in ED patients. If emergency physicians were to evaluate every patient with at least one risk factor for SEA, 50 MRI’s would be performed for every one SEA patient identified. SEA is a rare disease that most commonly presents with back pain and fever, but with the exception of one study, the diagnostic accuracy (sensitivity, specificity, likelihood ratios) of history, physical exam, routine labs, and imaging studies are unknown because all of the literature is case-series which do not evaluate disease-negative subjects.
One diagnostic algorithm has been evaluated and tested in a before-after observational trial (Figure 1).
Neither the reliability nor the accuracy of this algorithm were tested by Davis et al. but diagnostic delays were reduced from 83.6% before the guidelines were introduced to 9.7% after the guidelines were introduced. In addition, they prospectively enrolled a convenience sampling of patients for 9-months demonstrating the diagnostic test characteristics noted in this table.
|Risk Factor||Sensitivity (%)||Specificity (%)||LR +||LR –|
|Temp > 100.4°F||7||98||4||1|
None of these risk factors reduce the post-test probability of an epidural abscess in their absence. However, the presence of an indwelling catheter, intravenous drug abuse, a distant site of infection, or a spine fracture all increase the risk of infection. These investigators also prospectively evaluated the diagnostic accuracy of ESR and CRP in those with and without spinal epidural abscesses. Although they reported only dichotomous likelihood ratios, they presented sufficient data in Figure 4 and 5 on page 769 to compute interval likelihood ratios which are much more meaningful for continuous data. Based upon these interval likelihood ratios, a CRP above 10 or an ESR above 60 mm/hour significantly increase the post-test probability of spinal epidural abscess and an ESR in the 0-20 range significantly reduces the likelihood but no other results are diagnostically helpful.
|Interval LR||CRP Range||Interval LR|
|0 – 20||0||0 – 5||0.22|
|20 – 40||0.74||5 – 10||0.67|
|40 – 60||0.89||> 10||Infinity|
More prospective consecutive sample studies are needed to better understand the sensitivity, specificity and LR’s of traditional spinal epidural abscess risk factors. In addition to assessing a consecutive sampling of SEA-positive and SEA-negative patients using STARD criteria to establish diagnostic accuracy, future trials, should establish cost-effective diagnostic strategies, perhaps in the form of clinical decision rules.