- Email: [email protected]
Low yield of ED magnetic resonance imaging for suspected epidural abscess
American Journal of Emergency Medicine (2011) 29, 978–982
www.elsevier.com/locate/ajem
Original Contribution
Low yield of ED magnetic resonance imaging for suspected epidural abscess☆ Mazen El Sayed MD ⁎, Michael D. Witting MD, MS Emergency Medicine, University of Maryland Medical Center, Baltimore, MD 21201, USA Received 23 March 2010; revised 6 May 2010; accepted 8 May 2010
Abstract Purpose: The aim of this study was to estimate the yield of emergency department (ED) magnetic resonance imaging (MRI) in detecting spinal epidural abscess (SEA) and to identify clinical factors predicting positive MRI results. Basic Procedure: We examined a cohort of patients who underwent MRI to rule out SEA, followed by a nested case-control comparison of those with positive results and a sample with negative results. A positive result was defined as osteomyelitis, discitis, or SEA. Predictor variables included temperature, presenting complaint, drug abuse status, history of SEA or back surgery, midline back tenderness, neurologic deficit, MRI level, mean white blood cell count, erythrocyte sedimentation rate, and C-reactive protein level. Main Findings: Fourteen of the 120 available MRIs were excluded; 7 (6.6%) of the remaining 106 were positive. Temperature was 1.1°C higher in cases than controls (95% CI, 0.6-1.7). Conclusion: Emergency department MRI for suspected SEA has a low yield. Clinical guidelines are needed to improve efficiency. © 2011 Elsevier Inc. All rights reserved.
1. Introduction Spinal epidural abscess (SEA) is a rare disease that affects 0.2 to 2 of every 10 000 hospitalized patients [1-3]. In patients presenting to the emergency department (ED) with severe back or neck pain, a prevalence of spinal infection as high as 29% has been reported [4]. Spinal epidural abscess affects all age groups but is more prevalent in patients who are 30 years and older [1]. Risk factors for SEA include diabetes mellitus, intravenous drug use, ☆ Presented at University of Maryland Emergency Medicine Research Day, June 12, 2009. ⁎ Corresponding author. One Boston Medical Center Place, Dowling 1 South, Emergency Medicine, Boston, MA 02118, USA. Tel.: +1 617 343 1384. E-mail address: [email protected] (M. El Sayed).
0735-6757/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ajem.2010.05.004
immunosuppressive therapy, cancer, HIV or AIDS, trauma to the spine, abnormality of the vertebral column, chronic renal failure, and alcoholism [5,6]. Clinical features of SEA include spinal or back pain, fever, and neurologic deficits. This clinical triad of symptoms is present in only 10% to 15% of patients [7,8]. Physical examination findings vary from spinal tenderness to neurologic deficits, including bowel or bladder dysfunction. Laboratory studies available in the ED may reveal leukocytosis and elevation of other inflammatory markers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) [9]. Although this information may suggest SEA, it is not specific for it. Magnetic resonance imaging (MRI) with gadolinium is highly sensitive (more than 90%) in detecting SEA [3] and remains the diagnostic imaging method of choice because it is less invasive than computed tomographic (CT) myelography. The diagnosis of SEA is difficult to make in the ED
MRI for suspected epidural abscess and relies on high clinical suspicion and the availability of MRI capabilities. Small EDs that lack MRI capabilities are compelled to transfer patients who require MRI. In addition, difficulties in completing the MRI in the ED range from factors inherent to patient characteristics (claustrophobia, agitation) to technical factors such as the need for additional resources (nurse accompanying the patient, monitored sedation). Magnetic resonance imaging is a time-consuming study that indirectly contributes to ED overcrowding by delaying patient discharge. However, a delay in diagnosing SEA and, therefore, in initiating treatment can result in high morbidity and mortality [10,11]. The objectives of this study were (1) to estimate the yield of MRI in ED patients with suspected spinal epidural abscess and (2) to estimate the association between factors available in the ED and positive MRI results.
2. Methods 2.1. Study design A retrospective cohort design was used to estimate the yield of ED MRI, and a nested case-control design was used to examine associations between predictive variables and positive MRI results.
2.2. Study setting This study was conducted in an inner-city ED with an annual patient volume of approximately 60 000 visits. This study was deemed exempt by our institutional review board.
2.3. Chart review methods Both investigators queried the hospital's radiology database until they both agreed on the same search terms. We then reviewed all spine MRI studies done between January 1, 2004, and January 1, 2009, and retrieved those that mentioned the following search terms: age N17 years; ED location; and comment field indicating “abscess,” “epidural,” “IVDA” (intravenous drug abuse), or “IVDU” (intravenous drug use). We chose these keywords because we believed that an emergency physician suspecting a spinal infection would include one of them in the comments section of the MRI study request. Both investigators independently reviewed the list of MRI comments and excluded MRIs for which the comment fields indicated that the imaging was done primarily to assess recent surgical interventions or results of instrumentation (lumbar puncture) or to evaluate the patient for trauma, cancer, or disc disease rather than SEA. We included only the first eligible MRI for each patient. Any discrepancies in eligibility were resolved by consensus of the investigators, who were blinded to the MRI results at the time.
979 We considered MRI results to be positive if they indicated a need for emergent therapy and were part of the SEA disease spectrum: osteomyelitis, discitis, or epidural abscess [12,13]. We considered results negative if neurosurgical consultants did not recommend immediate treatment. Both authors reviewed MRI results independently, discussing discrepant cases until they agreed on the appropriate classification. We then conducted a nested case control study comparing all patients with positive results with a random sample with negative results, chosen in a 3:1 control:case ratio. The first author (MES) abstracted the following data from the charts: body temperature at presentation, chief complaint of pain (yes/no), chief complaint of neurologic dysfunction (yes/no), IVDA status (yes/no), history of SEA (yes/no), history of back surgery (yes/no), presence of midline back tenderness (yes/no), neurologic deficit on physical exam (yes/no), MRI level (focused vs. unfocused), white blood cell (WBC) count, erythrocyte sedimentation rate (ESR), and CRP level. Our standardized record prompts the physician to document the presence or absence of illegal drug use but not of less common entities, such as a history of SEA. If a clinical finding was not documented as positive, it was considered to be absent. After reviewing the MRI results of the positive cases, we did an additional analysis of the radiologic findings to assess whether some cases may have been detected by CT. For patients with records indicating that osteomyelitis was noted on the MRI, we reviewed the medical record to determine whether osteomyelitis was also seen in CT scans used to assess response to therapy.
2.4. Data analysis The primary outcome, i.e., the yield of MRI in regard to detection of SEA, was calculated by dividing the number of positive MRI reports by the number of eligible MRI reports. Confidence intervals for this proportion were determined by normal binomial approximation. For predictor variables, P values were calculated using Fisher exact test for dichotomous data and the Mann–Whitney U test for continuous data, with statistical significance defined at P b .05. Erythrocyte sedimentation rate and CRP results were dichotomized using our laboratory's reference values for the upper limits of normal. We calculated 95% confidence intervals based on normal binomial approximation for proportions and the t distribution for continuous variables.
3. Results During the study period, 120 MRI studies were obtained using the search approach described above. Fourteen of these MRI reports were excluded from the study for the following reasons: imaging was ordered in 6 patients who recently sustained trauma in a fall or motor vehicle collision, in 2 patients with a history of disc disease to look for worsening
980
M. El Sayed, M.D. Witting
Table 1
MRI results in 7 patients with spinal epidural abscess
Case MRI findings
Bone abnormality seen on CT
1 2
Yes Yes
3
4 5 6 7
T11-T12 discitis/osteomyelitis L4 and left S1 vertebral body osteomyelitis Fracture of T3 with osteomyelitis; soft tissue collection within the spinal canal at T3-T4 Epidural abscess from C4 through T2 Osteomyelitis/discitis at T12-L1 Epidural abscess at T1-T2 (0.3 × 0.7 × 1 cm) L1-L2 discitis/osteomyelitis with phlegmon material
Table 3 Comparison of continuous variables between patients with positive MRI results (n = 7) and a random sample of with negative MRI results (n = 20) Predictor
Cases Controls Mean difference [95% CI], P a
Mean age 49.0 Temperature (C) 37.6 8.3 Mean WBC count (1000/μL)
Yes
No Yes No
43.4 36.4 7.1
5.6 [(−)5.81 − (+)17.01], .3 1.1 [0.6 − 1.7], .001 1.2 [(−)1.8 − (+)4.1], .3
a
95% CI was calculated using the t distribution; P values were calculated using the Mann-Whitney U test.
Yes
T indicates thoracic; L, lumbar; C, cervical.
disease, in 2 patients as follow-up to surgery, in 2 patients to rule out epidural hematoma following lumbar puncture, and in 2 patients with a history of cancer to rule out metastasis. Of the remaining 106 MRI reports, 99 showed no evidence of infection, and 7 were positive for spinal infection. Two of the 99 negative MRI scans were suspicious for infection, but those patients were discharged from the ED after the neurosurgical service was consulted. We counted these as negative results in the calculation of diagnostic yield. The resulting MRI yield in ED patients with suspected spinal epidural abscess or infection was 6.6% (95% CI, 3.0– 13.6). No follow-up data were obtained on the 99 patients with negative MRIs. The positive results are summarized in Table 1. Osteomyelitis, with or without discitis, was noted in 5 patients, and 2 patients had epidural abscess without bone changes. All 7 patients were admitted for intravenous antibiotics or surgery. Among the clinical findings, only temperature differed statistically between cases and controls: the mean maximum temperature for cases was 1.1 C (95% CI, 0.6–1.7, P = .001) Table 2 Association between dichotomous predictor variables and treatable findings on MRI Characteristic
Cases (%)
Controls (%)
P
Male Chief complaint of back pain IVDA Previous epidural abscess Previous back surgery Midline tenderness Neurologic deficit ESRN12 mm/h a CRPN1.0 mg/L a CRPN3.0 mg/L a
3/7 (42) 7/7 (100)
13/20 (65) 18/20 (90)
.4 1.0
5/7 (71) 1/7 (14) 2/7 (29) 3/7 (43) 2/7 (29) 3/3 (100) 4/5 (80) (2/5) (40)
10/20 (50) 1/20 (5) 5/20 (25) 12/20 (60) 3/20 (15) 7/8 (85.71) 4/6 (66.67) 3/6 (50)
.6 .9 1.0 .7 .8 1.0 1.0 1.0
IVDA, intravenous drug abuse; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein. a ESR and CRP not ordered in some cases.
higher than that for controls. The other predictor factors that were available in the ED (Tables 2 and 3) were similar between the 2 groups. In the 7 positive cases, all patients came to the ED with the chief complaint of back pain (100%). Among the 20 control subjects, 18 presented with back pain, and 2, with neurologic deficit. A combination of symptoms was documented in the larger study group, but all of the 7 cases and the 20 controls that we reviewed had only one complaint. In the last part of the analysis, we looked at whether the bony abnormalities would have shown on CT scan. In 5 of the seven patients (71%) admitted for treatment (Table 1), bone abnormalities were identified on inpatient CT scan, and the effects of treatment were followed up by CT imaging.
4. Discussion Our study demonstrates a low yield (6.6%) of MRI in suspected cases of SEA; we were unable to find another study designed to estimate the yield of MRI requested for SEA in an ED. As in other areas of medicine, clinicians may have requested MRI when suspicion for SEA was present but only slight, leading to a low yield of diagnostic imaging. In our ED, MRI is available 24 hours per day, and it is the study of choice when physicians feel that the suspicion for SEA is high enough. Spinal epidural abscess is challenging to diagnose in the ED because the history and physical are often inconclusive, and it can be difficult to ascertain the current phase of the condition. These uncertainties are coupled with the awareness that missing the condition can have catastrophic consequences. In 1948, Heusner described the stages of progression of SEA [14,15]. In stage I, patients have back pain, fever, or tenderness. In stage II, patients develop radicular pain, nuchal rigidity or stiffness, and reflex changes. Progression to stage III involves sensory abnormalities, motor weakness, and bowel and bladder dysfunction. Stage IV is the irreversible stage of paralysis. The time frame for progression through these stages has not been clearly defined; however, there is a clear advantage in detecting SEA in its earliest clinical stages, when back pain (found in 100% of our cases) and spinal
MRI for suspected epidural abscess tenderness (described by 43% of our patients) are by far the most frequent findings [15,16]. Most of our patients (71%) were in this clinical stage. Patients with stage I symptoms may be considered for urgent, rather than emergent, imaging. Being in a later clinical stage at presentation predicts poorer outcome despite treatment [17-19]. Due to the retrospective nature of this study, we do not have accurate information about symptom onset for many of our cases. All patients with confirmed SEA required hospitalization for treatment with intravenous antibiotics or surgery. Their outcomes after treatment were not ascertained. Among our cases, several had bone abnormalities; 5 of the 7 patients (71%) with SEA might have been identified with helical CT. When MRI is not available, it may be reasonable to start the workup with spinal CT, but MRI with gadolinium remains the diagnostic modality of choice in cases of suspected spondylodiscitis or vertebral osteomyelitis [20]. Computed tomography has numerous limitations: it presents a radiation risk, it has low specificity and sensitivity, and it is inadequate for detecting early SEA or discitis [21-23]. Myelography followed by CT, although it is invasive and presents the risks associated with ionizing radiation, has been proposed as an alternative imaging modality for the diagnosis of SEA because it is as sensitive as MRI (both are N90% sensitive for detecting spinal cord compression) [24-26]. Our findings suggest that helical CT may play an important role in the detection of SEA associated with bony abnormalities such as osteomyelitis. In the second part of our study, we found that the clinical variables that are considered risk factors for SEA are actually poor predictors of positive MRI results. This observation was expected, since previous studies have been consistent in describing the limitations of the history and physical examination findings in diagnosing SEA [7,8,27]. The mean age of our patients was 49 years, which is situated at the lower end of the age range with the highest prevalence of the disease, i.e., the fifth and seventh decades [2,3,5,28]. This difference may be related to our patients' IVDU, which causes the infection to emerge in younger people. Chuo et al [29] found that “IV drug users with spinal infections were much younger and healthier than the typical spinal infection patient.” The mean age in their series was 44 years. In our study, temperature was the only variable with a statistically significant relationship with positive MRI results. The patients with MRI-confirmed SEA had a higher body temperature at triage (99.6°F) than did the control subjects (97.6°F). Previous studies reported that the absence of fever is not a reliable reason to exclude SEA and that patients with SEA may be normothermic or have mildly elevated temperature [9,30,31]. We did not take into account the possibility that patients may have taken antipyretics before coming to the ED; nonetheless, low-grade fever was found to be a significant clinical predictor for spinal infections. Several limitations of our study can be noted. First, the small sample size may have affected the power to detect a statistically significant difference in the analysis of the
981 predictive factors. Second, the data collection was done through the computerized imaging database of the ED by using specific keywords as search terms. This search method may have missed some emergency MRIs (suspected and confirmed cases of SEA). By limiting our search to patients whose physicians specifically documented their suspicion of SEA in the comments field, we expect that we isolated a group of patients with a relatively high likelihood of a positive MRI scan. Another limitation is inherent to retrospective reviews of medical records. Since there are no specific ED clinical guidelines for the workup of patients with suspected SEA other than ordering MRI of the spine with gadolinium, laboratory results (CRP and ESR) were not obtained for a number of our patients. This lack of data may have affected our ability to draw conclusions regarding the utility of laboratory studies in the workup of suspected SEA.
5. Conclusion Magnetic resonance imaging yield is low in the evaluation of ED patients in whom spinal infection is suspected. Evidence of spinal infection, ranging from discitis to osteomyelitis, was found in 6.6% of patients who underwent MRI for suspected SEA. Our findings suggest a need for decision rules for the evaluation of suspected SEA. Temperature may be an important predictive variable. Better selection criteria of high-risk ED patients with suspected SEA should be developed prospectively to avoid unnecessary imaging.
Acknowledgment The manuscript was copyedited by Linda J. Kesselring, MS, ELS, the technical editor/writer in the Department of Emergency Medicine at the University of Maryland School of Medicine.
References [1] Reihsaus E, Waldbaur H, Seeling W. Spinal epidural abscess: a metaanalysis of 915 patients. Neurosurg Rev 2000;23:175-204. [2] Sampath P, Rigamonti D. Spinal epidural abscess: a review of epidemiology, diagnosis, and treatment. J Spinal Disord 1999;12: 89-93. [3] Hlavin ML, Kaminski HJ, Ross JS, Ganz E. Spinal epidural abscess: a ten-year perspective. Neurosurgery 1990;27:177-84. [4] Shroyer SR, Mehta S. Screening strategies for early identification of spinal infections in patients presenting to emergency departments with severe back or neck pain. Ann Emerg Med 2009;54:S104-5. [5] Rigamonti D, Liem L, Sampath P, et al. Spinal epidural abscess: contemporary trends in etiology, evaluation, and management. Surg Neurol 1999;52:189-97. [6] Darouiche RO. Spinal epidural abscess. N Engl J Med 2006;355: 2012-20. [7] Davis DP, Wold RM, Patel RJ, et al. The clinical presentation and impact of diagnostic delays on emergency department patients with spinal epidural abscess. J Emerg Med 2004;26:285-91.
982 [8] Curry Jr WT, Hoh BL, Amin-Hanjani S, Eskandar EN. Spinal epidural abscess: clinical presentation, management, and outcome. Surg Neurol 2005;63:364-71. [9] Calder KK, Severyn FA. Surgical emergencies in the intravenous drug user. Emerg Med Clin North Am 2003;21:1089-116. [10] Weinberg J, Silber JS. Infections of the spine: what the orthopedist needs to know. Am J Orthop 1996;33:13-7. [11] Liou LM, Shih PY. Epidural abscess of the cervical spine with atypical manifestations: a report of two cases. Neurologist 2007;13:215-8. [12] Khan SH, Hussain MS, Griebel RW, Hattingh S. Comparison of primary and secondary spinal epidural abscesses: a retrospective analysis of 29 cases. Surg Neurol 2003;59:28-33. [13] Hadjipavlou AG, Mader JT, Necessary JT, Muffoletto AJ. Hematogenous pyogenic spinal infections and their surgical management. Spine 2000;25:1668-79. [14] Sendi P, Bregenzer T, Zimmerli W. Spinal epidural abscess in clinical practice. Q J Med 2008;101:1-12. [15] Heusner AP. Nontuberculous spinal epidural infections. N Engl J Med 1948;239:845-54. [16] Peterson JA, Paris P, Williams AC. Acute epidural abscess. Am J Emerg Med 1987;5:287-90. [17] Baker AS, Ojemann RG, Swartz MN, Richardson Jr EP. Spinal epidural abscess. N Engl J Med 1975;293:463-8. [18] Danner RL, Hartman BJ. Update on spinal epidural abscess: 35 cases and review of the literature. Rev Infect Dis 1987;9:265-74. [19] Kaufman DM, Kaplan JG, Litman N. Infectious agents in spinal epidural abscesses. Neurology 1980;30:844-50.
M. El Sayed, M.D. Witting [20] Rothman SL. The diagnosis of infections of the spine by modern imaging techniques. Orthop Clin 1996;27:15-31. [21] Stabler A, Reiser MF. Imaging of spinal infection. Radiol Clin North Am 2001;39:115-35. [22] Jarvik JG, Deyo RA. Diagnostic evaluation of low back pain with emphasis on imaging. Ann Intern Med 2002;137:586-97. [23] Wong D, Raymond NJ. Spinal epidural abscess. N Z Med J 1998;111: 345-7. [24] Carmody RF, Yang PJ, Seeley GW, et al. Spinal cord compression due to metastatic disease: diagnosis with MR imaging versus myelography. Radiology 1989;173:225-9. [25] Butler JS, Shelly MJ, Timlin M, Powderly WG, O'Byrne JM. Nontuberculous pyogenic spinal infection in adults: a 12-year experience from a tertiary referral center. Spine 2006;31:2695-700. [26] Gerstein N. Spinal epidural abscess. N Engl J Med 2007;356:638-9. [27] Broder J, Snarski JT. Back pain in the elderly. Clin Geriatr Med 2007;23:271-89. [28] Soehle M, Wallenfang T. Spinal epidural abscesses: clinical manifestations, prognostic factors, and outcomes. Neurosurgery 2002;51:79-87. [29] Chuo C, Fu Y, Lu Y, Chen J, Shen W, Yang C, et al. Spinal infection in intravenous drug abusers. J Spinal Disord Tech 2007;20: 324-8. [30] Pradilla G, Pradilla GA, Hsu W, Rigamonti D. Epidural abscesses of the CNS. Lancet Neurol 2009;8:292-300. [31] Winters ME, Kluetz P, Zilberstein J. Back pain emergencies. Med Clin North Am 2006;90:505-23.
www.elsevier.com/locate/ajem
Original Contribution
Low yield of ED magnetic resonance imaging for suspected epidural abscess☆ Mazen El Sayed MD ⁎, Michael D. Witting MD, MS Emergency Medicine, University of Maryland Medical Center, Baltimore, MD 21201, USA Received 23 March 2010; revised 6 May 2010; accepted 8 May 2010
Abstract Purpose: The aim of this study was to estimate the yield of emergency department (ED) magnetic resonance imaging (MRI) in detecting spinal epidural abscess (SEA) and to identify clinical factors predicting positive MRI results. Basic Procedure: We examined a cohort of patients who underwent MRI to rule out SEA, followed by a nested case-control comparison of those with positive results and a sample with negative results. A positive result was defined as osteomyelitis, discitis, or SEA. Predictor variables included temperature, presenting complaint, drug abuse status, history of SEA or back surgery, midline back tenderness, neurologic deficit, MRI level, mean white blood cell count, erythrocyte sedimentation rate, and C-reactive protein level. Main Findings: Fourteen of the 120 available MRIs were excluded; 7 (6.6%) of the remaining 106 were positive. Temperature was 1.1°C higher in cases than controls (95% CI, 0.6-1.7). Conclusion: Emergency department MRI for suspected SEA has a low yield. Clinical guidelines are needed to improve efficiency. © 2011 Elsevier Inc. All rights reserved.
1. Introduction Spinal epidural abscess (SEA) is a rare disease that affects 0.2 to 2 of every 10 000 hospitalized patients [1-3]. In patients presenting to the emergency department (ED) with severe back or neck pain, a prevalence of spinal infection as high as 29% has been reported [4]. Spinal epidural abscess affects all age groups but is more prevalent in patients who are 30 years and older [1]. Risk factors for SEA include diabetes mellitus, intravenous drug use, ☆ Presented at University of Maryland Emergency Medicine Research Day, June 12, 2009. ⁎ Corresponding author. One Boston Medical Center Place, Dowling 1 South, Emergency Medicine, Boston, MA 02118, USA. Tel.: +1 617 343 1384. E-mail address: [email protected] (M. El Sayed).
0735-6757/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ajem.2010.05.004
immunosuppressive therapy, cancer, HIV or AIDS, trauma to the spine, abnormality of the vertebral column, chronic renal failure, and alcoholism [5,6]. Clinical features of SEA include spinal or back pain, fever, and neurologic deficits. This clinical triad of symptoms is present in only 10% to 15% of patients [7,8]. Physical examination findings vary from spinal tenderness to neurologic deficits, including bowel or bladder dysfunction. Laboratory studies available in the ED may reveal leukocytosis and elevation of other inflammatory markers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) [9]. Although this information may suggest SEA, it is not specific for it. Magnetic resonance imaging (MRI) with gadolinium is highly sensitive (more than 90%) in detecting SEA [3] and remains the diagnostic imaging method of choice because it is less invasive than computed tomographic (CT) myelography. The diagnosis of SEA is difficult to make in the ED
MRI for suspected epidural abscess and relies on high clinical suspicion and the availability of MRI capabilities. Small EDs that lack MRI capabilities are compelled to transfer patients who require MRI. In addition, difficulties in completing the MRI in the ED range from factors inherent to patient characteristics (claustrophobia, agitation) to technical factors such as the need for additional resources (nurse accompanying the patient, monitored sedation). Magnetic resonance imaging is a time-consuming study that indirectly contributes to ED overcrowding by delaying patient discharge. However, a delay in diagnosing SEA and, therefore, in initiating treatment can result in high morbidity and mortality [10,11]. The objectives of this study were (1) to estimate the yield of MRI in ED patients with suspected spinal epidural abscess and (2) to estimate the association between factors available in the ED and positive MRI results.
2. Methods 2.1. Study design A retrospective cohort design was used to estimate the yield of ED MRI, and a nested case-control design was used to examine associations between predictive variables and positive MRI results.
2.2. Study setting This study was conducted in an inner-city ED with an annual patient volume of approximately 60 000 visits. This study was deemed exempt by our institutional review board.
2.3. Chart review methods Both investigators queried the hospital's radiology database until they both agreed on the same search terms. We then reviewed all spine MRI studies done between January 1, 2004, and January 1, 2009, and retrieved those that mentioned the following search terms: age N17 years; ED location; and comment field indicating “abscess,” “epidural,” “IVDA” (intravenous drug abuse), or “IVDU” (intravenous drug use). We chose these keywords because we believed that an emergency physician suspecting a spinal infection would include one of them in the comments section of the MRI study request. Both investigators independently reviewed the list of MRI comments and excluded MRIs for which the comment fields indicated that the imaging was done primarily to assess recent surgical interventions or results of instrumentation (lumbar puncture) or to evaluate the patient for trauma, cancer, or disc disease rather than SEA. We included only the first eligible MRI for each patient. Any discrepancies in eligibility were resolved by consensus of the investigators, who were blinded to the MRI results at the time.
979 We considered MRI results to be positive if they indicated a need for emergent therapy and were part of the SEA disease spectrum: osteomyelitis, discitis, or epidural abscess [12,13]. We considered results negative if neurosurgical consultants did not recommend immediate treatment. Both authors reviewed MRI results independently, discussing discrepant cases until they agreed on the appropriate classification. We then conducted a nested case control study comparing all patients with positive results with a random sample with negative results, chosen in a 3:1 control:case ratio. The first author (MES) abstracted the following data from the charts: body temperature at presentation, chief complaint of pain (yes/no), chief complaint of neurologic dysfunction (yes/no), IVDA status (yes/no), history of SEA (yes/no), history of back surgery (yes/no), presence of midline back tenderness (yes/no), neurologic deficit on physical exam (yes/no), MRI level (focused vs. unfocused), white blood cell (WBC) count, erythrocyte sedimentation rate (ESR), and CRP level. Our standardized record prompts the physician to document the presence or absence of illegal drug use but not of less common entities, such as a history of SEA. If a clinical finding was not documented as positive, it was considered to be absent. After reviewing the MRI results of the positive cases, we did an additional analysis of the radiologic findings to assess whether some cases may have been detected by CT. For patients with records indicating that osteomyelitis was noted on the MRI, we reviewed the medical record to determine whether osteomyelitis was also seen in CT scans used to assess response to therapy.
2.4. Data analysis The primary outcome, i.e., the yield of MRI in regard to detection of SEA, was calculated by dividing the number of positive MRI reports by the number of eligible MRI reports. Confidence intervals for this proportion were determined by normal binomial approximation. For predictor variables, P values were calculated using Fisher exact test for dichotomous data and the Mann–Whitney U test for continuous data, with statistical significance defined at P b .05. Erythrocyte sedimentation rate and CRP results were dichotomized using our laboratory's reference values for the upper limits of normal. We calculated 95% confidence intervals based on normal binomial approximation for proportions and the t distribution for continuous variables.
3. Results During the study period, 120 MRI studies were obtained using the search approach described above. Fourteen of these MRI reports were excluded from the study for the following reasons: imaging was ordered in 6 patients who recently sustained trauma in a fall or motor vehicle collision, in 2 patients with a history of disc disease to look for worsening
980
M. El Sayed, M.D. Witting
Table 1
MRI results in 7 patients with spinal epidural abscess
Case MRI findings
Bone abnormality seen on CT
1 2
Yes Yes
3
4 5 6 7
T11-T12 discitis/osteomyelitis L4 and left S1 vertebral body osteomyelitis Fracture of T3 with osteomyelitis; soft tissue collection within the spinal canal at T3-T4 Epidural abscess from C4 through T2 Osteomyelitis/discitis at T12-L1 Epidural abscess at T1-T2 (0.3 × 0.7 × 1 cm) L1-L2 discitis/osteomyelitis with phlegmon material
Table 3 Comparison of continuous variables between patients with positive MRI results (n = 7) and a random sample of with negative MRI results (n = 20) Predictor
Cases Controls Mean difference [95% CI], P a
Mean age 49.0 Temperature (C) 37.6 8.3 Mean WBC count (1000/μL)
Yes
No Yes No
43.4 36.4 7.1
5.6 [(−)5.81 − (+)17.01], .3 1.1 [0.6 − 1.7], .001 1.2 [(−)1.8 − (+)4.1], .3
a
95% CI was calculated using the t distribution; P values were calculated using the Mann-Whitney U test.
Yes
T indicates thoracic; L, lumbar; C, cervical.
disease, in 2 patients as follow-up to surgery, in 2 patients to rule out epidural hematoma following lumbar puncture, and in 2 patients with a history of cancer to rule out metastasis. Of the remaining 106 MRI reports, 99 showed no evidence of infection, and 7 were positive for spinal infection. Two of the 99 negative MRI scans were suspicious for infection, but those patients were discharged from the ED after the neurosurgical service was consulted. We counted these as negative results in the calculation of diagnostic yield. The resulting MRI yield in ED patients with suspected spinal epidural abscess or infection was 6.6% (95% CI, 3.0– 13.6). No follow-up data were obtained on the 99 patients with negative MRIs. The positive results are summarized in Table 1. Osteomyelitis, with or without discitis, was noted in 5 patients, and 2 patients had epidural abscess without bone changes. All 7 patients were admitted for intravenous antibiotics or surgery. Among the clinical findings, only temperature differed statistically between cases and controls: the mean maximum temperature for cases was 1.1 C (95% CI, 0.6–1.7, P = .001) Table 2 Association between dichotomous predictor variables and treatable findings on MRI Characteristic
Cases (%)
Controls (%)
P
Male Chief complaint of back pain IVDA Previous epidural abscess Previous back surgery Midline tenderness Neurologic deficit ESRN12 mm/h a CRPN1.0 mg/L a CRPN3.0 mg/L a
3/7 (42) 7/7 (100)
13/20 (65) 18/20 (90)
.4 1.0
5/7 (71) 1/7 (14) 2/7 (29) 3/7 (43) 2/7 (29) 3/3 (100) 4/5 (80) (2/5) (40)
10/20 (50) 1/20 (5) 5/20 (25) 12/20 (60) 3/20 (15) 7/8 (85.71) 4/6 (66.67) 3/6 (50)
.6 .9 1.0 .7 .8 1.0 1.0 1.0
IVDA, intravenous drug abuse; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein. a ESR and CRP not ordered in some cases.
higher than that for controls. The other predictor factors that were available in the ED (Tables 2 and 3) were similar between the 2 groups. In the 7 positive cases, all patients came to the ED with the chief complaint of back pain (100%). Among the 20 control subjects, 18 presented with back pain, and 2, with neurologic deficit. A combination of symptoms was documented in the larger study group, but all of the 7 cases and the 20 controls that we reviewed had only one complaint. In the last part of the analysis, we looked at whether the bony abnormalities would have shown on CT scan. In 5 of the seven patients (71%) admitted for treatment (Table 1), bone abnormalities were identified on inpatient CT scan, and the effects of treatment were followed up by CT imaging.
4. Discussion Our study demonstrates a low yield (6.6%) of MRI in suspected cases of SEA; we were unable to find another study designed to estimate the yield of MRI requested for SEA in an ED. As in other areas of medicine, clinicians may have requested MRI when suspicion for SEA was present but only slight, leading to a low yield of diagnostic imaging. In our ED, MRI is available 24 hours per day, and it is the study of choice when physicians feel that the suspicion for SEA is high enough. Spinal epidural abscess is challenging to diagnose in the ED because the history and physical are often inconclusive, and it can be difficult to ascertain the current phase of the condition. These uncertainties are coupled with the awareness that missing the condition can have catastrophic consequences. In 1948, Heusner described the stages of progression of SEA [14,15]. In stage I, patients have back pain, fever, or tenderness. In stage II, patients develop radicular pain, nuchal rigidity or stiffness, and reflex changes. Progression to stage III involves sensory abnormalities, motor weakness, and bowel and bladder dysfunction. Stage IV is the irreversible stage of paralysis. The time frame for progression through these stages has not been clearly defined; however, there is a clear advantage in detecting SEA in its earliest clinical stages, when back pain (found in 100% of our cases) and spinal
MRI for suspected epidural abscess tenderness (described by 43% of our patients) are by far the most frequent findings [15,16]. Most of our patients (71%) were in this clinical stage. Patients with stage I symptoms may be considered for urgent, rather than emergent, imaging. Being in a later clinical stage at presentation predicts poorer outcome despite treatment [17-19]. Due to the retrospective nature of this study, we do not have accurate information about symptom onset for many of our cases. All patients with confirmed SEA required hospitalization for treatment with intravenous antibiotics or surgery. Their outcomes after treatment were not ascertained. Among our cases, several had bone abnormalities; 5 of the 7 patients (71%) with SEA might have been identified with helical CT. When MRI is not available, it may be reasonable to start the workup with spinal CT, but MRI with gadolinium remains the diagnostic modality of choice in cases of suspected spondylodiscitis or vertebral osteomyelitis [20]. Computed tomography has numerous limitations: it presents a radiation risk, it has low specificity and sensitivity, and it is inadequate for detecting early SEA or discitis [21-23]. Myelography followed by CT, although it is invasive and presents the risks associated with ionizing radiation, has been proposed as an alternative imaging modality for the diagnosis of SEA because it is as sensitive as MRI (both are N90% sensitive for detecting spinal cord compression) [24-26]. Our findings suggest that helical CT may play an important role in the detection of SEA associated with bony abnormalities such as osteomyelitis. In the second part of our study, we found that the clinical variables that are considered risk factors for SEA are actually poor predictors of positive MRI results. This observation was expected, since previous studies have been consistent in describing the limitations of the history and physical examination findings in diagnosing SEA [7,8,27]. The mean age of our patients was 49 years, which is situated at the lower end of the age range with the highest prevalence of the disease, i.e., the fifth and seventh decades [2,3,5,28]. This difference may be related to our patients' IVDU, which causes the infection to emerge in younger people. Chuo et al [29] found that “IV drug users with spinal infections were much younger and healthier than the typical spinal infection patient.” The mean age in their series was 44 years. In our study, temperature was the only variable with a statistically significant relationship with positive MRI results. The patients with MRI-confirmed SEA had a higher body temperature at triage (99.6°F) than did the control subjects (97.6°F). Previous studies reported that the absence of fever is not a reliable reason to exclude SEA and that patients with SEA may be normothermic or have mildly elevated temperature [9,30,31]. We did not take into account the possibility that patients may have taken antipyretics before coming to the ED; nonetheless, low-grade fever was found to be a significant clinical predictor for spinal infections. Several limitations of our study can be noted. First, the small sample size may have affected the power to detect a statistically significant difference in the analysis of the
981 predictive factors. Second, the data collection was done through the computerized imaging database of the ED by using specific keywords as search terms. This search method may have missed some emergency MRIs (suspected and confirmed cases of SEA). By limiting our search to patients whose physicians specifically documented their suspicion of SEA in the comments field, we expect that we isolated a group of patients with a relatively high likelihood of a positive MRI scan. Another limitation is inherent to retrospective reviews of medical records. Since there are no specific ED clinical guidelines for the workup of patients with suspected SEA other than ordering MRI of the spine with gadolinium, laboratory results (CRP and ESR) were not obtained for a number of our patients. This lack of data may have affected our ability to draw conclusions regarding the utility of laboratory studies in the workup of suspected SEA.
5. Conclusion Magnetic resonance imaging yield is low in the evaluation of ED patients in whom spinal infection is suspected. Evidence of spinal infection, ranging from discitis to osteomyelitis, was found in 6.6% of patients who underwent MRI for suspected SEA. Our findings suggest a need for decision rules for the evaluation of suspected SEA. Temperature may be an important predictive variable. Better selection criteria of high-risk ED patients with suspected SEA should be developed prospectively to avoid unnecessary imaging.
Acknowledgment The manuscript was copyedited by Linda J. Kesselring, MS, ELS, the technical editor/writer in the Department of Emergency Medicine at the University of Maryland School of Medicine.
References [1] Reihsaus E, Waldbaur H, Seeling W. Spinal epidural abscess: a metaanalysis of 915 patients. Neurosurg Rev 2000;23:175-204. [2] Sampath P, Rigamonti D. Spinal epidural abscess: a review of epidemiology, diagnosis, and treatment. J Spinal Disord 1999;12: 89-93. [3] Hlavin ML, Kaminski HJ, Ross JS, Ganz E. Spinal epidural abscess: a ten-year perspective. Neurosurgery 1990;27:177-84. [4] Shroyer SR, Mehta S. Screening strategies for early identification of spinal infections in patients presenting to emergency departments with severe back or neck pain. Ann Emerg Med 2009;54:S104-5. [5] Rigamonti D, Liem L, Sampath P, et al. Spinal epidural abscess: contemporary trends in etiology, evaluation, and management. Surg Neurol 1999;52:189-97. [6] Darouiche RO. Spinal epidural abscess. N Engl J Med 2006;355: 2012-20. [7] Davis DP, Wold RM, Patel RJ, et al. The clinical presentation and impact of diagnostic delays on emergency department patients with spinal epidural abscess. J Emerg Med 2004;26:285-91.
982 [8] Curry Jr WT, Hoh BL, Amin-Hanjani S, Eskandar EN. Spinal epidural abscess: clinical presentation, management, and outcome. Surg Neurol 2005;63:364-71. [9] Calder KK, Severyn FA. Surgical emergencies in the intravenous drug user. Emerg Med Clin North Am 2003;21:1089-116. [10] Weinberg J, Silber JS. Infections of the spine: what the orthopedist needs to know. Am J Orthop 1996;33:13-7. [11] Liou LM, Shih PY. Epidural abscess of the cervical spine with atypical manifestations: a report of two cases. Neurologist 2007;13:215-8. [12] Khan SH, Hussain MS, Griebel RW, Hattingh S. Comparison of primary and secondary spinal epidural abscesses: a retrospective analysis of 29 cases. Surg Neurol 2003;59:28-33. [13] Hadjipavlou AG, Mader JT, Necessary JT, Muffoletto AJ. Hematogenous pyogenic spinal infections and their surgical management. Spine 2000;25:1668-79. [14] Sendi P, Bregenzer T, Zimmerli W. Spinal epidural abscess in clinical practice. Q J Med 2008;101:1-12. [15] Heusner AP. Nontuberculous spinal epidural infections. N Engl J Med 1948;239:845-54. [16] Peterson JA, Paris P, Williams AC. Acute epidural abscess. Am J Emerg Med 1987;5:287-90. [17] Baker AS, Ojemann RG, Swartz MN, Richardson Jr EP. Spinal epidural abscess. N Engl J Med 1975;293:463-8. [18] Danner RL, Hartman BJ. Update on spinal epidural abscess: 35 cases and review of the literature. Rev Infect Dis 1987;9:265-74. [19] Kaufman DM, Kaplan JG, Litman N. Infectious agents in spinal epidural abscesses. Neurology 1980;30:844-50.
M. El Sayed, M.D. Witting [20] Rothman SL. The diagnosis of infections of the spine by modern imaging techniques. Orthop Clin 1996;27:15-31. [21] Stabler A, Reiser MF. Imaging of spinal infection. Radiol Clin North Am 2001;39:115-35. [22] Jarvik JG, Deyo RA. Diagnostic evaluation of low back pain with emphasis on imaging. Ann Intern Med 2002;137:586-97. [23] Wong D, Raymond NJ. Spinal epidural abscess. N Z Med J 1998;111: 345-7. [24] Carmody RF, Yang PJ, Seeley GW, et al. Spinal cord compression due to metastatic disease: diagnosis with MR imaging versus myelography. Radiology 1989;173:225-9. [25] Butler JS, Shelly MJ, Timlin M, Powderly WG, O'Byrne JM. Nontuberculous pyogenic spinal infection in adults: a 12-year experience from a tertiary referral center. Spine 2006;31:2695-700. [26] Gerstein N. Spinal epidural abscess. N Engl J Med 2007;356:638-9. [27] Broder J, Snarski JT. Back pain in the elderly. Clin Geriatr Med 2007;23:271-89. [28] Soehle M, Wallenfang T. Spinal epidural abscesses: clinical manifestations, prognostic factors, and outcomes. Neurosurgery 2002;51:79-87. [29] Chuo C, Fu Y, Lu Y, Chen J, Shen W, Yang C, et al. Spinal infection in intravenous drug abusers. J Spinal Disord Tech 2007;20: 324-8. [30] Pradilla G, Pradilla GA, Hsu W, Rigamonti D. Epidural abscesses of the CNS. Lancet Neurol 2009;8:292-300. [31] Winters ME, Kluetz P, Zilberstein J. Back pain emergencies. Med Clin North Am 2006;90:505-23.