ACADEMIC EMERGENCY MEDICINE 2011; 18:719–725 © 2011 by the Society for Academic Emergency Medicine
Objectives: The objectives were to evaluate the presenting signs and symptoms of spinal cord and cauda equina compression (SCC) and to determine the incidence of emergency department (ED) misdiagnosis.
Methods: This was a retrospective chart review at an urban, tertiary care hospital of patients discharged from an inpatient stay (April 2008 through July 2009) with an International Classification of Diseases, Ninth Revision (ICD-9) code indicating spinal disease, who had visited the ED for a related complaint within the previous 30 days, and who had a final diagnosis of new SCC. Trauma and transferred patients were excluded. The authors defined a misdiagnosis as no ED-documented diagnosis of SCC and failure to perform an appropriate diagnostic study either prior to arrival, in the ED, or immediately upon admission.
Results: Of 1,231 charts reviewed, 63 met inclusion criteria. The most common presenting symptoms in patients with SCC were pain (44, 70%), difficulty ambulating (38, 60%), and weakness (35, 56%). On physical examination, motor deficits (45, 71%) were more common than sensory deficits (27, 43%); however, 15 (24%) patients had no motor or sensory deficit, and 13 (23%) patients only had unilateral findings. Impaired gait was present in 14 patients of only 20 tested, three of whom had no associated motor or sensory deficit. SCC was misdiagnosed in 18 (29%, 95% confidence interval [CI] = 19% to 41%) cases, which resulted in a significant delay to diagnosis (median = 54 hours, interquartile range [IQR] = 38 to 77 vs. 5.3 hours, IQR = 3.0 to 15) in these patients.
Conclusions: SCC can have a subtle presentation with absent or unilateral motor and sensory deficits, but gait ataxia may be an additional finding. ED misdiagnosis of SCC in nontrauma patients is common.
ACADEMIC EMERGENCY MEDICINE 2011; 18:1–7 © 2011 by the Society for Academic Emergency Medicine
Acute spinal cord and cauda equina compression (SCC) has the potential to cause devastating neurologic impairment; therefore, timely and accurate diagnosis in the emergency department (ED) setting is paramount. Prompt diagnosis of acute SCC is critical because optimal patient outcomes are predicated on early treatment. Previous studies have shown that neurologic function at the time of treatment is an important predictor of final outcome, and if diagnosis is missed or delayed, patients may have further neurologic deterioration.1–4 Specifically, preoperative ability to walk has been shown to independently predict the likelihood of postoperative ambulation.3 Accurate diagnosis of early-stage SCC can be quite challenging, because patients may present with only minimal neurologic complaints or deficits. One study investigating the abilities of neurosurgical residents to accurately diagnose cauda equina syndrome based on clinical examination in a subset of patients referred to them with suspected spinal cord pathology reported a sensitivity and diagnostic accuracy of 80 and 56%, respectively.5 The diagnostic difficulty in the ED is increased due to the inherent pressures of maintaining the expeditious throughput of patients, constraints on resources such as magnetic resonance imaging (MRI), and the large “denominator” of patients with back and neck pain, the overwhelming majority of whom have other benign etiologies.
Spinal cord and cauda equina compression is caused by vertebral fracture, spinal epidural abscess and hematoma, metastatic or primary spine tumors, disc herniation, spondylolisthesis, and other etiologies. Regardless of the exact cause, the first priority should be to determine whether or not a patient has acute SCC, as this determines the urgency with which the rest of the workup should proceed. Much of the existing literature on SCC reviews specific causes rather than the initial ED presentation and thus provides only limited value to emergency physicians (EPs).6,7 We are unaware of previous published studies that have systematically investigated SCC of all nontraumatic causes in the ED setting.
We undertook this study to investigate the ED presentation of SCC from any nontraumatic etiology. We sought to characterize the presentation of nontraumatic SCC as a diagnostic entity and to determine the incidence and possible causes of misdiagnosis in the ED.
This was a retrospective case series designed to evaluate the ED presentation of sequential cases of nontraumatic SCC. The institutional review board approved the study protocol and waived the requirements for informed consent.
Study Setting and Population
This study evaluates the diagnosis of nontraumatic SCC at an urban, university-affiliated, tertiary care center with an ED volume of over 52,000 annual visits, staffed by residents and board-certified or board-eligible emergency physicians. All patients 18 years of age or older who were discharged from an inpatient stay between April 1, 2008, and July 31, 2009, with an International Classification of Diseases, Ninth Revision (ICD-9), code indicating spinal disease (Figure 1 lists all of the ICD-9 codes used as part of the inclusion criteria) were screened for the following inclusion criteria: 1) an ED visit to our facility for a related complaint within the previous 30 days; 2) diagnosis of SCC, defined as either a diagnosis of spinal cord or cauda equina compression recorded in the medical record, or radiologic evidence of spinal cord deformity from an extrinsic source by either MRI or computed tomography (CT) myelogram; 3) no evidence or documentation of recent or chronic SCC in this location. We excluded patients with acute traumatic mechanisms that resulted in activation of the trauma team and patients who were transferred to our ED with an existing diagnosis of SCC. For patients with more than one hospitalization meeting inclusion criteria, only the first visit within the study time frame was included.
One author screened patients with the appropriate ICD-9 codes for inclusion criteria. Included cases were reviewed by a second author, to ensure appropriate selection. Two of the authors, who were unblinded, extracted patient data from our hospital’s electronic medical record system (webOMR), which included both ED and inpatient documentation. We established the data points and abstraction process prior to study initiation and entered the data directly into a standardized, closed entry, Microsoft Access (Microsoft Corp., Redmond, WA) database form. Although the vast majority of the data abstraction was performed by one author, both abstractors met weekly to confirm compliance. After completion, we rechecked the electronic data abstraction forms once more for accuracy, and discrepancies were resolved with reevaluation of the patient chart and author consensus.
The source for the majority of the data was the EPs’ chart, including the history of present illness, past medical history, and initial physical exam. Notes from consultants who had evaluated the patient in the ED were included as part of the ED record for additional information in these areas, but the EPs’ charts were used if disagreement existed. History of present illness and physical exam findings were listed as present, absent, or not recorded/not performed. Inpatient medical records supplied additional information on past medical history, treatment, hospital course, predischarge physical examination findings and ambulation status, and final disposition. We included treatments instituted during that admission only.
Figure 2 details how we defined a “correct” ED diagnosis of SCC. If none of the requirements elaborated in Figure 2 were met, we determined that an ED misdiagnosis had occurred. Time to diagnosis is calculated as the elapsed time between when the patient registered in the ED, and the time the diagnostic imaging was performed.
We analyzed the data for the entire study population, and in two groups, based on whether they were correctly diagnosed or misdiagnosed. Continuous variables are reported as medians with interquartile range (IQR) to account for nonnormal distributions, and categorical variables are reported as total counts or ratios with percentages. We used the modified Wald method to estimate 95% confidence intervals (CIs) for proportions (QuickCalcs, GraphPad Software, La Jolla, CA).
Between April 1, 2008, and July 31, 2009, a total of 1231 patients were discharged from the hospital with an applicable ICD-9 code and had an ED visit within the previous 30 days. Of these, we identified 63 patients with new SCC meeting the inclusion and exclusion criteria. The majority of patients were excluded due to a lack of a new SCC diagnosis. The median age of the patients was 65 (IQR = 52–73) years, and 34 (54%) were male. Table 1 details the baseline characteristics and past medical history of the patients presenting with SCC.
|Baseline Characteristics||Total (n = 63)||Misdiagnosis (n = 18)||Correct Diagnosis (n = 45)|
|Age (yr)||65 (52–73)||66 (56–72)||63 (50–77)|
|Male||34 (54)||8 (44)||26 (58)|
|Past medical history|
|Previous SCC||6 (10)||0 (0)||6 (13)|
|Spine surgery||11 (17)||5 (28)||6 (13)|
|Diabetes mellitus||15 (24)||4 (22)||11 (24)|
|Current cancer||10 (16)||3 (17)||7 (16)|
|Treated cancer||7 (11)||4 (22)||3 (7)|
|HIV||0 (0)||0 (0)||0 (0)|
|Intravenous drug use||1 (2)||0 (0)||1 (2)|
|Time to diagnosis (hours)||9.3 (3.9–34)||54 (38–77)||5.3 (3.0–15)|
Among the entire cohort, SCC was most commonly located in the cervical region, followed by lumbar or cauda equina, and finally, the thoracic cord. The most common causes of compression were disc protrusion and spondylosis, followed by tumor, and spinal epidural abscess. Table 2 summarizes the locations and causes of SCC.
|Characteristic||Total (n = 63)||Misdiagnosis (n = 18)||Correct Diagnosis (n = 63)|
|Disk herniation/ protrusion||19 (30)||3 (17)||16 (36)|
|Spondylosis||16 (25)||7 (39)||9 (20)|
|Tumor||10 (16)||3 (17)||7 (16)|
|Spinal epidural abscess||7 (11)||3 (17)||4 (9)|
|Hematoma||5 (8)||0 (0)||5 (11)|
|Vertebral fracture||3 (5)||1 (6)||2 (4)|
|OPLL||1 (2)||0 (0)||1 (2)|
|Spondylolisthesis||1 (2)||0 (0)||1 (2)|
|Epidural lipomatosis||1 (2)||1 (6)||0 (0)|
|Cervical||32 (51)||11 (61)||21 (47)|
|Thoracic||12 (19)||2 (11)||10 (22)|
|Lumbar/cauda equina||17 (27)||4 (22)||13 (29)|
|Multiple sites||2 (3)||1* (6)||1† (2)|
The most common presenting symptoms in patients with SCC were pain in 44 (70%) patients, weakness in 35 (56%) patients, and difficulty ambulating in 38 (60%) patients. On physical examination, motor deficits were present in 45 (71%) patients compared to sensory deficits which were elicited in only 27 (43%) patients. However, almost one-quarter of the patients (15, 24%) had no motor or sensory deficits. Table 3 outlines the remaining elements of the history and physical examination. Figures 3 and 4 demonstrate the overall number of positive findings of history and physical examination elements, as well as the frequency with which they were documented in the ED record.
|Finding||Total (n = 63)||Misdiagnosis (n = 18)||Correct Diagnosis (n = 45)|
|History of present illness|
|Pain||44/63 (70)||12/18 (67)||32/45 (71)|
|Difficulty with ambulation||38/63 (60)||12/18 (67)||26/45 (58)|
|Weakness||35/63 (56)||8/18 (44)||27/45 (60)|
|Numbness||22/63 (35)||4/18 (22)||18/45 (40)|
|Tingling||19/63 (30)||3/18 (17)||16/45 (36)|
|Unilateral symptoms||20/54 (37)||6/13 (46)||14/41 (34)|
|Urinary incontinence or retention||15/45 (33)||3/9 (33)||12/36 (33)|
|Fever||5/44 (11)||2/13 (15)||3/31 (10)|
|Bowel incontinence or retention||6/42 (14)||0/8 (0)||6/34 (18)|
|Lumbar puncture||1/14 (7)||1/2 (50)||0/12 (0)|
|Recent weight loss (unintentional)||3/13 (23)||0/2 (0)||3/11 (27)|
|Median duration of symptoms (days)*||5 (1–14)||3 (1–14)||7 (1–14)|
|Motor deficit(s)||45/63 (71)||9/18 (50)||36/45 (80)|
|Sensory deficit(s)||27/63 (43)||5/18 (28)||22/45 (49)|
|Normal motor and sensory†||15/63 (24)||8/18 (44)||7/45 (16)|
|Unilateral exam findings||13/57 (23)||3/12 (25)||10/45 (22)|
|Decreased rectal tone||5/32 (16)||0/3 (0)||5/29 (17)|
|Midline spinal tenderness||20/31 (65)||3/8 (38)||17/23 (74)|
|Proprioception deficit||11/28 (39)||2/7 (29)||9/21 (43)|
|Saddle anesthesia||2/28 (7)||0/4 (0)||2/24 (8)|
|Abnormal gait||14/20 (70)||5/6 (83)||9/14 (64)|
|Normal motor/sensory† with abnormal gait||3/20 (15)||3/6 (50)||0/14 (0)|
|Urinary retention‡||3/17 (17)||0/3 (0)||3/14 (21)|
Spinal cord and cauda equina compression misdiagnosis occurred in 18 (29%, 95% CI = 19% to 41%) of the 63 patients. The misdiagnosed and correctly diagnosed groups were similar in terms of age, sex, and past medical history. As expected, median time to diagnosis was longer in those misdiagnosed (54 hours, IQR = 38–77 hours) than in those correctly diagnosed (5.3 hours, IQR = 3.0–15 hours); however, the misdiagnosed group had a shorter median duration of symptoms prior to presentation (3 [IQR 1–14] days vs. 7 [IQR 1–14] days). When these two times were combined, the median number of days from symptom onset until diagnosis was similar, but slightly shorter in the correctly diagnosed (7 days, IQR = 1–14 days) than misdiagnosed (8 days, IQR = 3–16 days) patients. Misdiagnosed patients were more likely to have normal motor and sensory function (not including proprioception) on examination (44% vs. 24%). Although the number of patients with gait testing was low, misdiagnosed patients were more likely to have an isolated gait deficit in the setting of normal motor and sensory function on examination (50% vs. 0%).Data Supplement S1 (available as supporting information in the online version of this paper) briefly describes the presentation of each misdiagnosed patient and shows there was often an alternate diagnosis used to explain the patients’ pain, weakness, or difficulty ambulating.
The majority of patients were treated with surgery (33 patients, 52%), while others were treated with radiation, antibiotics, bracing, or simply physical therapy. Upon discharge from the hospital, 41 (65%) patients had a persistent motor deficit, and 23 (43%) had a persistent sensory deficit. There was no obvious difference in neurologic outcomes (motor or sensory function, ability to ambulate, or need for bladder catheterization at hospital discharge) between the misdiagnosed and correctly diagnosed groups. Table 4 shows the treatments and outcomes.
|Characteristic||Total (n = 63)||Misdiagnosis (n = 18)||Correct Diagnosis (n = 45)|
|Surgery||33 (52)||7 (39)||26 (58)|
|Radiation||3 (5)||2 (11)||1 (2)|
|IV antibiotics||3 (5)||1 (6)||2 (3)|
|Cervical collar alone||14 (22)||5 (28)||9 (20)|
|Physical therapy alone/none||10 (16)||3 (17)||7 (16)|
|Ambulatory to nonambulatory*||2 (3)||1 (6)||1 (2)|
|Nonambulatory to ambulatory*||5 (8)||1 (6)||4 (9)|
|Motor deficit at hospital discharge||41 (65)||10 (56)||31 (69)|
|Sensory deficit at hospital discharge||23 (37)||4 (22)||19 (42)|
|Foley catheter at hospital discharge‡||10 (16)||3 (17)||7 (16)|
From the ED perspective, SCC is a critical diagnosis that presents with a dynamic array of symptoms and signs varying in degree, number, and timing. Studies focused on specific etiologies provide an incomplete perspective of SCC. We are not aware of other published studies that have systematically investigated the presentation of nontraumatic SCC as a single entity or attempted to estimate the misdiagnosis rate of nontraumatic SCC in the ED, regardless of cause.
The patient demographics and etiology of nontraumatic SCC will vary by location based on incidence of risk factors for epidural abscess and other disease processes. In our population, disk disease, such as herniated disk and spondylosis, was the most common cause, followed by tumor, epidural abscess, and epidural hematoma. A 2009 study from Singapore of patients evaluated by urgent MRI found 56% of SCC caused by metastasis, followed by disk herniation (16%), epidural hematoma (12%), infective spondylodiscitis (8%), and vertebral fracture (8%).8 Although this study was not specifically an ED population, the comparative increase of SCC caused by disk disease in our study is striking. This difference could be due to higher disease prevalence, or possibly, that in our hospital, with an active oncology practice, patients with metastatic SCC may be recognized and directly admitted from the oncology outpatient clinic without being seen in the ED.
The cervical spine is the most common site of SCC in this cohort of patients. This stands in contrast to the emphasis placed on cauda equina compression in both the emergency medicine literature and in physician education. Perhaps this emphasis on cauda equina compression is derived from the frequency with which emergency medicine providers treat lumbar back pain or the likely false belief that cauda equina compression has a subtler presentation than SCC. As the current literature focuses on diagnosis-specific SCC, there is little comparison in the literature to frame our findings on location of SCC. Previous studies on metastatic SCC show the most common site is the thoracic spine (60%), followed by lumbosacral (25%), and then cervical (10%).9 Likewise for spinal epidural abscesses, the most common site is thoracic (35%), followed by lumbosacral (30%), and cervical (19%).10 The proportion of cervical disease in this population is likely due to the increased prevalence of disk disease. Although cauda equina compression is an important mechanism of spinal pathology, it is not the most common site of compression, and the emphasis of future studies and education should be balanced among the other spinal sections.
Nearly one-quarter of our cohort had no motor or sensory deficit on physical exam in the ED and an additional 23% had unilateral findings on exam. This low proportion of physical exam findings in patients with radiologically confirmed SCC emphasizes the importance of the history of present illness, past medical history, and clinical suspicion in establishing the patient’s epidemiologic context.11 One possible explanation for the lack of symptoms in our population is inclusion of patients with mild SCC, which is supported by the fact that 16% of our patients received only physical therapy at the time of discharge from the hospital. The preponderance of disk disease in our population, which may resolve without surgery, could also explain the low rate of surgical intervention. All patients included in this study had radiologic evidence of new external impingement on the spinal cord (not merely the nerve roots) and had symptoms that correlated with the radiologic findings. Two misdiagnosed patients were discharged home due to minimal neurologic symptoms and lack of a diagnosis of SCC, only to return within 3 days with worsening neurologic deficits, ultimately requiring surgical intervention. Because a given patient’s course is difficult if not impossible to predict, even cases of early or mild cord compression need to be recognized to receive appropriate treatment and close observation to avoid deterioration of the patient’s neurologic status.
The misdiagnosis rate of 29% confirms the difficulties of recognizing this entity in the ED. Although there are no estimates of SCC misdiagnosis rates in the literature comparable in scope to ours, previous investigations that have specifically focused on the delayed diagnoses of spinal epidural abscesses and cauda equina syndrome found misdiagnosis rates of 75 and 50%, respectively.6,7 Even though these studies did not address the complete entity of SCC, they do provide additional evidence that the misdiagnosis of SCC in the ED is a common phenomenon.
Pain was the most common element of history of present illness, with difficulty with ambulation (present in 60% of patients) more common than sensations of weakness (56%) and numbness (35%). Similarly, 14 patients (22% of the cohort) had an abnormal gait on exam. Only 20 patients were even evaluated for gait disturbance; therefore, it is possible that the actual number of patients with abnormal gait was even higher. Difficulty with gait on both history and exam is known to be associated with SCC, although it is not always asked about or tested in the ED. Three of our patients presented with ataxia, despite having normal motor and sensory exams, all of whom were misdiagnosed. Prior published case reports and a systematic review of metastatic cord compression have documented other cases in which ataxia was the predominant neurologic finding.12,13 In a systematic review, 9 of 130 patients presented with isolated ataxia, two of whom had no back pain. Physicians may be less likely to diagnose SCC in patients whose gait abnormality cannot be explained by motor or sensory deficits. However, given the anatomy of the spinal cord, it is not surprising that ataxia can be an important initial sign of SCC.
Upon closer analysis of the misdiagnosed cases in our cohort, most patients had additional acute and chronic medical problems. Patients with multiple comorbidities or multiple acute conditions in the ED seemed more likely to have subjective complaints such as weakness, unsteadiness, or an inability to get out of bed, initially attributed to one of their chronic or acute non-SCC medical conditions. Due to diagnostic anchoring, their EP may have failed to reconsider SCC as the workup progressed. In some cases, patients with early SCC may have normal neurologic exams, with the only clue of spinal cord pathology being a nonspecific history of unsteadiness, recent falls, or acute changes in ambulation status.
Alternatively, misdiagnosed patients may present with milder disease. The misdiagnosed group received less surgical intervention, possibly due to fewer neurologic findings and less severe disease. Despite a prolonged time to diagnosis in the misdiagnosed group, the outcomes in both groups were similar. Although previous studies have shown that increased time to diagnosis is associated with worse outcomes, the misdiagnosed group also had a decreased duration of symptoms at presentation to the ED, equalizing the time from symptom onset to diagnosis.2 The literature also shows that neurologic outcome is based on the neurologic status at the time of diagnosis, and the misdiagnosed group tended to have few neurologic findings at presentation.1 Although we neither powered nor designed our study to show a difference in outcome between the misdiagnosed and correctly diagnosed groups, early diagnosis, before additional neurologic deterioration occurs, could theoretically lead to improved outcomes.
Additional research focused on the ED presentation of patients diagnosed with SCC should be a high priority. In particular, future research should focus on the prospective determination of ED misdiagnosis from multiple centers to better understand the full scope of the problem, the effect of misdiagnosis on patient morbidity and mortality, the causes of misdiagnosis, and specific measures that can be implemented to improve diagnostic accuracy.
We relied exclusively on documentation present in each patient’s medical record, and therefore our data were subject to any inaccuracies or biases inherent in a retrospective chart review. We attempted to reliably reflect the record by recording variables as present, not present, or not recorded; however, it is impossible to know for sure whether information that was not documented in a patient’s record was not elicited or was elicited but not recorded. Patients with more severe symptoms or abnormalities on physical examination may have been more likely to have this information documented, because the treating physician may have had a greater concern for these patients and, thus, had more incentive to perform a thorough workup or to document more completely. This may have also affected our outcome analysis, since the patients with documentation in our population may have been more likely to have severe disease.
Furthermore, the definitions we used to differentiate correct diagnoses from misdiagnoses may have inaccurately categorized some patients, thus potentially resulting in misclassification bias. As a result of the specific methodology used, this study detects cases of SCC based on inpatient discharge diagnosis. Therefore, it would not detect patients with SCC who were misdiagnosed and discharged from the ED and then sought follow-up care at another hospital. The possibility also exists that the ICD-9 codes used to identify potential patients did not capture all cases of SCC during the study time period, either because the wrong ICD-9 codes were used for some patients or because only ICD-9 codes reflecting the patient’s non-SCC diagnoses were used. The fact that 1,231 patients were reviewed to find 63 cases of cord compression suggests that the included ICD-9 codes were quite broad and inclusive in scope. All patients in this study were evaluated at a single center, which could limit the generalizability of the results to other locations with a different demographics or disease patterns. Finally, the small size of our study population is another factor that may have impacted our results.
The ED presentation of spinal cord and cauda equina compression may be subtle, increasing the diagnostic difficulty. In this population, 29% of patients were misdiagnosed in the ED. Greater awareness that early spinal cord and cauda equina compression may present with unilateral motor or sensory deficits, or as subjective ambulatory dysfunction without other neurologic deficits, might improve diagnostic accuracy.