Determination of potential risk characteristics for cauda equina compression in emergency department patients presenting with atraumatic back pain: a 4-year retrospective cohort analysis within a tertiary referral neurosciences centre

OBJECTIVE: Atraumatic back pain is an increasingly common presentation to the ED. A minority of these cases will have significant structural pathology, resulting in acute cauda equina compression (CEC). Although clinicians often look for 'red flags' to identify potential CEC, the prognostic accuracy of these presenting symptoms and clinical examination findings is unclear. We sought to evaluate the accuracy of individual clinical features in a large cohort of ED patients with atraumatic backpain and reference standard imaging, for the diagnosis of CEC.

METHODS: A retrospective case note review from 2014 to 2018 within an established ED atraumatic back pain pathway, undertaken at the largest tertiary spinal referral centre in the UK. We analysed routine data, collected prospectively by treating clinicians within a structured electronic health record clinical proforma. Data on signs and symptoms in 996 patients with suspected CEC referred for definitive MRI over a 4-year study period were extracted and compared against a final reference standard diagnosis.

RESULTS: We identified 111 patients with radiological evidence of CEC within the cohort referred for definitive imaging (111/996, 11.1%), of whom 109 (98.2%) underwent operative intervention. Patients with CEC were more likely to present with bilateral leg pain (OR=2.2), dermatomal sensory loss (OR 1.8) and bilateral absent ankle or ankle and knee jerks (OR=2.9). Subjective weakness was found to be associated with CEC on univariate but not multivariate analysis. We found no relationship between digital rectal examination findings and the diagnosis of CEC.

CONCLUSIONS: In our cohort, factors independently associated with CEC diagnosis on MRI included bilateral leg pain, dermatomal sensory loss. Loss of lower limb reflexes was strongly suggestive of CES (likelihood ratio 3.4 on multivariate logistic regression). Our findings raise questions about the diagnostic utility of invasive digital rectal examination.