An independent validation of a clinical prediction rule for the diagnosis of cervical radiculopathy with radicular pain

BACKGROUND: A clinical prediction rule (CPR) for cervical radiculopathy (CR), published in 2003, is widely used in practice and recommended in clinical practice guidelines. To date, this CPR has not been independently validated. OBJECTIVE: To perform an independent broad validation of the 2003 CPR using magnetic resonance imaging (MRI) as reference standard, and to investigate whether an alternative test cluster has stronger diagnostic utility for identifying CR with radicular pain. METHODS: This prospective diagnostic accuracy study was conducted following the updated STARD 2015 guidelines. Eighty-five individuals (27 with CR) were included from 109 consecutive patients. The diagnosis of CR was based on cervical spine MRI findings consistent with the patient's clinical features, symptoms, and neurological examination, interpreted by a neurosurgeon. Twelve clinical tests were performed by the same examiner blinded to diagnosis. RESULTS: Validation of the 2003 CPR (cervical distraction, Spurling's test, upper limb neurodynamic test (ULNT) 1, and cervical rotation <60°) produced diagnostic values comparable to the original study. Our independent cluster based on backwards stepwise regression identified three variables: (1) modified (passive) shoulder abduction test, (2) Spurling's arm pain test, and (3) ≥2 of 4 ULNTs positive. With this new cluster, having all three tests positive provided an infinite LR+ and a 100% post-test probability, identifying 37.0% of CR with radicular pain cases compared to 17.9% in the original 2003 cluster (4/4 positive tests). CONCLUSION: This study findings corroborate the original 2003 CPR and identified a new cluster that had stronger diagnostic utility for CR with radicular pain.