The Canadian CT Head Rule (CCHR), New Orleans Criteria (NOC), Canadian C-Spine Rule (CCR), and NEXUS criteria are widely used to guide computed tomography (CT) after minor head injury, but many contemporary patients fall outside their original derivation cohorts. We aimed to externally validate these four rules in an extended, real-world emergency department cohort. We conducted a prospective multicenter observational study (HELMET trial) at five emergency and critical care centers. Adults (≥18 years) presenting within 24 h of blunt head trauma with Glasgow Coma Scale (GCS) scores of 13–15 were eligible. We excluded patients with GCS 2 in any non-head region, prior intracranial injury or cranial surgery, penetrating trauma, pregnancy, or refusal. Primary outcomes were radiographic intracranial injury on head CT and radiographic cervical injury on cervical CT. Diagnostic accuracy of each rule was estimated in complete-case cohorts and with multiple imputation for missing components. Of 3278 eligible patients, 1706 formed the complete-case cohort (median age 71 years). Traumatic findings were present in 143 head CTs (8.4%) and 14 cervical CTs (0.8%), and 44 patients (2.6%) underwent neurosurgical intervention. For head CT, CCHR sensitivity/specificity were 91.1%/33.8%, and NOC 95.8%/17.9%. For cervical CT, CCR sensitivity/specificity were 93.1%/37.0%, and NEXUS 71.4%/75.5%. In this extended minor head injury cohort, CCHR, NOC, CCR, and NEXUS did not achieve near-perfect sensitivity and had limited specificity. These rules have limited utility as stand-alone exclusion tools for reducing CT use and may serve as adjuncts to clinical judgment.
Kimoto et al. (Sun,) studied this question.