The Canadian C-Spine rule performs better than unstructured physician judgment

Abstract

Study objective We compare the predictive accuracy of emergency physicians' unstructured clinical judgment to the Canadian C-Spine rule. Methods This prospective multicenter cohort study was conducted at 10 Canadian urban academic emergency departments. Included in the study were alert, stable, adult patients with a Glasgow Coma Scale score of 15 and trauma to the head or neck. This was a substudy of the Canadian C-Spine and CT Head Study. Eligible patients were prospectively evaluated before radiography. Physicians estimated the probability of unstable cervical spine injury from 0% to 100% according to clinical judgment alone and filled out a data form. Interobserver assessments were done when feasible. Patients underwent cervical spine radiography or follow-up to determine clinically important cervical spine injuries. Analyses included comparison of areas under the receiver operating characteristic (ROC) curve with 95% confidence intervals (CIs) and the κ coefficient. Results During 18 months, 6,265 patients were enrolled. The mean age was 36.6 years (range 16 to 97 years), and 50.1% were men. Sixty-four (1%) patients had a clinically important injury. The physicians' κ for a 0% predicted probability of injury was 0.46 (95% CI 0.28 to 0.65). The respective areas under the ROC curve for predicting cervical spine injury were 0.85 (95% CI 0.80 to 0.89) for physician judgment and 0.91 (95% CI 0.89 to 0.92) for the Canadian C-Spine rule ( P<.05). With a threshold of 0% predicted probability of injury, the respective indices of accuracy for physicians and the Canadian C-Spine rule were sensitivity 92.2% versus 100% ( P<.001) and specificity 53.9% versus 44.0% ( P<.001). Conclusion Interobserver agreement of unstructured clinical judgment for predicting clinically important cervical spine injury is only fair, and the sensitivity is unacceptably low. The Canadian C-Spine rule was better at detecting clinically important injuries with a sensitivity of 100%. Prospective validation has recently been completed and should permit widespread use of the Canadian C-Spine rule.