Detected and overlooked cervical spine injury in comatose victims of trauma: report from the Pennsylvania Trauma Outcomes Study

Abstract

This study was undertaken to determine whether a clinically useful rule could be formulated for identifying the presence of traumatic brain injury (TBI) in patients who are at exceptionally low risk of cervical spine injury. The Pennsylvania Trauma Outcomes Study database was searched for cases of TBI in which the admission Glasgow Coma Scale (GCS) score was less than or equal to 8. Cases of cervical injury were identified based on diagnostic codes. Associations between cervical injury and various clinical variables were tested using chi-square analysis. The probability of cervical injury was modeled using logistic regression. Decision tree models were constructed. Statistical determinants of overlooked cervical injury were examined. The prevalence of cervical injury among 41,142 cases of TBI was 8%. Mechanism of injury, thoracolumbosacral (TLS) fracture, age, limb fracture, admission GCS score, hypotension, and facial fracture were associated with cervical injury and were incorporated into the following logistic regression model: probability = 1/(1 + exp[4.248 - 0.417 X mechanism -0.264 X age - 0.678 X TLS - 0.299 X limb - 0.218 X GCS - 0.231 X hypotension - 0.157 X facial]). The results of applying this model provided a rule for cervical spine clearance applicable to 28% of the cases with a negative predictive value (NPV) of 97.0%. Decision tree analysis yielded a rule applicable to 24% of the cases with an NPV of 98.2%. The prevalence of overlooked cervical injury in all individuals with severe TBI was 0.3%; the prevalence of overlooked cervical injury in patients with cervical injury was 3.9%. Overlooked cervical injury was less common in patients with associated TLS fractures (odds ratio 0.453, 95% confidence interval 0.245-0.837). This analysis identified no acceptable rule to justify relaxing vigilance in the search for cervical injury in patients with severe TBI. Provider vigilance and consequent rates of overlooked cervical injury can be affected by environmental cues and presumably by other behavioral and organizational factors.