O0021: Derivation of a Pediatric Cervical Spine Injury Decision Rule

Background: Cervical spine injuries (CSIs) in children are uncommon but potentially devastating. Indiscriminate neck imaging after trauma, however, unnecessarily exposes children to ionizing radiation. A robust decision rule for identifying children at risk for CSIs after blunt trauma is lacking. The purpose of this study was to derive a decision rule for CSI in a large multi-center prospective cohort of injured children.

Methods: Children ages 0-17 years who experienced blunt trauma and were evaluated at any of 18 Pediatric Emergency Care Applied Research Network emergency departments (EDs) were eligible for enrollment. To meet inclusion criteria, children must have presented from the scene of injury via emergency medical services, received trauma team evaluation and/or had cervical spine imaging ordered. Children with solely penetrating trauma were excluded. Data were collected prospectively from treating ED providers at the time of the encounter. Risk factors assessed included mechanisms of injury, predisposing conditions, neck and neurologic complaints, and physical examination findings. Medical records were reviewed for neck imaging 21-28 days after the ED encounter, and if present, the imaging reports were systematically reviewed for CSIs. If no imaging was obtained, guardians were contacted to determine if a CSI was subsequently diagnosed in another setting. A classification rule was constructed by first identifying a subset of variables associated with high likelihood of CSI. Next a classification and regression tree (CART) analysis was performed on the remaining participants to predict CSI. Measures of test performance (sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV)) were calculated.

Results: Of 11,870 children enrolled in the derivation cohort, 274 (2.3%) had CSIs. Glasgow Coma Scale of 3-8 or unresponsiveness to pain; abnormal airway, breathing, or circulation findings; and focal neurologic deficits were highly associated with CSI based on bivariable analysis. Of 930 children with at least one of these findings, 118 (12.7%) had CSIs. Bivariable analysis using the remaining 10,940 participants identified 28 CSI risk factors (Table 1). CART identified self-reported neck pain, altered mental status, substantial torso injury, and neck tenderness on examination to be associated with CSI. A decision rule using the combined risk factors had 92.7% sensitivity, 59.5% specificity, 5.1% PPV, and 99.7% NPV for CSI. If these factors had been used in decision-making for imaging within the cohort, (i.e., highest risk factors trigger CT and remaining risk factors trigger x-ray), x-ray use would have decreased from 45.2% to 33.9% and CT use from 21.4% to 7.8%.

Conclusion: We derived a parsimonious and highly accurate decision rule for identifying children at risk for CSI after blunt trauma. Future directions include validation of the decision rule in an independent cohort and designing and implementing a risk-centered decision aid for pediatric CSI assessment.