3 Derivation and internal validation of a clinical decision rule to guide whole body computed tomography scanning in trauma.

BACKGROUND: There are no widely accepted validated clinical decision rules for the use of WBCT in trauma. Given the potential risks and costs, there is a clear need for a clinical decision rule (CDR) to safely guide targeted use of WBCT. We aimed to derive a CDR to guide clinical decisions on WBCT utilisation by detecting patients at high and low risk of multi-region trauma.

METHODS: We retrospectively identified consecutive patients who had presented to a major trauma centre with suspected major trauma. Study took place at Aintree University Hospital, Merseyside. After extracting data, we derived a clinical decision rule for detection of multi-region trauma by logistic regression and recursive partitioning. The primary outcome was defined as injuries of AIS≥2 in two or more body regions, while the secondary outcome was the presence of two injuries of AIS≥3 in two or more body regions. This rule was cross-validated on the original sample using bootstrapping.

RESULTS: 1608 patients were included in the study. The derived model combined a bespoke physiological score with mechanistic and anatomical factors. The physiological score identified the risk of multi-region injury at various cut-offs of age, systolic blood pressure, GCS, heart rate and respiratory rate. Patients were further categorised according to mechanism of injury and clinical findings, and specific physiological scores were applied to each category to determine which patients in these categories required WBCT. 'High risk' injury mechanisms included high falls and unprotected road traffic collisions. Clinical signs of injury were categorised by body region, including the head, chest, abdomen and pelvis (figure 1). The overall sensitivity of the clinical decision rule for the primary objective was 96.0% (95% CI:94.8 to 97.2) while the specificity was36.1% (95% CI:33.3 to 39.0). The negative likelihood ratio was 0.11. For the secondary objective the sensitivity was 98.5%, the negative likelihood ratio 0.04.emermed;34/12/A861-a/F1F1F1Figure 1 CONCLUSION: This study derived a two stage CDR which was highly sensitive for identifying patients at high risk of multiregion injury. A prospective external validation study is now required to further refine and improve this model. This could provide a useful screening tool in the future.