Physicians are inaccurate in predicting blunt trauma severity prior to imaging results
Harrison N, Ball M, Grahf D, Miller M, Klausner HA, Otero R, and Ventura A. Physicians are inaccurate in predicting blunt trauma severity prior to imaging results. Acad Emerg Med 2019; 26:S118-S119.
Acad Emerg Med
Background: Routine whole-body computed tomography (WBCT, or “pan-scanning”) was shown to reduce mortality in numerous observational studies for patients with major trauma, but failed to show a benefit in the only randomized controlled trial to date. Still, routine WBCT use in suspected major trauma remains frequent and controversial in the United States (US). Major trauma has been defined by an Injury Severity Score (ISS)>15, which is retrospectively calculated after imaging results and accounts for only about 1% of US traumas. We hypothesized that physician prediction of blunt trauma severity is too inaccurate to guide routine WBCT ordering for suspected ISS>15. Methods: We prospectively enrolled a convenience sample of 561 blunt trauma patients at two US Level I trauma centers. The treating EM physicians were asked before imaging to predict the patient's ISS in one of three ordinal categories: 0-1, 2-15, or >15. ISS prediction was calculated by simple proportion of correct vs. incorrect (primary outcome). Incorrect predictions were further stratified by overestimates vs. underestimates (secondary outcome). The primary and secondary outcomes were compared to clinical variables hypothesized to affect prediction accuracy, with unpaired T-tests for continuous variables and odds ratios (OR) with 95%Cis for binary variables. Results: 145 patients were excluded because the physician was too uncertain to identify a single ISS category, leaving 416 patients in the sample. Physicians predicted ISS correctly 61.3% of the time. When incorrect, physicians overestimated severity twice as often as underestimated (67.1% vs 32.9%, respectively). Chronic opiate use (OR=2.2; 95%CI 1.2-3.9), ICU disposition (OR=6.0; 95%CI 1.2-29.2), and higher ISS (3.0 vs. 2.2, p=0.02) were associated with incorrect prediction. The prevalence of ISS>15 traumas was 1.6%. If a protocol for routine WBCT in all patients with ISS expected >15 had been implemented in our cohort, there would have been significantly more inappropriate patient selection for WBCT than appropriate decisions (OR=5.6; 95%CI 2.4-13.2) Conclusion: Physicians are inaccurate in predicting blunt trauma severity before CT imaging, and tend to overestimate. CT imaging decisions based on perceived severity, such as routine WBCT for suspected major blunt trauma, are likely to lead to inappropriate CT ordering.