285 Predictive Accuracy of Adding Shock Index to the American College of Surgeons' Major Resuscitation Criteria for Adult Trauma Triage

Abstract

The American College of Surgeons' Major Resuscitation Criteria (MRC) are used to define when a trauma surgeon should be involved in a patient’s care in the emergency department (ED). While hypotension is strongly predictive of critical illness or the need for emergent intervention, hypotension may not occur during early stages of significant hemorrhage or in some populations (eg, young) during later stages of hemorrhagic shock. Our objective was to assess the added predictive accuracy of shock index (SI) to the MRC. Design: Retrospective cohort study. Setting: Urban level 1 trauma center with an approximate annual ED census of 60,000 adult patients, 600 of which have major multi-systems injuries. Population: All patients (≥18 years of age) who presented for care following trauma and included in our trauma registry. Data collection: We used prospectively collected trauma registry data from September 1, 1993 through November 30, 2006 as part of a structured evaluation of trauma triage. Outcomes: Emergent operative intervention (EOI), defined as operative intervention by a trauma surgeon within 1 hour of arrival; emergent procedural intervention (EPI), defined as cricothyrotomy or emergency department thoracotomy; and injury severity score (ISS) > 15. All outcomes were confirmed by physician abstractors blinded to the purpose of the study. Analyses: Sensitivities, specificities, and 95% confidence intervals (CIs); subgroup analyses by age strata. 20,872 patients met criteria for inclusion with a median ISS of 9 (interquartile range [IQR]: 4 - 16) and a median SI of 0.7 (IQR: 0.6 - 0.9). Prevalences of individual MRC were: SBP < 90 mm Hg: 1,760 (8%); respiratory compromise: 3,111 (15%); GSW to the neck, chest, or abdomen: 791 (4%); GCS score < 8: 2,561 (12%); transfer from another hospital and requiring blood: 68 (0.3%); and physician discretion: 139 (0.7%). Of all patients, 4,663 (22%) met at least 1 MRC criterion. EOI or EPI was required in 1,099 (5%; 95% CI: 5% - 6%) and 5,552 (27%; 95% CI: 26% - 27%) had an ISS > 15. Sensitivity and specificity of the MRC only for EOI or EPI was 86% (95% CI: 83% - 88%) and 81% (95% CI: 80% - 81%), respectively; sensitivity and specificity of the MRC only for ISS > 15 was 57% (95% CI: 56% - 58%) and 90% (95% CI: 89% - 91%), respectively. Addition of SI thresholds modified predictive accuracies with MRC plus SI ≥ 0.8 significantly improving sensitivities (86% to 92%; +6%, 95% CI: 4% - 9%, p<0.001 for EOI/EPI, and 57% to 73%; +16%, 95% CI: 14% - 18%, p<0.001 for ISS > 15) but with a substantial decreases in specificity (81% to 61%; -20%, 95% CI: -19% - -21%, p<0.001 for EOI/EPI, and 89% to 68%; -21%, 95% CI: -20% - -22%, p<0.001) (Figure). No significant differences were identified when stratified by age. Among a large adult trauma population, the addition of SI to the MRC significantly changed predictive accuracies for EOI, EPI, and ISS > 15. Addition of a SI threshold ≥ 0.8 significantly improved sensitivities but with a relatively greater reduction in specificities. Adoption of this MRC plus SI trauma triage strategy would increase trauma activations by approximately 33% while decreasing the number of under-triaged patients with EOI/EPI by 6% and ISS > 15 by 15%.