Shock Index versus Systolic Blood Pressure for Mortality Prediction in Traumatic Hemorrhage: A Comparative Prognostic Evaluation in a Middle-Income Country.

Abstract

Purpose: This study aims to evaluate and compare the prognostic performance of shock index (SI) and systolic blood pressure (SBP) for predicting mortality among trauma patients at risk of hemorrhage in South Africa, a high-trauma burden setting with resource constraints.

Patients and methods: This is a secondary analysis of data from the EpiC study, a prospective, multicenter observational investigation of trauma patients in the Western Cape, South Africa. Adult patients with blunt or penetrating injuries and no significant head injury were included. The predictor variables-SI and SBP at facility arrival-and the outcome (30-day mortality) were analyzed through a sequential analytical framework. Segmented logistic regressions identified inflection points; ROC analyses with Youden's J statistic determined optimal thresholds. Model comparisons employed DeLong's test, calibration plots, Brier scores, and decision curve analysis. Subgroup analyses were conducted by injury mechanism and severity.

Results: Among 3609 patients, 6.1% died within 30 days. SI demonstrated an inflection point at 0.69 and SBP at 82 mmHg. ROC analysis yielded similar AUCs of 0.62 (SI) and 0.60 (SBP) (p=0.418). However, SI showed superior calibration (Brier score: 0.054 vs 0.058), a higher positive likelihood ratio (2.70 vs 2.01; p<0.001), and greater net benefit across a broader range of predicted-risk thresholds on decision curve analysis. The optimal SI cutoff of ≥1.21 yielded 85.1% specificity and 40.3% sensitivity.

Conclusion: Both SI and SBP demonstrated only modest discrimination for 30-day mortality, reflecting the inherent limitations of single vital-sign predictors in complex trauma populations. Within these constraints, SI showed marginally better calibration and rule-in performance than SBP; however, neither predictor alone is sufficient for definitive mortality risk prediction. An SI threshold of ≥1.21 may serve as a practical screening tool to flag high-risk patients warranting closer monitoring or escalated care, particularly in resource-limited settings where laboratory and imaging resources are constrained.