The elderly is an increasing proportion of all cases treated at trauma centers. Shock index (SI) calculated as heart rate (HR) divided by systolic blood pressure (SBP), has been shown to be a good predictor of mortality and transfusion in injured patients. One limitation of SI is that its accuracy in different age groups, especially the elderly has not been fully evaluated.

We studied the accuracy of admission SI in predicting early, 48-hour and in-hospital mortality, and major interventions (massive transfusion, ICU admission and surgery in 24 hours) in trauma patients admitted to a major trauma center. We examined whether age, injury severity, injury type, blood alcohol and comorbidities affected the predictive accuracy of SI. Of particular interest is the accuracy of SI in the elderly. We also compared the predictive accuracy of SI, HR and SBP. Optimal cut-points for SI were determined.

SI had acceptable accuracy in predicting mortality outcomes, and ICU admission overall. Accuracy was good in the prediction of massive transfusion, and poor in the prediction of surgery in 24 hours. SI was better than HR or SBP in predicting mortality outcomes (all ages, elderly, and younger patients). However, in older patients, accuracy of SI in predicting major interventions was not different from that of SBP. Accuracy of SI in predicting 48-hour and all in-hospital mortality, and ICU admission was better in younger patients. Accuracy was also better among those with lower injury severity than in those who were more severely injured. Accuracy of SI in predicting massive transfusion was similar in older and younger trauma patients. Optimal cut-offs for predicting outcomes were lower for older patients (0.5-0.7 for mortality and major interventions) than in younger patients (0.6-0.9 for mortality and 0.6-0.8 for major interventions). Accuracy of SI in predicting all in-hospital death and massive transfusion was less among patients with elevated blood alcohol while comorbidities did not affect accuracy.

In conclusion, SI is less accurate in in predicting mortality among older patients and is less accurate in predicting mortality and massive transfusion among blood alcohol-positive patients, potentially affecting its utility in triage and clinical management.