RESUMEN
BACKGROUND: Trauma is the cause of 1.2 million deaths in India annually. Injury severity scores play an important role in trauma research and care because these scores enable the adjustment of trauma severity when comparing mortality outcomes. The generalizability of the International Classification of Diseases Injury Severity Score (ICISS) between different populations is not fully known, and the validity of the ICISS has not been assessed in the Indian context. The aim of this study was to assess the predictive performances of three international versions of the ICISS, derived from data from Australia, New Zealand and pooled data from seven different high-income countries, in trauma patients admitted to four public hospitals in urban India. MATERIAL AND METHODS: We used patient data from an Indian cohort of 16,047 trauma patients. The patients were assigned an ICISS based on International Classification of Diseases codes using survival risk ratios from publicly available data sets from Australia and New Zealand and with pooled data from seven different high-income countries. Predicted mortality based on the ICISS was compared with observed patient mortality, and the predictive performance was assessed in terms of discrimination and calibration. RESULTS: Discrimination and calibration did not reach the threshold for predictive performance in any of the ICISS versions used. The threshold value used was 0.8 for discrimination, which was not significantly different from one for the calibration slope and not significantly different from zero for the calibration intercept. CONCLUSIONS: None of the international versions of the ICISS adequately predicted mortality within the study population, indicating the need for an ICISS version specifically adapted to the Indian context.