Predictive power of RTS, ASCOT, NISS, ISS AND TRISS scores in severe trauma patients from Republic of Moldova. Preliminary data

Abstract

Introduction: Attempts to predict the evolution course of the pathological conditions led to the traumatic scores’ elaboration based on the distinct medical systems realities that differ in many respects from the local one. Thus, it is necessary to identify the optimal one for the trauma centers in the country. Purpose: Comparative evaluation of the five most common traumatic scores described in the literature in the conditions of a trauma center in the Republic of Moldova to identify the score with maximum predictive abilities. Material and methods: In the retrospective study (654 patients with severe trauma), RTS, ISS, NISS, ASCOT, TRISS scores were applied to assess patient’s survival rate. The prediction results were compared and statistically analyzed by logistic regression. Results: The comparative evaluation of the five traumatic scores indicated the superior predictive abilities of the ASCOT score as indicated by its coefficient of determination (21.6%), calibration (χ² = 9,171, df = 8, р = .328) and discrimination (area under the curve RОС 0.727). It was followed by the modified TRISS score with the coefficient of determination (20.3%), calibration (χ² = 8,824, df = 8, р = .357) and discrimination (area under the RОС 0.719 curve). Conclusions: The comparison of the proposed scores determined the ASCOT score as one with the highest accuracy in prediction in the conditions of a trauma center in the Republic of Moldova.