Predicting mortality in severe polytrauma with limited resources.

Background: Objective evaluation of the severity of injured patients is crucial for the adequate triage, decision-making, operative and intensive care management, prevention, outcome studies and system quality assessment. This study aimed to compare six, widely- used, trauma scores as predictors of mortality, and to identify the most powerful among them in limited-resources settings.

Methods: Seventy-five polytraumatized patients, admitted to the ICU of the Clinic for Emergency Surgery (Level 1 trauma center, CSS Belgrade) from June 2018-August 2020, were included in the study. The inclusion criteria were age≥16, ISS≥16 and SOFA≥5 points. Scores were evaluated using logistic regression model and analysis of areas under the receiver operating characteristic (ROC) curve (AUC).

Results: During the 26 months period, highly selected cases, mostly of blunt trauma (97.3%), due to a road traffic accident (68%) and free-falls (25.3%) were included. Surgery was indicated in 56 (74.7%) and non-operative treatment in 19 (25.3%) cases, with overall mortality rate at 36%. Logistic regression analysis demonstrated that all six trauma scores (ISS, NISS, APACHE2, SOFA, TRISS, KTS) were significant mortality predictors (p<0.001). Observed cut-off values for ISS: 39.5, NISS: 42, APACHE 2: 25, SOFA 6.5 points are predictive for mortality in non-survivors. A multivariate analysis showed that the most powerful mortality predictors are TRISS and APACHE 2 with AUCs: 0.9 and 0.866.

Conclusion: According to our study the most powerful mortality predictors are APACHE 2 and TRISS, even in limited-resources hospital settings, while statistically significant KTS, did not perform as expected. We propose the appliance of the KTS, as the tool for exploiting 'golden hour', ISS or NISS during admission stage and APACHE 2 or TRISS for use in the first 24 hours after admission to ICU.