The evaluation of accuracy of prediction of survival probability in trauma patients according to APACHE II, ISS and TRISS

Overview

The target of the study was to compare the accuracy of survival probability according to scoring systems APACHE II (Acute Physiologic and ChronicHealth Evaluation), ISS (Injury Severity score) and TRISS (Trauma and Injury Severity Scoring) in trauma patients hospitalized in Department ofAnaesthesiology and Intensive Care, University Hospital Pilsen from January 1, 1995 to December 31, 1997. The patients were divided into threegroups according to the type of diagnosis. First group of patients consisted of polytrauma patients without head injury (P, n = 165), the second groupconsisted of polytrauma patients with head injury (P+H, n = 109), the third group consisted of patients with isolated head-brain injury (H, n = 147).In all patients we recorded age, length of hospitalization, duration of mechanical ventilation, score points, survival prediction and actual survival.Calibration of each scoring system on evaluated population was evaluated with the use of ROC (Receiver Operating Characteristic) curves. Theevaluation of prediction accuracy of scoring systems used according to the area below ROC curve shows that APACHE II is unreliable in prognosisprediction in trauma patients, ISS has good discrimination capability in polytrauma patients without head injury, while TRISS is feasible for prognosisprediction in all evaluated diagnostic groups.

Key words:
scoring system – prediction