Improved Predictions from A Severity Characterization of Trauma (ASCOT) over Trauma and Injury Severity Score (TRISS): Results of an Independent Evaluation

摘要: Objective : In 1986, data from 25,000 major trauma outcome study patients were used to relate Trauma and Injury Severity Score (TRISS) values survival probability. The resulting norms have been widely used. Motivated by TRISS limitations, A Characterization of (ASCOT) was introduced in 1990. objective this evaluate compare ASCOT probability predictions using carefully collected independently reviewed not the development those norms. Design This a prospective collection for consecutive admissions four level I centers participating study. Materials Methods Data 14,296 admitted sites between October 1987 through 1989 indices evaluated measures discrimination (disparity, sensitivity, specificity, misclassification rate, area under receiver-operating characteristic curve) calibration [Hosmer-Lemeshow goodness-of-fit statistic (H-L)]. Measurements Main Results For blunt-injured adults, has higher sensitivity than (69.3 vs. 64.3) meets criterion model (H-L < 15.5) needed accurate z W scores. does meet = 30.7). adults with penetrating injury, substantially lower H-L value (20.3 138.4), but neither criterion. Areas ROC curves are significantly different exceed 0.91 0.95 injury. pediatric patients, sensitivities (near 77%) areas (both 0.96) comparable, both models satisfy Conclusions age health care decisions influenced evaluations, ASCOT's more precise description anatomic injury its improved actual outcomes argue adoption as standard method prediction.