LAUSR

OBJECTIVE To assess the ability of the acute physiology and chronic health evaluation II (APACHE II) and trauma-injury severity score (TRISS) in predicting mortality in intensive care unit (ICU) trauma patients. METHODS Databases of PubMed, Cochrane Library, SinoMed, CNKI were retrieved from January 1980 to December 2020. The ability of the APACHE II and the TRISS to predict mortality in the ICU trauma patients was compared in the retrieval literatures. The relevant literatures were screened by two researchers independently. The data of the included literatures were extracted, and the quality of the included literatures was evaluated. MetaDiSc 1.4 software was used to test the heterogeneity among studies. Meta-analysis was performed on diagnostic accuracy indicators and the summary receiver operator characteristics curve (SROC curve) was fitted. The area under SROC curve (AUC) of the two scores was compared. Deek test was used to analyze literature publication bias. RESULTS Six studies were selected with 4 054 patients involved with medium and high quality. Meta-analysis results showed that APACHE II and TRISS had low sensitivity [the pooled sensitivity and 95% confidence interval (95%CI) was 0.48 (0.41-0.55) and 0.51 (0.41-0.62)], high specificity [the pooled specificity and 95%CI was 0.96 (0.93-0.97) and 0.98 (0.95-0.99)], the pooled diagnostic odds ratio (DOR) and 95%CI was 20 (14-28) and 46 (18-120), and overall good performance in terms of AUC [the AUC and 95%CI was 0.79 (0.75-0.82) and 0.80 (0.76-0.83)] in predicting the prognosis of ICU trauma patients. There was no statistical difference in AUC between the two scores (Z = 1.542, P > 0.05). Deek funnel plot showed little publication bias. CONCLUSIONS Both APACHE II and TRISS scores could accurately predict mortality in ICU trauma patients.