Association between a single-pass whole-body computed tomography policy and survival after blunt major trauma: a retrospective cohort study

Abstract

Abstract Introduction Single-pass, whole-body computed tomography (pan-scan) remains a controversial intervention in the early assessment of patients with major trauma. We hypothesized that a liberal pan-scan policy is mainly an indicator of enhanced process quality of emergency care that may lead to improved survival regardless of the actual use of the method. Methods This retrospective cohort study included consecutive patients with blunt trauma referred to a trauma center prior to (2000 to 2002) and after (2002 to 2007) the introduction of a liberal single-pass pan-scan policy. The overall mortality between the two periods was compared and stratified according to the availability and actual use of the pan-scan. Logistic regression analysis was employed to adjust mortality estimates for demographic and injury-related independent variables. Results The study comprised 313 patients during the pre-pan-scan period, 223 patients after the introduction of the pan-scan policy but not undergoing a pan-scan and 608 patients undergoing a pan-scan. The overall mortality was 23.3, 14.8 and 7.9% (P < 0.001), respectively. By univariable logistic regression analysis, both the availability (odds ratio (OR) 0.57, 95% confidence interval (CI): 0.36 to 0.90) and the actual use of the pan-scan (OR 0.28, 95% CI: 0.19 to 0.42) were associated with a lower mortality. The final model contained the Injury Severity Score, the Glasgow Coma Scale, age, emergency department time and the use of the pan-scan. 2.7% of the explained variance in mortality was attributable to the use of the pan-scan. This contribution increased to 7.1% in the highest injury severity quartile. Conclusions In this study, a liberal pan-scan policy was associated with lower trauma mortality. The causal role of the pan-scan itself must be interpreted in the context of improved structural and process quality, is apparently moderate and needs further investigation with regard to the diagnostic yield and changes in management decisions. (The Pan-Scan for Trauma Resuscitation [PATRES] Study Group, ISRCTN35424832 and ISRCTN41462125)