Autopsy Data in the Peer Review Process Improves Outcomes Analysis

Abstract

The value of autopsy findings has been questioned in peer review at mature trauma centers. We sought to determine the impact of autopsy data on the peer-review process. This was a retrospective study. Data analyzed included mortality type (immediate/dead on arrival [DOA]; early [< or =48 hours]; late [>48 hours]), Injury Severity Scale (ISS) score, Trauma and Injury Severity Score-generated probability of survival (PS), peer-review judgment of preventability, and findings at autopsy. Deaths were assigned to a category; then pre- and postautopsy ISS score, PS, and outcomes of the peer- review process (percent nonpreventable [%NP]) were compared. Paired t tests (alpha = 0.05) were performed to determine whether changes in ISS score and PS were statistically significant. All descriptive and inferential analyses were based on cases with pre- and postautopsy data for the relevant variables. Of the 170 deaths, 126 deaths had an autopsy performed (74.1%) and 112 autopsy reports were available (89.9%). Autopsy data resulted in statistically significant changes in ISS score for each mortality category and in PS for the immediate/DOA and early categories. There were also autopsy- related changes in peer-review outcomes for immediate/DOA and late deaths but not for early deaths. The proportion of overall agreement between pre- and postautopsy outcomes for the immediate/DOA category was 94.3% (50/53); three deaths initially deemed NP were reclassified as potentially preventable (PP) after autopsy. Overall agreement for the late category was 87% (20/23); one PP was reclassified as NP and two NPs were reclassified as PP. Autopsy data enhanced peer review in immediate/DOA and late death after injury but did not impact peer review in early deaths. Autopsy data were most important to the analysis of late deaths. Targeting autopsy performance to these mortality categories is an effective strategy for centers with constrained access to autopsy data.