Traditional autopsy versus computed tomography imaging autopsy in trauma: A case of "synergistic disagreement"

Decreases in the rates of traditional autopsy (TA) negatively impact traumatology, especially in the areas of quality improvement and medical education. To help enhance the understanding of trauma-related mortality, a number of initiatives in imaging autopsy (IA) were conceived, including the postmortem computed tomography ("CATopsy") project at our institution. Though IA is a promising concept, few studies directly correlate TA and IA findings quantitatively. Here, we set out to increase our understanding of the similarities and differences between key findings on TA and IA in a prospective fashion with blinding of pathologist and radiologist evaluations.

A prospective study of TA versus IA was conducted at an Academic Level I Trauma Center (June 2001-May 2010). All decedents underwent a postmortem, whole-body, noncontrast computed tomography that was interpreted by an independent, blinded, board-certified radiologist. A blinded, board-certified pathologist then performed a TA. Autopsy results were grouped into predefined categories of pathologic findings. Categorized findings from TA and IA were compared by determining the degree of agreement (kappa). The χ(2) test was used to detect quantitative differences in "potentially fatal" findings (eg, aortic trauma, splenic injury, intracranial bleeding, etc) between TA and IA.

Twenty-five trauma victims (19 blunt; 9 female; median age 33 years) had a total of 435 unique findings on either IA or TA grouped into 34 categories. The agreement between IA and TA was worse than what chance would predict (kappa = -0.58). The greatest agreement was seen in injuries involving axial skeleton and intracranial/cranio-facial trauma. Most discrepancies were seen in soft tissue, ectopic air, and "incidental" categories. Findings determined to be "potentially fatal" were seen on both TA/IA in 48/435 (11%) instances with 79 (18%) on TA only and 53 (12%) on IA only. TA identified more "potentially fatal" solid organ and heart/great vessel injuries, while IA revealed more spine injuries, "potentially fatal" procedure-related findings, and the presence of ectopic air/fluid.

This limited study does not support substitution of noncontrast, computed tomography-based IA for TA. Our quantitative analyses suggest that TA and IA evaluations may be complementary and synergistic when performed concurrently. There are potential benefits to using IA in trauma process/quality improvement and in educational settings. Further research should focus on the value (and limitations) of the information provided by IA in the absence of TA.