Structuring medico-legal autopsy data for public health surveillance using the 11th Revision of the International Classification of Diseases: a retrospective cohort study.
Source: PubMed, NCBI / U.S. National Library of Medicine
Medico-legal autopsy case files contain prevention-relevant information on causes and circumstances of death but remain underused for surveillance because key detail is recorded as narrative text and terminology varies across jurisdictions. We evaluated whether the 11th Revision of the International Classification of Diseases can generate structured, surveillance-ready outputs from these records, while signalling where coding cannot preserve medico-legal meaning. In this retrospective study at the Institute of Forensic and Traffic Medicine, Heidelberg University Hospital, Germany, we screened consecutive medico-legal autopsy case files from 1 January 2022 to 31 December 2024 and analysed all eligible reports. We generated three output layers: causes of death, injury phenotypes, and external-cause dimensions for non-natural deaths. An expert-validated quality-flag framework identified representational limitations affecting surveillance interpretation. Of 935 screened files, 921 were analysed. Overall, 196 of 921 autopsies (21.3%) had one or more quality flags. Flags affected 119 of 921 cases (12.9%) in cause-of-death coding, 60 of 921 (6.5%) in injury phenotyping, and 35 of 423 non-natural deaths (8.3%) in external-cause dimensions. Flags were most prevalent among suicides (23 of 58, 39.7%); injury-related flags were most frequent in neck injuries (32 of 137, 23.4%). Among non-natural deaths, place and mechanism were codable in over 92% of cases. The 11th Revision of the Inter
Abstract
Medico-legal autopsy case files contain prevention-relevant information on causes and circumstances of death but remain underused for surveillance because key detail is recorded as narrative text and terminology varies across jurisdictions. We evaluated whether the 11th Revision of the International Classification of Diseases can generate structured, surveillance-ready outputs from these records, while signalling where coding cannot preserve medico-legal meaning. In this retrospective study at the Institute of Forensic and Traffic Medicine, Heidelberg University Hospital, Germany, we screened consecutive medico-legal autopsy case files from 1 January 2022 to 31 December 2024 and analysed all eligible reports. We generated three output layers: causes of death, injury phenotypes, and external-cause dimensions for non-natural deaths. An expert-validated quality-flag framework identified representational limitations affecting surveillance interpretation. Of 935 screened files, 921 were analysed. Overall, 196 of 921 autopsies (21.3%) had one or more quality flags. Flags affected 119 of 921 cases (12.9%) in cause-of-death coding, 60 of 921 (6.5%) in injury phenotyping, and 35 of 423 non-natural deaths (8.3%) in external-cause dimensions. Flags were most prevalent among suicides (23 of 58, 39.7%); injury-related flags were most frequent in neck injuries (32 of 137, 23.4%). Among non-natural deaths, place and mechanism were codable in over 92% of cases. The 11th Revision of the International Classification of Diseases can support structured cause-of-death, injury, and external-cause outputs from medico-legal autopsy records for public health surveillance. However, routine use requires explicit signalling of residual uncertainty and classification limitations to preserve interpretability.