External Examination versus Full Autopsy for Determining Cause and Manner of Death: A Systematic Review of Diagnostic Concordance and Forensic Implications

Abstract In many jurisdictions, death certificates are issued after external postmortem examination (EPME) without internal dissection, despite the full autopsy remaining the diagnostic reference standard. We systematically evaluated how accurately EPME determines the cause of death (COD) and manner of death (MOD) compared with a full forensic autopsy. Following PRISMA 2020 and PRISMA-DTA principles, we searched MEDLINE, Embase, Web of Science, Scopus, and the Cochrane Library (January 2000–October 2025), screened in duplicate, and assessed risk of bias with a QUADAS-2 adaptation. We included paired studies in which the same decedents underwent EPME and full autopsy, plus closely related comparators (a reader study simulating “no-autopsy” decisions in suspected intoxications and a physician-certified verbal-autopsy study). Given heterogeneity in case-mix and reporting, we performed narrative synthesis without meta-analysis. Primary outcomes were COD and MOD agreement; secondary outcomes included the proportion undetermined on EPME but resolved by autopsy and qualitative patterns of misclassification. Nine studies met inclusion criteria, five of which were paired case-level comparisons; across these, ≥1044 decedents had both EPME and autopsy. COD agreement varied widely by setting and case-mix: In sudden deaths <45 years agreement was low, with autopsy revealing occult cardiovascular or toxicological causes not apparent externally; in fatal head trauma EPME identified the lethal mechanism in ~60%, while ~20% remained “unclear/no cause” on external certification; in mixed natural-death cohorts, external COD was wrong in ~25%–30%, often due to default attribution to “cardiac” causes; in suspected drug deaths, simulated external-only decisions were correct in ~73%, underscoring the need for toxicology. Across series, undetermined after EPME was common and frequently resolved by autopsy. MOD agreement tended to be higher than COD but still missed a small yet consequential proportion of nonnatural deaths. Risk-of-bias was generally low for the reference standard (autopsy), but blinding of EPME to autopsy results was rarely explicit. EPME is indispensable for scene assessment and initial documentation but insufficient as a stand-alone determinant of COD/MOD in key scenarios, particularly young sudden deaths, suspected intoxications, low-force trauma with potential intracranial injury, and any case remaining undetermined after EPME. A risk-stratified triage is supported: EPME for low-risk witnessed naturals; EPME + targeted postmortem imaging where appropriate; and full autopsy with ancillary testing for high-risk or uncertain cases. Training, quality management (including structured precremation second examinations), and access to imaging/toxicology should be strengthened to safeguard diagnostic accuracy and public-health integrity.