Background and aims Rates of traditional medical autopsy are low in Japan. In particular, obtaining consent for autopsy of children from parents is difficult. Although postmortem imaging of adults has been well studied, this is not the case in children. Few studies have investigated the accuracy of postmortem imaging in diagnosing the causes of neonatal deaths. We aimed to identify the accuracy of postmortem computed tomography (CT), and compared postmortem CT findings to clinical diagnosis and autopsy in the neonatal intensive care unit (NICU).

Methods Twenty-five patients died in our NICU from 2010 to 2012. Consent for autopsy was obtained for 10 cases (40%) and consent for postmortem imaging was obtained for 19 (76%). Both postmortem imaging and autopsy were able to be performed for 10 cases (40%).

Results The concordance rate between cause of death from postmortem CT and that from clinical diagnosis was 74% (14/19), while the rate between cause of death from postmortem CT and that from autopsy was 70% (7/10). Moreover, postmortem CT uncovered unrecognised diagnoses such as pericardial emphysema that remained undetected from the clinical course or autopsy.

Conclusions Postmortem imaging is reliable and valid in NICU settings, and the combination of postmortem imaging and autopsy could improve the accuracy of determining causes of neonatal deaths.