BackgroundThere has been increasing use of invasive techniques, such as extracorporeal organ support, in intensive care units (ICU), and declining autopsy rates. Thus, new measures are needed to maintain high-quality standards. We investigated the potential of computed tomography (CT)-based virtual autopsy to substitute for medical autopsy in this setting. MethodsWe investigated the potential of virtual autopsy by post-mortem CT to identify complications associated with medical devices in a prospective study of patients who had died in the ICU. Clinical records were reviewed to determine the number and types of medical devices used, and findings from medical and virtual autopsies, related and unrelated to the medical devices, were compared. ResultsMedical and virtual autopsies could be performed in 61 patients (Group M/V), and virtual autopsy only in 101 patients (Group V). In Group M/V, 41 device-related complications and 30 device malpositions were identified, but only with a low inter-method agreement. Major findings unrelated to a device were identified in about 25% of patients with a high level of agreement between methods. In Group V, 8 device complications and 36 device malpositions were identified. ConclusionsDevice-related complications are frequent in ICU patients. Virtual and medical autopsies showed clear differences in the detection of complications and device malpositions. Both methods should supplement each other rather than one alone for quality control of medical devices in the ICU. Further studies should focus on the identification of special patient populations in which virtual autopsy might be of particular benefit. Clinical trial registrationNCT01541982.