Factors associated with medico-legal autopsy of decedents with psychiatric disorders

Introduction: Autopsy rates are declining worldwide, resulting in increasing selectivity in referral for forensic autopsy and increased uncertainty about the validity of assigned causes of death. Persons with psychiatric disorders have high rates of premature death but not all are referred for forensic autopsies. Knowledge is needed on which decedents with psychiatric disorders are chosen for forensic autopsy to determine whether causes of death are at risk of being misclassified among certain subgroups of decedents. Methods: We conducted a nationwide register-based case-control study including all decedents with psychiatric disorders in Denmark in the period 1998-2015. Using multivariate logistic regression, we examined associations between demographic and socioeconomic factors, comorbidities, healthcare utilization, and referral for forensic autopsy, overall and stratified by age at death (<45, 45-64, >= 65 years). Results: Of the 152,799 decedents in the study population, 7043 (4.61 %) had a forensic autopsy. Decedents referred for forensic autopsy were more likely to be young, have a history of substance use, and have schizophrenia or an affective disorder (factors listed in diminishing order of strength of association). Increasing severity of comorbidities as measured by the Charlson comorbidity index was associated with decreasing likelihood of being autopsied. Patterns of association with sex, alcohol use, habitation and education did not vary by age at death. Schizophrenia and drug use were most strongly associated with forensic autopsy in decedents < 45 years of age, whereas death early in the study period was more strongly associated with autopsy in the oldest age groups. Discussion: The decision to refer a decedent for forensic autopsy was predominantly based on the decedent's age, history of drug use, and the absence of non-psychiatric comorbidities. Causes of death in decedents with comorbidities or recent contact with the healthcare system and decedents > 65 years may be more likely to be inaccurate, particularly in drug users.