Latent classes in diagnoses among psychiatric inpatients predicting mortality and imprisonment - a nationwide cohort study

Purpose: Identify risk factors of death or imprisonment within classes defined by demographic factors and diagnoses within one year of first psychiatric admission. Methods: Nationwide data was obtained from hospital registers from psychiatric hospitals in Iceland 1983-2007. Mortality and cause of death as well as information about imprisonments during the study period, and discharge diagnoses for the first year after initial admission were obtained for each individual. Individuals aged 18 during the study period with at least one year of follow-up were included. Latent Class Analysis was used to identify groups with distinguishable risk of either being alive, dead or having been imprisoned at the end of follow-up. Results: Among psychiatric patients, 4677 were included, average age was 27 years (range 18-43). Four latent classes were identified with different risks of adverse outcomes. Class B (16%), predominantly males with substance use disorder (SUD) diagnoses, had highly increased risk of imprisonment and death accounting for 85 and 34% of these outcomes, respectively. Class A (12%), all with alcohol use disorder, had similar mortality rate as the general population and no imprisonments. Class C (23%) were younger at first admission with some SUD and increased risk of mortality. Class D (46%) had increased mortality rate, SUDs were rare but depression common. Conclusions: Risk of mortality and criminal trends among psychiatric inpatients can be described as distinct clusters of risk factors present at first admission to a psychiatric hospital. Treatment and interventions to reduce mortality and criminality should take these risk differences into account. (C) 2016 Elsevier Masson SAS. All rights reserved.