Mental hospitalisation and its impact on mortality

This paper investigates how severe mental problems (mental hospitalisation) affects mortality for males. We account for the selection into mental hospitalisation (both admittance and discharge) by using a correlated multistate model for the mental hospitalisation process and mortality. The empirical analyses, based on data for men from the Swedish Military Conscription Data linked to administrative Swedish registers, reveal that ignoring the endogeneity of the mental hospitalisation process would substantially overestimate the effect of being in a mental hospital on the mortality hazard and (slightly) underestimate the effect of hospital experience on mortality. We also investigate how the effect of mental hospitalisation changes by birth cohort and by education level. The empirical analyses show that the effect of mental hospitalisation increases over time and that high educated men are affected the most by mental hospitalisation. To investigate whether mental hospitalisation affects different causes of death differently we also estimate a timing-of-events model with cause-specific mortality rates, distinguishing five different causes of death. The model is an extension of the timing-of-events model. The empirical results reveal that not only suicide and death due to external causes are affected by mental hospitalisation but also death due to cardiovascular diseases, cancer, and death due to other natural causes.