The methodological and conceptual aspects of mortality studies in psychiatry.

An extensive literature documented a mortality differential for natural causes between psychiatric patients and the general population. Less clear is the pattern for cancer diseases. Methodological problems arise when trying to explain such mortality gap: selection bias and reverse causation; time-dependent confounders that are also intermediate variables; complex relationships within a life course have to be considered. We try to explain such problems in terms of causal graphs. Excess risk for causes of death which are not attributable to higher prevalence of risk factors or treatment side-effects and higher mortality rates for avoidable causes have been also documented. These findings underline the need for research on health promotion and preventive programs targeted to psychiatric patients.