An increased mortality risk associated with mental disorder has been reported for patients, but there are few studies are based on random samples with interview-based psychiatric diagnoses. Part of the increased mortality for those with mental disorder may be attributable to worse somatic health or hazardous health behaviour – consequences of the disorder – but somatic health information is commonly lacking in psychiatric samples. This study aims to examine long-term mortality risk for psychiatric diagnoses in a general population sample and to assess mediation by somatic ill health and hazardous health behaviour.

We used a double-phase stratified random sample of individuals aged 18–65 in Stockholm County 1970–1971 linked to vital records. First phase sample was 32 186 individuals screened with postal questionnaire and second phase was 1896 individuals (920 men and 976 women) that participated in a full-day examination (participation rate 88%). Baseline examination included both a semi-structured interview with a psychiatrist, with mental disorders set according to the 8th version of the International Classification of Disease (ICD-8), and clinical somatic examination, including measures of body composition (BMI), hypertension, fasting blood glucose, pulmonary function and self-reported tobacco smoking. Information on vital status was obtained from the Total Population Register for the years 1970–2011. Associations with mortality were studied with Cox proportional hazard analyses.

A total of 883 deaths occurred among the participants during the 41-year follow-up. Increased mortality rates were found for ICD-8 functional psychoses (hazard ratio, HR = 2.22, 95% confidence interval (95% CI): 1.15–4.30); psycho-organic symptoms (HR = 1.94, 95% CI: 1.31–2.87); depressive neuroses (HR = 1.71, 95% CI: 1.23–2.39); alcohol use disorder (HR = 1.91, 95% CI: 1.40–2.61); drug dependence (HR = 3.71, 95% CI: 1.80–7.65) and psychopathy (HR = 2.88, 95% CI: 1.02–8.16). Non-participants (n = 349) had mortality rates similar to participants (HR = 0.98, 95% CI: 0.81–1.18). In subgroup analyses of those with psychoses, depression or alcohol use disorder, adjusting for the potential mediators smoking and pulmonary function, showed only slight changes in the HRs.

This study confirms the increased risk of mortality for several psychiatric diagnoses in follow-up studies on American, Finnish and Swedish population-based samples. Only a small part of the increased mortality hazard was attributable to differences in somatic health or hazardous health behaviour measured at baseline.