Mortality in persons with recent primary or secondary care contacts for mental disorders in Finland

Excess mortality among persons with mental disorders has been consistently documented1, 2, but the mortality risk over a full spectrum of mental disorders treated both in primary and secondary care remains to be explored at a nationwide level.

Integration of mental health care in primary care services is considered a priority in low‐, middle‐, and high‐income countries ³ , and depression and anxiety are among the top ten most common reasons for visits in primary care ⁴ . The global shortage of mortality data concerning mental disorders in primary care may lead to an overestimation of the population‐wide burden of the full spectrum of treated mental disorders5, 6.

Excess mortality is related to a variety of risk factors at the individual, health system and social levels ⁷ . Mental disorders are associated with socioeconomic factors and an increased vulnerability to several physical conditions, with complex bi‐directional pathways ⁸ . Physical comorbidities contribute to the majority of life‐years lost in people with mental disorders, and low socioeconomic position (SEP) associates with mental disorders and physical conditions, as well as with mortality in the general population ⁹ .

This national register‐based open cohort study aimed to: a) assess the excess mortality in persons with mental disorders seen in both primary and secondary care, and compare these estimates with secondary care data only; b) determine the extent to which adjusting for physical comorbidities and individual‐level socioeconomic factors affects the estimates.

We used individual‐level register data concerning all citizens with Finnish background aged at least 20 years and living in Finland at some point between January 1, 2011 and December 31, 2017. We identified all deaths (using the Finnish Causes of Death Register), the dynamic population at risk of death (through Population Registers), and all mental health contacts (using Care Register for Health Care, in which primary care has been included since 2011) during that period. The ethical review board of the Finnish Institute for Health and Welfare approved the study protocol. Data were linked with the permission of Statistics Finland (TK‐53‐1696‐16) and the Finnish Institute of Health and Welfare. Informed consent is not required for register‐based studies in Finland.

A history of mental health related contacts was defined as having any contact with secondary care psychiatric inpatient or outpatient services, or with primary care, with a diagnosis of any mental disorder (i.e., ICD‐10 chapter V, or International Classification of Primary Care‐2 chapter P) within the previous year.

We collected data on the following individual‐level variables: sex, urbanicity of residence area, region of residence, living alone status, level of educational attainment, economic activity, and equivalized household net income deciles. Income measurement with a three‐year lag was used to account for potential reverse causation. Physical comorbidity was assessed using the Charlson Comorbidity Index (CCI), categorized by previously used cut‐points: none, 1‐3, and ≥4.

Three sets of data were collected and analyzed separately, concerning: a) individuals seen in primary and secondary care combined, compared with those without such contacts; b) individuals seen in primary and secondary care separately, compared with those without such contacts; c) individuals seen in secondary care only, compared to all individuals without such contacts (including individuals with possible primary care treatments), which is the traditional approach.

Mortality rate ratios (MRRs) were estimated using a Poisson regression model. Men and women were analyzed separately. To investigate the association between physical comorbidities and mortality, a stratified analysis for the CCI categories was performed. In addition, the ICD‐10 diagnostic blocks were analyzed separately. We performed sensitivity analyses using three‐ and five‐year histories of mental health related contacts. R and Stata were used for the analyses.

During the period between 2011 and 2017, we observed 4,417,635 individuals (51.3% women), contributing 28,049,912 person‐years. Along that period, 860,287 (19.5%) of all observed individuals had mental health related contacts, more commonly in primary care. Mood disorders was the most commonly used ICD‐10 diagnostic block. Altogether, 357,119 persons died (50.3% women), of whom 44,364 (12.4%) had had some contact with psychiatric secondary or primary care within the previous year.

Age and calendar year adjusted MRRs of 2.83 (95% CI: 2.79‐2.87) and 1.79 (95% CI: 1.76‐1.82) were observed for men and women with a one‐year history of primary or secondary care mental health contacts, compared to those without. After SEP adjustments, MRRs of 2.17 (95% CI: 2.13‐2.20) and 1.71 (95% CI: 1.68‐1.74) were observed. After further adjustments for physical comorbidi­ties, the estimates decreased to 1.63 (95% CI: 1.60‐1.65) and 1.20 (95% CI: 1.18‐1.22), respectively. These SEP and comorbidity adjusted MRR estimates were 27% and 42% lower, respectively, compared to the MRRs of 2.24 (95% CI: 2.19‐2.30) and 2.07 (95% CI: 2.01‐2.12) obtained with the traditional approach considering secondary care only.

In diagnosis‐specific analysis, the highest SEP and comorbidity adjusted MRRs were observed in disorders related to substance use. Excess mortality varied by age and turned to decrease in both men and women starting from the age of 35 years (see supplementary information).

Individuals with recent primary care mental health contacts had more commonly diagnosed physical comorbidities than individuals treated in psychiatric secondary care (24.5% vs. 18.1% of person‐time). The presence of physical comorbidities modified the association between mortality and a one‐year history of mental health contact: excess mortality related to mental disorders was the highest in people without comorbidities, and the lowest in people with multiple comorbidities. Sensitivity analysis with three‐ or five‐year histories of treated mental disorders, ­instead of one year, showed only a little difference (see supplementary information).

These findings confirm the previously reported evidence of an excess mortality in people with mental disorders, but also suggest that the previously published MRR estimates would have been considerably lower if primary care had been included in those analyses. As mental disorders are commonly treated in primary care, the current results are likely to have generalizability, especially in high‐income countries. They provide a more optimistic view of the burden of mental disorders and highlight the diversity of these disorders in the population.