Main findings

We found that people with a psychiatric hospital admission record for depression in Scotland had a more than three-fold increased risk of all-cause mortality. Severe depression was associated with three-fold and ten-fold increased risks of deaths from natural and unnatural causes of death, respectively, compared with the general population. Although the relative excess death was greatest for unnatural causes of death, deaths from circulatory disease, neoplasms and respiratory disease were the most common causes of death. Among these leading causes of death, the excess relative mortality was highest for circulatory and respiratory diseases. Among circulatory deaths, the excess relative mortality was largest for cerebrovascular-specific deaths. Although patterns were similar in males and females for all-cause mortality and, in general, for cause-specific mortality, the excess relative mortality from suicide and self-harm was markedly higher in females than in males.

Comparison with findings from other studies

Previous studies have provided limited insight into cause-specific mortality among people with depression specifically. The majority of previous studies have combined several types of mental illness,^{Reference Lawrence, Hancock and Kisely5,Reference John, McGregor, Jones, Lee, Walters and Owen7,Reference Korhonen, Moustgaard, Tarkiainen, Östergren, Costa and Urhoj10} making it challenging to understand the specific association between depression and mortality. Our findings provide understanding of the association between severe depression and cause-specific mortality by using a more comprehensive list of causes of death, including subtypes of circulatory disease. A previous study conducted in Scotland between 1986 and 2009 reported an SMR of 2.01 (95% CI 1.96–2.05) for all-cause mortality among individuals with depression identified from psychiatric hospital admission records. This was lower than our comparable estimate for 2000 to 2019 of 3.09 (95% CI 3.02–3.16) from the sensitivity analysis after excluding people with other SMI, suggesting that the relative effects of severe depression on mortality may have increased over time.^{Reference Ajetunmobi, Taylor, Stockton and Wood9} The expected number of deaths in the previous Scottish study was roughly 20% higher than our estimate, which is unsurprising given the improvement in general population mortality rates over time.¹⁷

Two other studies have previously reported on all-cause mortality and a few specific causes of death among people with depression identified from primary and secondary care records.^{Reference Berardi, Stivanello, Chierzi, Musti, Perlangeli and Pandolfi6,Reference Pan, Yeh, Chan and Chang8} Two studies conducted in Italy^{Reference Berardi, Stivanello, Chierzi, Musti, Perlangeli and Pandolfi6} and Taiwan,^{Reference Pan, Yeh, Chan and Chang8} with 10 year and 3 year follow-up periods, respectively, reported lower SMRs for all-cause mortality in people with depression than that found in our study (SMR 1.89 [95% CI 1.84–1.94] and 1.83 [95% CI 1.81–1.86], respectively). Direct comparison of relative mortality to these studies was not appropriate owing to differences in reference populations. Nonetheless, these studies consistently reported that composite SMI^{Reference John, McGregor, Jones, Lee, Walters and Owen7} or depression^{Reference Berardi, Stivanello, Chierzi, Musti, Perlangeli and Pandolfi6,Reference Pan, Yeh, Chan and Chang8} was associated with a higher risk of mortality relative to the general population, with CVD and suicide being the most common natural and unnatural causes, respectively. In the UK, a Welsh study that included 29 797 individuals with SMI, including schizophrenia, bipolar disorder, or other mood-related disorders and other non-organic psychotic disorders identified from primary and/or secondary care records, reported broadly similar findings to those of our study.^{Reference John, McGregor, Jones, Lee, Walters and Owen7} However, direct comparisons are difficult as the Welsh study included depression within a composite mood disorder category. During the study period of 2004 and 2013, 1565 deaths occurred in this group, with a lower all-cause SMR (2.2, 95% CI 2.1–2.3) than that observed in our study for depression only. This is likely to be due to the ascertainment of mood disorder from both primary and secondary care records. The authors reported a higher all-cause SMR for all SMI conditions combined when restricting analyses to participants with hospital admission records as the only source of SMI diagnosis; this aligned with the SMR for depression and all-cause mortality in our study. As in our study, SMRs for all-cause mortality were slightly higher in males than females. Comparisons of cause-specific mortality are also difficult owing to the Welsh study reporting on cause-specific mortality for all SMI conditions combined.^{Reference John, McGregor, Jones, Lee, Walters and Owen7} However, the overall patterns of results were broadly similar, including findings by sex. SMRs for natural-cause-specific deaths were generally higher in males than in females (albeit with confidence intervals often overlapping). There were, however, differences in the magnitude of the sex-specific SMRs across various specific causes of death, including respiratory disease, cerebrovascular disease and, for women only, mental and behavioural disorders. The SMRs for these causes of death in our study were notably higher (with confidence intervals often not overlapping) than those in the Welsh study. For example, although SMRs for IHD were similar across the two studies, the overall SMR for cerebrovascular disease was greater in our study (SMR 3.0, 95% CI 2.77–3.22) than in the Welsh study (SMR 2.3, 95% CI 2.1–2.6). Finally, the higher relative risk of death from suicide, self-harm and undetermined intent among females compared with males aligned with findings from the Welsh study,^{Reference John, McGregor, Jones, Lee, Walters and Owen7} and from the Taiwanese study, which reported similar findings for depression in relation to suicide.^{Reference Pan, Yeh, Chan and Chang8} In Scotland, this sex difference is largely explained by the higher absolute suicide mortality rates in males compared with females in the general Scottish population (22.8 deaths from suicide per 100 000 males v. 7.5 for females in 2019).¹⁸

Strengths and limitations

Our study had several strengths. To our knowledge, this is the first study to describe all-cause and cause-specific mortality by sex in people with severe depression using a reasonably comprehensive list of causes of death. We were able to use large-scale national linked data over a recent 20 year period, allowing for a longer follow-up time from admission to mortality than the majority of previous studies. Moreover, we included people with a diagnosis of depression from psychiatric hospital admission records regardless of diagnosis with other mental disorders, with a sensitivity analysis among people with depression in the absence of other SMI diagnoses. Our study expands on a previous Scottish study^{Reference Ajetunmobi, Taylor, Stockton and Wood9} to provide additional insight into the mortality experience of people with depression specifically, by examining cause-specific death in far more detail and using contemporaneous data.

A key limitation of our study was that it was based on people who were admitted to hospital and so were likely to have more severe depression than people whose depression was diagnosed in primary care or out-patient clinics. Although this does not limit the internal validity of our study, our findings are not likely to be generalisable to the wider population of people with a diagnosis of depression. A small positive dose–response relationship has previously been described between severity of depression and premature mortality, with increased risk of mortality still observed for less severe depression.^{Reference White, Zaninotto, Walters, Kivimäki, Demakakos and Shankar19} Second, although the SMR is a useful measure of mortality on a population level, it is a relative measure and therefore cannot be used to make direct comparisons with other settings or populations. Finally, it was outside the scope of our study to examine mortality trends over time or to investigate the potential effect of multiple hospital admissions as a further proxy for severity of depression.

Interpretation and implications

The mechanisms underlying the association between depression and mortality are not fully understood but are complex and multifactorial. Depression has been linked to increased risk of various chronic physical conditions, including CVD, respiratory diseases and diabetes, all of which can increase the risk of mortality.^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers20–Reference Atlantis, Fahey, Cochrane and Smith22} In our study, among people with severe depression, the most common cause of death was circulatory disease, followed by neoplasms and respiratory diseases, similar to the pattern observed in the general Scottish population.¹⁷ Depression is associated with a range of behavioural and lifestyle factors that can affect physical health, such as unhealthy diet, lack of exercise and substance misuse.^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers20} In addition, people with depression may be less likely to receive optimal medical care or adhere to treatment for physical health conditions.^{Reference Firth, Siddiqi, Koyanagi, Siskind, Rosenbaum and Galletly3} In Scotland, as in other settings, severe depression is associated with a 22% reduction in receipt of coronary revascularisation following a myocardial infarction.^{Reference Fleetwood, Wild, Smith, Mercer, Licence and Sudlow23}

Despite the introduction of suicide prevention strategies in Scotland over the past 20 years,²⁴ our findings suggest that the risk of death from suicide among people with severe depression remains markedly higher than that of the general population. Several strategies have been proposed to address the increased mortality risk associated with depression and other SMI.^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers20,Reference Riblet, Shiner, Young-Xu and Watts25–Reference Arango, Díaz-Caneja, McGorry, Rapoport, Sommer and Vorstman27} Many of these strategies aim to prevent the onset of mental disorders, reduce case fatality and optimise prognosis, particularly among those with coexisting physical diseases. In general, interventions in the proposed strategies included improving access to evidence-based treatments, adopting collaborative care models, addressing social determinants of health and advocating for policy changes to reduce disparities in health outcomes between people with and without mental health disorders.^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers20,Reference Riblet, Shiner, Young-Xu and Watts25–Reference Arango, Díaz-Caneja, McGorry, Rapoport, Sommer and Vorstman27}

In some countries, these strategies have been helpful in improving general health outcomes in people with mental health disorders, including improved cardiovascular risk profiles and lower suicide rates.^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers20,Reference Mann, Michel and Auerbach26,Reference Arango, Díaz-Caneja, McGorry, Rapoport, Sommer and Vorstman27} Despite widespread recognition of the excess risks of physical diseases and premature mortality observed in individuals with mental illness over recent decades, it is disappointing to report evidence of a possible widening gap in mortality between people with severe depression compared with the general population of Scotland. Further investigation is needed to establish whether the mortality gap has indeed widened in this setting and for which specific causes of death.

Summary and future directions

In Scotland, compared with the general population, people with severe depression have a substantially higher risk of all-cause mortality, with relative mortality varying by specific cause of death. Relative mortality is similar by sex, except for deaths by suicide, self-harm and injuries of undetermined intent. Further research is needed to identify the contributions of socioeconomic status, age, physical disease and other factors, including time trends, to the higher risk of all-cause and cause-specific mortality among people with severe depression. In addition, further insight from qualitative research would help to identify barriers and enablers to delivering optimal physical healthcare to this vulnerable group. Future healthcare models should ensure equitable and optimal provision of care for the physical health of people with severe depression and include strategies to reduce the burden of preventable death in this high-risk population group.