Elsevier

The Lancet

Volume 394, Issue 10211, 16–22 November 2019, Pages 1827-1835
The Lancet

Articles
A comprehensive analysis of mortality-related health metrics associated with mental disorders: a nationwide, register-based cohort study

https://doi.org/10.1016/S0140-6736(19)32316-5Get rights and content

Summary

Background

Systematic reviews have consistently shown that individuals with mental disorders have an increased risk of premature mortality. Traditionally, this evidence has been based on relative risks or crude estimates of reduced life expectancy. The aim of this study was to compile a comprehensive analysis of mortality-related health metrics associated with mental disorders, including sex-specific and age-specific mortality rate ratios (MRRs) and life-years lost (LYLs), a measure that takes into account age of onset of the disorder.

Methods

In this population-based cohort study, we included all people younger than 95 years of age who lived in Denmark at some point between Jan 1, 1995, and Dec 31, 2015. Information on mental disorders was obtained from the Danish Psychiatric Central Research Register and the date and cause of death was obtained from the Danish Register of Causes of Death. We classified mental disorders into ten groups and causes of death into 11 groups, which were further categorised into natural causes (deaths from diseases and medical conditions) and external causes (suicide, homicide, and accidents). For each specific mental disorder, we estimated MRRs using Poisson regression models, adjusting for sex, age, and calendar time, and excess LYLs (ie, difference in LYLs between people with a mental disorder and the general population) for all-cause mortality and for each specific cause of death.

Findings

7 369 926 people were included in our analysis. We found that mortality rates were higher for people with a diagnosis of a mental disorder than for the general Danish population (28·70 deaths [95% CI 28·57–28·82] vs 12·95 deaths [12·93–12·98] per 1000 person-years). Additionally, all types of disorders were associated with higher mortality rates, with MRRs ranging from 1·92 (95% CI 1·91–1·94) for mood disorders to 3·91 (3·87–3·94) for substance use disorders. All types of mental disorders were associated with shorter life expectancies, with excess LYLs ranging from 5·42 years (95% CI 5·36–5·48) for organic disorders in females to 14·84 years (14·70–14·99) for substance use disorders in males. When we examined specific causes of death, we found that males with any type of mental disorder lost fewer years due to neoplasm-related deaths compared with the general population, although their cancer mortality rates were higher.

Interpretation

Mental disorders are associated with premature mortality. We provide a comprehensive analysis of mortality by different types of disorders, presenting both MRRs and premature mortality based on LYLs, displayed by age, sex, and cause of death. By providing accurate estimates of premature mortality, we reveal previously underappreciated features related to competing risks and specific causes of death.

Funding

Danish National Research Foundation.

Introduction

The prevalence of mental disorders is high, with estimates indicating that one in three individuals experiences a mental disorder during their lifetime.1, 2, 3 Additionally, there is robust evidence indicating that mental disorders are associated with premature mortality.3 Many studies on mental disorders report risk ratios such as standardised mortality ratios and mortality rate ratios (MRRs), which are useful for comparing how different types of disorders affect premature mortality and to identify specific causes of death.3 Health metrics can also provide information on the age of death or differential life expectancy associated with mental disorders.4 While previous literature has provided a range of health metrics for people with mental disorders,4, 5, 6 there is a need for a harmonised panel of mortality-related health metrics that includes a comprehensive range of mental disorders.

Life expectancy is a widely used health metric, which is readily understood by experts and non-experts alike, and comparing the difference in average lifespan between people with mental disorders and the general population is an intuitive way to measure premature mortality. The best known metric for premature mortality is years of life lost (YLLs), which is used by the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD).7 When considering deaths attributable to specific causes, YLLs estimate the expected remaining life expectancy from age at death on the basis of an ideal standard life expectancy derived from the lowest age-specific mortality rates in any population in the world greater than 5 million. Other studies have used methods based on remaining life expectancy by applying assumptions related to age of onset of the disorder or age of the cohort for follow-up. For example, an Australian study found the life expectancy gap for all mental disorders to be 15·9 years for men and 12·0 years for women.8 A Danish study found similarly large estimates of reduced life expectancy for people with schizophrenia (18·7 years for men and 16·3 years for women).9 However, these studies assumed that all cases of mental disorders had onset at age 15 years.

Research in context

Evidence before this study

Systematic reviews have consistently shown that people with mental disorders have an increased risk of premature mortality. Traditionally, this evidence has been based on relative risks (eg, mortality rate ratios [MRRs]) or crude estimates of life expectancy that do not incorporate variation in the age of onset of the disorder. Recently, a new method has been introduced to estimate life-years lost (LYLs), a measure that incorporates the precise age of onset of the disorder. A literature review indicates that this method has not been applied to a comprehensive range of mental disorders, nor contextualised with traditional mortality-related measures such as MRRs.

Added value of this study

In this register-based cohort study, we use a variety of mortality-related metrics, presented by age, sex, and cause of death, to investigate mortality associated with different types of mental disorders. Besides MRRs, we report premature mortality based on LYLs. We found that people with mental disorders had higher mortality rates than those without, with the largest difference observed at age 33 years (MRR 7·25, 95% CI 7·06–7·44). When looking at specific types of disorders, being diagnosed with all types was associated with higher mortality rates and shorter life expectancies, both in males and females. The largest reduction in remaining life expectancy after diagnosis was observed for substance use disorders. When examining specific causes of death, we found that men with all types of mental disorder had fewer LYLs associated with neoplasm-related deaths compared with the general population, which indicates that the general population loses more years of life due to neoplasms than do those with mental disorders. This finding relates to the ability of the LYL to accommodate competing causes of death. While those with mental disorders had higher rates of dying from neoplasms, they had even higher rates of mortality from other causes, thus reducing their risk of dying from neoplasms. Finally, through an interactive data visualisation website, we allow fine-grained interrogation of our results.

Implications of all the available evidence

Mental disorders are associated with both increased mortality rates and reduced life expectancy when assessed with LYLs. Our new methods provide more accurate estimates of premature mortality and reveal previously underappreciated features related to competing risks and specific causes of death. Our findings reinforce the need to optimise the coordinated care of general medical conditions in those with mental disorders.

Recent advances in health metrics can now take into account the observed age of onset (or age of first administrative onset) of the disorder of interest.4, 10 The life-years lost (LYLs) method focuses on the observed age at diagnosis, differing in this way from GBD's YLLs, which focus on age at death. By using the age at diagnosis, the LYL method allows quantification of the reduction in life expectancy for any type of disease, and not only those defined as potential causes of death, as in the GBD estimations. With LYLs, it has been estimated that the average life expectancy for people with any mental disorder is 10·2 years shorter for men and 7·3 years shorter for women.4 These findings, and a recent study applying these methods to schizophrenia,11 indicate that past estimates might have overestimated the actual contribution of mental disorders to premature mortality. Another advantage of using the LYL metric is that it allows investigation of the association between particular disorders and premature mortality, regardless of the main underlying cause of death. It also allows the decomposition of the total reduction in life expectancy into specific causes of death. Given that differential life expectancy in people with mental disorders is used as a proxy indicator of the effectiveness of health care, which can influence resource allocation,12 it is essential that valid metrics related to differential life expectancy be available.

The aim of this study was to investigate mortality associated with different types of mental disorders. We used high-quality Danish registers to provide an integrated atlas of different estimates related to premature mortality, which can be interrogated by age, sex, type of mental disorder, and specific cause of death.

Section snippets

Study population and assessment of mental disorders

We designed a population-based cohort study including all people younger than 95 years living in Denmark at some point between Jan 1, 1995, and Dec 31, 2015. Since 1968, the Danish Civil Registration System13 has maintained information on all residents, including sex, date of birth, continuously updated information on vital status, and a unique personal identification number that can be used to link information from various national registries. Information on mental disorders was obtained from

Results

7 369 926 people (3 671 564 males and 3 698 362 females) were followed up for 111 027 388·7 person-years (table). Overall, 762 419 (10·3%) people were diagnosed with at least one mental disorder (347 972 [9·5%] males and 414 447 [11·2%] females). During the 21-year follow-up period, 1 122 351 (15·2%) people died (564 301 [15·4%] males and 558 050 [15·1%] females). We have included all data on an interactive data visualisation website to facilitate fine-grained interrogation of the data and

Discussion

To our knowledge, this study provides the most comprehensive analysis of mortality-related metrics in people with mental disorders, providing estimates for each group of mental disorders separately. The highest MRR was found for substance use disorders; however, all types of mental disorders were associated with higher mortality rates. The point estimates varied by sex, with higher estimates for males than females for all disorders except for developmental disorders and intellectual

References (30)

  • D Lawrence et al.

    The gap in life expectancy from preventable physical illness in psychiatric patients in Western Australia: retrospective analysis of population based registers

    BMJ

    (2013)
  • PK Andersen

    Life years lost among patients with a given disease

    Stat Med

    (2017)
  • TM Laursen et al.

    Cause-specific life years lost among persons diagnosed with schizophrenia: is it getting better or worse?

    Schizophr Res

    (2018)
  • NH Liu et al.

    Excess mortality in persons with severe mental disorders: a multilevel intervention framework and priorities for clinical practice, policy and research agendas

    World Psychiatry

    (2017)
  • CB Pedersen

    The Danish Civil Registration System

    Scand J Public Health

    (2011)
  • Cited by (316)

    View all citing articles on Scopus

    Contributed equally

    View full text