Summary of findings

Our findings reveal persistent large disparities in life expectancy of those with SMI compared with the general population across the lifespan. In Scotland, between 2000 and 2019, life expectancy in the general population increased. Although life expectancy similarly improved among those with a psychiatric hospital admission record for major depression, the life expectancy gap did not narrow. In contrast, the gap for people with schizophrenia appears to have widened. Among men with schizophrenia, LYL due to natural causes decreased, whereas LYL due to unnatural causes increased, with no changes observed among women. The pattern is less clear for bipolar disorder and, although life expectancy may have improved slightly, there is no evidence that the gap is narrowing.

Comparison with other studies

Our estimates of the life expectancy gap for each of schizophrenia, bipolar disorder and major depression are lower than reported in some previous studies, including UK studies. ^{Reference Hjorthoj, Sturup, McGrath and Nordentoft9,Reference Chan, Tong, Wong, Chen and Chang10,Reference Chang, Chesney, Teng, Hollandt, Pritchard and Shetty12,Reference Ajetunmobi, Taylor, Stockton and Wood15} This may reflect differences in study setting, period and methodology, including our adoption of the new LYL method, which takes account of the variation in age at onset of psychiatric illness, providing more conservative life expectancy estimates. ^{Reference Plana-Ripoll, Canudas-Romo, Weye, Laursen, McGrath and Andersen22} Our estimates generally align with other recent studies from Denmark and Hong Kong that also used the LYL method. ^{Reference Erlangsen, Andersen, Toender, Laursen, Nordentoft and Canudas-Romo18–Reference Yung, Wong, Chan, Chen and Chang21,Reference Plana-Ripoll, Pedersen, Agerbo, Holtz, Erlangsen and Canudas-Romo30}

In contrast to our study, previous contemporaneous studies from Denmark and Hong Kong employing the LYL method found no change in the gap over time for people with schizophrenia compared with the general population. ^{Reference Laursen, Plana-Ripoll, Andersen, McGrath, Toender and Nordentoft19–Reference Yung, Wong, Chan, Chen and Chang21} This could partly reflect differences in ascertainment method, with Danish studies defining schizophrenia based on both out-patient and in-patient hospital records. Where mood disorders were analysed, studies did not disaggregate bipolar disorder and major depression, ^{Reference Plana-Ripoll, Weye, Momen, Christensen, Iburg and Laursen20} but estimates of life expectancy and lack of change in the gap over time aligned with our findings. Interestingly, where examined in previous Danish studies, authors found that reductions in unnatural causes of death were offset by increasing deaths from natural causes. ^{Reference Erlangsen, Andersen, Toender, Laursen, Nordentoft and Canudas-Romo18–Reference Plana-Ripoll, Weye, Momen, Christensen, Iburg and Laursen20} This contrasts with our findings, where we generally found a decrease in LYL from natural causes of death and no change or an increase in LYL from unnatural causes of deaths. This may reflect the underfunding of mental health services in the UK during this time period. ³⁴ The only other contemporary UK-based study found a narrowing of the life expectancy gap from 2008–2012 to 2013–2017 for all SMI disorders combined ^{Reference Chang, Chesney, Teng, Hollandt, Pritchard and Shetty12} ; although confidence intervals of estimates for the two periods overlapped, and this narrowing was not observed across all gender and mental disorder combinations. The study also included a single highly urbanised area, limiting generalisability to the rest of England and the wider UK, and the comparator was UK-wide life expectancy rather than life expectancy of the defined geographical area of the study population. Our findings differ from those of two studies based on earlier decades of data from Australia ^{Reference Lawrence, Hancock and Kisely16} and Nordic countries, ^{Reference Wahlbeck, Westman, Nordentoft, Gissler and Laursen17} but apparent inconsistencies may reflect the differing time periods and settings, health systems and study methodology.

Mortality rate ratios in people with schizophrenia and bipolar disorder relative to that of the general population have been reported to be widening between 2000 and 2014 in the UK. ^{Reference Hayes, Marston, Walters, King and Osborn5} Relative measures are sensitive to decreases in the reference population mortality rates over time, resulting in larger relative differences between groups, but are essentially consistent with the changes in absolute differences in life expectancy over time that we have reported.

Strengths and limitations

A key strength of our study is the use of recent national level data spanning two decades, which enabled assessment of change in life expectancy over a substantial period of time with stratification by sex. Given the use of whole population data during a long study period, we were able to analyse life expectancy by individual SMI, and to distinguish between bipolar disorder and major depression, which have often previously been analysed as a composite group of all mood disorders, potentially masking differences between disorders. Finally, we adopted the recently introduced LYL method that accounts for the age of diagnosis distribution and allows decomposition of LYL by cause of death. Given there is sometimes a delay between onset of psychiatric disorder and diagnosis, the true gap in life expectancy for those with SMI is likely to lie between estimates generated by this more conservative approach and earlier, less conservative, methods.

Our study has some limitations. Ascertainment of SMI through hospital admission data means we are likely to have included people with SMI at the more severe end of the spectrum or with particularly challenging/complex health and social care needs, and our findings may not be generalisable to all people with SMI. Our study included people with both prevalent SMI (diagnosed before 2000) and incident SMI (diagnosed from 2000). Previous studies have indicated that individuals with SMI have higher mortality following discharge from psychiatric hospital, ^{Reference Musgrove, Carr, Kapur, Chew-Graham, Mughal and Ashcroft35} and hence our inclusion of prevalent cases may underestimate mortality in people with SMI. Similarly, by using first hospital admission date as a proxy of diagnosis date, our estimates of life expectancy from disease onset may be slightly underestimated. Our findings on the widening life expectancy gap for schizophrenia may reflect changes in practice related to psychiatric hospital admission in Scotland, with number of psychiatric hospital admissions having steadily decreased from around 30 000 in 2000 to around 20 000 from 2010 onward. ³⁶ It is possible that people admitted in later years have more severe illness than those admitted in earlier years, which may contribute to the widening gap. Although we included nationwide data, the number of people included with bipolar disorder was smaller than in the other groups, and LYL and excess LYL estimates were therefore more uncertain. Also, we did not have information on emigration, and so people whose death was registered outside Scotland would have been lost to follow-up. Although emigration from Scotland is relatively low ³⁷ (and likely to be lower among those with SMI, because of a healthy immigrant effect), we may have underascertained deaths in those with SMI, thus underestimating differences in life expectancy. Unfortunately, we did not have access to data for the period 2020–2024, and so cannot comment on mental health disparities in life expectancy during these most recent years, which may have worsened as a consequence of the COVID-19 pandemic, given reports of higher COVID-19 mortality in people with SMI. ^{Reference De Hert, Mazereel, Stroobants, De Picker, Van Assche and Detraux38} Finally, a number of questions were beyond the scope of our study and warrant investigation. These include further examination of the role of deprivation, with previous studies indicating an effect of SMI on excess mortality that is over and above that of deprivation ^{Reference Das-Munshi, Chang, Dregan, Hatch, Morgan and Thornicroft14} ; closer investigation of changes in cause-specific mortality over time, particularly alcohol-related deaths ^{Reference Moustgaard, Tarkiainen, Östergren, Korhonen, Zengarini and Costa11} ; and the role of changes in mental health services (with fewer in-patient beds available).

Implications

Excess deaths due to natural causes account for much of the observed premature mortality in people with SMI in both our study and a Danish study, ^{Reference Erlangsen, Andersen, Toender, Laursen, Nordentoft and Canudas-Romo18} reflecting the poorer physical health of people with SMI, in whom there is a high burden of multimorbidity. ^{Reference Launders, Kirsh, Osborn and Hayes39} The cause of this premature mortality is complex, multifactorial and multi-level, ^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers40,Reference O’Connor, Worthman, Abanga, Athanassopoulou, Boyce and Chan41} with key risk factors operating at the individual through to societal level, and encompassing modifiable lifestyle behaviours, such as smoking, overweight/obesity and physical inactivity; multimorbidity and substance misuse; psychological factors; socioeconomic disadvantage; healthcare access, including year-on-year underfunding of mental health services and diagnostic overshadowing; iatrogenic harms of some psychotropic medication; structural inequalities; and stigma and marginalisation. ^{Reference O’Connor, Worthman, Abanga, Athanassopoulou, Boyce and Chan41} Addressing these complex causes requires multi-level and synergistic solutions. ^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers40,Reference O’Connor, Worthman, Abanga, Athanassopoulou, Boyce and Chan41} These have recently been structured around three key principles: integration of mental and physical healthcare, prioritisation of prevention of physical disease while strengthening treatment of mental illness, and optimisation of intervention synergies across social-ecological levels. ^{Reference O’Connor, Worthman, Abanga, Athanassopoulou, Boyce and Chan41} Current health system structures generally create disconnected care that particularly disadvantages patients with complex needs, such as those with SMI. In particular, there is a need for better integration across primary, secondary and community healthcare settings and between generalist and specialist providers, along with coordinated support from social care and allied health professionals. ^{Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers40,Reference Shidhaye, Lund and Chisholm42} In addition to regular physical health monitoring in those with SMI, a greater focus on primary prevention strategies is required to tackle key modifiable risk behaviours in this group. ^{Reference Ilyas, Chesney and Patel43}

In conclusion, our national study highlights the persistent and potentially widening gap in life expectancy experienced by people with SMI. This entrenched disparity reflects ingrained inequalities at multiple levels that affect the health and lifespan of this marginalised subgroup of the population. Future success in narrowing this shameful gap relies on government commitment to introducing equitable social, economic and health policies, significant healthcare re-structure and improved resourcing, and investment in tailored, well-implemented and appropriately evaluated interventions and strategies for primary and secondary prevention of SMI and associated comorbidities.