The research team, led by Mike Trott of the University of Queensland together with David Lawrence, Sean Halstead, Claudia Bull, Dan Siskind and Steve Kisely, linked hospital admission records from the Queensland Hospital Admitted Patient Data Collection with death registrations from the state’s Registry of Births, Deaths and Marriages. The final cohort comprised 255,448 adults aged 18 or over who had at least one hospital admission with a principal mental health diagnosis between January 2009 and December 2023. Individuals with dementia were excluded because of their fundamentally different mortality profile. Each person was assigned a single principal diagnosis using a severity-based hierarchy spanning seven categories: schizophrenia, affective psychoses, other psychotic disorders, alcohol or substance use disorders, neurotic disorders, stress and adjustment reactions, and depressive disorders. Over a median follow-up of 8.6 years, the cohort accumulated more than 2.2 million person-years of observation and 28,764 deaths.

The methodological heart of the study lies in its use of abridged period life tables, constructed separately for each diagnostic group, sex and three-year calendar period using the Chiang actuarial method. Life expectancy at age 18 was extracted as the primary outcome, and the gap was calculated against Australian Bureau of Statistics life tables for the general Queensland population. Because life expectancy cumulates mortality across all ages, it captures the survival impact of conditions that kill early, something standardised mortality ratios alone can miss. Confidence intervals were derived through parametric bootstrapping with 500 iterations, and a three-point moving average smoothed age-specific mortality rates to stabilise cells containing few deaths. Temporal trends were tested with linear regression, and a sensitivity analysis fixed the diagnostic composition of the cohort at its pooled distribution to check that shifts in who was being hospitalised were not driving the results.

The headline numbers are striking. Life expectancy for women in the psychiatric cohort rose from 71.5 years in 2009–2011 to 76.1 years in 2021–2023; for men, it climbed from 64.2 to 69.3 years. Over the same period, general population life expectancy increased by less than a year. As a result, the life expectancy gap for men narrowed from 16.7 to 12.1 years, and for women from 13.4 to 9.4 years, with both trends statistically significant at p < 0.001. Standardised mortality ratios told the same story: the overall ratio fell from 3.12 to 2.10 in men and from 2.35 to 1.73 in women. Crucially, the narrowing was driven by genuine gains in survival within the psychiatric cohort rather than by any deterioration in the health of the general population, and the sensitivity analysis confirmed that changing diagnostic composition did not explain the improvement.

Most diagnostic groups shared in the gains. The steepest narrowing occurred among people with alcohol and substance use disorders, whose life expectancy gap for men fell from 20.6 to 16.0 years, and among those with neurotic, stress-related and depressive disorders, whose gaps shrank to between roughly 4 and 8 years by the end of the study. But the exceptions are as informative as the successes. Neither men nor women with schizophrenia showed any statistically significant change in their life expectancy gap, which remained stuck at around 14 years for men and even edged upward to 14.8 years for women by 2021–2023. Women with affective psychosis and both sexes with other psychotic disorders also showed no significant narrowing. The authors contrast this with the earlier foundational Australian evidence from Western Australia, which covered 1985 to 2005 and found the gap widening; the new data suggest that decades of mental health reform and improved physical health care have finally bent the curve, but that the benefits have bypassed those with the most severe diagnoses.

The cause-of-death analysis delivers perhaps the study’s most consequential message. While intentional self-harm carried the highest excess mortality rate of any single cause, the combined excess mortality from physical health conditions dwarfed it: roughly 440 excess deaths per 100,000 person-years in both sexes, which is about 2.7 times the self-harm rate in men and 5.3 times the rate in women. In men, intentional self-harm was the single largest contributor to excess mortality at 163.7 per 100,000 person-years, followed by cancer at 112.9 and coronary heart disease at 109.1. In women, coronary heart disease led at 101.7, followed by cancer and then self-harm. Attributable fractions were sobering: 87 per cent of self-harm deaths in men and 92 per cent in women were excess deaths, meaning they would not have been expected in the general population, while nearly two-thirds of deaths from physical causes in women were excess.

The mechanistic pathways behind premature death differ sharply across diagnoses, and the authors argue this demands tailored rather than one-size-fits-all interventions. In people with substance use disorders, excess mortality was dominated by the direct physical sequelae of use, most notably liver disease. In schizophrenia, by contrast, the excess concentrated in cardiovascular disease, consistent with the metabolic consequences of the illness itself and of its treatment. Second-generation antipsychotic medications, while effective at controlling psychiatric symptoms, carry well-documented adverse effects including weight gain, dyslipidaemia and impaired glucose tolerance, which may have paradoxically raised cardiometabolic risk in some patients. Layered on top of these biological vulnerabilities are structural inequities: people with mental illness receive cardioprotective medications less often, participate less in cancer screening programmes, and are less likely to receive guideline-concordant treatment for physical conditions.

The study’s strengths rest on its population-level scope. By capturing every hospitalised psychiatric admission across an entire Australian state over 14 years, the design minimises selection bias and provides the statistical power to detect trends within individual diagnostic subgroups, something the fragmented international literature, drawn from heterogeneous studies with differing methods, has struggled to do. The findings partially resolve a genuinely contested question: Scandinavian register studies had suggested modest narrowing of the mortality gap, while recent meta-analyses found no significant reduction in years of potential life lost and even hinted the relative gap might be widening. The authors suggest that broad time intervals in previous syntheses may have masked finer-grained trends, and that their hospital-ascertained cohort, being more clinically severe than community-based samples, naturally shows larger absolute gaps.

Limitations temper the optimism. The cohort captures only people with hospital-based psychiatric admissions, excluding those managed solely in primary or community care, so the results may not generalise to milder presentations. The general population comparator from ABS life tables inherently includes people with mental illness, modestly understating the true gap. Cause-of-death comparisons relied on Australia-wide rather than Queensland-specific reference rates, and single underlying-cause coding cannot capture the contribution of comorbid conditions. Changes in hospitalisation practice over time could also have altered who enters the cohort. A sensitivity analysis comparing pre- and post-COVID periods found no substantial differences, which the authors attribute to the continuity of mental health care in Queensland during the pandemic, including a rapid shift to telepsychiatry without significant reductions in antipsychotic prescribing.

The implications are clear-eyed rather than celebratory. Even after 14 years of improvement, a man with a hospitalised psychiatric diagnosis in Queensland can expect to die more than 12 years earlier than his peers, and a woman nearly a decade earlier. For people with schizophrenia, the gap has not moved at all. The authors call for sustained investment in physical health monitoring, metabolic risk management and equitable access to cancer and cardiovascular care for people with psychotic disorders, alongside integrated physical and mental health care as routine practice rather than aspiration. Suicide prevention, including lethal means counselling, remains a priority, particularly for men with psychosis. But the dominant lesson of the data is that the road to longer lives for people with mental illness runs primarily through their bodies as much as their minds, and that health systems which treat the two separately are still paying for it in years of life lost.

Subject of Research: Temporal trends in the life expectancy gap between people with hospitalised psychiatric disorders and the general population in Queensland, Australia

Article Title: Change in the life expectancy gap among people with mental illness: a 14-year population-based cohort study in Australia

Article References: Trott, M., Lawrence, D., Halstead, S., Bull, C., Siskind, D., & Kisely, S. (2026). Change in the life expectancy gap among people with mental illness: a 14-year population-based cohort study in Australia. The Lancet Regional Health – Western Pacific, Article 101978. https://doi.org/10.1016/j.lanwpc.2026.101978

Image Credits: AI Generated

DOI: 10.1016/j.lanwpc.2026.101978

Keywords: life expectancy, mental illness, schizophrenia, excess mortality, psychiatric disorders, cohort study, cardiovascular disease, cancer, suicide, substance use disorders, health inequality, Australia