A 225,000-Person Study Found Psychiatric Disorders Age You Faster at the Cellular Level

Biological age and chronological age can diverge in ways that affect health and longevity, and a new study suggests that psychiatric diagnoses may push them further apart than previously understood. Researchers analyzing plasma metabolite data from 225,212 UK Biobank participants found that people with substance use, psychotic, affective, and neurotic disorders tended to show a metabolomic profile older than their chronological age. The findings, published in BMJ Mental Health, add molecular-level detail to a long-standing observation: people with mental illness often die younger than the general population.

Psychosis and substance use disorders linked to the oldest metabolomic profiles

The research team, led by Mutz and colleagues, built a metabolomic ageing clock called MileAge, trained on plasma metabolite data from all 225,212 participants. The clock generates a "MileAge delta," the difference between the age a person's metabolites predict and their actual chronological age. A positive delta means the body's biochemistry looks older than the birthdate would suggest.

Of the 38,524 participants who had at least one mental disorder diagnosis before baseline, the clearest signal came from psychosis. That diagnosis was associated with a MileAge delta of 0.556 years on average (95% CI 0.250 to 0.861), making it the largest single-disorder effect in the dataset. Alcohol, hallucinogen, and tobacco use disorders also showed positive deltas, as did single-episode depression and stress-related disorders. The signal for tobacco use disorder remained comparable across age groups, while the association with affective disorders and single-episode depression was more pronounced in people under 65.

Not all diagnoses pointed toward faster ageing. Obsessive-compulsive disorder and eating disorders were associated with a metabolite-predicted age younger than chronological age, a pattern that the authors flag as requiring careful interpretation rather than dismissal.

Why accelerated metabolomic ageing raises serious clinical stakes

The metabolomic age gap matters because it is not merely a number on a chart. The study validated MileAge against mortality risk, frailty, shorter telomeres, and increased dementia risk, grounding it in outcomes that clinicians already track. Research on accelerated aging in serious mental disorders has consistently linked this biological trajectory to earlier onset of cardiovascular disease, metabolic syndrome, and all-cause mortality, conditions that shorten the lives of people with schizophrenia or severe depression by an estimated decade or more.

Increasing psychiatric comorbidity burden compounded the effect in this dataset. Participants carrying more than one diagnosis showed progressively higher metabolomic ages relative to their chronological peers, suggesting the biological toll may accumulate across conditions rather than plateauing. Studies on early mental illness and accelerated aging have raised the possibility that intervening earlier in the course of psychiatric illness could reduce this biological drift, though no intervention trial has directly tested that hypothesis.

How UK Biobank metabolomics made the analysis possible

The study's scale is what makes it unusual. Most ageing-clock research relies on epigenetic (DNA methylation) measures, which are expensive and harder to collect at population scale. Plasma metabolomics, used here, captures real-time biochemical activity across hundreds of circulating molecules and can be run on stored biobank samples. The team extracted mental disorder diagnoses from a combination of primary care records, hospital inpatient data, death registries, and self-reported physician diagnoses, using ICD-10 codes F10 through F99 and excluding organic disorders.

Sex and age stratification revealed meaningful heterogeneity. The association between psychosis and older metabolomic age reached significance in males (beta = 0.765, p less than 0.001) but not in females. Behavioral syndromes showed a younger metabolite-predicted age particularly in females. These patterns resist a single explanation and may reflect differences in illness course, treatment exposure, or underlying biology between sexes.

A separate analysis using a related UK Biobank cohort studying biological aging and depression found that accelerated biological ageing contributes to the risk of developing depression and anxiety, adding a potential bidirectional dimension: ageing may worsen mental health while mental health conditions accelerate ageing.

Polygenic scores tell a partly different story

The study also examined whether genetic liability for psychiatric conditions predicted the same metabolomic ageing patterns as clinical diagnoses, and the results were not always parallel. Higher polygenic scores for major depression, autism, and ADHD were associated with an older metabolomic age, with effect sizes (beta range 0.020 to 0.047) in the same direction as diagnosed depression. But polygenic scores for psychosis and tobacco use disorder predicted a younger metabolomic age, running counter to the older profile seen in people who actually carry those diagnoses.

The researchers interpret this discrepancy as evidence that non-genetic factors, including illness duration, treatment history, lifestyle, and social circumstances, drive much of the metabolomic ageing signal observed in people with active diagnoses. The median time from first diagnosis to UK Biobank baseline ranged from about 7 years for substance use disorders to nearly 19 years for childhood and adolescent disorders, meaning many participants had been living with their conditions for a long time before measurement.

Open questions about causality, generalisability, and medication effects

The study's authors are careful about causal claims, and that caution is warranted. The cross-sectional design at baseline means the data cannot establish whether psychiatric illness accelerates ageing, whether an older biological profile predisposes people to mental disorder, or whether both directions operate simultaneously. As research on metabolomics for psychiatry has noted, replication across independent cohorts and longitudinal validation remain critical gaps in this field.

Medication is another unmeasured variable. Many psychiatric medications alter metabolic function, and the study did not adjust for prescription data, leaving open the possibility that some ageing signal reflects drug effects rather than disorder biology. Sensitivity analyses excluding self-reported diagnoses or adjusting for smoking produced negligible changes, but adjusting for BMI attenuated the associations for eating disorders and schizophrenia, suggesting body composition mediates part of the effect. The UK Biobank's well-documented healthy volunteer bias also limits how far these findings generalize to people with more severe or untreated illness.

MileAge itself has moderate prediction accuracy, and the polygenic scores were derived predominantly from European ancestry genome-wide association studies, limiting their applicability to other population groups. Despite these constraints, the consistency of findings across multiple diagnostic categories in a sample of more than 200,000 people gives the broad pattern credibility. Future work connecting MileAge trajectories to treatment response and long-term health outcomes could eventually clarify whether slowing metabolomic ageing is a viable target for psychiatric care.

Evidence rating

EMERGING — 78/100

Caveats:

• Large cross-sectional observational study — shows association, not causation; bidirectional relationships cannot be ruled out.
• Funding not disclosed; conflicts of interest cannot be fully assessed.
• MileAge clock was developed by the same research group, introducing potential validation circularity.
• UK Biobank healthy volunteer bias limits generalisability to the broader population.
• Polygenic scores derived predominantly from European ancestry GWAS, reducing transferability to other ancestry groups.
• Smoking and BMI were excluded from primary models as potential mediators, leaving residual confounding possible.

Large cross-sectional observational study using plasma metabolomic profiles from 225,212 UK Biobank participants. Associations between mental disorder diagnoses and metabolomic age are correlational; the design cannot establish causation or rule out reverse causation. The MileAge clock was developed by the same authors, and findings have not yet been independently replicated at scale.

References

1. Julian Mutz, Lachlan Gilchrist, Andrea G Allegrini, Sandra Sanchez Roige, Cathryn M Lewis (2026). Metabolomic ageing across mental and behavioural disorders. BMJ Mental Health. https://doi.org/10.1136/bmjment-2025-302181
2. Owen M. Wolkowitz (2018). Accelerated biological aging in serious mental disorders. World Psychiatry. https://doi.org/10.1002/wps.20546
3. Mental illnesses in early life linked to faster aging and worse health in later years. NIH. https://www.nia.nih.gov/news/mental-illnesses-early-life-linked-faster-aging-and-worse-health-later-years
4. Biological aging and the risk of depression and anxiety: a prospective cohort study of 500,000 adults. Nature. https://www.nature.com/articles/s41467-023-38063-4
5. Daniele Cavaleri, Carlo Bassetti, Giorgio Cucchi, et al. (2026). Metabolomics biomarkers for precision psychiatry. Frontiers in Psychiatry. https://doi.org/10.3389/fpsyt.2026.1736206