This piece discusses mortality, including suicide as a cause-of-death category. If you’re struggling, Samaritans are available free, 24/7, on 116 123.
A gap measured in years of life, not quality of it
Yasmin Rahali, Charlotte Dennison, Anita Thapar and Ajay Thapar, at Cardiff University’s Wolfson Centre for Young People’s Mental Health, have published a narrative review in European Child & Adolescent Psychiatry synthesising what population-scale research now shows about ADHD and premature mortality. This is the heaviest subject the Directory covers, and it deserves stating soberly and precisely, because the numbers are not ambiguous and the people they describe are not abstractions.
The review focuses on nine studies from Denmark, Sweden, Finland, the UK, Taiwan and the USA — whole-population cohorts and national registries, so the estimates apply to general populations rather than clinical samples. Eight of the nine found elevated all-cause mortality in people with ADHD. Hazard ratios ranged from 1.07 to 4.44, with the spread largely explained by how much each model adjusted for and how long it followed people; a recent meta-analysis puts the pooled figure at 2.18 times. The most recent UK health-records study estimates that adults with diagnosed ADHD live around seven to nine years less than matched controls — 6.78 years for men, 8.64 for women.
Where the deaths come from matters for understanding what the gap is. Most studies found non-natural causes — accidents and suicide — to be the most prevalent, with unintentional injury the more common and suicide carrying the higher relative risk. Studies following people into older age found natural causes growing in importance, consistent with the elevated cardiovascular and metabolic burden ADHD carries. Recent work from the same Cardiff group suggests smoking and obesity mediate part of the link to premature mortality — which matters because both are modifiable.
The reflex reading of these numbers is that ADHD itself is lethal — impulsivity as a death sentence written into the neurotype. The review’s own findings resist that reading at every turn, and the resistance is the story. Risk was not uniform. It concentrated — dramatically — in the late-diagnosed, in women, and in those with accumulated psychiatric comorbidity. Those are not three random subgroups. They are the three faces of the same person: the one the detection apparatus found last, supported least, and left to accumulate the consequences of navigating an unaccommodating world unrecognised. The mortality gap tracks the support gap.
The comorbidity gradient makes the accumulation visible. In the Swedish registry, ADHD alone carried a hazard ratio of 1.41 — elevated, but modest. ADHD with four or more psychiatric comorbidities carried a hazard ratio of 25.22. Substance use disorder was the deadliest single addition. The condition is not doing that arithmetic by itself; a lifetime of unmet need is compounding into it.
The later the system finds you, the steeper the risk becomes
The review’s most important finding is a dose-response relationship between age at diagnosis and mortality risk, and it deserves its exact numbers. In the Swedish cohort, people diagnosed at twelve or under carried an adjusted hazard ratio of 1.33 — not significantly different from no elevation at all. Diagnosed between thirteen and seventeen: 2.41. Diagnosed at eighteen or later: 9.12. The Danish registry shows the same staircase — diagnosis in early childhood, 1.86; in school age, 1.58; in adulthood, 4.25.
The authors offer the conventional explanation — late-diagnosed samples may contain more persistent, severe cases — and it surely carries some weight. But sit with what else a late diagnosis means. It means a childhood of being corrected rather than understood, an adolescence of failing in systems built for other minds, and an early adulthood accumulating exactly the comorbidities — depression, anxiety, substance use — that the review shows multiplying the risk. The undiagnosed years are not neutral waiting time. They are the years in which the compounding happens.
Read that way, the dose-response curve is the detection-failure thread of recent weeks arriving at its highest possible stakes. I have written about the diagnostic apparatus finding women only when their compensation collapses in perimenopause. This review prices that failure in years of life. A child found at eight carries essentially baseline risk. The same neurotype found at twenty-eight carries nine times the hazard — not because the person changed, but because two decades of unsupported navigation were allowed to do their work first. Late diagnosis is not an administrative inconvenience. On this data it is a mortality risk factor in its own right.
A female excess that dissolves into the comorbidity the delay produced
Most studies reporting sex differences found women with ADHD at higher risk than men — the UK life-expectancy figures above run nearly two years worse for women. The review’s treatment of why is the most quietly damning passage in it.
In the Swedish data, the raw female hazard ratio was 6.46 against 5.00 for males. Adjust for psychiatric and neurodevelopmental comorbidity, and the female figure collapses to 1.25 while the male falls to 1.47 — the excess risk in women is very largely carried by the comorbid conditions they have accrued. And the review then assembles how they accrue them: women are diagnosed on average four years later than men; the diagnostic criteria are historically male-symptom focussed; girls are referred for evaluation less often than boys with equivalent difficulty; and women are less likely to be prescribed medication unless they present with prominent, externalising issues — that is, unless they present like boys.
So the female mortality excess is not a biological property of being a woman with ADHD. It is the downstream shape of a detection system calibrated to male presentation — later recognition, less treatment, more years unsupported, more depression and anxiety and substance use accumulated in the interval, and then a steeper hazard ratio at the end of it, which the apparatus can point to as female vulnerability. The Kini-Seery finding from three weeks ago — women surfacing in clinics only when hormonal transition strips their masking — is this same pipeline observed at midlife. The review shows where the pipeline terminates.
Protection exists, which makes the waiting list a clinical variable
The review’s final contribution is the one that converts all of the above from tragedy into indictment: the risk is modifiable, and we know at least one thing that modifies it. Three quasi-experimental studies — from Taiwan, Sweden and Canada — found ADHD medication associated with reduced all-cause mortality, hazard ratios 0.61 to 0.89. The Swedish JAMA study found methylphenidate initiation associated specifically with fewer deaths from unnatural causes — the accidents and suicides that dominate the gap. The findings are clearer for stimulants than non-stimulants, and the authors are properly cautious: these are quasi-experimental designs, not randomised trials.
I report the medication finding straight, because it is an honest and important result whatever one’s wider view of prescribing. But it should be read at its proper width. Medication in these datasets is not merely a molecule — it is a marker of having been found, diagnosed, and retained in care. The protective signal is the support signal, read from the pharmacological side. Which makes the converse equally true: every year on a waiting list, every right-to-choose obstruction, every shared-care refusal is not administrative friction. On this evidence, it is exposure.
One more honesty note the authors themselves raise: most of these registries are Scandinavian, where ADHD recognition and provision are comparatively strong. The true risks in countries with poorer provision — the UK’s current assessment crisis being the obvious case — may be higher than these figures. The 3:1 male-to-female ratio across the studies also means the female estimates rest on thinner samples, even as they run worse.
The clinical translation the authors land on is the right one, and it is structural, not individual: earlier recognition, equitable access, integrated physical-health monitoring, and treatment where appropriate, because the highest-risk people — women, the late-diagnosed, the comorbid — are precisely those the current system serves worst. The mortality gap is real. But it is not written into the neurotype. It is written into the years the system leaves people unfound — and years, unlike neurotypes, are something policy can actually change.
Citations
Rahali, Y., Dennison, C. A., Thapar, A. & Thapar, A. K. (2026) — Attention deficit hyperactivity disorder (ADHD) and premature mortality: A narrative review of the literature — European Child & Adolescent Psychiatry
O’Nions, E., El Baou, C., John, A. et al. (2025) — Life expectancy and years of life lost for adults with diagnosed ADHD in the UK: matched cohort study — British Journal of Psychiatry
Sun, S., Kuja-Halkola, R., Faraone, S. V. et al. (2019) — Association of Psychiatric Comorbidity With the Risk of Premature Death Among Children and Adults With ADHD — JAMA Psychiatry
Dalsgaard, S., Østergaard, S. D., Leckman, J. F. et al. (2015) — Mortality in children, adolescents, and adults with attention deficit hyperactivity disorder: a nationwide cohort study — The Lancet
Li, L., Zhu, N., Zhang, L. et al. (2024) — ADHD Pharmacotherapy and Mortality in Individuals With ADHD — JAMA
Skoglund, C., Sundström Poromaa, I., Leksell, D. et al. (2024) — Time after time: failure to identify and support females with ADHD — Journal of Child Psychology and Psychiatry
