News · Brain & Mental Health
Mental illness before 35 tracked with dementia after 50
A Molecular Psychiatry study of 2.5 million Swedes compared siblings and found the dementia link runs through a general vulnerability to psychiatric illness, not any one diagnosis.
Based on a peer-reviewed cohort study in Molecular Psychiatry
- Swedish national registers covering 2,543,621 people and 1,485,880 full-sibling pairs, published in Molecular Psychiatry.
- Exposures were six psychiatric diagnoses recorded by age 35; outcomes were all-cause dementia and Alzheimer's disease recorded after age 50.
- Every one of the six diagnoses tracked with higher dementia risk, roughly two to three and a half times.
- Comparing siblings with each other cut the estimates only modestly, so shared family background does not explain them away.
- Almost all of the signal loaded onto a general psychopathology factor, a shared vulnerability across conditions, rather than any specific diagnosis.
- After accounting for that general factor, only the psychotic-specific factor still carried extra risk.
- The general factor's own effect is modest, around 1.25 times the odds within sibling pairs.
- This is register data. Diagnoses recorded by 35 may partly reflect very early dementia processes rather than a separate cause.
That psychiatric illness earlier in life goes with dementia later is not news. Which part of the illness carries the risk is. A study in Molecular Psychiatry used two and a half million Swedes and their siblings to pull the two apart.
The associations were largely attributable to the general psychopathology factor: not depression, not anxiety, but whatever the diagnoses share.
The question underneath the correlation
The pattern is well established and, until now, badly understood. Multiple psychiatric disorders are associated with later dementia, but it remains unclear whether these associations reflect a shared liability toward all psychiatric conditions or diagnosis-specific effects.
The distinction is not academic. If depression specifically damages the brain, treating depression is the lever. If a broad vulnerability drives both, then depression is a marker rather than a mechanism, and the lever is somewhere else entirely.
Two and a half million people, and their brothers and sisters
The study drew on Swedish registers covering 2,543,621 individuals and 1,485,880 full-sibling pairs.
Exposures were six psychiatric diagnoses recorded by age 35, along with a model separating one general and three specific factors covering internalizing, externalizing and psychotic conditions. Outcomes were all-cause dementia and Alzheimer’s disease recorded after age 50.
The sibling structure is what raises this above the usual cohort study. Brothers and sisters share half their genes and most of their upbringing, so a link that survives a within-family comparison cannot be blamed on the family.
Every diagnosis looked risky on its own
Taken one at a time, the results are stark. All psychiatric diagnoses were significantly associated with increased risk of all-cause dementia, with hazard ratios ranging from about two to three and a half.
Those numbers held up under the family comparison, dropping only modestly. Shared background explains some of the pattern. It does not explain most of it.
The part that dissolved under scrutiny
Then the individual diagnoses stopped mattering. After accounting for the general factor, only the psychotic-specific factor remained associated with dementia and Alzheimer’s disease.
Everything else, the depression, the anxiety, the substance problems, contributed through what they had in common rather than through anything of their own.
The general factor’s own effect is modest: odds of about 1.25 within sibling pairs. This is a finding about structure, not about magnitude.
What register data cannot show
The most difficult alternative is timing rather than confounding. Dementia begins decades before anyone diagnoses it, so a psychiatric diagnosis at 30 might occasionally be the earliest visible sign of a process already underway. A sibling comparison does nothing about that.
Registers also record diagnoses, not illnesses. People who never reached psychiatric care are counted as unaffected.
The authors phrase the implication carefully: liability toward general psychopathology and, independently, psychotic conditions, by early adulthood might be early markers of increased dementia risk and targets for timely identification and prevention.
Markers, in other words, before mechanisms. Which is the honest version of a finding this size.
People also ask
What is a general psychopathology factor?
It is the statistical observation that psychiatric diagnoses cluster together more than chance allows. Someone with depression is more likely than average to also have an anxiety disorder, a substance problem or a psychotic episode. Modelling a shared factor beneath all of them separates what the conditions have in common from what is specific to each. This study found the dementia link sat almost entirely in the shared part.
Why compare siblings?
Because siblings share about half their genes and most of their childhood environment. If a link between early psychiatric illness and later dementia survives a comparison between brothers and sisters, it cannot be explained by the family background they had in common. The estimates here shrank when siblings were compared, but did not disappear, which is what you would expect if family factors explain part but not all of it.
Does psychiatric illness cause dementia?
This cannot establish that. The most awkward alternative is timing: dementia processes begin decades before diagnosis, so a psychiatric diagnosis at 30 might sometimes be the earliest visible sign rather than a separate risk factor. Register data records when a diagnosis was made, not when a disease started. The sibling design handles family confounding, not reverse causation.
Which diagnosis carries the most risk?
That framing is the one the study argues against. All six diagnoses looked risky on their own, but once the shared vulnerability was accounted for, only the psychotic-specific factor added anything further. The practical reading is that the number and breadth of psychiatric problems by early adulthood may matter more than which label they carry.
How large is the effect?
The raw diagnosis-level figures look dramatic, at two to three and a half times the risk. The general factor's own contribution within sibling pairs is much smaller, around 1.25 times the odds. Dementia is also a late-life outcome with many contributors, so this sits alongside the established modifiable risks rather than above them.