Who a 17-year-old sat next to shows up in their psychiatric record a decade later

Design

nationwide register-based cohort study, Cox proportional hazards models

Population

604,819 Finnish residents born 1985 to 2000, followed from age 17 for a median 11.7 years

Primary outcome

time to incident mental disorder diagnosis after age 17

Effect

peer genetic risk, externalising disorders HR 1.34 (1.29 to 1.38); peer diagnoses, internalising HR 1.17 (1.15 to 1.18)

A Finnish register cohort of 604,819 people born between 1985 and 2000 assigned each person to four peer networks — lower secondary school, upper secondary school, postal code, and 1,000-metre square — and followed them from age 17 to first psychiatric diagnosis, emigration, death or the end of 2023. Two exposures were tested: peers' family-based genetic risk scores, built from diagnoses in first- to fifth-degree relatives, and peers' own diagnoses. Models adjusted for the person's own genetic risk score, sex, birth year, parental education and income, and network size.

Over a median 11.7 years, 234,117 were diagnosed. Peer genetic risk was associated with the same disorder later, strongest for externalising disorders in upper secondary school (HR 1.34; 95% CI 1.29 to 1.38). Peer diagnoses mattered most for internalising disorders, again in upper secondary school (HR 1.17; 1.15 to 1.18). Cross-disorder associations ran in both directions, though peer externalising diagnoses predicted the person's internalising outcomes only weakly.

The effect sizes are small and the mechanism is unidentified — shared environment, assortment into schools and genuine social transmission are all live explanations. What it argues for is that late-adolescent settings are a plausible target for population-level prevention, which is a commissioning point rather than a consulting-room one.

• Take a peer and school context history in late adolescence, not only a family history
• Note that the strongest signal sat in upper secondary school — the years many services discharge from
• Read this as support for school-based prevention programmes rather than individual risk scoring
• Do not use peer information to predict an individual's risk

Why it matters

It puts a modifiable environment — the upper secondary school year group — in the frame for prevention.

Don't overread it

This is observational register data; it cannot show that peers cause the later diagnoses.

The statistics, in plain English

A hazard ratio of 1.34 across 600,000 people is a real but small shift: it moves population risk, not any one person's prognosis. The confidence intervals are extremely narrow because the cohort is huge, which makes the estimates precise without making them large. Register studies also cannot separate social transmission from families choosing the same schools.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.