Peer environment at seventeen tracked mental disorder risk for the next decade

Design

nationwide register-based cohort, Cox proportional hazards

Population

604,819 Finnish residents born 1985-2000, followed from age 17, median 11.7 years

Primary outcome

time to first incident mental disorder diagnosis after age 17

Effect

peer genetic risk and externalising disorders HR 1.34 (95% CI 1.29-1.38); peer diagnoses and internalising disorders HR 1.17 (1.15-1.18)

A nationwide Finnish register study took everyone born between 1985 and 2000 - 604,819 people - and assigned each to four peer network contexts: lower secondary school, upper secondary school, postal code area, and the 1,000-metre square around their home. Follow-up ran from age 17 to first psychiatric diagnosis, emigration, death or the end of 2023, a median of 11.7 years. Two exposures were tested: peers' own diagnoses, and peers' family-based genetic risk scores, estimated from diagnoses in first- to fifth-degree relatives. Models adjusted for the participant's own genetic risk score, sex, birth year, parental education and income, and network size.

Of the cohort, 234,117 received a diagnosis. Peer genetic risk predicted the same disorder later, most strongly for externalising disorders in upper secondary school (hazard ratio 1.34, 95% CI 1.29-1.38). Peer diagnoses predicted most strongly for internalising disorders in the same setting (HR 1.17, 95% CI 1.15-1.18). Cross-disorder effects appeared in both directions, but asymmetrically: peers' internalising diagnoses predicted both internalising and externalising outcomes, while peers' externalising diagnoses predicted mainly externalising ones.

The design is the interesting part. Adjusting for the participant's own genetic risk and then finding an effect of their *peers'* genetic risk is a way of isolating a social pathway that genetic confounding cannot easily explain. It does not identify the mechanism, and Finnish schools sort students in ways Indian ones do not, so the size of the effect will not transfer. What transfers is the question to ask a worried parent: not only what is happening at home, but what is happening in the class.

• Ask specifically about the peer group and the school year when assessing a new adolescent presentation, not only family history
• Note that the strongest signal was in the later school years, when peer groups are self-selected rather than assigned
• Treat a cluster of presentations from one school or class as a signal worth acting on rather than a coincidence
• Keep the effect size in proportion - a hazard ratio of 1.17 to 1.34 is a population-level finding, not a prediction for an individual

Why it matters

It argues that adolescent risk is partly social and partly located outside the family, which is where most of our history-taking stops.

Don't overread it

This is an observational register study - it shows association with peer environment, not that changing a peer group changes risk.

The statistics, in plain English

Hazard ratios of 1.17 to 1.34 are small at the individual level and substantial at the population level - a 1.34 hazard ratio spread across a whole school cohort produces many more cases than the same ratio means for one patient. The confidence intervals are extremely tight because the cohort is very large, so these estimates are precise; precision is not the same as freedom from confounding, and unmeasured features of schools and neighbourhoods could still drive part of the association.

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.