# Childhood household dysfunction and psychiatric, criminal, and social outcomes in emerging adulthood. A cousin comparison study

## TL;DR

Childhood household dysfunction is linked to poor psychiatric, criminal, and social outcomes in adulthood, but some of these links may be due to unmeasured family factors.

## Contribution

The study uses cousin comparisons to estimate the impact of unmeasured familial confounding on the association between childhood household dysfunction and adult outcomes.

## Key findings

- Population-level models showed hazard ratios of 1.3 to 2.5 for associations between childhood dysfunction and adverse outcomes.
- Cousin comparisons reduced these associations by an average of 12%, indicating some role of unmeasured familial confounding.
- Dose–response relationships were observed, but associations were attenuated in cousin comparisons.

## Abstract

Childhood household dysfunction is a well-known risk factor for adverse medical and social outcomes. However, less is known about the extent to which such associations are affected by unmeasured familial confounding.

This cohort study is based on Finnish register data on birth cohorts 1987–2000 (n = 835 987). We considered parental hospital-presenting substance use and psychiatric disorders, prison sentences, death, means-tested social assistance, and union dissolution at ages 0–14 as indicators of childhood household dysfunction. The study participants were followed from age 15 until the end of 2020 for hospital-presenting psychiatric disorders and substance use, psychotropic medication purchases, violent and property crime arrests, and not being in education, employment, or training. The associations were estimated using Cox regression, and cousin comparisons were used to account for unmeasured confounders shared within extended families (n = 87 500).

All the exposures were associated with the outcomes in the population-level models, with hazard ratios ranging from 1.3 (95% confidence interval 1.3–1.4) to 2.5 (2.4–2.6). The associations attenuated in the cousin comparisons, on average 12% but with a wide range from −2% to 39% [hazard ratios ranging from 1.2 (1.1–1.4) to 1.9 (1.6–2.3)]. A dose–response relationship between the exposures and the outcomes was observed in the population-level models and the cousin comparisons, with attenuated associations in the latter.

Our findings show systematic associations between childhood household dysfunction and subsequent outcomes. Unobserved confounding likely creates upward bias in these associations, but the extent of this confounding depends on the specific exposure-outcome pairs.