Familial factors inflate the estimated effects of genetic liability to ADHD on behavioural outcomes

Background: People with ADHD have worse health, behavioural and social outcomes, widely interpreted as evidence that ADHD causes these harms, justifying early identification and treatment. But ADHD clusters in families, and conventional studies cannot separate the effect of a person's own ADHD from the family environment in which it arises. We used a family-based genetic design to distinguish the two. Methods: We used up to 22 genome-wide significant variants associated with ADHD (38,691 cases, 186,843 controls) to estimate effects of genetic liability to ADHD on 16 health and behavioural outcomes. We compared population-based estimates with family-based estimates from within-sibship analyses of up to 164,220 siblings. We tested for heterogeneity between the designs, and applied sensitivity analyses. We involved individuals with lived experience of ADHD in interpreting the findings and shaping how the clinical/public-health implications were described. Findings: Population-based analyses implied that greater genetic liability to ADHD increased smoking, alcohol use and depressive symptoms, and reduced educational attainment, age at first child and cognitive performance. In family-based analyses, these effects attenuated. For example, the effect on number of depressive symptoms fell from 0.059 SD (95% CI 0.039, 0.080) to 0.002 SD (-0.052, 0.055), and age at first child increased from -0.045 SD (-0.058, -0.033) to 0.003 SD (-0.040, 0.047), with significant differences between designs (p=0.048 and p=0.034). Findings were consistent across sensitivity analyses. Interpretation: Much of the excess health and behavioural burden attributed to ADHD may reflect the family and environmental context in which ADHD occurs, rather than an individual's own ADHD. Interventions targeting the individual might not reduce these outcomes as much as expected. Distinguishing such effects requires family-based designs. Funding: Medical Research Council, UK Research and Innovation, Research Council of Norway, US National Institute of Mental Health, and Australian National Health and Medical Research Council.