heterogeneity including sex differences and disease subtypes in the overall sample, and the loss of power in stratified analyses that use only a portion of the sample. Likewise, Martin and colleagues 95 hypothesized that under a liability threshold model, females with ADHD would have a higher genetic threshold (as represented by higher PRS) than do males with ADHD. However, no significant sex differences in the SNP‐based heritability of ADHD was found (female h 2 SNP = 0.123, SE = 0.025; male h 2 SNP = 0.247, SE = 0.021), nor was there an association between ADHD PRS with sex in cases, and the odds ratio for the PRS was the same in males and females. Consistent with expectations of the multifactorial polygenic threshold model they did find an increased risk of ADHD in relatives of females compared with males in a separate registry analysis. 95 Reconciliation of discrepant findings from genomic and genetic epidemiologic analyses in these studies of depression and ADHD and other conditions will be an important future direction to gain understanding of the mechanisms underlying sex differences.
