Estimating heritability and shared environmental effects for refractive error in twin and family studies

4,602 subjects (1,152 monozygotic and 1,149 dizygotic twin pairs), aged 16-82

TwinsUK Adult Twin Registry

Classical twin design with maximum-likelihood variance-component modelling of autorefraction data, explicitly designed to test whether the shared family environment that family studies attribute up to a third of variance to actually exists. Equal-environments assumption applies. Adult sample, so the estimate describes the trait, not the developmental window in which schooling operates.

Best-fit heritability of spherical equivalent refractive error was 77% (95% CI 68-84%). Shared environment explained 7% (95% CI 0-15%) and unique environment 16% (95% CI 15-18%). Including age effects in the model reduced the shared-environment component to about 2%. Refractive error is therefore one of the most heritable common human traits with essentially no detectable household effect - the family-study estimates attributing a third of variance to shared environment are reinterpreted as cohort and cross-generational effects that twins, being perfectly age-matched, do not pick up.

This IS the genetic-confound source. It establishes that variation in who becomes myopic is dominated by genotype, which is why every observational correlation between schooling practices, near work, outdoor time and child myopia is presumptively confounded, and why only the randomized outdoor trials in this topic carry evidentiary weight.