Heritability of sleep duration and quality: A systematic review and meta-analysis

23 papers / 19 independent samples, 45,328 twins aged 6 months to 88 years for sleep duration; 13 papers / 10 samples, 39,020 twins for sleep quality

General-population twin samples across the lifespan, with age-stratified estimates including childhood and adolescence

Systematic search of five databases with independent double screening of 5,644 abstracts; restricted to general-population twin studies. Classical twin methodology carries the usual equal-environments assumption. Actigraphy- and polysomnography-based estimates rest on few small samples and are the weakest part of the evidence; the headline numbers are mostly self- and parent-report.

Pooled heritability is 46% for sleep duration and 44% for sleep quality. The age gradient is the load-bearing result for this topic: 17% in infancy, 20-52% in childhood, 69% IN ADOLESCENCE, and 42-45% in adulthood. Heritability also varies sharply by reporter — 8% for parent-report versus 38-52% for self-report. Shared environment plays a role for childhood sleep duration but the remaining variance is mostly non-shared environment. In other words, at exactly the age at which school start times are debated, roughly seven-tenths of the variance in how long adolescents sleep is genetic.

This source IS the confound estimate. It establishes that observational associations between how much an adolescent sleeps and how well that adolescent does at school are contaminated by a trait that is ~69% heritable at that age, and that self-selected sleep duration is therefore a poor proxy for a manipulable input. It is precisely why the district-level bell-time designs matter so much: they assign the constraint independently of the genotype that would otherwise determine sleep.