Background Severe depression symptoms (depression) frequently cooccur with short and long sleep durations. Although short and long sleep are commonly thought of as a symptom of depression, a growing literature suggests they may be a cause. While each represents a process of mutual influence, directionality between them remains uncertain. Using polygenetic scores (PGS), a measure of an individuals’ genetic propensity for a trait, we investigate the direction of associations between overall sleep duration, short sleep, long sleep, and depression over time. It was hypothesised that polygenic predisposition for overall sleep duration, short sleep, and long sleep would be unidirectionally associated with the onset of depression.

Methods Data from the English Longitudinal Study of Ageing comprised participants from the general population aged 65 years on average (SD=9, range=50–99). PGSs for sleep duration, short sleep, and long sleep were calculated using UK Biobank summary statistics. PGS for depression was calculated from a consortium of genome-wide association studies (GWAS) of major depressive disorders. Sleep duration and depression (Centre for Epidemiological Studies Depression Scale) were measured at baseline and over a 12-year follow-up period (mean=9.5; SD=2.5; range=7–12). Respondents with baseline depression, or short sleep and long sleep were excluded from analyses leaving two analytic samples (n=6521/6070). All analyses were adjusted for baseline age, sex, age squared (to account for non-linearity), and genetic ancestry (to control for population stratification).

Results One standard deviation increase in PGS for short sleep was associated with 14% higher odds of incident cases of depression (CI=1.035–1.255, p=0.008). But PGSs for overall sleep duration (OR=0.916, CI=0.839–1.001, p=0.053) and long sleep (OR=0.973; CI=0.890–1.063, p=0.544) were not associated with depression onset. Across the same period, PGS for depression was not associated with overall sleep duration (β=-0.020; CI=-0.041–0.001, p=0.061), short sleep (RRR=1.055; CI=0.970–1.148, p=0.212), or long sleep (RRR=0.966; CI=0.846–1.103, p=0.607).

Conclusion In this large population-representative sample of older adults, we provide novel evidence that genetic predisposition to short sleep was strongly associated with severe depression symptoms over a 12-year period, but genetic predisposition to overall sleep duration and long sleep was not. This suggests that different mechanisms underlie the associated genetic markers in symptom onset. Evidence is also provided of directionality, since genetic predisposition to depression was not associated with sleep duration, short sleep, or long sleep. Here, we lay important groundwork for future investigations on the intersection of genetics, sleep, and depression, with insights on the genetic basis for inter-individual variation in depression.