Abstract
Sleep problems are common in Parkinson's disease. Several lines of research have linked insufficient sleep to neurodegenerative processes, raising the possibility that sleep disturbances may serve as aetiological risk factors for neurodegenerative diseases. However, most cohort studies on sleep traits and Parkinson's disease have been too short to disentangle risk factors from prodromal symptoms. This study investigated the association of insomnia and sleep duration with incident Parkinson's disease in a Finnish cohort with long follow-up and modelling for the competing risk of death. The study included 73 281 Parkinson's disease-free participants with a mean age of 45.7 (12.3) years from the National FINRISK Study surveys conducted every 5 years from 1972 to 2012. Insomnia (never, sometimes and often) and night-time sleep duration (<7 h, 7-8 h, >8 h) were self-reported at baseline and linked with health register data on incident Parkinson's disease. We applied both a cause-specific hazard model (Poisson) and a subdistribution hazard model (Fine-Gray) accounting for the competing risk of death. Additionally, we examined associations between polygenic risk scores for the sleep traits and Parkinson's disease. During 1 806 843 person-years of follow-up (mean 24.6 ± 12.4 years), 2679 participants developed Parkinson's disease. The cause-specific hazard model showed an association of insomnia reported as sometimes (IRR 1.14, 95% CI 1.05-1.24) and often (IRR 1.54, 95% CI 1.35-1.75) with Parkinson's disease, and a subdistribution hazard model showed an association of insomnia often (IRR 1.26, 95% CI 1.11-1.43). In sensitivity analyses excluding individuals with <20 years of follow-up, the association for insomnia symptoms often remained (IRR 1.37, 95% CI 1.13-1.66). Neither self-reported short nor long sleep duration was associated with Parkinson's disease. Accordingly, polygenic risk score for insomnia (IRR 1.13, 95% CI 1.05-1.23) but not for short or long sleep was associated with incident Parkinson's disease. In this long-term cohort study, both self-reported insomnia and genetic liability to insomnia were associated with an increased risk of incident Parkinson's disease, suggesting insomnia is a potential risk factor rather than solely a prodromal symptom. However, the attenuated association of insomnia with the cumulative incidence of Parkinson's disease in the subdistribution hazard model accounting for the competing risk of death suggests a modest real-life effect.