Sleep Characteristics and Risk of Stroke and Dementia: An Observational and Mendelian Randomization Study

Abstract

Background and objectives: Sleep disturbances are implicated as risk factors of both stroke and dementia. However, whether these associations are causal and whether treatment of sleep disorders could reduce stroke and dementia risk remain uncertain. We aimed to evaluate associations and ascertain causal relationships between sleep characteristics and stroke/dementia risk and MRI markers of small vessel disease (SVD).

Methods: We used data sets from a multicenter population-based study and summary statistics from genome-wide association studies (GWASs) of sleep characteristics and outcomes. We analyzed 502,383 UK Biobank participants with self-reported sleep measurements, including sleep duration, insomnia, chronotype, napping, daytime dozing, and snoring. In observational analyses, the primary outcomes were incident stroke, dementia, and their subtypes, alongside SVD markers. Hazard ratios (HRs) and odds ratios (ORs) were adjusted for age, sex, and ethnicity, and additional vascular risk factors. In Mendelian randomization (MR) analyses, ORs or risk ratios are reported for the association of each genetic score with clinical or MRI end points.

Results: Among 502,383 participants (mean [SD] age, 56.5 [8.1] years; 54.4% female), there were 7,668 cases of all-cause dementia and 10,334 strokes. In longitudinal analyses, after controlling for cardiovascular risk factors, participants with insomnia, daytime napping, and dozing were associated with increased risk of any stroke (HR 1.05, 95% CI 1.01-1.11, p = 8.53 × 10^-3; HR 1.09, 95% CI 1.05-1.14, p = 3.20 × 10^-5; HR 1.19, 95% CI 1.08-1.32, p = 4.89 × 10^-4, respectively). Almost all sleep measures were associated with dementia risk (all p < 0.001, except insomnia). Cross-sectional analyses identified associations between napping, snoring, and MRI markers of SVD (all p < 0.001). MR analyses supported a causal link between genetically predicted insomnia and increased stroke risk (OR 1.31, 95% CI 1.13-1.51, p = 0.00072), but not with dementia or SVD markers.

Discussion: We found that multiple sleep measures predicted future risk of stroke and dementia, but these associations were attenuated after controlling for cardiovascular risk factors and were absent in MR analyses for Alzheimer disease. This suggests possible confounding or reverse causation, implying caution before proposing sleep disorder modifications for dementia treatment.

Figure 1. Longitudinal Analyses for the Association of Sleep Measures With Stroke and Dementia

Associations are shown for sleep duration, insomnia, chronotype, daytime napping, dozing, and snoring with stroke (A–C, stroke, ischemic stroke, intracerebral hemorrhage) and dementia (D–F, all-cause dementia, vascular dementia, Alzheimer disease). The x-axis represents the hazard ratio, and solid squares indicate significant associations after FDR correction. The results are controlled for covariates including both demographic variables and cardiovascular risk factors. FDR = false discovery rate; HR = hazard ratio.

Figure 2. MR Results for the Association Between Sleep Measures With Clinical End Points (Stroke, Dementia) and SVD Imaging Traits

Red indicates a positive β coefficient while blue indicates a negative β coefficient. * indicates that the results were significant before FDR correction, and ** indicates that the results remained significant after FDR correction for multiple comparisons. AD = Alzheimer disease; AIS = all ischemic stroke; AS = all stroke; CES = cerebral embolic stroke; FA = fractional anisotropy; FDR = false discovery rate; LAS = large-artery stroke; MD = mean diffusivity; NC_SVS = neuroimaging-confirmed lacunar stroke; PSMD = peak width of skeletonized mean diffusivity; SVD = small vessel disease; SVS = small vessel stroke; WMH = white matter hyperintensity.