Seasonality of human sleep: Polysomnographic data of a neuropsychiatric sleep clinic

Abstract

While short-term effects of artificial light on human sleep are increasingly being studied, reports on long-term effects induced by season are scarce. Assessments of subjective sleep length over the year suggest a substantially longer sleep period during winter. Our retrospective study aimed to investigate seasonal variation in objective sleep measures in a cohort of patients living in an urban environment. In 2019, three-night polysomnography was performed on 292 patients with neuropsychiatric sleep disturbances. Measures of the diagnostic second nights were averaged per month and analyzed over the year. Patients were advised to sleep "as usual" including timing, except alarm clocks were not allowed. Exclusion criteria: administration of psychotropic agents known to influence sleep (N = 96), REM-sleep latency > 120 min (N = 5), technical failure (N = 3). Included were 188 patients: [46.6 ± 15.9 years (mean ± SD); range 17-81 years; 52% female]; most common sleep-related diagnoses: insomnia (N = 108), depression (N = 59) and sleep-related breathing disorders (N = 52). Analyses showed: 1. total sleep time (TST) longer during winter than summer (up to 60 min; not significant); 2. REM-sleep latency shorter during autumn than spring (about 25 min, p = 0.010); 3. REM-sleep longer during winter than spring (about 30 min, p = 0.009, 5% of TST, p = 0.011); 4. slow-wave-sleep stable winter to summer (about 60-70 min) with 30-50 min shorter during autumn (only significant as % of TST, 10% decrease, p = 0.017). Data suggest seasonal variation in sleep architecture even when living in an urban environment in patients with disturbed sleep. If replicated in a healthy population, this would provide first evidence for a need to adjust sleep habits to season.

Keywords: REM; circadian; clinical data; human; polysomnography; season; sleep; slow wave sleep.

FIGURE 1

Moving averages [window: 90 days; slide: 7 days (centered on Thursdays)] for selected sleep parameters. Displayed are: total sleep time (TST), rapid eye movement sleep time (REM), and slow wave sleep time (SWS) in minutes (min). Dotted vertical lines indicate the meteorological start for each season (winter, spring, summer, and autumn). Dotted horizontal line indicates a break in the y-axis. Residuals of all sleep parameters are not independently distributed (Ljung-Box Test: p < 0.001, lag: 1–30 weeks) and therefore exhibit serial correlation. * y-axis: REM’s monthly means show a significant effect for “month of record” within a mixed linear model analysis of variance. * x-axis: The months’ post hoc displays a significant difference to at least one other month.

FIGURE 2

Moving averages [window: 90 days; slide: 7 days (centered on Thursdays)] for selected sleep parameters. Displayed are total sleep time in min (TST, left), rapid eye movement as a percentage of TST (REM, middle), and slow wave sleep as a percentage of TST (SWS, right). Gray outward-going lines represent calendar weeks, with the last Thursday of the month being thicker. *The sleep parameters’ monthly means show a significant effect for “month of record” within a mixed linear model analysis of variance. **The months’ post hoc displays a significant difference to at least one other month.