Little is known about human entrainment under natural conditions, partly due to the complexity of human behavior, torn between biological and social time and influenced by zeitgebers (light-dark cycles) that are progressively "polluted" (and thereby weakened) by artificial light. In addition, data about seasonal variations in sleep parameters are scarce. We, therefore, investigated seasonal variation in cross-sectional assessments of sleep/wake times of 9765 subjects from four European populations (EGCUT = Estonian Genome Centre, University of Tartu in Estonia; KORA = Cooperative Health Research in the Region of Augsburg in Germany; KORCULA = The Korcula study in Croatia; and ORCADES = The Orkney Complex Disease Study in Scotland). We identified time-of-year dependencies for the distribution of chronotype (phase of entrainment assessed as the mid-sleep time point on free days adjusted for sleep deficit of workdays) in cohorts from Estonia (EGCUT) and Germany (KORA). Our results indicate that season (defined as daylight saving time - DST and standard zonetime periods - SZT) specifications of photoperiod influence the distribution of chronotype (adjusted for age and sex). Second, in the largest investigated sample, from Estonia (EGCUT; N = 5878), we could detect that seasonal variation in weekly average sleep duration was dependent on individual chronotype. Later chronotypes in this cohort showed significant variation in their average sleep duration across the year, especially during DST (1 h advance in social time from the end of March to end of October), while earlier chronotypes did not. Later chronotypes not only slept less during the DST period but the average chronotype of the population assessed during this period was earlier than during the SZT (local time for a respective time zone) period. More in detail, hierarchical multiple regression analyses showed that, beyond season of assessment (DST or SZT), social jetlag (SJl; the discrepancy between the mid sleep on free and work days - which varied with age and sex) contributed to a greater extent to the variation in sleep duration than chronotype (after taking into account factors that are known to influence sleep duration, i.e. age, sex and body mass index). Variation in chronotype was also dependent on age, sex, season of assessment and SJl (which is highly correlated with chronotype - SJl was larger among later chronotypes). In summary, subjective assessments of sleep/wake times are very reliable to assess internal time and sleep duration (e.g. reproducing sleep duration and timing tendencies related to age and sex across the investigated populations), but season of assessment should be regarded as a potential confounder. We identified in this study photoperiod (seasonal adaptation) and SJl as two main factors influencing seasonal variation in chronotype and sleep duration. In conclusion, season of assessment, sex and age have an effect on epidemiological variation in sleep duration, chronotype and SJl, and should be included in studies investigating associations between these phenotypes and health parameters, and on the development of optimal prevention strategies.
