Factors contributing to sleep timing and sleep restriction in daily life include chronotype and less flexibility in times available for sleep on scheduled days versus free days. There is some evidence that these two factors interact, with morning types and evening types reporting similar sleep need, but evening types being more likely to accumulate a sleep debt during the week and to have greater sleep extension on weekend nights. The aim of the present study was to evaluate the independent contributions of circadian phase and weekend-to-weekday variability to sleep timing in daily life. The study included 14 morning types and 14 evening types recruited from a community-based sample of New Zealand adults (mean age 41.1 ± 4.7 years). On days 1-15, the participants followed their usual routines in their own homes and daily sleep start, midpoint and end times were determined by actigraphy and sleep diaries. Days 16-17 involved a 17 h modified constant routine protocol in the laboratory (17:00 to 10:00, <20 lux) with half-hourly saliva samples assayed for melatonin. Mixed model ANCOVAs for repeated measures were used to investigate the independent relationships between sleep start and end times (separate models) and age (30-39 years versus 40-49 years), circadian phase [time of the dim light melatonin onset (DLMO)] and weekday/weekend schedules (Sunday-Thursday nights versus Friday-Saturday nights). As expected on weekdays, evening types had later sleep start times (mean = 23:47 versus 22:37, p < .0001) and end times (mean = 07:14 versus 05:56, p < .0001) than morning types. Similarly on weekend days, evening types had later sleep start times (mean = 00:14 versus 23:07, p = .0032) and end times (mean = 08:56 versus 07:04, p < .0001) than morning types. Evening types also had later DLMO (22:06 versus 20:46, p = .0002) than morning types (mean difference = 80.4 min, SE = 18.6 min). The ANCOVA models found that later sleep start times were associated with later DLMO (p = .0172) and weekend-to-weekday sleep timing variability (p < .0001), after controlling for age, while later sleep end times were associated with later DLMO (p = .0038), younger age (p = .0190) and weekend days (p < .0001). Sleep end times showed stronger association with DLMO (for every 30 min delay in DLMO, estimated mean sleep end time occurred 14.0 min later versus 10.19 min later for sleep start times). Sleep end times also showed greater delays on weekends versus weekdays (estimated mean delay for sleep end time = 84 min, for sleep start time = 28 min). Comparing morning types and evening types, the estimated contributions of the DLMO to the mean observed differences in sleep timing were on weekdays, 39% for sleep start times and 49% for sleep end times; and on weekends, 41% for sleep start times and 34% of sleep end times. We conclude that differences in sleep timing between morning types and evening types were much greater than would be predicted on the basis of the independent contribution of the difference in DLMO on both weekdays and weekend days. The timing of sleep in daily life involves complex interactions between physiological and psychosocial factors, which may be moderated by age in adults aged 30-49 years.
