O015 Longer wake predicts shorter subsequent sleep across ~450 000 free-living episodes

Abstract Background Sleep duration is jointly regulated by the circadian pacemaker and a homeostatic process that promotes increased sleep propensity with longer wakefulness. Counterintuitively, longer periods of wakefulness typically precede shorter sleep durations for individuals in time-isolated environments. Sleep episodes are also shorter when initiated closer to the circadian minimum for these individuals. Whether these phenomena generalize to free-living individuals is not clear. Methods We analysed ~11 million hours of data from accelerometers and light sensors worn by 85 955 free-living UK Biobank participants (62.3 ± 7.9 years, 58% female). We identified ~450 000 sleep–wake episodes, estimating their associated circadian minima by inputting light data into a computational model of the central circadian pacemaker. Results Longer wakefulness was associated with shorter subsequent sleep duration (linear mixed model fit, p<.0001). A one-hour longer wake duration predicted a 0.42-hour shorter subsequent sleep duration. Sleep duration also depended upon estimated circadian phase at sleep onset. Sleep episodes initiated at -8, -4, 0, and 4 hours since the estimated circadian minimum were 10.23, 7.53, 5.06, and 3.13 hours in duration, respectively (mixed model fitted values; p<.0001). These relationships were independent of age, sex, and employment. Discussion Longer wakefulness predicted shorter subsequent sleep duration, and sleep duration depended on estimated circadian phase of sleep onset, in free-living individuals. These findings are strikingly similar to observations of sleep–wake patterns under time-isolation. Extending sleep duration may be easier to achieve by targeting an earlier bedtime in preference to a later wake up time.