Abstract
Investigations of Deep sleep homeostasis, the process by which the amount of Deep sleep is increased following a night of reduced sleep, often involve controlled intentional sleep deprivation experiments in service of understanding mechanistic physiology. We tested the hypothesis that a homeostatic increase in Deep sleep is detectable after relative sleep loss arising in naturalistic settings. In this retrospective observational study, we analyzed participants who provided informed consent to participate in the Apple Heart and Movement Study and elected to contribute sleep data (n = 44,564). Instances of relative sleep loss, defined as >=2 hours below each participant's median duration, occurred in 92.9% of participants, most often in isolation, and with a median duration of just over 4 hours. The Deep sleep rebound was proportional to the amount of sleep loss, for short night definitions ranging from 30 minutes to >=3 hours less. Focusing on short nights that were at least 2 hours below the median duration, 58.8% of participants showed any increase in subsequent Deep sleep, with a median increase of 12% (absolute increase of 5 minutes). In addition, the variability in Deep sleep after short nights markedly increased in a dose response manner. The Deep sleep homeostatic response showed little correlation to sleep duration, timing, consistency, or sleep stages, but was inversely correlated with Deep sleep latency (Spearman R = -0.28), another proxy for homeostatic response to sleep loss. The results provide evidence for homeostatic responses in a real-world setting. Although the Deep sleep responses to sleep loss are modest, naturalistic short nights are a milder perturbation compared to experimental sleep deprivation, and reactive behaviors potentially impacting sleep physiology are uncontrolled, leading to wide variance. The findings illustrate the utility of longitudinal sleep tracking to assess real-world correlates of sleep phenomenology established in controlled experimental settings.