Personalized definition of short sleep using long-term wearable sleep distributions

To evaluate a framework using wearable data to personalize the definition of 'short sleep', comparing its temporal and functional characteristics against a fixed threshold.

462 healthy adults wore sleep trackers and provided daily ecological momentary assessments for a year. Short sleep was defined using either a fixed threshold of <6 h/night (fSS) or personalized thresholds anchored to individual sleep-duration distributions (pSS). Temporal patterns of consecutive short-sleep nights were characterized. Linear mixed-effects models examined associations between accumulating short-sleep nights and short- and long-term markers. Reproducibility was tested by evaluating short sleep patterns in seven country-level samples, including an independent Singapore sample.

Although pSS and fSS identified a similar overall proportion of short-sleep nights, the individuals and nights identified as 'short' differed between the approaches. pSS showed larger effect estimates for short-term outcomes, including alertness, sleep satisfaction, stress, motivation, sadness, sleep heart rate, sleep HRV, sedentariness and 2 levels of physical activity. Effects increased with successive short-sleep nights. These ten outcomes remained significant after Benjamini-Hochberg false discovery rate correction (q < 0.05). The proportion of pSS nights showed larger effect estimates for blood pressure and arterial stiffness. These effects also survived multiple comparison corrections. Isolated short nights were common, whereas longer runs were uncommon. Personalized thresholds distinguished stable short sleepers with few pSS nights from individuals experiencing recurrent sleep shortfall and highlighted vulnerability among those achieving 'recommended' sleep duration but with high variability. Despite marked cross-country differences in sleep habits, the distribution of short-sleep runs, and termination patterns were remarkably similar.

Anchoring short sleep to an individual's habitual sleep distribution captures relative sleep shortfall beyond absolute duration, better characterizing the functional impact of 'short' sleep. Preventive strategies may benefit from limiting pSS accumulation together with addressing sporadic inadequate sleep.

Sleep Health, Workplace Stress and Wellbeing in NUS Staff: the NUS1000 Staff Edition Study (NUS1000SE), NCT06594549, https://clinicaltrials.gov/study/NCT06594549.