Daylight Saving Time, Acute Sleep Disruption, and Return-to-Baseline Patterns in All of Us Fitbit Data
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Seasonal daylight saving time (DST) transitions abruptly alter the alignment between clock time, environmental light, and sleep. In preregistered analyses, we used longitudinal Fitbit data from the All of Us Research Program to compare sleep around 13 seasonal transitions in Colorado, New Mexico, and Utah with sleep on the same dates in Arizona, where clocks do not change. The sample included 1,522 participants and 6,658 participant-transition windows. Following the spring transition, sleep duration changed more from one night to the next in Mountain states than in Arizona (14.6 minutes, 95% CI [7.2, 22.0]), even though sleep duration itself on day +1 showed little evidence of a DST-related reduction (-8.4 minutes, 95% CI [-17.3, 0.6]). Sleep timing shifted later after the spring transition and about 28 minutes earlier after the fall transition. Among Mountain-state observations with a large post-transition departure from usual sleep duration, the median return to the usual range was two nights. Nearly the same two-night pattern appeared in Arizona after similarly large sleep departures, indicating that this return pattern was not specific to DST. Greater baseline sleep-duration variability predicted a greater likelihood of remaining outside the usual range at days +7 and +14. Seasonal clock changes produced acute sleep disruption, while the Arizona comparison shows that a rapid return to baseline should not be interpreted as the duration of recovery from DST itself.
Statement of Significance
Seasonal daylight saving time changes are often studied with before-and-after comparisons, which can mix clock-change effects with ordinary variation across the week. Using Arizona as a calendar-matched no-clock-change comparison, we found that the Mountain-versus-Arizona difference in sleep duration on spring day +1 was similar to that on day -1, while the change from the preceding observed night was larger in Mountain states. In a separate Mountain-state analysis, large post-transition departures usually returned to the pre-transition ±30-minute window within two nights. The same return pattern appeared on the same spring and fall dates in Arizona as well. This distinction shows why acute exposure effects and threshold-defined return trajectories should be evaluated separately in wearable studies.
