P019 No mythical effect of moon cycles on sleep

Abstract Introduction The lunar cycle effects behaviours in invertebrates. In humans, lunar cycle effects on sleep have been reported, but with mixed and inconclusive findings. This study sought to leverage a very large longitudinal consumer device dataset to more definitively investigate potential moon cycle effects on human sleep. Methods We used sleep data recorded between March 2017 and September 2023 from two consumer devices, using location and date data to determine moon phase and total sleep time from all available data covering at least 80% of a moon cycle. One device was a pneumatic under-mattress bed sleep sensor with 124,544 users and 81 million nights of data. The other device was a wrist-worn sleep sensor with 232,576 users and 96 million nights of data. Both a Cosinor and Genearlized Additive Model (GAM) were used to test for moon phase effects on sleep. Results Univariate analysis using the Cosinor model showed no discernible moon effect, a poor model fit for over 99% of users (effect size <0.33 hrs and R-squared <0.25) and a random distribution of the cosine peak across the moon cycle. GAM adjusted for multiple confounders also showed no evidence of moon phase effects on total sleep time (effect size <0.02 hrs). Multiple sensitivity analyses with different sleep related outcomes and model structures showed no evidence of moon phase effects on sleep. Conclusions Comprehensive analysis of two very large independent data sources of sleep time estimates do not support any moon cycle effects on total sleep time.