Patterns of Sleep Duration and Metabolic Biomarkers Across the Menstrual Cycle.

Abstract

Along the menstrual cycle, associations between inconsistent sleep duration and levels of metabolic biomarkers are uncertain and could involve fluctuations in estrogen concentrations. To examine associations between patterns of sleep duration and metabolic biomarkers across two menstrual cycles within a cohort of premenopausal women. The BioCycle Study was conducted in New York between 2005-2007, enrolling 259 premenopausal women over two menstrual cycles. This micro-longitudinal cohort study involved intensive data collection including daily sleep diaries and biomarker assessments of leptin, insulin, and glucose at 16 key points timed to menstrual cycle phases. We considered dynamic sleep duration, as hours slept one night or as mean hours slept during the two nights prior to each biomarker assessment. Variability in habitual sleep duration, i.e., reported daily sleep duration, summarized across both menstrual cycles. Variation in habitual sleep duration was computed using L-moments, a robust version of dispersion, skewness, and kurtosis. To examine associations between patterns of sleep duration and metabolic biomarkers, we fitted a series of linear mixed models with random intercepts and inverse probability weighting. These models were adjusted for potential demographic, lifestyle, health confounders, and menstrual cycle phase. Sleep duration one night or two nights prior to clinic visits were not associated with metabolic biomarker measures we assessed. However, overall variability (dispersion) in habitual sleep duration was associated with lower mean insulin HOMA-IR levels, but not glucose. Moreover, extreme short or long bouts of sleep duration was associated with higher mean levels of leptin, insulin, and HOMA-IR. These data suggest that variation in habitual sleep duration along the menstrual cycle may be associated with metabolic function.