0262 Empirically-Derived Subgroups in Self-Reported Sleep Outcomes Across the Menstrual Cycle

Abstract Introduction While the extant research suggests objective sleep measures do not appear to vary across the menstrual cycle, there is mixed support for a modest decrease in self-reported sleep quality in the luteal phase. However, emerging evidence on hormone sensitivity suggests strong individual differences in how hormonal flux across the cycle impacts affect and behavior, giving rise to distinct temporal symptom patterns. This dimensional framework may explain prior mixed results or modest effect sizes. The current study examined potential subgroups of change in self-reported sleep outcomes across the menstrual cycle. Methods Methods: Participants were N=125 naturally-cycling females recruited for elevated borderline personality disorder symptoms to participate in an ongoing study examining ovarian steroid effects on mood outcomes across the cycle. Participants completed daily diaries every morning for 2 menstrual cycles which included questions asking about sleep the prior night (time to fall asleep, hours slept, number of awakenings, overall sleep quality). They also completed LH-surge tests to determine ovulation. Cycles were scaled using the menstrualcycleR package in R. Generalized Additive Mixed Models (GAMMs) were conducted to examine fluctuation in sleep outcomes across the scaled cycle in the full sample; Smoothing Mixture Models (SMMs) were conducted to test for empirically-derived subgroups. Results In full-sample GAMMs, only sleep duration varied significantly across the cycle (luteal peak), but substantial heterogeneity emerged in both baseline levels and deviations from the average trajectory across all outcomes. SMMs identified distinct subgroups for each outcome: four for time to fall asleep (modest post-ovulatory increase, perimenstrual peak, periovulatory peak, late-luteal peak); three for sleep quality (periovulatory, midluteal, perimenstrual peaks); and two each for sleep duration (modest perimenstrual increase, midluteal peak) and awakenings (perimenstrual, midluteal peaks). Subgroups differed on baseline levels, timing, severity of fluctuations, with most displaying significant nonlinear trends. Conclusion Results demonstrate distinct subgroups of patterns of change in sleep outcomes across the cycle, highlighting the importance of considering the cycle in clinical practice. Failure to examine these individual differences when modeling sleep across the cycle will potentially obscure effects. Future directions include applying these approaches to objective sleep measures. Support (if any) K23MH136325 (Allison Stumper) R01MH126940 (Jessica Peters)