Sleep Health and Anxiety Symptoms in Midlife Women: The Study of Women's Health Across the Nation (SWAN).

ABSTRACT

Purpose: To investigate the associations between anxiety symptoms in midlife women and sleep features later in life, the aim is to test the hypothesis that poor sleep, as measured by each of six individual dimensions (4 objective actigraphy measures, 2 self-reports) of sleep health, is associated with higher levels of anxiety symptoms in midlife women. Participants and Methods: The participants in this longitudinal analysis included women from the SWAN Sleep I Study, a subcohort of the community-dwelling midlife women participating in the core Study of Women's Health Across the Nation (SWAN), which was initiated in 1996. Of the 370 participants enrolled in the Sleep Study, 270 were included in the analytic sample, and 100 who did not meet the inclusion criteria were excluded. Baseline measures of six dimensions of multidimensional sleep health (actigraphy measures: efficiency, duration, mid-sleep timing, regularity; self-report measures: alertness, satisfaction) were obtained between 2003 and 2005, corresponding to SWAN core annual/biennial assessments 5-8. Associations of each dimension with self-reported anxiety symptoms (Generalized Anxiety Disorder - 7-item scale; GAD-7), collected during visits 12 (2009-2011), 13 (2011-2013), and 15 (2015-2017), were examined using mixed models. The GAD-7 outcome was measured both continuously and as a categorical variable due to its skewed distribution. Results: No statistically significant associations were found between any of the six baseline sleep health dimensions and the GAD-7 score after adjustment for covariates. Conclusion: The reasons for the lack of support for our hypothesis, despite previous evidence supporting an association between sleep and anxiety, are unclear. There is considerable overlap between anxiety and sleep symptoms, which may complicate the interpretation of our the findings. Thus, the failure to identify associations is likely multifactorial, and more studies with shorter follow-up intervals are warranted to better understand these relationships.