Association Between Insomnia, Stress Events, and Other Psychosocial Factors and Incident Atrial Fibrillation in Postmenopausal Women: Insights From the Women's Health Initiative

Abstract

Background The association between psychosocial factors and atrial fibrillation (AF) is poorly understood. Methods and Results Postmenopausal women from the Women's Health Initiative were retrospectively analyzed to identify incident AF in relation to a panel of validated psychosocial exposure variables, as assessed by multivariable Cox proportional hazard regression and hierarchical cluster analysis. Among the 83 736 women included, the average age was 63.9±7.0 years. Over an average of 10.5±6.2 years follow-up, there were 23 954 cases of incident AF. Hierarchical cluster analysis generated 2 clusters of highly correlated psychosocial variables: the Stress Cluster included stressful life events, depressive symptoms, and insomnia, and the Strain Cluster included optimism, social support, social strain, cynical hostility, and emotional expressiveness. Incident AF was associated with higher values in the Stress Cluster (hazard ratio [HR], 1.07 per unit cluster score [95% CI, 1.05-1.09]) and the Strain Cluster (HR, 1.03 per unit cluster score [95% CI, 1.00-1.05]). Of the 8 individual psychosocial predictors that were tested, insomnia (HR, 1.04 [95% CI, 1.03-1.06]) and stressful life events (HR, 1.02 [95% CI, 1.01-1.04]) were most strongly associated with increased incidence of AF in Cox regression analysis after multivariate adjustment. Subgroup analyses showed that the Strain Cluster was more strongly associated with incident AF in those with lower traditional AF risks (P for interaction=0.02) as determined by the cohorts for heart and aging research in genomic epidemiology for atrial fibrillation score. Conclusions Among postmenopausal women, 2 clusters of psychosocial stressors were found to be significantly associated with incident AF. Further research is needed to validate these associations.

Keywords: atrial fibrillation; hierarchical cluster analysis; psychosocial clusters; strain; stress; women's health.

Figure 1. Dendrogram of grouping individual psychosocial constructs into 2 clusters by hierarchical cluster analysis.

Figure 2. Kaplan‐Meier AF‐free survival by quartiles of the Stress Cluster (A) and the Strain Cluster (B).

AF indicates atrial fibrillation; and WHI, Women's Health Initiative.

Figure 3. Continuous psychosocial clusters on incident atrial fibrillation (AF), stratified by baseline subgroups.

Subgroup hazard ratios (HRs), 95% CIs, and interaction P values are from a proportional hazards regression, with AF as a function of the Stress Cluster or the Stress Cluster by subgroup interaction, the Strain Cluster, and the Strain Cluster by subgroup interaction, stratified by Women's Health Initiative (WHI) component (clinical trial/observational study), and adjusted for age, ethnicity, race, education, waist–hip ratio (WHR), physical activity, smoking, alcohol use, hypertension, diabetes, heart failure, and myocardial infarction (MI). *HRs and corresponding CIs are for an increase of 1 point in the given cluster score. ^† P value is for the subgroup by cluster interaction. ^ǂCluster by age interaction terms from a separate model with linear trend over age groups, the Stress Cluster, Stress Cluster by linear trend over age groups interaction, the Strain Cluster, and Strain Cluster by linear trend over age groups interaction. ^§ White=non‐Hispanic White; Black=non‐Hispanic Black; Hispanic=Hispanic, all races; Asian=non‐Hispanic Asian. ^ǁHistory (Hx) of coronary artery disease (CAD) subgroup model is not adjusted for history of MI. ^#Cohorts for heart and aging research in genomic epidemiology for atrial fibrillation (CHARGE‐AF) (a validated score, , , encompassing traditional AF risk factors to predict incident AF within 5 years in diverse patient populations) subgroup model is stratified by WHI component (clinical trial/observational study) and adjusted for ethnicity, race, education, WHR, physical activity, and alcohol use. Ann%, annual percent event rate.