Cortisol dynamics and sleep quality: the role of sex and oral contraceptive use 111 comparison to also compare sleep quality and sleep disruptions between NC women and men as a secondary outcome measure. To address this aim, sleep quality (PSQI sum score) was compared between the groups using linear regression models. The distribution of the reported sleep disruption (PSQI item 5B) between the groups (OC users, NC women, men) were compared using a probabilistic index model (PIM) (Thas et al., 2012). These PIMs are used to estimate the probabilistic index (denoted π) which can be interpreted here as the probability that a randomly selected person from one group will score higher, and thus have more frequent sleep disruptions, than a randomly selected person from the reference group. π=0.5 corresponds to the null hypothesis: i.e., lack of between-group differences whereas π=0 or 1 indicates perfect separation between the groups. Each analysis was conducted without covariates (the marginal model) and with an additional covariate for age (the age-controlled model). For the analyses comparing the OC and NC groups, which was the primary comparison aim, we adjusted the resulting p-values from the age-controlled model for two repeated tests. To assess whether CAR AUCi were different between the groups, we used linear regression analyses. We conducted adjusted analyses comparing the CAR between groups through addition of age, smoking, day of saliva sampling (i.e., work day, free day or test day) and cortisol analysis batch as covariates to the linear regression models, creating the covariate-controlled model. The second aim of this study was to conduct a mediation or moderation analysis to examine the role of the CAR within groups in sleep quality and sleep disruptions. Due to the lack of significant group differences in sleep disruptions in the OC users vs. NC women and the men vs. NC women, as well as in sleep quality in the men vs. NC women, we opted to conduct a moderation analysis assessing the role of the CAR in these differences. The association between reported sleep disruptions and CAR were compared between the groups by using a PIM analysis with PSQI item 5B as the outcome variable, and the group (NC women compared to OC users and men) in interaction with CAR as predictors. Associations between sleep quality and the CAR were compared between the groups using a linear regression analysis with the PSQI sum score as the outcome variable, and the group (NC women compared to men) in interaction with the CAR as
RkJQdWJsaXNoZXIy MTk4NDMw