Sex hormones, insomnia, and sleep quality: subjective sleep in the first year of hormone use 141 score of the PSQI (scale 0–21), which can be used as a proxy of reported sleep quality. The PSQI is used as a broader measure of sleep quality, since it also inquiries about sleep- and wake duration, and has the ability to discriminate “good” from “poor” sleepers using a cutoff score of 5: a score of 5 or lower indicates good quality sleep, while a score of higher than 5 indicates poor quality sleep (Buysse et al., 1989). Additionally, the PSQI also asks about the time spent trying to fall asleep (i.e. the sleep onset latency or SOL), the total number of hours of sleep per night (total sleep time or TST), and the total time spent in bed, which can be used to calculate sleep efficiency (or SE; i.e. the % of time in bed that is actually spent sleeping). 2.5. Lifestyle and medication Drug and alcohol use were either self-reported in questionnaires (Amsterdam) or asked by healthcare professionals and recorded in medical files (Ghent, Tel Aviv, Florence). Body Mass Index (BMI) was calculated using participants’ weight, which was measured in light indoor clothing without shoes at clinical visits using a digital floor scale, and their height, also recorded during clinical visits. Medication use was recorded in medical files. Due to the possible effects of psychotropic medications on sleep, for this study the psychotropic medication use was grouped based on WHO-defined ATC codes into the use of antidepressants (ATC code N06A), stimulants (ATC code N06B), antipsychotics (ATC code N05A), sedative medication use (ATC codes N05B and N02A), or mood stabilizers (ATC code N05AN). 2.6. Moderation variables In order to test the possible moderation effect of hot flashes on insomnia, reported hot flashes were recorded using item 1 from the menopause rating scale (MRS; Heinemann et al., 2004). This item was dichotomized into either no hot flashes (e.g. “None” or “Mild”) or hot flashes (e.g. “Moderate”, “Severe”, “Very Severe”), and included in our models as an interaction term with the duration of GAHT. In our preregistration, we also proposed to analyze interaction effects of cycle regulation use and having undergone mastectomy surgery. However, due to underpowered samples in cycle regulation analyses and the high likelihood of selection bias in mastectomy surgeries, the results of both analyses are only reported in the supplementary materials.
RkJQdWJsaXNoZXIy MTk4NDMw