Chronotype changes after sex hormone use 203 2.5. Chronotype The primary outcome of this study was change in chronotype, measured by changes in midpoint of sleep at baseline and after 3 months of GAHT and by changes in sleep duration at baseline and after 3 months of GAHT. The sleep-corrected Midpoint of Sleep on Free days (MSFsc) and sleep duration on free days were measured using the ultra-short Munich ChronoType Questionnaire (µMCTQ) (Ghotbi et al., 2020). The µMCTQ is a validated short questionnaire used to determine sleep-wake-behavior in a regular week, containing questions about sleep onset and wake times on workdays and work-free days, use of an alarm clock on free days, number of workdays, and doing shift work. We used two main outcomes from the µMCTQ: reported sleep duration and MSFsc. Additionally, we used the items on the number of work days one has in a week and use of an alarm clock on free days. Chronotype is determined based on sleep timing on work-free days, since these are assumed to be relatively free of constraints on sleep-wake behavior, such as an early work schedule. Chronotype is typically represented as the mid-point of sleep on work-free days (MSF), which is calculated based on the midpoint between the time of sleep onset and the time of sleep end on work-free days. Since people with late chronotypes tend to accumulate sleep debt throughout the week, the chronotype can be corrected for the accumulated sleep debt, creating the “sleep debt corrected” midpoint of sleep on work-free days (or MSFsc). The MSFsc is calculated by weighing the average sleep duration on work days compared to the sleep duration on work-free days for people who sleep longer on work-free days than on work days. In people who do not sleep longer on work-free days than on work days, the MSF equals the MSFsc. Sleep duration is the weighted average of the time between sleep onset and sleep offset on work days and on work-free days, corrected by the ratio of work to workfree days. 2.6. Statistical analyses R studio (version 4.0.3) was used for all statistical analyses. Data analysis was performed using linear mixed effect models using the R packages lme4 (Bates et al., 2015) and lmerTest (Kuznetsova et al., 2017). To account for
RkJQdWJsaXNoZXIy MTk4NDMw