Chapter 7 204 repeated measures in the same person, a random intercept was used for each subject. Analyses were conducted separately in the TM and TF groups. To estimate the changes in midpoint of sleep and sleep duration after 3 months of GAHT, we used the measurement time point (e.g. baseline or 3-month follow-up) as a fixed predictor in the unadjusted model, as displayed below. Secondly, we incorporated a possible confounder (work status: working more or less than 3 days a week) into account in the adjusted model. Thirdly, to account for the use of alarm clocks in the cohort, we conducted a sensitivity analysis where we included only the subgroup of the cohort who reported not using an alarm clock on free days. All models are displayed below. Unadjusted model: Outcome a. ~ measurement time point + (1|Participant ID) Adjusted model: Outcome a. ~ measurement time point + Work status b. + (1|Participant ID) Sensitivity model in subgroup c.: Outcome a. ~ measurement time point + Work statusb. + (1|Participant ID) a. The following outcomes were tested: sleep duration, MSFsc b. Work status: whether participant reports working or going to school 3 or more days per week or working or going to school less than 3 days per week. c. This model was conducted on a subgroup of participants who reported not using an alarm on free days. 3. Results 3.1. Demographic characteristics The sociodemographic characteristics of the entire study sample at baseline and after 3 months of GAHT are reported in Table 7.1. At baseline, TM participants had a median age of 23 ± 5 years and 63% had more than 3 days of work or school per week. At the start of GAHT, testosterone gel was the most commonly utilized type of GAHT in TM participants (90%), short-acting testosterone injections were used by the rest of the group (10%), and 44% of TM participants used cycle regulation medication, of
RkJQdWJsaXNoZXIy MTk4NDMw