Chapter 5 142 2.7. Statistical analyses Statistical analyses were conducted using Rstudio (version 4.1.2), combined with the lme4 package (Bates et al., 2015) and the lmerTest package (Kuznetsova et al., 2017). We used descriptive statistics to describe the prevalence of insomnia and poor sleep at the start of GAHT in both trans men and trans women, describing the number of participants qualifying for scores indicating (subclinical) insomnia and poor sleep quality using total numbers and percentages. To analyze the changes in ISI and PSQI scores, we used linear mixed models with a random intercept per participant and a random intercept per center, to account for the repeated measures per participant and nested participants within the participating centers. Analyses were conducted separately for trans men and trans women. Missing values in the PSQI and ISI variables were deleted listwise per model analysis, and results were interpreted using the fixed effect estimates and 95% confidence intervals. In the first unadjusted model, time of measurement was used as the only fixed predictor in the unadjusted model. Secondly, the variables age, BMI, alcohol use (categorized into 0 drinks per week, 1 to 7 drinks per week, or more than 7 drinks per week), and psychotropic medication use (dichotomized in use and no use) at each measurement point were incorporated as fixed factors in the models to see if they improved model fits. This resulted in a final adjusted model, which included alcohol use and psychotropic medication use as confounder variables, also displayed below as the adjusted model (see Table 5.1). For post-hoc analyses, we tested the aforementioned moderators as interaction terms together with the time of measurement in separate models to see whether the interaction term affected the association between sleep outcomes and the duration of GAHT use. This resulted in the models as reported in Table 5.1, analyzed in trans men and women separately.
RkJQdWJsaXNoZXIy MTk4NDMw