Changes in depression symptom profile with gender-affirming hormone use in transgender persons 77 The Dutch version of the 30-item Inventory of Depressive SymptomatologySelf Report (IDS-SR; Rush et al., 1996) was used to assess depressive symptoms. It inquires about depressive symptoms in the last seven days, with items ranging from 0 to 3 (0=lowest severity, 3=highest severity). In line with the standardized scoring, items 11 and 12 and items 13 and 14 were rescored into two items (“change in appetite” and “change in weight”, respectively), resulting in a 28-item total (range 0 to 84). Cutoff scores for the IDS-SR indicate no depression (≤13 points), mild (14 to 25 points), moderate (26 to 38 points), severe (39 to 48 points), and very severe depressive symptoms (≥49 points) (Rush et al., 1996). The IDS-SR has adequate internal consistency in the cisgender population (Cronbach's alpha=0.88; Rush et al., 1996). 2.5. Statistical methods Data cleaning All statistical analyses were conducted in Rstudio (version 1.3.1093). Participants’ IDS-SR questionnaire was excluded if more than 10% of items were missing. 19 measurements had 1 or 2 missing items (17 ENIGI study measurements and 2 RESTED study measurements). Imputation of these missing items was performed using the “mice” package for Rstudio (Buuren & Groothuis-Oudshoorn, 2011). Symptom clusters To determine the symptom clusters of the IDS-SR at baseline, we used exploratory factor analysis (EFA) with maximum likelihood (ML) estimation. EFA analysis was conducted using the “psych” package (Revelle, 2017). To determine the number of factors to fit the data, we used parallel analysis, using Scree plot inspection combined with Kaiser's rule. Due to the high likelihood of between-item correlations, oblimin rotation was used for extraction. Items with a loading of 0.4 or higher were included in each factor, and if items were loading on two factors they were included in the factor with the highest loading. Four EFA models were fit with solutions for 2, 3, 4, and 5 factors. Model fit was interpreted using Tucker Lewis Indices (TLI), Root Mean Square Error of Approximation (RMSEA), and these measures are also reported in Supplementary Materials 3.3. The symptom clusters were used as subscales of the IDS-SR in subsequent analyses.
RkJQdWJsaXNoZXIy MTk4NDMw