Ires Ghielen

45 Replication factor analysis in neuropsychiatric PD patients Anxiety The BAI is a self-report instrument that consists of 21 items, through which patients can report the symptoms of anxiety that they have experienced in the previous week. Items are scored on a 4-point Likert scale, which ranges from 0 (not at all) to 3 (severely), with a total score range from 0 to 63 [17]. The cut-off score used for clinically relevant anxiety in PD is > 12 [19]. Statistical analyses All analyses were performed using SPSS Statistics 22 for Windows with a two-sided significance level of p < 0.05. The acceptability of missing values of the BAI and BDI was set to less than three missing items, i.e. 16.67%. If this criterion was not met, the patient’s data was excluded from further analysis. Mean imputation was used for residual missing data of the BAI and BDI. Data were excluded pairwise when the total MoCA score and/or the total UPDRS-III score was not available since imputation was not considered to be reliable. PCA was used to assess the dimensionality of the BAI. In PCA, items that share the most common explained variance cluster together in factors. In order to determine the number of factors that can be reliably extracted, the ‘scree plot’ criterion and the Guttman-Kaiser Eigenvalue greater-than-one rule were used. Oblimin rotation was used since it was expected that the different factors correlate with each other. The resulting factors of the BAI can be considered subscales of the BAI. The scores on these subscales were used in further analyses. The associations between the BAI and its subscales, BDI, MoCA, and UPDRS-III were investigated by conducting multiple linear regression analyses. Assumptions of normality and homoscedasticity of residuals were checked. Multicollinearity was evaluated by calculating the variance inflation factor (VIF) and investigation of the correlation matrix. In the first set of regression analyses, the total BAI score was the dependent variable. In the second set, the scores on the subscales of the BAI derived from the PCA were the dependent variables. The independent variables were the total score on the BDI, UPDRS-III, and MoCA. First, we investigated the association between the dependent and independent variables in an unadjusted model. Next, we adjusted the model stepwise for age and gender (model 1), use of dopaminergic medication (model 2), and the two other independent variables of interest, i.e. the BDI, MoCA, and UPDRS-III (model 3). 3