Ires Ghielen

87 Longitudinal relationships between anxiety and motor symptoms Figure 2. Crude analyses results of the univariate associations 1) between FoG and FoF, 2) between FoG and anxiety, and the multivariate association 3) between FoG (dependent) and FoF and anxiety (independent variables). Abbreviations: FOGQ = Freezing of Gait Question- naire; FES = Falls Efficacy Scale; HADSA = Hospital Anxiety and Depression Scale; β = beta regression coefficient; FoG = Freezing of Gait; FoF = Fear of Falling. The regression coefficient between FoG and FoF (analysis 1) showed the greatest change when adjusted for disease characteristics. The greatest change in regression coefficient in the association between FoG and anxiety (analysis 2) was seen after adjustments for medication and adverse effects. Table 3 shows the results of all crude and adjusted univariate associations. When analysing FoF and anxiety together as independent variables (analysis 3, see table 4), the association between FoG and FoF remained as in the univariate analysis (analysis 1), while the association between FoG and anxiety weakened (as compared to analysis 2). In this multivariate association model, as in the separate univariate associations, the regression coefficient between FoF and FoG decreased the most after adjustment for disease characteristics. In addition, the regression coefficient between anxiety and FoG decreased the most after adjustment for medication and adverse effects, as was also seen in the univariate association. Discussion To our knowledge, this is the first study to investigate the longitudinal associations between FoG, FoF, and anxiety in a large cohort of PD patients. As expected, crude associations between FoG and FoF, and between FoG and anxiety were positive and significant. Adjustments for disease characteristics, medication and adverse effects showed the greatest decrease in the strength of these associations. When analysing 5