Ires Ghielen

89 Longitudinal relationships between anxiety and motor symptoms FoF and anxiety together as independent variables to investigate which variable is contributing the most to the association with FoG, the regression coefficient of the contribution of anxiety decreased. In the small but evident contribution of FoF to FoG (analysis 1), the confounding effects of disease characteristics and medication and adverse effects may be explained by the fact that disease characteristics influence both FoG and FoF, and that FoG may occur in the ‘on’, ‘transition’ and ‘off’ state [8]. In their cross-sectional study comparing PD patients (n=58) with age-matched healthy controls, Adkin et al. [49] reported that FoF correlated positively with gait and balance instability in PD. Current FoG treatment proves to be insufficient, since 70% [8] of the fall incidents occur during a FoG episode. Clearly, FoF and FoG are highly related to one another, and it appears that the state (‘on’ and ‘off’) plays an important role in this association. The positive association between FoG and anxiety (analysis 2) remained significant after adjusting for demographics; disease characteristics; medication and adverse effects. This finding, using a longitudinal design, is in line with the experimental study of Ehgoetz Martens et al. [3], who showed that anxiety directly resulted in FoG. Using a cross-sectional design, Lieberman [50] found a correlation between the presence of FoG, anxiety and panic scores in PD patients (n=109). In our PD cohort, a quarter of the patients reported clinically relevant anxiety symptoms, which rises to the assumption that even without high anxiety scores, the impact of anxiety on FoG is evident and might even cause episodes of FoG. When analysing the multivariate association of FoF and anxiety with FoG (analysis 3), the contribution of anxiety diminished as the contribution of FoF remained the same in comparison with the univariate associations. Therefore, the univariate association between FoG and anxiety is largely explained by FoF. One might speculate about the direction of the associations, where the most logical direction resulting from our analyses would be from anxiety to FoF to FoG. As Rahman and colleagues [17] showed, anxiety is a predictor for avoidance behaviour in PD patients who experience FoF. FoF [11] and anxiety [51] both can lead to avoidance and restriction of activities and social life, and we speculate that this behavior will aggravate FoF and lead to even more physical decline. Further research is necessary to investigate the possible mediating effect of anxiety on the relationship between FoF and FoG. We found that crude associations between FoG and FoF and between FoG and anxiety (analyses 1 & 2) remain significant after adjustment for demographics and disease characteristics. Adjusting the multivariate model (analysis 3) for 5