Ires Ghielen

31 BAI factor analysis in PD patients S1. The confounding variables were age and the BDI, SCOPA-AUT and UPDRS-III score. Although, as demonstrated in Table 2, these variables were all significantly correlated with each other, VIF remained within acceptable bounds. Discussion In this study, clinically relevant symptoms of anxiety were present in 45% of patients in our sample, which corresponds with previously reported rates [1, 2]. PCA of the BAI revealed one affective and four somatic symptom dimensions ( thermoregulation, hypotension, hyperventilation and trembling ) of anxiety. The finding of a distinct affective factor and multiple somatic factors is in line with previous research on dimensionality of the BAI in non-PD samples [11, 13, 18, 19]. In these studies, less than five factors were found, probably because the samples consisted of psychiatric patients with a mean age under 40 years. These patients are less likely to experience somatic symptoms than the patients in our sample, who have a variety of PD-related motor and non-motor symptoms that overlap with somatic anxiety equivalents. Moreover, differences in the methodology of the factor analysis, such as the choice of rotation method, can lead to a different number of obtained factors. The BDI significantly influenced all subscales of the BAI, as well as the BAI total score. Moreover, there was a high correlation between the BAI and BDI. This suggests that the large overlap between anxiety and depression that is found in psychiatric patients [20], is present in PD patients as well. In previous studies, anxiety and depressive disorders coexisted in 19 to 40% of patients with PD, which is higher than in matched controls [21-23].Autonomic symptoms significantly influenced the total score on the BAI and on the hypotension, thermoregulation and hyperventilation subscales, while severity of motor symptoms influenced the score on the trembling subscale. The most obvious explanation for these observations is that the BAI is designed to measure episodic anxiety, such as in panic disorder, which is accompanied by somatic equivalents of anxiety [19]. Alternatively, symptoms of PD-related autonomic and motor dysfunction might be misinterpreted as anxiety, resulting in overdiagnosis of anxiety disorders in PD patients [24]. In this study, we demonstrated that the score on the affect subscale was not influenced by the SCOPA- AUT or the UPDRS-III score. One might therefore conclude that the affect subscale constitutes the most reliable measure of anxiety in PD, because it is not affected by ‘noise’ generated by motor and autonomic symptoms. The use of questionnaires excluding somatic symptoms for the screening of psychiatric disorders in the elderly has been advocated for this reason [25]. In PD patients, however, attempts 2