Ires Ghielen

67 Replication factor analysis in neuropsychiatric PD patients Network comparison analysis The networks for both groups are displayed in figure 1. In total, the network represents the partial correlations between 50 items (i.e. motor and anxiety symptoms) and since it concerns a parsimonious network, all visual correlations are considered relevant. The network architecture of both the high-anxiety and low-anxiety group shows clustering of motor symptoms (i.e. UPDRS-III items) on the left side and clustering of anxiety symptoms (i.e. BAI items) on the right side. Resting tremor as well as action tremor items are somewhat separated from the rest of the UPDRS-III items. The symptoms within the networks of both patient groups are mainly positively correlated, which is represented by the green connective lines. There is a significant difference between the high-anxiety and low-anxiety patient group networks in global strength (S=7.33, p <0.001), with the high-anxiety group showing higher global strength (20.16) compared to the low-anxiety group (12.83). This difference is mostly driven by the higher global strength (i.e. more and stronger connections) amongst the BAI items, as can be seen in figure 1. Also, the connections between the UPDRS-III items appear stronger in the high-anxiety patient group. Explorative visual inspection of the specific connections between motor and anxiety symptoms shows correlations between BAI item 12 (i.e. trembling hands) and resting tremor items of the UPDRS-III. Also, BAI item 8 (i.e. unsteadiness) is associated with UPDRS-III items related to gait and posture. We found no statistically significant differences between the two patient groups concerning correlations between the motor and anxiety items. Concerning the stability of the networks, bootstrapped confidence intervals are wide (figures S1 and S2 in the supplementary materials) in both patient group networks. Discussion We investigated the differences in symptom network architecture focusing on the connections between motor and anxiety symptoms in PD patients with high levels of anxiety versus PD patients with low levels of anxiety. The high-anxious compared to the low-anxious PD group networks differed in global strength. This was expected, since we selected the groups on the amount and severity of anxiety symptoms, resulting in more and stronger connections especially amongst the BAI items of the high-anxiety group network. Previous research 4