Ires Ghielen

66 Chapter 4 It is important to notice that, considering this statistical method, our sample sizes are small and these analyses should therefore be considered explorative [17]. Results Subjects Of the 649 assessed PD patients, 80 were excluded due to missing data. Using the cut-off score of 12 on the BAI, 316 patients were assigned to the high-anxiety group and 253 patients were assigned to the low-anxiety group. We initially matched the two patient groups on age and sex using the matchit package, since the NCT is most reliable when two networks of equal sample size are compared. However, this resulted in very small sample sizes (n=139 per group) with still significant differences in age and sex between the groups. To retain the highest possible power we decided to perform our analyses with the largest, unmatched sample sizes. Descriptive statistics Descriptive and clinical measures of the two patient groups are displayed in table 1. The high-anxiety group was older, included more females, and showed significantly higher severity of motor impairment and cognitive decline. Since we divided the two groups based on anxiety level, there was also an expected difference in severity of anxiety symptoms. Table 1. Characteristics of the anxious and non-anxious PD patient groups. Anxious Non-anxious Mean N 316 253 Age (sd) 67.3 (9.8)* 65.2 (11.0)* Sex 44% female** 33% female** BAI total (sd) 25.5 (10.1)*** 6.7 (3.5)*** UPDRS-III total (sd) 30.4 (20.2)*** 23.0 (11.4)*** MMSE total (sd) 26.7 (3.6)*** 27.7 (3.2)*** * p<0.05; ** p<0.01; *** p<0.001 sd = standard deviation; BAI = Beck Anxiety Inventory; UPDRS = Unified Parkinson’s Disease Rating Scale; MMSE = Mini Mental State Examination