Ires Ghielen

104 Chapter 6 Within the second analysis, only MS patients were investigated in the included studies. Regarding the treatments of interest, four studies investigated a CBT-based treatment [43-46] and three examined an MBT [47-49]. Overall, the quality of the included RCTs was low, based on the scores on the risk of bias assessment tool. Blinding of participants/researchers was impossible due to the nature of studies on psychotherapeutic interventions, and was therefore always considered as risk of bias. As allocation concealment was often not well reported, two studies had a risk of detection bias [41, 47]. The study by Okai and colleagues [41] also showed an attrition bias, as did the study by Calleo and colleagues [39]. In the first analysis, four studies showed good quality [31, 33, 38, 42], as shown by a total risk of bias of 1 (only risk of performance bias). In the second analysis, only the study by Carletto and colleagues had good quality (score of 1 on the risk of bias assessment tool: only blinding of participants was not achieved) [49]. Treatment Effects Meta-analysis 1: psychological therapy versus TAU or waitlist condition Figure 2 displays the forest plot of the standardized effect sizes of psychological therapies on psychological distress in PD and MS patients, compared with a waitlist or TAU condition. The mean effect size ( g ) was 0.51 (95% CI = 0.22 — 0.80) with a heterogeneity estimate (I²) of 66 (95% CI = 27 — 80). As a post-hoc analysis, the studies of Okai et al. [41], Ghielen et al. [40], and Kiropoulos et al. [35] were excluded in a separate meta-analysis. These studies were considered outliers since the effect sizes with their 95% confidence intervals were outside the 95% confidence interval of the pooled main effect size. The effect size decreased to g = 0.31 (95% CI = 0.13 — 0.48) and heterogeneity decreased to I² = 0 (95% CI = 0 — 56) when these three studies were removed (see table 2). To investigate the treatment effects on the different types of outcome measure separately, three meta-analyses were conducted on anxiety, depression, and general psychological distress outcome measures. The treatment effect on general mental health outcomes was highest ( g = 0.79, 95% CI = 0.32 — 1.25 with I² = 66, 95% CI = 0 – 85), followed by the effect on anxiety symptoms ( g = 0.36, 95% CI = 0.03 — 0.66 with I² = 59, 95% CI = 0 — 79), and depressive symptoms ( g = 0.33, 95% CI = 0.05 — 0.62 with I² = 60, 95% CI = 0 — 78) (see table 2).