78 Chapter 4 in outcome assessment selection between the included studies. The secondary outcome data was represented using several scales: Yale-Brown-Cornell Eating Disorder Scale (26), Yale-Brown Obsessive-Compulsive Scale (Y-BOCS) (27), Hamilton Anxiety Rating Scale (HAM-A) (28), and Hamilton Depression Rating Scale (HAM-D) (29). Quality of life was assessed using the Quality of Life Scale (30), SF36 (31), and Eating Disorder Quality of Life (32). Next to effectiveness data, we also extracted safety data including adverse effects and complications. Statistical analyses Main analyses We used random effect models with Comprehensive Meta-Analysis V3 for our quantitative data synthesis. We used Hedges’ g for continuous results, with a 95% confidence interval and two-tailed p-values. Sensitivity analyses were performed using pre-operative BMI measures of Villalba et al. instead of the calculated reference BMI value (25). Forest plots with I2 statistics were used to examine any study heterogeneity. Publication bias A funnel plot was plotted to assess publication bias. The classic and Orwin’s fail-safe N, Begg and Mazumdar rank correlation, and Egger’s regression intercept were calculated. Should it be deemed necessary Duval and Tweedie’s trim-and-fill method was used to report adjusted values. RESULTS Study selection The literature search provided a set of 290 articles. 276 articles were excluded based on title and abstract (Figure 1). The full text of the fourteen remaining articles was reviewed. Duplicates and articles that turned out to be protocols were removed from analysis. Four of these fourteen articles were included for meta-analysis. Figure 1 shows the process of study selection.
RkJQdWJsaXNoZXIy MjY0ODMw