Page 37 - The efficacy and effectiveness of psychological treatments for eating disorders - Elske van den Berg

Chapter 2 37
Eating disorder pathology
Thirteen comparisons were included in the meta-analysis of eating disorder pathology. No significant difference between psychological treatment and control treatment on eating disorder pathology was found, g = 0.06, 95% CI [-0.10, 0.21]. Examining the effects for BMI and kg separately does not lead to a different finding. A very high between-study heterogeneity was detected including all 16 comparisons, I2 = 87, 95% CI [80, 91]. After removing three outliers, heterogeneity lowered, I2 =17, 95% CI [0, 56] (Table 2). Examining potential publication bias using Duvall and Tweedie’s trim and fill procedure resulted in g = 0.10, 95% CI [-0.07, 0.26], with one study imputed. Testing the symmetry of the funnel plot using the Begg & Mazumbar’s test (Kendall’s T = 0.064, p = .760) and Egger’s test (T = 0.358, df = 11, p = .727), did not reveal significant publication bias.
Sensitivity analyses including only the largest or smallest effect size of each study, including only high-quality studies, or including only outpatient studies, did not result in significant differences between treatment and control conditions. Neither when the effects were examined for Eating Disorder Examination-Interview and Eating Disorder Inventory separately. Finally, a series of subgroup analyses were conducted, but none of the included moderators was significantly associated with eating disorder pathology. The meta-regression analysis showed that publication year was not significantly associated with the effect on eating disorder pathology, b = 0.002, 95% CI [-0.04, 0.02], p = .626.