Page 30 - The efficacy and effectiveness of psychological treatments for eating disorders - Elske van den Berg

30 Chapter 2
mixed effects model, which pools studies within subgroups with the random effects model and tests for significant differences between subgroups with the fixed effects model. Given the modest evidence for family therapy for younger anorexia nervosa patients, living with their family and an illness duration of less than three years (Hay et al., 2014) subgroup analyses were performed on studies including patients under 18 and on studies including family therapy. Given the efficiency of the control condi- tion SSCM (NICE, 2017), subgroup analyses were performed on studies including SSCM. Additional subgroup analyses for the study characteristics baseline BMI, age of illness onset, duration of illness, therapy setting, number of treatment sessions, availability of treatment manual, and training of the therapists were exploratory.
Finally, we conducted meta-regression analyses using the mixed effects model, to assess whether the continuous variable year of publication predicted the effect sizes, indicated by a z value and an associated p value.
Power calculation
We conducted a power calculation according to the procedures described by Borenstein, Hedges and Rothstein (2009), to examine how many studies would have to be included in order to have sufficient statistical power to identify significant effects. We estimated the number of studies needed to identify an effect size of 0.3. The power calculation indicated at least 20 studies with a mean sample size of 30 (15 participants per condition) had to be included, for being able to detect an effect size of d = .30 (conservatively assuming a medium level of between-study variance, τ2, a statistical power of .80, and a significance level of α = .05). Alternatively, we would need 15 studies with 40 participants each or 14 studies with 50 participants to detect an effect size of d = .30.