Page 32 - The efficacy and effectiveness of psychological treatments for eating disorders - Elske van den Berg
P. 32

  32 Chapter 2
 Study characteristics
The 17 randomized controlled trials included a total of 1,279 participants: 761 in the psychological treatment conditions and 518 in the control treatment condi- tions (see Appendix C). Some studies reported on multiple psychological treatment conditions versus one control condition, resulting in a total of 24 arms. Most studies included adult patients (n = 9), three reported on adolescents and five reported on both adolescents and adults. The number of patients in the experimental conditions ranged from 10 to 80 per study. Mean pre-treatment BMI ranged from 15.0 to 18.1. The majority of the treatment conditions for adults reported on cognitive (behaviour) therapy (n = 5), Maudsley Model of Anorexia Nervosa Treatment for Adults (n = 3) or cognitive analytical psychotherapy or focal psychodynamic therapy (n = 4). In studies in which only adolescents were treated, all treatment conditions included family interventions (n = 3). The majority of the control conditions was defined as treatment as usual (TAU) (n = 6) or SSCM (n = 5). Three of the 17 studies were inpatient studies, although hospitalization was offered to a minority of patients in several outpatient studies. In one study, both an inpatient and an outpatient treatment were compared with TAU (Gowers et al., 2007). Besides the family therapy conditions, all conditions but one (Pillay & Crisp, 1981) were individually based treatments. The offered outpa- tient treatment doses ranged from four sessions (Motivational Interviewing; Wade, Frayne, Edwards, Robertson, & Gilchrist, 2009) to a little over 40 sessions (Cognitive Behavioral Therapy-Enhanced; Byrne et al. (2017) and Focal Psychodynamic Therapy; Zipfel et al. (2014). The mean number of outpatient sessions was 23.2 (range 0-58.3) and dropout rates per arm ranged from 7% (Hall & Crisp, 1987) to 100% (Serfaty, Heap, Ledsham, & Jolley, 1999).
Six studies were assessed as having a low risk of bias based on the Cochrane domains (Higgins et al., 2011), meaning that 35% of the included studies are regard- ed high-quality studies. A high risk of bias with regard to the domain incomplete data was the most commonly found risk of bias (n = 4). With regard to allocation conceal- ment, this domain was unclear in seven studies and was in one study found to be a high risk of bias. All relevant study characteristics can be found in Appendix C.
Weight gain
Twenty comparisons were included in the meta-analysis of weight gain (Figure 2). The pooled effect size indicating the difference between psychological treatment and control condition on weight gain at post-treatment was not significant, g = 0.07, 95% CI [-0.09, 0.23]. Similarly, no significant differences were found when the effects were examined for BMI and kg separately (Table 1). Including all 23 comparisons from 16 studies resulted in a very high between-study heterogeneity, I2 = 85, 95% CI [78, 89].





























































































   30   31   32   33   34