Page 46 - The efficacy and effectiveness of psychological treatments for eating disorders - Elske van den Berg

46 Chapter 2
it was decided to exclude them because their presence influenced subsequent subgroup analyses and thus, it could not be concluded whether subgroup differences were related to the particular subset of studies (e.g., high-quality trials) or due to the presence of these two outliers.
The finding of high heterogeneity can also be indicative of a large variety in included patient groups. In addition, the variety in used outcome measurements, probably assessing slightly different constructs, may have increased heterogeneity.
This large variety in measures led to a limited number of studies that could be included in the subgroup analyses. Because of these small sample sizes, we may have lacked adequate power to detect effect sizes.
In the included studies, both in the treatment and in control arms, a large variety of interventions is used. There are only few direct replications, and these replication studies have been done with different patient populations, leading to small subgroup samples.
Finally, in addition to previous mentioned risk of biases, in some studies princi- pal investigators were also responsible for the development of treatment interven- tions used in the RCT, so researcher allegiance cannot be excluded.
Suggestions for further research
In general, by using meta-analysis, multiple trials showing non-significant or only minor benefits may result in a significant difference between products as the power of the individual studies is magnified. Alternatively, the significant benefit observed in a few trials may be outweighed by multiple trials showing no difference between products. Therefore, using individual patient data meta-analyses instead, might produce a better estimate of the pooled effect size.
Standardized outcome measures are essential in the assessments of the treatment effects and should preferable be chosen in line with other, high-quality RCTs. In addi- tion, replication studies in which similar psychological interventions are replicated for homogenous patient groups may reduce heterogeneity.