146 Chapter 6 Limitations of this thesis Conducting our studies in, and staying close to, routine clinical practice implicated an important limitation: the difference in dosage between DBT-BED and the more intensive CBT+. This may have disadvantaged the DBT-BED group and limits the reach of our conclusions about the observed differences in outcome. E.g. it prevents us from generalizing our findings to less intensive forms of CBT, such as CBT-E (Fairburn, 2008). Also, the issue whether differences in outcome are related to specific CBT/DBT elements or to the intensity of treatment cannot be resolved. In addition, staying close to routine clinical practice meant that we stuck to the original manuals and did not control for content. Therefore, some overlap in interventions may have occurred between the two treatments. This may have compromised differential effects of both therapies. Next, all variables were assessed with self-report questionnaires at a few timepoints. Although validated instruments were used, findings may be limited by recall bias, self-bias or blind spots. Ecological momentary assessment (EMA) in the week before the start of treatment could for instance provide a more reliable and valid image of the relevant predictor/moderator variables (Engel et al., 2016). On the other hand, self-report questionnaires are more likely to be implemented in routine clinical practice than EMA. Also, some remarks can be made regarding sample size. Despite the somewhat limited sample size in the randomized study (Chapter 3) we did find significant differences between treatments on both primary measures. The power issues may have prevented us from finding more, and more robust, differences in favour of CBT+ as all nonsignificant differences favoured CBT+. A similar issue is related to the combination of the large CBT sample (n = 133) and the smaller DBT-BED sample (n = 42) in the non-randomized effectiveness study (Chapter 4). As such, the sample size provides adequate power to detect large differences between groups, but the smaller group limits the ability to detect small or moderate differences. Yet, compared to most other DBT-BED effectiveness studies, this study had a relatively large sample size. And finally, although the moderation-study (Chapter 5) constitutes a large group at baseline and EOT, some of the FU effects may have been obscured by the smaller sample size at FU. More significant variables might have showed up at FU, had the FU-group been larger than it was. We could have further optimized the assessment of BED pathology by either making use of the Eating Disorder Examination interview (EDE) instead of the EDE-Q (Berg et al., 2011) or by providing a specific definition of binge eating when administering the EDE-Q (as suggested by Celio and colleagues, 2004).
RkJQdWJsaXNoZXIy MjY0ODMw