Behavior Change Techniques Within Digital Interventions for the Treatment of Eating Disorders: Systematic Review and Meta-Analysis (Preprint)

Abstract

Previous systematic reviews of digital eating disorder interventions have demonstrated effectiveness at improving symptoms of eating disorders; however, our understanding of how these interventions work and what contributes to their effectiveness is limited. Understanding the behavior change techniques (BCTs) that are most commonly included within effective interventions may provide valuable information for researchers and developers. Establishing whether these techniques have been informed by theory will identify whether they target those mechanisms of action that have been identified as core to changing eating disorder behaviors. It will also evaluate the importance of a theoretical approach to digital intervention design.

This study aims to define the BCTs within digital self-management interventions or minimally guided self-help interventions for adults with eating disorders that have been evaluated within randomized controlled trials. It also assessed which of the digital interventions were grounded in theory and the range of modes of delivery included.

A literature search identified randomized controlled trials of digital intervention for the treatment of adults with eating disorders with minimal therapist support. Each digital intervention was coded for BCTs using the established BCT Taxonomy v1; for the application of theory using an adapted version of the theory coding scheme (TCS); and for modes of delivery using the Mode of Delivery Ontology. A meta-analysis evaluated the evidence that any individual BCT moderated effect size or that other potential factors such as the application of theory or number of modes of delivery had an effect on eating disorder outcomes.

Digital interventions included an average of 14 (SD 2.6; range 9-18) BCTs. <i>Self-monitoring of behavior</i> was included in all effective interventions, with <i>Problem-solving</i>, <i>Information about antecedents</i>, <i>Feedback on behavior</i>, <i>Self-monitoring of outcomes of behavior</i>, and <i>Action planning</i> identified in &gt;75% (13/17) of effective interventions. <i>Social support</i> and <i>Information about health consequences</i> were more evident in effective interventions at follow-up compared with postintervention measurement. The mean number of modes of delivery was 4 (SD 1.6; range 2-7) out of 12 possible modes, with most interventions (15/17, 88%) being web based. Digital interventions that had a higher score on the TCS had a greater effect size than those with a lower TCS score (subgroup differences: <i>χ</i>²₁=9.7; <i>P</i>=.002; <i>I</i>²=89.7%) within the meta-analysis. No other subgroup analyses had statistically significant results.

There was a high level of consistency in terms of the most common BCTs within effective interventions; however, there was no evidence that any specific BCT contributed to intervention efficacy. The interventions that were more strongly informed by theory demonstrated greater improvements in eating disorder outcomes compared to waitlist or treatment-as-usual controls. These results can be used to inform the development of future digital eating disorder interventions.

PROSPERO CRD42023410060; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=410060