Which behaviour change techniques make digital interventions for eating disorders effective?

Eating disorders (EDs) are highly common in Western countries, with recent reports from US data suggesting that, by the age of 40, 1 in 5 women and 1 in 7 men will experience some type of ED (Ward et al., 2019).

Even though EDs are linked to high psychological and financial costs, as well as high mortality (Arcelus et al., 2011), it is estimated that lots of people with EDs still do not receive the treatment they need (Striegel Weissman & Rosselli, 2017). That’s why digital interventions, with their easy, low-threshold access and low cost to the user, are a potential game changer.

Digital interventions for ED symptoms are promising (Linardon et al., 2020), but we still don’t know which factors underlie their effectiveness. The first step to understanding these factors is to figure out which behaviour change techniques (BCTs) are included in the interventions. BCTs are the elements designed to change the processes that lead to behaviour – in other words, the “active ingredients” of an intervention (Kok et al., 2016; Michie et al., 2013), such as giving feedback or reinforcement.

In this paper, Thomas and colleagues (2024) set out to investigate which BCTs are included in effective digital interventions for EDs.

Eating disorders are on the rise, with data simulations suggesting that 1 in 5 women and 1 in 7 men will experience some type of eating disorder by the age of 40.

Methods

Following a well-documented search strategy conforming to PRISMA guidelines, the authors searched 8 databases and identified 17 randomised controlled trials (RCTs) on digital interventions for adults with mild-to-moderate ED symptoms.

They coded the interventions for their theoretical background using an adapted version of the Theory Coding Scheme; mode of delivery using the Mode of Delivery Ontology; and their BCTs using the established BCT Taxonomy (Michie et al., 2013). Risk of bias was assessed with the Cochrane Risk of Bias tool.

The authors also conducted a meta-analysis using a random effects model on the effectiveness of the digital interventions, using the Eating Disorder Examination Questionnaire (EDE-Q; Fairburn & Beglin, 2008) as primary outcome.

Results

Study characteristics

Of the 17 included studies, 12 (71%) compared a digital intervention to a waitlist control (WC) or treatment-as-usual (TAU), whereas the other five (29%) used an active treatment as control.

The total sample across studies was large (n = 5,254). Participants were between 22.1 and 43.2 years old on average, and predominantly female, with only two (12%) studies including more than 10% male participants. Eight of the 17 studies targeted people with any ED symptoms, whereas six targeted binge eating and three targeted bulimia or ED-Not Otherwise Specified.

Digital intervention characteristics

Most (16/17, 94%) studies reported that the digital intervention was guided by a specific theoretical background, which mostly consisted of Cognitive Behavioural Therapy (CBT) and the transdiagnostic theory of EDs.

The most common modes of delivery were websites (11/17, 61%), and less often mobile apps (2/17, 12%) or a combination of the two (4/17, 24%). Video or audio functionalities were not often used (<5/17, 29%). Some interventions (4/17, 24%) were completely self-guided, but the majority (13/17, 76%) contained some form of human interaction, such as e-mail check-ins with a therapist.

Behaviour change techniques (BCTs)

Thirty-eight (41%) of the 93 BCTs described in the BCT Taxonomy (Michie et al., 2013) were identified across interventions. But which BCTs characterised the effective digital interventions?

While the study cannot pinpoint effects to specific BCTs, the authors noted that over three quarters of the digital interventions included the following BCTs:

• Self-monitoring of behaviour (e.g., letting users keep a food diary)
• Self-monitoring of outcomes of behaviour (e.g., weekly weight monitoring)
• Feedback on behaviour (e.g., providing bar charts that visualise users’ progress)
• Action planning (e.g., facilitating setting up meal schedules)
• Problem-solving (e.g., providing tips for relapse prevention)
• Information about antecedents (e.g., psychoeducation about what may precede an episode).

About half of the effective interventions also included:

• Behavioural practice/rehearsal (e.g., on the basis of made-up scenarios)
• Framing/reframing (e.g., challenging food-related cognitive distortions)
• Prompts/cues (e.g., reminders to record daily progress)
• Exposure (e.g., mirror confrontation exercises).

Effectiveness

The meta-analysis included 10 studies (five for follow-up) and showed that the digital interventions were more effective than waiting list control or treatment as usual in reducing ED behaviours, such as bingeing and purging. The effect was moderate, with a mean difference of -0.57 (95% CI [-0.080 to -0.39]; Z = 4.77, p <.001) in favour of the intervention at post-intervention and -0.33 (95% CI [-0.049 to -0.180; Z = 4.27, p <.001) at follow-up (> 8 weeks; though the authors note some concerns of bias for the follow-up data). Subgroup analyses showed that the digital interventions with the strongest theoretical background were the most effective.

Digital interventions were moderately effective in reducing eating disorders symptoms, compared to waitlist or treatment-as-usual. Over three quarters of the effective interventions contained the same six behavioural change techniques.

Conclusions

According to the authors,

[there] is increasing evidence for the effectiveness of digital interventions for the treatment of people with mild to moderate EDs, with improved outcomes at postintervention and sustained outcomes at follow-up.

The effective interventions seemed to rely on the same BCTs. Although there is no evidence that any one of the techniques by itself is responsible for the improvement in ED symptomatology, the presence of self-monitoring in all interventions suggests that it is important in driving change in ED behaviours and should therefore be considered in clinical practice.

According to Thomas et al. (2024, p. 16), “[effective] digital ED interventions mostly used the same specific [behaviour change techniques] informed by theory.” This indicates an important avenue for further investigation.

Strengths and limitations

When interpreting the results of this study, there are some limitations that we need to keep in mind.

First, the meta-analysis only included 10 studies, which restricts its statistical power. And whereas the pooled sample was generally large (> 5000 participants, with ~2000 included in the meta-analysis) and recruited from the community, the overwhelming majority of participants were women. In addition, the reviewed studies were conducted in Western countries, with most not reporting on participants’ ethnicity. The results may thus not generalise to men, non-binary people, or non-Western cultures.

Second, drop-out ranged between 6.7% and 58%, and was higher for the digital interventions that included minimal or no therapist support. Whereas 58% may seem high, other analyses of user engagement with popular mental health apps show a drop-out closer to 90% one month after app installation (Baumel et al., 2019). Of course, it is possible that people with a certain symptomatology are more likely to engage with digital interventions that are relevant to them as compared to the general public trying out mental health apps; however, drop-out in digital interventions does remain an issue and can weaken the power of follow-up analyses.

Finally, as Thomas et al. (2024) also mention, it is not entirely clear whether the statistically significant effects shown in the meta-analysis translate to clinically relevant outcomes. In other words, it is not clear to what degree the decrease in questionnaire scores has a practical meaning for digital intervention users.

At the same time, the study also has various strengths. The methodology is well-described and thus replicable; there was a high inter-rater agreement between the researchers who coded the studies; and there was no concerning risk of bias for the post-intervention data. And of course, the reviewed studies were RCTs, which is the design offering the highest quality of information when it comes to treatment effectiveness.

Due to the populations of the studies included in this systematic review and meta-analysis, findings may not generalise to men, non-binary people, and/ or non-Western cultures.

Implications for practice

Digital interventions for various conditions are here to stay, and there is an evolving body of research focused on their effectiveness and the parameters of that effectiveness. This is an important topic for us at The Mental Elf; for example, we have recently blogged about a digital intervention for bulimia, and a review of smartphone apps for symptoms of depression and anxiety.

This paper by Thomas et al. (2024) adds to this growing body of research and shows that digital interventions can help decrease symptoms of EDs in adults. The paper also highlights smartphone apps, which were considerably fewer than websites in this review, as an important avenue for future intervention development and research.

Importantly, Thomas et al. (2024) go beyond effectiveness to also look at the behaviour change techniques (BCTs) that drive it. This way, their paper has important implications for the selection of digital interventions to use. Therapists of clients with EDs, for example, who wish to blend their face-to-face treatment with a digital intervention as a support tool, may choose websites or apps that allow clients to self-monitor their behaviour – given that self-monitoring was a technique shown to be consistently present in pretty much all effective digital interventions.

In addition, the paper has implications for designers of digital interventions for adults with ED symptoms, as it points to BCTs that may be relevant to include in new interventions. Whereas BCT taxonomies such as the one by Michie et al. (2013), its updated version (Marques et al., 2023), or alternative taxonomies (e.g., Kok et al., 2016), are used to code the content of existing interventions, they can also be used to guide the development of new interventions. Using such a taxonomy to select BCTs that are grounded in theory is likely to lead to greater behaviour change. In the future, it may even be possible to link BCTs to user profiles (e.g., people with certain symptoms or demographic characteristics) in order to personalise content and this way maximise user engagement and – ideally – intervention outcomes.

Improved digital interventions may support face-to-face treatment with a healthcare professional, or fill the gap until such treatment becomes available, with immediate, low-cost and low-threshold intervention. Ideally, they may even lead to a decrease in symptoms, such that further treatment may be unnecessary. With the rise in prevalence of EDs and their devastating consequences, this will be especially relevant.

Knowing which behaviour change techniques are found in effective digital interventions makes it easier to select digital interventions – either as a therapist or as a client.

Statement of interests

This elf has no conflicts of interest to report.

Links

Primary paper

Thomas, P. C., Curtis, K., Potts, H. W., Bark, P., Perowne, R., Rookes, T., & Rowe, S. (2024). Behavior Change Techniques Within Digital Interventions for the Treatment of Eating Disorders: Systematic Review and Meta-Analysis. JMIR Mental Health, 11, e57577.

Other references

Arcelus, J., Mitchell, A. J., Wales, J., & Nielsen, S. (2011). Mortality rates in patients with anorexia nervosa and other eating disorders: A meta-analysis of 36 studies. Archives of General Psychiatry, 68(7), 724–731.

Baumel, A., Muench, F., Edan, S., & Kane, J. M. (2019). Objective user engagement with mental health apps: Systematic search and panel-based usage analysis. Journal of Medical Internet Research, 21(9), 1–15.

Fairburn, C. G., & Beglin, S. J. (1994). Eating Disorder Examination Questionnaire (EDE-Q). APA PsycTests.

Ferreira, A. J. (2024). Digital self-help for bulimia recovery: encouraging results for waiting list management. The Mental Elf.

Kok, G., Gottlieb, N. H., Peters, G. J. Y., Mullen, P. D., Parcel, G. S., Ruiter, R. A. C., Fernández, M. E., Markham, C., & Bartholomew, L. K. (2016). A taxonomy of behaviour change methods: An Intervention Mapping approach. Health Psychology Review, 10(3), 297–312.

Linardon, J., Shatte, A., Messer, M., Firth, J., & Fuller-Tyszkiewicz, M. (2020). E-mental health interventions for the treatment and prevention of eating disorders: An updated systematic review and meta-analysis. Journal of Consulting and Clinical Psychology, 88(11), 994–1007.

Marques, M. M., Wright, A. J., Corker, E., Johnston, M., West, R., Hastings, J., Zhang, L., & Michie, S. (2023). The Behaviour Change Technique Ontology: Transforming the Behaviour Change Technique Taxonomy v1. Wellcome Open Research, 8(May).

Michie, S., Richardson, M., Johnston, M., Abraham, C., Francis, J., Hardeman, W., Eccles, M. P., Cane, J., & Wood, C. E. (2013). The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: Building an international consensus for the reporting of behavior change interventions. Annals of Behavioral Medicine, 46(1), 81–95.

Striegel Weissman, R., & Rosselli, F. (2017). Reducing the burden of suffering from eating disorders: Unmet treatment needs, cost of illness, and the quest for cost-effectiveness. Behaviour Research and Therapy, 88, 49–64.

Valentine, L. (2024). Apps for depression and anxiety: big new meta-analysis supports effectiveness. The Mental Elf.

Ward, Z. J., Rodriguez, P., Wright, D. R., Austin, S. B., & Long, M. W. (2019). Estimation of eating disorders prevalence by age and associations with mortality in a simulated nationally representative US cohort. JAMA Network Open, 2(10), 1–12.

Photo credits