ABSTRACT Objective While it is assumed that digital interventions are more suitable for milder eating disorder presentations, the evidence supporting this assumption remains unclear. We conducted meta‐regressions testing whether baseline severity at the study level was associated with treatment effects from digital interventions for eating disorders. Method Databases were searched to identify RCTs of digital interventions compared to controls in individuals with diagnostic, subthreshold, or self‐reported eating disorders. Baseline severity was operationalised as global levels of eating disorder psychopathology and objective binge eating. We conducted bivariable and multivariable meta‐regressions to test whether baseline severity was associated with treatment versus control effect sizes and dropout. Results Thirty‐one trials were included. Bivariable meta‐regressions showed that higher average objective binge eating was associated with greater treatment effect sizes and lower dropout rates. These associations remained significant in most sensitivity analyses, but not in the multivariable meta‐regressions. Average global eating disorder psychopathology levels at baseline were not associated with treatment effects in bivariable meta‐regressions, but was significantly and positively associated with effects in the multivariable model that adjusted for other characteristics. Conclusion Higher baseline severity may not limit the effectiveness or acceptability of digital interventions and in some contexts be associated with larger effects. Trial Registration https://osf.io/j9chm/overview
Liu et al. (Fri,) studied this question.