Health At Every Size intervention® under real-world conditions: the rights and wrongs of program implementation.

Implementation integrity is known to be critical to the success of interventions. The Health At Every Size® (HAES®) approach is deemed to be a sustainable intervention on weight-related issues. However, no study in the field has yet investigated the effects of implementation on outcomes in a real-world setting. Objective: This study aims to explore to what extent does implementation integrity moderate program outcomes across multiple sites. Methods: One hundred sixty-two women nested in 21 health facilities across the province of Québec (Canada) were part of a HAES® intervention and completed questionnaires at baseline and after the intervention. Participant responsiveness (e.g. home practice completion) along with other implementation dimensions (dosage, adherence, adaptations) and providers' characteristics (n = 45) were assessed using a mix of qualitative and quantitative data analysis. Adaptations to the program curriculum were categorized as either acceptable or unacceptable. Multilevel linear modeling was performed with participant responsiveness and other implementation dimensions predictors. Intervention outcomes were intuitive eating and body esteem. Results: Unacceptable adaptations were significantly associated with providers' self-efficacy (rs (23) = .59, p = .003) and past experience with facilitating the intervention (r(23) = .47, p = .03). Participant responsiveness showed a significant interaction between time and home practice completion (B = .07, p < .05) on intuitive eating scores. Conclusion: Except for participant responsiveness, other implementation dimensions did not moderate outcomes. Implications for future research and practice are discussed.