Fidelity evaluation of the Water First intervention for promoting tap water consumption in school-age children

A cluster-randomized controlled trial examining the impact of the Water First water promotion and access program in 18 elementary schools found that the intervention increased water consumption and prevented overweight. This study describes a secondary analysis of implementation fidelity and its association with change in students' beverage intake and body mass index (BMI) z-score in the intervention arm.

Researchers observed Water First fidelity (visible water promotional material, quality of study-provided water sources and drinking vessels in classrooms, school cafeterias, high-traffic areas) 2 times/week during the intervention period, with data collected via yes/no questions. Responses indicating high fidelity were assigned a value of 1 and low fidelity were assigned 0; these were averaged to generate a summary score for each location (0 to 1). Home-engagement activity completion (%) was calculated per student based on returned class assignments. Anthropometrics and past-week frequency of beverage intake (instrument adapted from the validated Beverage and Snack Questionnaire) were assessed at baseline, 7-months, and 15-months. Mixed-effects linear regression models examined associations of fidelity with the primary (water intake) and secondary (sugar-sweetened beverage [SSB], milk, 100% fruit juice intake, BMI z-score) outcomes. The primary predictor was the fidelity score (0.1-unit increase) by time interaction. Beverage frequencies were log-transformed, and regression coefficients were exponentiated to derive the percent change in outcomes.

Fidelity was highest in the cafeteria (mean: 0.87, SD: 0.06) compared to high-traffic water station areas (0.66, SD: 0.15) and classrooms (0.73, SD: 0.13). Mean home-engagement activity completion rate was 0.41 (SD: 0.33). There were no associations of fidelity with water intake. Each 0.1 unit increase in classroom fidelity was associated with a decreased intake of 100% fruit juice (0.95; Cl: 0.91-0.99). Higher high-traffic area fidelity was associated with decreased intake of SSB (0.94; Cl: 0.91-0.98) and plain milk (0.92; Cl: 0.89-0.96). Higher cafeteria fidelity was associated with 0.027 decrease in BMI z-score (Cl: -0.046, -0.008). No significant associations were observed for home-engagement activity completion.

The intervention was largely implemented with high fidelity, with scores ranging from 0.66 to 0.87. Higher cafeteria fidelity was associated with reductions in BMI z-score, while home-engagement activity completion was not associated with changes in beverage outcomes or BMI z-score.

The trial was preregistered (2017-06-09) at clinicaltrials.gov (NCT03181971).