A process evaluation of the Supermarket Healthy Eating for Life (SHELf) randomized controlled trial.

BACKGROUND
Supermarket Healthy Eating for Life (SHELf) was a randomized controlled trial that operationalized a socioecological approach to population-level dietary behaviour change in a real-world supermarket setting. SHELf tested the impact of individual (skill-building), environmental (20 % price reductions), and combined (skill-building + 20 % price reductions) interventions on women's purchasing and consumption of fruits, vegetables, low-calorie carbonated beverages and water. This process evaluation investigated the reach, effectiveness, implementation, and maintenance of the SHELf interventions.

METHODS
RE-AIM provided a conceptual framework to examine the processes underlying the impact of the interventions using data from participant surveys and objective sales data collected at baseline, post-intervention (3 months) and 6-months post-intervention. Fisher's exact, χ (2) and t-tests assessed differences in quantitative survey responses among groups. Adjusted linear regression examined the impact of self-reported intervention dose on food purchasing and consumption outcomes. Thematic analysis identified key themes within qualitative survey responses.

RESULTS
Reach of the SHELf interventions to disadvantaged groups, and beyond study participants themselves, was moderate. Just over one-third of intervention participants indicated that the interventions were effective in changing the way they bought, cooked or consumed food (p < 0.001 compared to control), with no differences among intervention groups. Improvements in purchasing and consumption outcomes were greatest among those who received a higher intervention dose. Most notably, participants who said they accessed price reductions on fruits and vegetables purchased (519 g/week) and consumed (0.5 servings/day) more vegetables. The majority of participants said they accessed (82 %) and appreciated discounts on fruits and vegetables, while there was limited use (40 %) and appreciation of discounts on low-calorie carbonated beverages and water. Overall reported satisfaction with, use, and impact of the skill-building resources was moderate. Maintenance of newly acquired behaviours was limited, with less than half of participants making changes or using study-provided resources during the 6-month post-intervention period.

CONCLUSIONS
SHELf's reach and perceived effectiveness were moderate. The interventions were more effective among those reporting greater engagement with them (an implementation-related construct). Maintenance of newly acquired behaviours proved challenging.