Is the level of implementation linked with intervention outcomes? The process evaluation of the Veggie Van intervention to increase fruit and vegetable consumption in underserved communities.

Background: The Veggie Van (VV) model is a mobile produce market intervention previously shown to increase fruit and vegetable (F&V) consumption among lower-income individuals. Although the model was recently evaluated in a multi-state effectiveness trial, preliminary findings suggested challenges to implementing the model with fidelity. This study examined whether variation in implementation fidelity was associated with participant-level dietary, food security, and market utilization outcomes. Methods: This study involved a secondary analysis of participant outcome data from a longitudinal RCT conducted with 9 implementing organizations operating 17 mobile market sites (n=699 participants). Previously collected participant data included baseline and 12-month measures of F&V consumption from 24-hour dietary recalls and self-reported surveys, as well as food security assessed using the USDA 10-item Adult Food Security Survey Module. Participant outcome data were aggregated at the site level and linked with organizational implementation data collected throughout the intervention period. Implementing organizations completed monthly process measures surveys assessing adherence to core VV model components; responses were scored and aggregated into site-level implementation fidelity scores. Correlational analyses and generalized linear models examined associations between implementation fidelity and changes in F&V consumption, food security, and market utilization outcomes over 12 months. Results: The mean implementation fidelity score across sites was 40.6 out of 55 possible points. Higher overall fidelity was positively associated with greater increases in mean F&V consumption over 12 months (Pearson correlation coefficient [PCC]=0.56, p=0.02). Fidelity to the nutrition education component demonstrated a particularly strong association with change in F&V consumption (PCC=0.81, p<0.0001). These associations remained significant in regression analyses. In addition, sites operating markets for ≥10 months annually had 76% greater odds of participants shopping at the mobile market at least once during the intervention period (p=0.04). No significant associations were observed between implementation fidelity and changes in food security. Conclusions: Greater fidelity to the VV model, particularly implementation of nutrition education components, was associated with improved dietary outcomes in this secondary analysis of participant-level RCT data. Findings highlight the importance of evaluating implementation alongside effectiveness outcomes in community-based nutrition interventions and suggest that tailored implementation support strategies may strengthen delivery of intervention components most strongly associated with dietary change. Trial registration: The larger randomized controlled trial, the Veggie Van study, was registered at https://clinicaltrials.gov/ on January 29, 2020 (NCT04246593).