Introduction
Food insecurity affects millions of households, prompting reliance on food pantries for essential nourishment. Clients of these pantries are disproportionately vulnerable to diet-related chronic diseases, partially due to high exposure to ultra-processed foods (UPFs) prevalent in food assistance settings. UPFs, classified as Nova 4 in the Nova food classification system, are industrial formulations high in energy density, fats, sugars, and additives, and are associated with adverse health outcomes including increased mortality risk. Behavioral economics-based interventions employing “nudging” strategies have been proposed to improve food selection by encouraging healthier choices without restricting client autonomy. This article examines the secondary analysis of the SuperShelf intervention which aimed to improve healthy food supply and demand in Minnesota food pantries and assesses its effect on the nutritional quality of client food selections categorized by the Nova system.
Study Design and Methods
This study utilized secondary data from a group-randomized controlled trial involving 11 food pantries in Minnesota, randomized into intervention (n=5) and control (n=6) groups. The SuperShelf intervention was delivered over two phases: an initial supply-focused phase to increase availability and cultural appropriateness of healthy foods, followed by a behavioral economics phase to nudge clients toward healthier choices using shelf arrangement, signage, and product prominence.
Participants included a convenience sample of 187 adult food pantry clients (85 intervention, 102 control) who selected their food choices independently. Client food selections at baseline and one-year post-intervention were recorded and analyzed using the Nutrition Data System for Research (NDSR). Foods were classified into four Nova categories: (1) unprocessed or minimally processed, (2) processed culinary ingredients, (3) processed foods, and (4) ultra-processed foods.
Energy share—the percentage of total calories contributed by each Nova category—was calculated for each client’s food cart. Linear mixed-effects models adjusted for baseline pantry food profiles, client demographics, pantry characteristics, and clustering at the pantry level, assessed post-intervention differences between groups. Additionally, sensitivity analyses at the item level using mixed-effects logistic regression evaluated the odds of selecting UPFs.
Key Findings
Descriptive analysis revealed that post-intervention, unprocessed or minimally processed foods (Nova 1) represented approximately 34.7% of energy intake in the intervention group versus 33.8% in controls. Ultra-processed foods (Nova 4) accounted for 41.1% in intervention and 43.6% in control groups, mirroring baseline distributions. Linear mixed models showed no statistically significant differences in energy shares among Nova categories between groups, with fully adjusted p-values exceeding 0.4 across all categories.
Sensitivity analysis corroborated these findings; the odds of a selected item being ultra-processed did not differ significantly between intervention and control clients after adjustment.
Subgroup analysis highlighted common energy sources such as bread, meats, processed foods, and sweets across both groups without meaningful shifts attributable to the intervention.
Expert Commentary
The lack of significant impact on the selection of UPFs versus less processed foods suggests that although environmental and behavioral “nudges” can improve some aspects of food choice, they may be insufficient when not specifically tailored to address food processing levels. The SuperShelf intervention’s focus on general healthfulness possibly failed to influence client preferences for convenience and culturally relevant items often present in the UPF category.
This aligns with the complexities acknowledged in food pantry contexts, where pantry inventory variability, client time constraints, and cultural food preferences necessitate nuanced approaches. Also notable is the challenge in classifying foods where some nutrient-dense alternatives (e.g., fortified whole grain bread) fall under UPF, complicating simplistic exclusion strategies.
Existing literature from interventions such as the UnProcessed Pantry Project suggests that targeted efforts addressing UPF specifically may hold greater promise. Incorporating nutrition education, culturally sensitive food availability, and interventions addressing client knowledge could enhance effectiveness.
Limitations
Limitations include the secondary nature of the analysis, the relatively small sample size, and lack of longitudinal tracking of the same clients pre- and post-intervention. Food pantry supply fluctuations and donations contribute to inventory variability, which was not controlled for longitudinally. Self-selection and external food sources for clients also confound estimates of actual UPF consumption.
Conclusion
The behavioral economics intervention within the SuperShelf program did not significantly influence the nutritional quality of food pantry clients’ selections based on the Nova food classification system. Future research should prioritize interventions explicitly targeting reductions in ultra-processed food selection and enhancing the availability and appeal of unprocessed, minimally processed foods in food pantries. Multicomponent strategies integrating education, environmental modifications, and culturally responsive approaches may be necessary to positively impact dietary quality and reduce diet-related health disparities among food-insecure populations.
References
Rossi K, Gombi Vaca MF, Schwartz MB, Caspi C. Nutritional quality of foods according to the Nova food classification system after a behavioral economics intervention in food pantries. Front Public Health. 2025 Aug 29;13:1613200. doi: 10.3389/fpubh.2025.1613200. PMID: 40904930; PMCID: PMC12403370.
