Purpose: Restaurant meals are becoming normative in the U.S. and are higher in calories and sodium compared to meals prepared at home. Research suggests sodium warning labels are associated with reductions in total sodium ordered while calorie labeling is minimally effective. This is the first study assessing the moderating role of health literacy on the effect of menu labels and ordering behavior. Understanding whether individuals can interpret and utilize these labels is pivotal in improving food labeling policy.

Methods: 2,829 participants from Amazon’s Turk Prime US sample completed an online survey, in which they selected: an entrée, side, beverage, and dessert from a Wendy’s menu. Participants were randomly assigned to view menus either displaying calorie information or not, featuring stoplight-style sodium warnings or not, and containing All-Natural labels or not (a 2×2×2 experimental design). Surveys included demographic and health-related questions, including the Newest Vital Sign (NVS) inventory to assess health literacy. Linear regression models adjusting for sociodemographic information, typical ordering behaviors, and frequency of fast-food consumption were used to evaluate the interactive effect between menu labeling condition and health literacy on the experimental food orders.

Results: Individuals’ health literacy significantly predicted their orders’ calories and sodium (p < .05) and moderated the effect of menu labels on food choice. NVS score had a significant negative correlation with total calories (r = -.24, p < .001) and total sodium (r = -.11, p < .001) of orders. Menu labels lowered calories and sodium among individuals with moderate literacy (p < .01), not those with high or low literacy. Having both sodium warning and calories information seems to be the most effective combination to improve people’s food choices.

Conclusions: Policy changes may produce differential outcomes depending on factors like individuals’ capacity to comprehend and apply health-related information in making more informed decisions. The interaction effect between menu labels and health literacy introduces a novel direction for interventions: prioritizing educational program to individuals with low health literacy to help them gain a moderate level of health knowledge might be the most pressing and effective way to improve population level food choices.

Purpose. Physical activity and time spent in nature have both been shown to positively affect physical and mental health. Nature prescriptions, written by a healthcare professional, are a potentially powerful way to increase time spent in greenspace and physical activity. However, uptake in the medical field has been slow. ParkRx America provides an easy-to-use platform to prescribe greenspace along with Continuing Medical Education (CME) training on why and how to write nature prescriptions. In this study, characteristics of people who had registered with Park Rx America and their prescription habits were examined. Methods. In October 2023, email invitations were sent to the 1,580 professionals registered with ParkRx America. Interested participants were taken to a Qualtrics survey and completed a 5–10-minute survey on their nature prescription habits, demographics, connectedness to and time spent in nature, and other related items. Results. The survey was completed by 168 health professionals (12% response rate) with 147 providing complete data. More than half (55.1% or 81 providers) had ever prescribed nature to their patients. However, of those, less than 42% prescribed nature to patients at least weekly. Providers who reported a stronger connection to nature, took more frequent breaks outside, and perceived nature prescriptions to be extremely valuable were more likely to prescribe nature (p < .05). Among those who had ever prescribed nature, the conditions for prescribing nature were: physical inactivity (72.8%), anxiety (72.8%), depression (72.8%) and general health (75.3%). Major reasons for not prescribing included: time constraints (43.5%), patient and family barriers (29.9%) and not knowing how to start the conversation (15.6%). Providers cited the following as facilitators to write nature prescriptions: embedding nature prescriptions into the EMR (58.5%), working with local park systems (52.4%) and hanging posters in the office (42.9%). Conclusions: Nature prescriptions have a strong potential to improve health, yet few American health care providers routinely issue nature prescriptions. More research is needed to study the effect of motivations (posters, posters, partnerships with local park systems, embedding nature prescriptions in EMR) on overcoming the cited barriers to issuing nature prescriptions. Lessons learned from other countries may inform future practices.

Purpose: The design of strategies to increase the affordability of healthy diets and guarantee the right to food requires the development of new methodological approaches to measure the costs of diets. The present work intended to analyze the evolution of the cost of a healthy food basket (HFB) in Uruguay, an emerging Latin American country, from January 2011 to October 2022, and to compare it to the evolution of the official Basic Food Basket (BFB) of the country.

Methods: The HFB for the Uruguayan population was constructed considering Uruguayan dietary guidelines, recommendations on nutrient intake, and the dietary patterns of the Uruguayan population. It included unprocessed, minimally processed, and processed foods and beverages consumed by at least 10% of the households, according to the 2016-2017 National Survey of Expenses and Incomes of Households. Linear programming was used to minimize the cost of the basket considering restrictions for nutrient intake and consumption of specific foods and groups of foods. Data from the National Institute of Statistics were used to calculate the cost of the HFB and to analyze differences with respect to the BFB.

Results/findings: The cost of the HFB in May 2017 was 4.93 USD per person per day (international dollars). Fruits and vegetables accounted for 25.9% of the total cost, while meats represented 17.0%. The nominal cost of the HFB was on average 25.2% higher than the cost of the BFB (minimum 19.3%, maximum 32.0%). Throughout the evaluated period, differences in nominal costs tended to be higher in December-January. The cost of the two baskets experienced similar trends in most of the series. Differences between the interannual increase in the costs of the baskets did not show a clear trend in favour of one over the other.

Conclusions: The present work contributes to the development of methodological approaches to measure the affordability of healthy diets and provides valuable insights for the design of public policies aimed at achieving SDG 2. Results enhance discussions regarding the cost of healthy diets and the criteria to determine food necessities and wellbeing indicators.

Purpose: Algorithm models are innovative and effective for predictive modeling with complex datasets. However, behavioral research primarily uses data models, leaving random forest algorithm approaches underexplored. This study aims to apply random forests to pinpoint individual, park, and neighborhood predictors of park use.
Methods: The current data were derived from the ParkIndex study, which occurred in 128 census block groups in Brooklyn(NY), Seattle(WA), Raleigh(NC), and Greenville County(SC). Using an innovative map-based survey, adult respondents (n=320) provided demographic/behavioral characteristics and indicated parks within one half-mile of their block group used within the past month. Parks (n=263) were audited using the Community Park Audit Tool. Measures were collected at the individual (park visitation, physical activity, sociodemographics), park (distance, quality, size), and block group/neighborhood (count of parks, population density, age structure, racial composition, walkability) levels. Random forests were employed to create an accurate predictive model that can handle complex datasets and provide insights into feature importance for making predictions. Random forest models were comprised of an ensemble of 100 decision trees with class weight adjustment. Python 3.10.9 and scikit-learn version 1.2.1 were used.
Results/findings: The random forest model ranked variables based on the significance of their impact on the model’s performance in predicting park use (feature importance (FI)): distance to parks (FI=0.24), park size (FI=0.21), park quality (FI=0.09), participant physical activity (FI=0.06), neighborhood age structure (FI=0.06), participant age (FI=0.06), count of parks (FI=0.05), neighborhood median household income (FI=0.05), population density (FI=0.05), neighborhood racial composition (FI=0.05), neighborhood walkability (FI=0.04), participant education (FI=0.02), participant race (FI=0.01), and participant gender (FI=0.01). The random model performed well in predicting park non-use (high precision and recall, F1-score=95%), with high overall accuracy (90%) to predict both park use and non-use. For park use, precision was sufficient (79%) but recall was lower (50%), resulting in a lower F1-score (61%).
Conclusions: Random forests proved effective in identifying crucial park use predictors. Enhancements in sample size, representativeness, and data quality can further improve predictive performance. Researchers looking to develop robust predictive models of park use, or other behavioral outcomes, should consider the use of random forests or other algorithm modeling techniques.

Objective: Many rural communities experience limited access to fruits and vegetables (FV) and may benefit from food environment interventions that increase the number of produce markets selling FV. Systems analysis is an innovative approach for informing policy, systems, and environmental (PSE) change interventions for the food retail environment. However, there has been little research. Methods: This study describes a new decision-making model that optimizes placement of new FV markets in a rural community in Texas. The decision-making model was based on an existing agent-based model for the same community and built using an open-source tool, Open Solver add-in, for Microsoft® Excel. Models used secondary sources of data. Three intervention factors were tested: recommended driving distance to nearest produce market, service capacity of new and existing FV markets, and financial resources. Each factor had three levels of low, normal, and high defined based on literature. Models estimated the potential effects on for three intervention outcomes: number of new FV markets, and ratio of fast food (FF) outlets to FV markets, and population coverage or proportion of local population with access to FV. In total, the analysis tested 27 different interventions and compared effects to a benchmark. Results: When no new FV markets were added to the local food environment or the benchmark, 46.4% of the population had access to FV markets and the ratio of FF outlets to FV markets was eight. The smallest increase in population coverage was 19% compared to benchmark, while other interventions increased population access to FV to 100%. Models showed that the largest relative gain in access to FV, from 29% to 37% for the local population, was at a lower level of financial resource availability ($15-20,000 per month). Conclusions: The most impactful interventions were less intensive with fewer financial resources or based on greater driving distances. Findings provide evidence for the potential effects of food environment changes for one rural Texas community. With this model, stakeholders can generate insights to inform context-specific decisions about their communities. In addition, this new decision-making model can be adapted for other communities to support PSE change interventions for nutrition.

Purpose: Crime continues to be a pervasive public health issue that disproportionately affects historically racialized communities in America. Studies have reported that food stores (e.g., grocery stores, corner stores, fast-food restaurants) can attract crime due to their high volume of cash transactions and lack of security measures. There is limited understanding of how exposure to crime in food stores affects the diets and health of community residents. This study aimed to fill this gap in knowledge by exploring the self-reported experiences of Black Americans.

Methods: In 2023, 502 Black-identifying adults (mean age: 47.2; 60.2% female) living in the U.S. completed a cross-sectional survey online. They self-reported their socio-demographic characteristics, height, weight, and prior experiences with crime at food stores in their community. Furthermore, they completed validated screeners for household food security status and fruit and vegetable (FV) intake. Crude and multivariable-adjusted regression models were used to examine associations between self-reported exposure to crime at food stores and the following variables: low food security status, obesity status, and daily servings of FVs (cups).

Results: Findings indicated that 150 (29.9%) survey respondents avoided one or more food stores in their community due to crime while 102 (20.5%) previously witnessed a crime at a food store in their community. After adjusting for age, sex, education level, and annual income, those that avoided food stores in their community due to crime had greater odds of low food security (OR: 2.16; 95% CI: 1.41 – 3.32) and obesity (OR: 2.12; 95% CI: 1.32 – 3.39) compared to others. In addition, they consumed fewer cups of FVs per day (p = 0.04). Those that witnessed a crime had significantly greater odds of low food security (OR: 3.78; 95% CI: 2.23 – 6.41) and consumed fewer cups of FVs per day compared to others (p = 0.03).

Conclusions: Exposure to crime in local food stores may have important health and nutritional implications for community residents. Future studies should explore these implications, particularly in racially minoritized populations that are disproportionately affected by crime.