Statistical Visualizations Suggest Food Environment Variables Play a Role in COVID-19 Outcomes in Wyoming
Team Nutritional Advantage: Ben Romanjenko, Mercedes Fermelia, Tom Hafner, & Gabby Flores
Preliminary Work & Researcher Standpoint
This project was part of a senior undergraduate Capstone course at the University of Wyoming. While R provides powerful statistical analysis and data visualizations, when viewing this data it is important to consider that the individuals that put this app together are not experts in R coding or statistics. This information can serve as a jumping point to fuel future research conducted by this team and other researchers looking at the effects of food environments and COVID-19 outcomes in rural areas.
While many conflate rurality with agricultural abundance, data show that food insecurity is as big of a problem in Wyoming as it is in the broader United States. Above is a histogram generated by our research team using the 2020 Map the Meal data set from Feeding America.
Introduction
With necessary closures to limit the spread of the pandemic, food networks have experienced interruptions (Patrick et al., 2020). Food inaccessibility is often disproportionate, leaving marginalized communities as the most impacted (Porter, 2020; Burman, 2020). Access to diets consisting of energy-rich, nutrient-poor foods have been shown to be linked with metabolic disorders like obesity and diabetes, and these comorbidities have further implications to severe COVID-19 complications (Kassir, 2020). Food insecurity can also be a considerable issue for children (Kane et al., 2015). Through a county-specific data analysis of Wyoming, the goal of this research is to elucidate the connections between food environment and the outcome of COVID-19 infections.
Specific Aims
- To identify associations between food environment, diet, obesity, and COVID-19 outcomes, we will acquire food inaccessibility data from the USDA ERS and Feeding America databases. We will acquire county-specific obesity (BMI), diabetes, overweight, and COVID-19 data by utilizing the WPHL’s chronic disease database and through John Hopkins database, respectively.
- We will also be using data sets from USDA ERS, Feeding America, and WPHL to identify associations between food insecurity, school-provided meals, children in low-income and impoverished households, and the effects of pandemic on food access.
- We will gather pre-COVID data sets from Census Household Pulse Survey accessed through Northwestern University to be able to compare to food accessibility, food networks, and school-provided meals regarding post-COVID changes.
- We will gather data to determine if there are statistically significant improvements in pre-existing health conditions, COVID-19 outcomes after participation in supplemental nutritional food programs, such as the National School Lunch Program, and community and home gardening programs.
- We will read data into R and perform statistical analysis, then overlay these data onto maps to visualize geographic distribution.
- We will use RShiny/ArcGIS software to create interactive visuals that will be uploaded to the University of Wyoming Microbiology Capstone website that can be accessed by the general public.
Objectives
- To better understand the relationships between food environment, diet, obesity, and the outcome severity of COVID-19 in Wyoming.
- To better understand how COVID-19 has impacted school-provided meals and food insecurity regarding children in low-income and impoverished households in Wyoming.
- To better understand why Fremont county and the WRIR are being inequitably affected by COVID-19.
- To determine if access to locally grown produce (through existing dietary supplement and food justice programs) on the WRIR is associated with multifaceted improvement to overall health.
- Through outreach, to engage and provide our stakeholders, including the Wyoming Public Health Lab, WRIR, Riverton Peace Mission, Fremont County Public Schools, and the Fremont County communities, with information and data visuals that work to attract investors to support existing food and health justice initiatives.
- To create information that is culturally mindful and aligns with the values of those most inequitably affected.
Our Initial Hypotheses Were
H1: Wyoming counties with more food inaccessibility, as defined by Feeding America, and poorer diets will positively correlate with more chronic diseases and more COVID-19 morbidity and mortality.
H2a: Food insecurity will be greater from March to August 2020 than before the COVID-19 shutdowns for low-income and impoverished households.
H2b: School-provided meals through programs like the National School Lunch Program and the School Breakfast Program increase food security in children from food-insecure households.
H3a: Native Americans living in Wyoming are disproportionately affected by COVID-19, and these disproportionate frequencies will increase for Fremont County as compared to the rest of Wyoming.
H3b: Native Americans living in Wyoming who are currently involved in community-based gardening and/or food distribution programs have a lower prevalence of COVID-19.
H1: Wyoming counties with more food inaccessibility, as defined by Feeding America, and poorer diets will positively correlate with more chronic diseases and more COVID-19 morbidity and mortality.
H2a: Food insecurity will be greater from March to August 2020 than before the COVID-19 shutdowns for low-income and impoverished households.
H2b: School-provided meals through programs like the National School Lunch Program and the School Breakfast Program increase food security in children from food-insecure households.
H3a: Native Americans living in Wyoming are disproportionately affected by COVID-19, and these disproportionate frequencies will increase for Fremont County as compared to the rest of Wyoming.
H3b: Native Americans living in Wyoming who are currently involved in community-based gardening and/or food distribution programs have a lower prevalence of COVID-19.
Methods
Results
By meeting a p-value of less than 0.05, the following findings were deemed as significant correlations.
Discussion
The relationships found between COVID-19 and food variables may be because of the following:
With decreasing access to stores there is an increased chance for infection from COVID-19. Those living in rural areas may have to travel long distances to access culturally appropriate food, and so many families buy items in large quantities. Buying in bulk makes it so these people have to spend more time in the stores when they are able to access them, increasing their time in the enclosed space and increasing chance for infection.
With increasing access to SNAP and WIC authorized stores, there is an increased chance acquiring COVID-19. If parents cannot afford childcare, they must bring their children with them grocery shopping, increasing exposure to COVID-19 for the entire family. This may also be because busy parents may need to buy groceries in bulk to accommodate babysitters, work schedules, etc. especially with school closures at the start of the pandemic.
For cases, deaths, and cases per 100,000, farmers markets were strongly correlated. This came as a surprise as outside spaces where farmers markets occur do not typically facilitate high transmissions, we questioned the activities that may occur before or after the market. Such activities could be eating inside restaurants, shopping, and socializing. These activities being in closed spaces would highly increase one’s chance for contracting and dying of COVID-19.
The relation of Deaths per 100,000 and Fatality rate are correlated to students eligible for free and reduced priced lunch. This may be because students that are eligible for these programs come from low-income families who are more likely to struggle with food security. Low-income families are likely overrepresented in essential worker positions, who typically maintain higher contact with the general public. Being low income also generally leads to less access to health care and testing services. Less access to testing means someone can be positive and not know, less access to healthcare if COVID-19 symptoms get worse. When school was canceled, a lot of these free and reduced lunch options were more difficult to access. Some programs were still available to ensure students were able to receive food, which creates a similar scenario to the farmer’s market situation. These factors may explain why this strong correlation was only observed with deaths per 100,000 and fatality rate.
Interestingly, diabetes had a significant correlation with COVID-19 deaths & fatality rate per 100,000 from two independent data sets (USDA-ERS and WPHL - % Diabetes/county). This correlation was also found in another article (Vas et al., 2020). It was interesting that obesity did not have a correlation because many articles suggested both diabetes and obesity (Alkhatib et al., 2020; Kassir, 2020; Vas et al,. 2020); however, diabetes is a very clinical chronic disease, whereas obesity, as measured by BMI, is essentially a mathematical equation to represent a more-extreme version of overweight and as a precursor to more significant health ailments.
Not indicated in the table were several non-findings. These included no correlations between COVID-19 Outcomes, % Adult Food Insecurity, % Child Food Insecurity, CSA Farms, Median Household Income, Poverty Rate, and Child Poverty Rate. We believe these non-findings to be rooted in cofounding variables that include underreporting or gaps in data.
With decreasing access to stores there is an increased chance for infection from COVID-19. Those living in rural areas may have to travel long distances to access culturally appropriate food, and so many families buy items in large quantities. Buying in bulk makes it so these people have to spend more time in the stores when they are able to access them, increasing their time in the enclosed space and increasing chance for infection.
With increasing access to SNAP and WIC authorized stores, there is an increased chance acquiring COVID-19. If parents cannot afford childcare, they must bring their children with them grocery shopping, increasing exposure to COVID-19 for the entire family. This may also be because busy parents may need to buy groceries in bulk to accommodate babysitters, work schedules, etc. especially with school closures at the start of the pandemic.
For cases, deaths, and cases per 100,000, farmers markets were strongly correlated. This came as a surprise as outside spaces where farmers markets occur do not typically facilitate high transmissions, we questioned the activities that may occur before or after the market. Such activities could be eating inside restaurants, shopping, and socializing. These activities being in closed spaces would highly increase one’s chance for contracting and dying of COVID-19.
The relation of Deaths per 100,000 and Fatality rate are correlated to students eligible for free and reduced priced lunch. This may be because students that are eligible for these programs come from low-income families who are more likely to struggle with food security. Low-income families are likely overrepresented in essential worker positions, who typically maintain higher contact with the general public. Being low income also generally leads to less access to health care and testing services. Less access to testing means someone can be positive and not know, less access to healthcare if COVID-19 symptoms get worse. When school was canceled, a lot of these free and reduced lunch options were more difficult to access. Some programs were still available to ensure students were able to receive food, which creates a similar scenario to the farmer’s market situation. These factors may explain why this strong correlation was only observed with deaths per 100,000 and fatality rate.
Interestingly, diabetes had a significant correlation with COVID-19 deaths & fatality rate per 100,000 from two independent data sets (USDA-ERS and WPHL - % Diabetes/county). This correlation was also found in another article (Vas et al., 2020). It was interesting that obesity did not have a correlation because many articles suggested both diabetes and obesity (Alkhatib et al., 2020; Kassir, 2020; Vas et al,. 2020); however, diabetes is a very clinical chronic disease, whereas obesity, as measured by BMI, is essentially a mathematical equation to represent a more-extreme version of overweight and as a precursor to more significant health ailments.
Not indicated in the table were several non-findings. These included no correlations between COVID-19 Outcomes, % Adult Food Insecurity, % Child Food Insecurity, CSA Farms, Median Household Income, Poverty Rate, and Child Poverty Rate. We believe these non-findings to be rooted in cofounding variables that include underreporting or gaps in data.
Conclusion & Future Work
In our research, although having many non-findings, our work allowed us to better understand the relationships between COVID-19, food environment factors and race/ethnicity. Our biggest finding was ensuring our work had strong validity when two seperate datasets (USDA-ERS and WPHL) for diabetes data both came back with significant correlation to COVID-19. In addition, we discovered that, while learning how to code is difficult, if you have the right resources and support you can still attain a visually stunning and relevant product.
With more time in the coming semester, we plan to look deeper into literature to better understand the reasoning behind our positive findings. In addition, we will scale up our analysis and include ArcGIS to help validate both the findings and the non-findings as well as look further into our H3a and H3b hypotheses.
With more time in the coming semester, we plan to look deeper into literature to better understand the reasoning behind our positive findings. In addition, we will scale up our analysis and include ArcGIS to help validate both the findings and the non-findings as well as look further into our H3a and H3b hypotheses.
Acknowledgements
Community Partner:
Wyoming Public Health Lab
Wyoming Public Health Lab
- Special thanks to Dr. Noah Hull, Dr. Heather Talbott, and Dr. Lynette Gumbleton
- Rachel Watson
- Ella DeWolf
- Sierra Jech
- Riverton Peace Mission
- Chesie Lee and L’Dawn Olsen
- Darrah Good-Voice Elk Perez
- Alma Law
- Erin Burman
- Jaynie Welsh
- Dr. Paddinton Hodza
- Dr. Christine Boggs
- Dr. Gerry Andrews
- USDA-ERS
- Feeding America
- John Hopkins COVID-19 Data
- Wyoming Public Health Labs
- Census Household Pulse Survey
