Identifying Areas of High Vulnerability for Rural Veteran Food Insecurity

Abstract

Introduction/Objectives:

Americans, including veterans, living in rural areas experience higher rates of food insecurity than the general population in the United States, but rural veteran food insecurity remains an understudied subject. Due to a lack of data on the subject, this project aimed to use geographic information systems (GIS) mapping of proxy variables for rural veteran food insecurity in order to identify areas of high vulnerability where future research could be targeted.

Methods:

Key factors which may indicate vulnerability to rural veteran food insecurity were identified in a review of the literature. These factors were mapped in ArcGIS Pro, including veteran population, community health, county rurality, and locations of food deserts.

Results:

After areas were identified that met search criteria, 3 sites were identified which serve a highly rural patient population and whose population has the highest vulnerability to rural veteran food insecurity based on search criteria.

Conclusions:

This project demonstrates how GIS mapping can be used to identify regions in the United States at high vulnerability for rural veteran food insecurity. Findings from this project will be used by the project team to develop interventions for rural veteran food insecurity which can be implemented at the local or national level.

Introduction

Up to 24% of veterans in the United States experience food insecurity, or inconsistent access to nutritionally adequate and safe foods.^1,2 The veteran population is at a much higher risk for food insecurity than the civilian population, with an overall prevalence rate of food insecurity twice that of the of the whole US population. ³ Food insecurity rates vary among sub-populations of veterans, but impact all ages of veterans, including post-9/11 veterans, working age veterans, and veterans who are older adults.^{2,4 -6} More specifically, research indicates that veterans with young children, veterans living with disabilities, veterans with severe mental illnesses, veterans struggling with housing security, and veterans struggling with unemployment may be at higher risk for food insecurity.^{7 -11} Food insecurity, and subsequent malnutrition, has been linked to a number of health conditions (including obesity and cardiovascular risk) and to poor health outcomes and poorer health status overall.^12,13 A number of factors, including geography, have been shown to have an important impact on the likelihood of food insecurity in a community. For example, rural communities experience food insecurity at higher rates than urban areas, ¹⁴ often due to a lower density of food access locations and higher rates of poverty. ¹⁵ Despite the high rates of food insecurity among both veterans and rural populations, food insecurity among rural veterans remain an understudied topic.

Assessing the risk of food insecurity in a community can help identify which sub-populations may require targeted interventions. One strategy, geographic information systems (GIS) mapping of food systems, has been used in both urban and rural contexts to spatially demonstrate inequalities in access and multifactorial causes of food insecurity. ¹⁶ The majority of this mapping focuses on the location of food deserts. However, as a sole indicator of access, this singular focus on food deserts has been criticized because it can miss other intersectional factors which contribute to inequalities in the food system. ¹⁷ Because food systems are complex and include a diverse range of stakeholders and processes, traditional GIS mapping of food systems can provide an insufficient picture of the multiple dimensions which can contribute to food insecurity. Previous studies have shown a need for more holistic assessments of food security as a systemic problem. ¹⁸

With this concern in mind, mapping of multiple indicators of food security has been used in some cases to approximate areas impacted by multiple deprivations which can contribute to population food insecurity^16,19 and impact food access.^20,21 Such strategies allow for a more nuanced approach for identifying areas of need by considering a range of factors which may contribute to food insecurity beyond the physical presence of food. Considering the various risk factors that make rural veterans more susceptible to food insecurity and the limitations of using GIS mapping of single indicators, there is a need to similarly map multiple factors in order to identify areas of high need for this population.

In this context, a Veterans Health Administration (VHA) quality improvement project was designed with the goals of: (1) determining the prevalence and risk factors associated with food insecurity in rural veterans screening positive for food insecurity during primary care visits at VHA facilities; (2) describing the food access resources and services, food environment, and structural limitations to food access available to rural veterans; and (3) describing the perspectives, experiences, barriers, and strategies of community program staff, VHA providers, and rural veterans experiencing food insecurity. In order to accomplish project objectives, geographic information systems (GIS) maps were first created to identify counties in the United States with high vulnerability to rural veteran food insecurity. This manuscript describes the GIS mapping portion of the quality improvement project, completed in 2022.

Methods

To determine the key factors that would be used to classify areas of high vulnerability to rural veteran food insecurity, the project team first conducted a literature review of relevant research manuscripts and gray literature ⁸ (eg, research reports, presentations, etc.) published between 2010 and 2021. Due to the limited available literature on rural veteran food security specifically, literature searches primarily focused on barriers and contributors to food insecurity among veterans and rural populations as a whole. Layers were then created in ArcGIS Pro (2.8.0) ²² for key variables identified through the literature search using publicly available spatial data in federal and university databases. Key variables used to create layers on the maps and where the data were accessed by been summarized in Table 1.

Table 1.

Data Sources Used for Mapping Layers.

Data layer	Location of data	Year data layer was created	When data layer was accessed
Locations of VHA facilities	Homeland Infrastructure Foundation-Level Data (HIFLD) Open Data Portal	2020	April 2021
	VA facility websites	2021	August 2021
Food desert locations	US Department of Agriculture (USDA) Food Access Research Atlas	2022	June 2023
Degree of rurality	USDA ERA Rural-Urban Continuum Code (RUCC)	2013	April 2021
Percentage of population who are veterans	US Census Bureau American Community Survey (ACS)	2020	March 2021
Low community health	University of Washington County Health Rankings & Roadmaps (CHR&R) Program	2019	April 2021
Presence of Rural Patient Aligned Care Team (PACT) Social Work Program	Veterans Health Administration	2021	August 2021

Data layers were created to show:

1. Locations of VHA facilities: The location of 1364 VHA medical facilities were derived first from an existing publicly available GIS layer available through the Homeland Infrastructure Foundation-Level Data (HIFLD) Open Data portal, ²³ run by the Department of Homeland Security (DHS). The HIFLD layer listed contact information, geospatial data, and other facility details for 1046 VA Medical Centers (VAMC), Community Based Outpatient Clinics (CBOC), Veterans Centers (VCTR), and VA Nursing Homes (VANH) locations across the United States. VA websites listing each state’s facilities were double checked to identify facilities missing from the list. Geospatial and facility data were added for 318 additional VAMC and CBOC locations based on state facility websites (accessed in August 2021), which were missing from the original HIFLD layer.

2. Presence of a food desert: Food desert data was accessed through the open data portal for the US Department of Agriculture. ²⁴ Areas are defined by the USDA as food deserts if they are a low-income tract more than 1 mile from a grocery store in an urban area or 10 miles from a grocery store in a rural area. ²⁵

3. Degree of rurality: Area rurality is based on 2013 USDA ERS Rural-Urban Continuum Codes (RUCC), a classification scheme that distinguishes metropolitan counties by the population size of their metro area, and nonmetropolitan counties by degree of urbanization and adjacency to a metro area. RUCC data is publicly available. ²⁶ For the purposes of FY21 mapping, areas were included that have a RUCC of 8 (“Nonmetro - Completely rural or less than 2500 urban population, adjacent to a metro area”) or 9 (“Nonmetro - Completely rural or less than 2500 urban population, not adjacent to a metro area”).

4. Areas with a high proportion of veterans: The percentage of the civilian population over 18 who are veterans was used to estimate the overall proportion of Veterans in an area. These data were accessed through publicly available American Community Survey (ACS) 2020 data, which is compiled by the US Census Bureau. ²⁷ For the purposes of FY21 mapping, areas were considered to have a high Veteran population if the percentage of the civilian population over the age of 18 who were Veterans exceeded 8%, the national average.

5. Areas with low community health: Community health mapping data is based on health rankings from the 2019 County Health Rankings & Roadmaps (CHR&R) program, run by the University of Washington Population Health Institutes. ²⁸ The CHR&R program collects county level health data including health behaviors (eg, tobacco use, diet and exercise), clinical care (eg, access to healthcare, quality of health care), social and economic factors (eg, education, employment, income), and physical environment (eg, environmental quality, housing and transit availability). These data are collected from federal and private health surveillance systems including the National Center for Health Statistics (CDC), Behavioral Risk Factor Surveillance System (CDC), Food Environment Atlas (USDA), Fatality Analysis Reporting System (US DOT), and American Community Survey (US Census Bureau), among others. County Health Rankings compare the overall health of counties relative to other counties in the same state. For this reason, CHR&R rankings were reclassified based on whether countries represented the top 25%, top 50%, top 75%, or top 100% of health rankings. For the purposes of FY21 mapping, counties were considered to have relatively low health if counties fell into the bottom 50% of CHR&R rankings for their state.

Additional data were used to identify VHA facility locations that were most appropriate for virtual site visits which would look at existing barriers and facilities in each region in much greater detail. These other variables included the rurality of the medical center’s patient population and the presence of a Rural Patient Aligned Care Team (PACT) Social Work program ²⁹ at the facility, as this program represented an important local partner for the team. This information was accessed internally at the VHA.

Results

Shown in Figure 1, all VHA facility locations were mapped first along with county rurality to limit the selection to facilities which were located in rural areas.

Figure 1.

Veterans Health Administration (VHA) locations, including VA Medical Centers (VAMC), Community Based Outpatient Clinics (CBOC), and Veteran Centers (VCTR), and county rurality based on Rural-Urban Continuum Codes (RUCC). Darkest areas are the most rural.

The next step was to identify rural food deserts, which was done using the intersect tool so that identified areas were considered a USDA food desert and a rural RUCC (Figure 2). Once rural food deserts were identified, the boundary tool was used to limit VHA locations to those within a 20-mile radius of a rural food desert.

Figure 2.

County rurality (based on Rural-Urban Continuum Codes) and the location of food deserts, defined by the USDA as low-income tracts at least 1 mile from a grocery store in urban areas or 10 miles from a grocery store in rural areas. Darkest areas are the most rural. Areas with hatched lines are food deserts.

Next, areas were identified with a higher-than-average percentage of the population who were veterans (greater than the national average of 8%; Figure 3). Using the intersect tool, VHA locations were limited to those which were in proximity to a rural food desert and located in a census tract with a high proportion of veterans. Similarly, level of community health by county was mapped in order to identify areas with low community health, based on Community Health Ranking health factor data (which describes health behaviors, access to and quality of clinical care, social and economic factors, and the physical environment; Figure 4). Using the intersect tool, VHA locations were then limited to those counties with the lowest levels of community health.

Figure 3.

VHA locations which are 20 miles or less from a rural food desert and percentage of the population who are veterans by census tract. Darkest areas have the highest percentage of the population who are veterans.

Figure 4.

VHA locations within 20 miles of a rural food desert and level of community health. Counties are shown with the highest and lowest ratings of health factors based on Community Health Ranking health factor data. Darker areas have lower community health.

Once these steps had been completed, 33 VHA locations were identified which were in proximity to a rural food desert, had a higher-than-average veteran population, and low community health (Figure 5). Since all of the locations meeting search criteria were CBOCs, which are satellite locations for VAMCs, the parent facilities of each CBOC were also added to the map. In addition, Veterans Integrated Services Network (VISN) regions were mapped, to ensure that selected sites were not in the same service area. The outcome of this step is shown in Figure 6. As a final step, to identify locations best suited to the project goals, locations were then limited to those which served a majority rural patient population (more than 50% of patient home locations were rural) and had a Rural Patient Aligned Care Team (PACT) Social Work program at the facility (Figure 7).

Figure 5.

VHA locations within 20 miles of a rural food desert, within counties with a higher-than-average veteran population and low community health.

Figure 6.

VAMCs of CBOCs meeting search criteria and VISN boundaries.

Figure 7.

VAMCs of CBOCs meeting search criteria, VISN boundaries, VHA facilities with Rural PACT Social Work Programs, and VAMCs serving a majority rural patient population.

Three locations met all search criteria and were selected by the project team for further investigation: Erie, PA; Muskogee, OK; and Tomah, WI. Facilities in each of these locations serve a highly rural patient population with the highest vulnerability to rural veteran food insecurity based on search criteria.

Discussion

In this quality improvement project, we used a variety of factors to identify areas of high risk for rural veteran food insecurity, including food desert locations, degree of rurality, proportion of the population who are veterans, and overall community health. The data sources used for this study have been used in previous studies of food systems, drawn primarily from federal data portals and VA sources, have been used in many mapping projects. However, to the best of our knowledge, no other project has combined these resources in order to assess vulnerability to rural veteran food insecurity. Since spatial data that specifically describes rural veteran food insecurity is not available, this strategy demonstrates the way in which available data on related factors can be used to identify areas where there may be a need for future intervention. Specifically, GIS maps were used to identify VHA facilities which were most appropriate for virtual site visits to explore local barriers and facilitators for rural veteran food insecurity in more detail.

The methods used in this quality improvement project are consistent with previous studies which have used ethnographic data like average household income, eligibility for benefits, and rates of disability as proxy variables, in addition to the location of food deserts, for assessing an area’s vulnerability to food insecurity. ²¹ When evaluating an area for overall access to food, studies have shown it is important to take additional factors into consideration beyond the physical presence of food. Studies which uncritically use only the presence or absence of a food desert to assess risk of food insecurity risk misrepresenting the entrenched structural inequalities which contribute to food insecurity as well as the level of food available in those areas.^{17,30 -34} Criticism of the use of food deserts as an indicator for vulnerability is not purely pedantic, as food deserts are also commonly used as a metric in the provision of funds for community interventions designed to increase population health. The use of other indicators, as in this study, reflects the intersectional issue of food insecurity, which allows for more nuanced and sustainable interventions.

There are several important limitations to this study. Variables mapped were intended to be used as proxies for mapping rural veteran food insecurity itself because data layers do not currently exist which show rates of rural veteran food insecurity. While variables were carefully selected to best represent vulnerability for rural veteran food insecurity, future research would benefit from data sets which specifically measure food insecurity among rural veteran populations. In addition, the mapping used in this project only provides a broad overview of risk for rural veteran food insecurity and does not provide details about the specific points of food access of each region beyond the location of grocery stores. Previous research has shown that doing can better reflect the lived experience of navigating food systems for food insecure households. ³¹ For this reason, future stages of the project must both incorporate detailed descriptions of local food systems and qualitative assessment of local food access through interviews with community stakeholders.^{35 -37}

Conclusion

The GIS mapping steps used in this project demonstrate how visualization of variables can help organizations like the VA identify areas of high vulnerability to issues like rural veteran food insecurity. By highlighting the regions which are in most need of targeted intervention, future phases of this research project will be able to conduct in-depth investigations of those local food systems and existing barriers and facilitators to food security in each region. Ultimately, the quality improvement project will result in recommendations for programmatic interventions which can address existing barriers to food access and best practices for maintaining rural veteran food security across the United States.