Skip to main content
Journal of General Internal Medicine logoLink to Journal of General Internal Medicine
. 2022 Jan 3;37(12):2941–2947. doi: 10.1007/s11606-021-07229-y

Mapping Rural and Urban Veterans’ Spatial Access to Primary Care Following the MISSION Act

Eliana Sullivan 1,, Whitney E Zahnd 2, Jane M Zhu 3, Erin Kenzie 1, Mary Patzel 1, Melinda Davis 1,4,5
PMCID: PMC9485404  PMID: 34981345

Abstract

Background

The 2018 MISSION Act sought to improve Veterans’ access to primary care by allowing Veterans living more than 30 min from VA care to utilize non-VA clinics. The impact of this legislation may vary for rural compared to urban Veterans.

Objective

Assess the extent to which the 2018 MISSION Act facilitates spatial access to primary care for Veterans living in rural versus urban Oregon.

Design

We identified locations of all VA and non-VA primary care clinics in Oregon then calculated 30-min drive-time catchment areas from census tract centroids to the nearest clinics. We compared measures of spatial access to primary care for Veterans in rural, micropolitan, and urban areas.

Participants

American Community Survey data representing Oregon adults.

Main Measures

Two measures of spatial access focusing on the number of clinics (supply), and an access index based on the two-step floating catchment area method (2SFCA) which accounts for number of clinics (supply) and population size (demand).

Key Results

Compared to only 13.0% of rural Veterans, 83.6% of urban Veterans lived within 30 min’ drive time of VA primary care. Given the MISSION Act’s eligibility criteria, 81.6% of rural Veterans and ~ 97% of urban and micropolitan Veterans had spatial access to primary care. When accounting for both supply and demand, rural areas had significantly higher access scores (p < 0.05) compared to urban areas.

Conclusions

Using MISSION Act guidelines for Veteran access to primary care, rural compared to urban Veterans had less spatial access based on clinic number (supply), but more access when considering clinic number and population size (supply and demand). Geographic Information System (GIS) spatial techniques may help to assess changes in access to care. However, these methods do not incorporate all dimensions of access and work is needed to understand whether utilization and quality of care is improved.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11606-021-07229-y.

Keywords: Spatial access, Veterans, Primary care, Rural

INTRODUCTION

Rural Veterans comprise a quarter of the 20 million Veterans in the USA1 and tend to have higher disease prevalence and lower quality of life scores than their non-rural counterparts.2 Due to a scarcity of health care clinics in rural regions, rural Veterans have limited access to primary care.3 Studies suggest that spatial access — whether an individual lives within a certain travel time or distance standard of at least one primary care clinic4 — is a key factor that influences whether or not Veterans receive primary care, especially when telehealth is not an option.57 Greater primary care access is associated with lower mortality rates, longer lifespans, increased access to specialty care, and improved health outcomes.810.

The Veterans Administration (VA) provides primary care for over nine million Veterans.11 However, long travel distances and lengthy wait times for VA appointments prompted legislation that expanded access to non-VA services (“community care”) under certain conditions.12 In 2014, the Veterans Choice Program allowed Veterans to utilize non-VA primary care if they met specific eligibility criteria, including living more than 40 miles from a VA clinic with a full-time primary care provider.13,14 Prior research found that the Veterans Choice Program did not improve primary care availability for Veterans.15 To further broaden access,16 regulations issued under the Maintaining Internal Systems and Strengthening Integrated Outside Networks (MISSION) Act of 2018 expanded access to non-VA primary care for Veterans living more than 30 min by car from a VA primary care clinic.17.

While research has explored the potential and observed impacts of the MISSION Act on specialty care,18,19 we are not aware of research that has examined its effects on Veterans’ spatial access to both VA and non-VA primary care services, particularly across geographies. Rigorous spatial approaches that account for supply and demand for health care have been used extensively to evaluate access to health services, such as primary care for the general population,20,21 but these have not been used to assess Veteran spatial access to primary care at areas smaller than the county level. Given the important associations between primary care and population health, research is needed to understand how spatial access to primary care for Veterans, particularly those in rural areas, may have changed following the MISSION Act.

Therefore, we examined how the MISSION Act affects Veterans’ spatial access to care using two measures of access across rural–urban designations in the state of Oregon, which has one of the largest concentrations of aging Veterans in the country and significant geographic dispersion.22 Specifically, we used Geographic Information System (GIS) analysis to (1) describe the geographic distribution of all VA and non-VA primary care clinics; (2) explore the relationship between primary care service location and Veterans’ residence in urban, micropolitan, and rural census tracts; and (3) utilize and compare two methods for assessing spatial access, one considering supply only (number of clinics) and the other incorporating both supply and demand (number of clinics and census tract population size). We hypothesized that urban Veterans would have better spatial access to primary care than rural Veterans using both methods.

METHODS

We used two methods to calculate Veterans’ spatial access to VA and non-VA primary care clinics across all populated census tracts (N = 822) in Oregon. These methods utilized measures of supply and demand and focused on differences across rural, micropolitan, and urban areas. In keeping with the MISSION Act’s eligibility criteria, we identified clinics within 30 min’ drive time of Veterans’ home census tract. This study was approved by the Veterans Affairs Portland Health Care System (VAPORHCS) and Oregon Health & Science University (OHSU) joint IRB (eIRB#20,843). Study data are available on request to the corresponding author at the discretion of the study Principal Investigator.

Data Sources

Primary Care Clinic Listing

Between September 2019 and February 2020, we prepared and validated a comprehensive list of all adult primary care clinics in Oregon (Fig. 1). We used the Oregon VA website to compile a list of VA primary care clinics in the state (N = 22).23 To identify non-VA primary care clinics, we used an existing list of primary care clinics engaged in projects with the Oregon Rural Practice-based Research Network (ORPRN)24 supplemented by information from state public health, health plan, and health system websites and other public clinic registries. We included clinics designated as Patient-Centered Primary Care Home (PCPCH),25 Vaccines for Children,26 School-based Health Center (SBHC, some of these clinics serve adults as well as youth),27 Federally Qualified Health Center (FQHC),28 Rural Health Clinic (RHC),29 and Indian and Tribal Health Services.30 Of 2,659 clinics identified, 1,049 unique clinics remained after removing duplicates. Another 317 clinics were excluded after online and direct phone verification found that they were pediatric-only locations (47%), did not provide primary care services (38%), or were closed or non-existent (15%). Our final sample included a total of 732 clinics — 22 VA and 710 non-VA — that were actively providing adult primary care.

Fig. 1.

Fig. 1

Flow selection diagram of Oregon primary care clinics serving adults, February 2020

Census Tract–Level Population Data and Rurality Classifications

We used the 2014–2018 5-year American Community Survey (ACS) Veteran Status and Demographic variables to identify the number of Veterans and other adults living in each Oregon census tract.31 Census tracts are administrative units nested within counties that generally include 1,200 to 8,000 people.32.

We applied the US Department of Agriculture’s Rural–Urban Commuting Area (RUCA) primary codes33 to classify these census tracts as urban, micropolitan, or rural. Primary RUCA codes range from 1 to 10 and categorize census tracts based on their location within an urbanized area or cluster and their commuting patterns.33 For this study, a census tract with a primary RUCA code of 1–3 was considered urban (≥ 50,000 residents), 4–6 micropolitan (between 10,000 and 49,999 population or a commuting pattern to larger areas), and 7–10 rural (< 10,000 people or limited commuting to an urbanized area); this approach aligns with definitions used in prior research.34,35 We used census tracts as the unit of analysis because it is the smallest geographic unit for which estimates of Veteran populations and a rural–urban metric are available.

Data Analysis

We used two methods to assess spatial access to primary care — the first method focuses on supply (30-min network buffer) and the second accounts for both supply and demand (two-step floating catchment area, 2SFCA).20 For both methods, we geocoded and mapped the locations of VA and non-VA primary care clinics using the ArcGIS Online World Geocoding Service.36.

Method #1: Access Based on Supply (30-min Drive Time Network Buffer Method)

To operationalize the MISSION Act primary care access definition, we used the ArcGIS Online drive-time analysis tool to construct 30-min network buffers (boundaries) around the VA and non-VA primary care clinics to capture areas within 30 min’ drive time of a given clinic. The Veteran population of a census tract was considered to be within 30 min of primary care if its population-weighted centroid fell within the buffer of at least one VA and/or non-VA primary care clinic. A population-weighted centroid accounts for population distribution within a census tract, which is particularly important in sparsely populated but geographically large rural census tracts in which the population may not be concentrated in the geographic center of the census tract.37.

Method #2: Access Based on Supply and Demand (Two-Step Floating Catchment Area Method, 2SFCA)

Consistent with Penchansky and Thomas’s multi-component definition of access, access consists of not only spatial dimensions (i.e., if a clinic exists), but also availability, accommodation, affordability, and acceptability.38 The drive-time method only incorporates the accessibility component of access, while other approaches, such as the two-step floating catchment area (2SFCA) method, consider both accessibility and availability. Thus, we employed the 2SFCA method which calculates a spatial accessibility score for each census tract based on proxy measures of supply (i.e., availability operationalized as the number and geocoded location of primary care clinics) and demand (i.e., those who may need to access primary care services operationalized as the Veteran or full adult population within a census tract).20 A strength of this approach is that it considers not only how far a Veteran lives from a health care service, but also how many others may need access to the same care.20 The 2SFCA method has been used extensively for policy-relevant assessments of spatial access to primary care.21,3941.

The first step in the 2SFCA method is to formulate a 30-min drive-time catchment area (buffer) around the geocoded location of each primary care clinic (per method #1), then calculate a clinic-to-population ratio (Veterans or all adults) for each clinic. This clinic-to-population ratio is determined by the populations of the census tracts whose centroid falls within the catchment area for the respective clinic. For example, if three census tract centroids with populations of 100 people were in the 30-min drive-time catchment area of a clinic, the clinic-to-population ratio would be 1:300. The second step is to create a 30-min drive-time catchment area for each census tract, centered on the population-weighted centroid, and to sum the clinic-to-population ratio of any primary care clinics that fall within it, yielding a spatial accessibility score for each census tract. In this step, if there were three clinics in the census tract’s catchment area, and each had a clinic-to-population ratio of 1:300, then the spatial accessibility score would be 3/300 (or 0.01).

A spatial accessibility score of zero indicates that the census tract centroid is not within 30 min of any primary care clinic. A score above zero indicates that there is at least one clinic within 30 min of the centroid. Higher scores signify a larger clinic-to-population ratio in that census tract.39 When calculating spatial accessibility scores for VA primary care clinics, we used the Veteran population as a proxy for demand, and when considering non-VA primary care clinics, we used the full adult population.

Spatial and Statistical Analyses

We created three maps to visualize our analyses: (1) 30-min network buffers around VA and non-VA primary care clinic locations (method #1), (2) overlap between primary care access within 30 min and level of rurality (bivariate map), and (3) distribution of spatial accessibility scores (method #2). We performed one-way analysis of variance (ANOVA) and post hoc Tukey’s tests to examine differences in spatial access to primary care across geographic categories (rural, micropolitan, urban). All statistical analyses were conducted in R version 3.6.3.

RESULTS

Rurality of Oregon Census Tracts and Primary Care Clinic Type

Nearly three-fourths of Oregon census tracts (73%) were urban, 17% were micropolitan, and 10% were rural. Approximately 288,450 Veterans lived in Oregon — 69% lived in urban regions, 20% in micropolitan regions, and 11% in rural areas. Only 7% of non-Veteran adults lived in rural regions, compared to 11% of Veterans, suggesting that Veterans are overrepresented in rural regions. Nearly all (97%) clinics were non-VA primary care clinics. Among these, 58% of rural and 82% of urban clinics had no federal designations (e.g., FQHCs, RHCs, VA clinic).

Access to Primary Care Based on Supply (Method #1)

As shown in Table 1, 84% of urban Veterans lived within 30 min of a VA primary care clinic, compared to just 65% of micropolitan and 13% of rural Veterans. In contrast, 97% of urban and micropolitan Veterans lived within 30 min of a non-VA primary care clinic, as did 82% of rural Veterans. Nearly one-fifth (18%) of rural Veterans still lived further than 30 min from any primary care clinic, compared to just 3% of urban and micropolitan Veterans. The clinic locations and their 30-min drive-time buffers are visualized in Appendix.

Table 1.

Veterans Living Within 30 min of Primary Care, by Rurality

VA primary care Non-VA primary care No primary care
Urban (N = 199,329) 83.61% 97.35% 2.65%
Micropolitan (N = 58,623) 64.91% 96.53% 3.47%
Rural (N = 30,588) 13.01% 81.63% 18.37%
Total (N = 288,540) 72.33% 95.52% 4.48%

Figure 2 shows both the geographic designation (urban, micropolitan, rural) of each census tract and whether it is within 30 min of VA (Fig. 2a) or non-VA (Fig. 2b) primary care, with greater spatial access to VA primary care in urban areas. The introduction of non-VA primary care shifts access from a majority red (no access) to a majority white/blue (access) across all geographic groupings.

Fig. 2.

Fig. 2

Bivariate map: spatial access to primary care (within 30 min) by rurality and census tract for Oregon Veterans. This map displays spatial access to primary care as determined by whether or not a census tract is within 30 min of primary care. It also visualizes the census tract’s rurality designation. Red tones indicate that a census tract is not within 30 min of primary care (no spatial access) while white/blue tones indicate that a census tract is within 30 min of primary care (spatial access). The color’s shade gradient indicates rurality — the lightest areas are urban, the darkest are rural, and those in the middle are micropolitan

Access to Primary Care Based on Supply and Demand (Method #2)

Figure 3 displays spatial accessibility scores, calculated using the 2SFCA method, from lowest to highest accessibility by quartile. When considering VA primary care clinics alone (Fig. 3a), micropolitan areas had the highest mean spatial accessibility score, which was 1.5 and 3.9 times greater than the mean scores of rural and urban areas, respectively. Upon including non-VA primary care (Fig. 3b), rural areas had the highest mean spatial accessibility score, which was 1.05 and 1.9 times greater than the scores of micropolitan and urban areas respectively. A Tukey’s post hoc analysis revealed a statistically significant difference between the mean spatial accessibility score of urban areas which was lower (less spatial access) than those of micropolitan and rural regions given utilization of VA or non-VA primary care (p < 0.05).

Fig. 3.

Fig. 3

Spatial accessibility scores for VA and non-VA primary care for Veterans by census tract. This map displays spatial accessibility scores as determined by the two-step floating catchment area method. Red census tracts have the lowest spatial accessibility scores while dark green ones have the highest. Orange and light green regions have spatial accessibility scores in the second and third quartiles respectively. 3a represents when Veterans can only utilize VA primary care, and 3b represents when Veterans can utilize non-VA primary care

Using one-way ANOVA tests, we identified that mean spatial accessibility scores increased significantly (p < 0.05 for all geographic groupings) when Veterans could access non-VA primary care compared to VA primary care only. Specifically, the mean spatial accessibility scores were 4.12, 1.95, and 2.93 times greater in urban, micropolitan, and rural areas, respectively, when Veterans could access non-VA primary care per the MISSION Act’s criteria.

DISCUSSION

Our findings show that a higher percentage of urban Veterans in Oregon live within 30 min of VA primary care, compared to their more rural counterparts (70.6 percentage point difference). Introducing access to non-VA care via community care initiatives, most recently the MISSION Act, significantly increased the number of Veterans across geographies who lived within 30 min of primary care. Interestingly, our two measures of spatial access yielded different results regarding whether spatial access improved more for urban or rural Veterans following the MISSION Act. The supply-only method suggests that urban Veterans benefited more because a greater percentage of urban census tracts are within 30 min of primary care. However, when potential demand is considered, rural census tracts had greater spatial accessibility scores.

These findings are consistent with other studies demonstrating that diverse geospatial analysis methods may be needed to identify areas with poor access to care. For example, people in urban areas typically have shorter average drive times to health care services than their rural counterparts,4245 and studies using drive time thresholds find generally greater spatial access in urban compared to rural areas. However, using methods that take into account service demand, as well as supply, may help to identify otherwise overlooked areas with relatively poorer access to care.43 Future research could apply these methods to non-Veteran populations to identify regions with insufficient access to primary care and prioritize the allocation of FQHCs and RHCs accordingly.

Our findings suggest that spatial access is likely a necessary but insufficient component of meaningful access to care. Prior studies found that the Veterans’ Choice Program did not change Veterans’ perceptions of access or availability of both VA- and non-VA care.15,46,47 Additionally, studies suggest that expansion of Veteran access to non-VA care may be associated with higher rates of hospitalization, care duplication, and increased costs stemming from care fragmentation when Veterans utilize parallel systems of VA and community care.48 Finally, the VA has a fixed budget for Veteran care; widespread utilization of non-VA primary care could rapidly deplete this budget, thereby reducing the capacity of VA services without improving the overall care that Veterans receive.49.

Beyond clinic presence and patient demand, factors such as appointment availability, insurance acceptance, transportation availability, quality of care, and care coordination may impact Veterans’ experiences and outcomes. Additionally, telehealth infrastructure has grown considerably across both VA and non-VA settings, expanding access to primary and specialty care, increasing the importance of technological access and digital literacy, and potentially decreasing the reliance on spatial access.50,51 While we identified that the MISSION Act was associated with greater numbers of accessible facilities, future research should incorporate additional dimensions of access to understand how increases in spatial access translate to realized utilization and health outcomes.

Our study has several limitations. First, as noted above, spatial analysis does not account for other important factors affecting access to care, including wait times, service hours, and insurance acceptance, nor does it evaluate downstream utilization. Second, we only considered clinics, census tracts, and drive time radii within Oregon which could underestimate access to care for Veterans who travel across state lines. Third, while Oregon has a high concentration of aging and rural Veterans,22 findings in one state may not be generalizable to other states. Fourth, we did not compare the new MISSION Act criteria to prior community care access criteria, as our focus was on current conditions; comparing accessibility with the MISSION Act to prior community care policies could be explored in subsequent research. Fifth, while drive-time is a more accurate measure of spatial access than distance, it is not representative for Veterans who lack access to transportation, especially in rural regions where public transit is scarce.52 Finally, the most recent RUCA codes were based on 2010 census rurality designations and may not reflect more recent demographic and population changes.53.

Taken together, our findings suggest that in order to most efficiently increase spatial accessibility, the supply or availability of clinics could be increased in rural regions, due to small population sizes and the relatively large number of census tracts with no spatial access to any primary care. However, increasing clinics in urban areas may reach a greater number of Veterans overall, a factor that reinforces structural urbanism — a bias in health care that favors large populations and disadvantages rural communities.54 Accordingly, future research could evaluate policies to expand Veterans’ access to primary care across the country by considering clinical capacity, availability, and quality.

Supplementary Information

ESM 1. (1MB, png)

(PNG 1.00 mb).

Acknowledgements

Thank you to Roselie Agulto, Jack Lazar, Stephanie Hyde, Julian Bermudez, Jean Campbell, Cort Cox, Kylie Lanman, and Caitlin Dickinson for support in creating and validating the primary care clinic listing.

Funding

Funding for this research was provided by the U.S. Department of Veterans Affairs (VA) Office of Rural Health (OMAT #15529). The views expressed in this manuscript are those of the authors and do not necessarily reflect the views of the VA or the US government. Visit www.ruralhealth.va.gov to learn more.

Declarations

Conflict of Interest

The authors declare that they do not have a conflict of interest.

Footnotes

Prior Presentations

None

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ESM 1. (1MB, png)

(PNG 1.00 mb).


Articles from Journal of General Internal Medicine are provided here courtesy of Society of General Internal Medicine

RESOURCES