National overview

Across the 50 U.S. States and the District of Columbia, there were 3,142 counties and county-equivalent jurisdictions in 2017, and only 131 (4.2 percent) had no markers of population risk in our analysis, with the remaining, vast majority of counties having between 1 and 13 markers. This suggests that nearly all U.S. counties have segments of their populations with one or more risk factors for health service access limitations and poor population health. Moreover, there were 206 U.S. counties (6.6 percent) with 10 or more markers, and 9 counties (0.3 percent) had the maximum, 13 markers, suggesting higher overall risk levels for populations within those counties. These 9 counties lay entirely within either HHS Region 4 (Alabama, 2 counties; Georgia, 1 county; Mississippi, 4 counties) or Region 6 (New Mexico, 1 county; Texas, 1 county), roughly the southeastern and south central United States, respectively.

Primary care HPSA counties

Looking specifically at pcHPSA counties, 89 percent of U.S. counties were classified as pcHPSA counties in our analysis (Table 2), as determined through a regulatory process using health care provider and delivery metrics. This proportion may appear high but likely reflects the recognized national-level deficit in primary care providers. [8, 9] The proportion varied by HHS Region, with the highest percentages of pcHPSA counties found in Region 9 (100 percent) and Region 10 (98 percent), which encompass the nation’s western-most states. In four additional Regions across the nation (Region 1, Region 4, Region 6, and Region 8), more than 90 percent of all counties were classified as pcHPSA counties. The lowest proportion of counties classified as pcHPSA counties was found in Region 3, encompassing the Mid-Atlantic states, where 79 percent of counties were considered pcHPSA counties.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of pcHPSA counties, by HHS Region, 2017.

https://doi.org/10.1371/journal.pone.0231443.t002

The picture becomes more complicated when we looked by HHS Region at the number of pcHPSA counties with more than six SDOH markers. This threshold is based on the 75^th percentile of the total number of markers across all U.S. counties, and reveals marked differences. Disparities were particularly evident in HHS Region 4 and Region 6, which include states in the southeastern and south central United States and where more than a third of the counties in each Region were pcHPSA counties with more than 6 of the 15 studied SDOH markers. Figs 1 and 2 and 3 illustrate the county-level variability in the number of markers, focusing first across all states, and then viewing HHS Region 4 and Region 6 more closely.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Markers of medical, economic, and geographic vulnerability in the United States, 2017.

Map prepared using R’s open source package, usmaps, developed by Paolo Di Lorenzo under a GNU General Public License (GPL), Version 3 (https://cran.r-project.org/package=usmap).

https://doi.org/10.1371/journal.pone.0231443.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Markers of medical, economic, and geographic vulnerability, HHS Region 4, 2017 Alabama, Florida, George, Kentucky, Mississippi, North Carolina, South Carolina, Tennessee.

Map prepared using R’s open source package, usmaps, developed by Paolo Di Lorenzo under a GNU General Public License (GPL), Version 3 (https://cran.r-project.org/package=usmap).

https://doi.org/10.1371/journal.pone.0231443.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Markers of medical, economic, and geographic vulnerability, HHS Region 6, 2017 Arkansas, Louisiana, New Mexico, Oklahoma, Texas.

Map prepared using R’s open source package, usmaps, developed by Paolo Di Lorenzo under a GNU General Public License (GPL), Version 3 (https://cran.r-project.org/package=usmap).

https://doi.org/10.1371/journal.pone.0231443.g003

SDOH marker co-occurrence and variability

Looking at the most common markers across pcHPSA counties by HHS Region, we see that each HHS Region reveals a signature (Table 3). For example, in HHS Regions 1, 2, and 3, one of the most frequent SDOH markers was the one for the number of pcHPSAs within the county (Marker 15, Table 1), with frequencies of 73 percent, 51 percent, and 33 percent, respectively. In Regions 1 and 3, having a high proportion of adults 65 and older was a marker in 32 percent and 37 percent of pcHPSA counties, respectively, and in Region 1, rurality was a common marker (26 percent of pcHPSA counties). Education was a dominant marker in Regions 4 and 6 (54 percent and 58 percent, respectively), while race/ethnicity was among the three most frequent markers in Regions 2, 6, and 9 (26 percent, 54 percent, and 59 percent, respectively). Unemployment was a consistent marker in Regions 5, 9, and 10 (25 percent, 41 percent, and 42 percent). In Regions 7 and 8, rurality (52 percent and 71 percent, respectively) together with sparse population (43 percent and 80 percent, respectively) were among the most frequent markers. Across all U.S. pcHPSA counties, the most common markers were rurality (36 percent of pcHPSA counties) and having multiple pcHPSAs in a county (30 percent of pcHPSA counties). Markers for education, low income, poverty, and sparse population were each present in 27 percent of pcHPSA counties nationwide. Marker frequencies for all markers and HHS Regions are provided in the S1 File.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Predominant SDOH markers in pcHPSA counties, by HHS Region, 2017.

https://doi.org/10.1371/journal.pone.0231443.t003

As this overview of co-occurring markers shows, certain markers often occurred in combination. Unsurprisingly, for example, 58 percent of pcHPSA counties that had the rurality marker were also characterized by sparse population. Similarly, markers for low income and poverty tended to co-occur, with 76 percent of pcHPSA counties having the poverty marker also having the low income marker.

In addition to examining marker frequencies, we also identified markers associated with the greatest variability across pcHPSA counties in each HHS Region and nationally. These findings help to elucidate individual SDOH factors associated with the greatest degree of disparity among county populations within a single Region (Table 4). For example, in Regions 1, 3, 5, 6, 7, and 8, poverty had the largest variability across county populations in each of these Regions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Markers with greatest variability in pcHPSA counties, by HHS Region, 2017.

https://doi.org/10.1371/journal.pone.0231443.t004

In fact, across HHS Regions, poverty was a dominant source of county-level population variability both nationally and in all HHS Regions except Regions 2 and 10. The sparse population marker was a major source of county-level variability in Region 2 and Region 4, while race/ethnicity was a contributor to variability in Regions 8 and 10, and the older population marker was a contributor to variability in Regions 5 and 10.

Principal findings

Our findings demonstrate that characteristics measured by markers of medical, economic, and geographic vulnerability were omnipresent in the national population in 2017. Nearly all U.S. county populations had one or more of the characteristics considered here, and, across the 89 percent of U.S. county populations in HRSA-designated pcHPSAs, nearly 90 percent had additional SDOH markers. A small but notable subset of U.S. county populations (206 counties, or 6.6 percent) had 10 or more SDOH markers, and 9 counties (0.3 percent) had the maximum, 13 markers–suggesting higher overall risk levels among residents in those counties in regard to limitations in health service access and poor population health. This latter group, with the most pervasive and complex patterns, lay entirely in HHS Region 4 and 6, roughly the southeastern and south central United States, and appears to reflect the recognized social, economic, and health disparities across Appalachia and in the South. [54, 55] Although, to some extent, each HHS Region had its own pattern of SDOH markers, the most common markers across pcHPSA county populations were rurality (the most common marker nationally), poverty, population sparsity, having a high proportion of older adults, lower education levels, greater racial/ethnic diversity, and higher levels of unemployment.

Many of these markers occurred in combination. Across all U.S. counties, poverty and low income were dominant contributors to the variability of SDOH markers for populations in pcHPSA counties. These findings are particularly important given the fact that poverty is one of the strongest predictors of health, and households in rural, underserved areas tend to experience higher rates of poverty and poorer overall health. [22, 25] Addressing shortages of health professionals in underserved areas is one key part of a broad array of efforts needed to better achieve health equity in these areas.

Several other points deserve emphasis here as well. First, the aim of our analyses is not to validate or critique the pcHPSA process. That process, defined in regulation, reflects health care provider and delivery metrics that are not captured by the makers considered here. This nation-wide shortage of primary care providers is recognized to have variable severity at the sub-national level [6, 31], and the alignment of our markers with this well-recognized mal-distribution of the primary care workforce is an important finding in itself. A measure that correctly identifies and elucidates a widely observed problem affecting most of the country can still be both a valid and a meaningful metric.

Second, the finding that both a high proportion of pcHPSA counties and a high number of counties overall have one or more SDOH markers does not mean that all parts of a particular county are beset by provider shortages or that everyone in a county experiences one or more health vulnerabilities. Rather, these patterns suggest that nearly all U.S. counties have segments of their populations with one or more risk factors, including shortages of primary care providers.

Perhaps more important than the ubiquity of risk factors are the regional differences and complexities in how these markers are distributed and in the attendant regional disparities reflected in those distributions. These disparities are particularly evident in HHS Region 4 and Region 6, roughly covering the southeastern and south central United States, where more than one third of pcHPSA counties in each Region have seven or more of the fifteen SDOH markers considered here. Our analyses help to underscore the population complexities of pcHPSA counties. For example, while many pcHPSA counties across the United States are characterized by elevated rates of poverty, these counties may differ markedly in how poverty is concentrated among children and older adults. For counties that are considered persistent poverty counties by CRS, unique and deeply rooted systemic challenges may present particular constraints. [45] Addressing provider shortages and delivering health care in these areas requires detailed understanding of population characteristics such as those captured by our analyses. [56] There is no “one-size-fits-all” approach to alleviating shortages in areas lacking adequate health provider capacity.

These analyses are not the first to overlay measures of social determinants of health. Indeed, recent and important work by Singh and Lin [57] takes a rigorous look at this issue by examining measures of deprivation across both time and space. What distinguishes our analysis from other recent work is our focus on pcHPSA counties and on how selected risk measures vary across pcHPSA counties, as well as by HHS Region. As discussed below, these analyses may help to inform primary care policy and planning at both federal and non-federal levels.

Policy and planning implications

Primary care has long been recognized often the first source of care for individuals and families, and as the foundation for maintaining health and wellness across broad health care systems both in the United States and worldwide. [4, 58, 59] Coordinated primary care, delivered through models such as the patient-centered medical home, is seen as a critical piece to improving the patient health care experience and population health, as well as reducing overall health care costs. [60] While the underlying economic and social factors are complex, access to high-quality primary care has been shown to improve longevity, improve quality of life, and help individuals avoid disability, while also reducing rates of emergency department visits and hospitalizations. [60] A recent epidemiological study demonstrated that, over the period of 2005 to 2015, for every 10 additional primary care physicians per 100,000 population, there was an associated increase in life expectancy of 51.5 days. [61] However, during that same period, although the absolute number of primary care physicians across the nation increased, the density of per-population primary care physicians actually decreased due to population increases and disproportionate losses of primary care physicians in some counties (which, in turn, may reflect recruitment/retention challenges in rural areas, urban preferences among some providers, and other factors. [62] This recent study by Basu et al. [61] highlights that access to high-quality primary care is highly uneven across the nation, and that there are notable population health impacts that may result from this access differential.

Access to health care services is a complex, multifactorial problem that is due at least in part to variable “resource” availability–including available sites of care such as hospitals and outpatient practices, and available specialty and primary care providers. [63] Primary care HPSAs are designations that indicate shortages in primary care providers that may be geographic-, population-, or facility-based in nature. [5] HRSA estimates that the population of individuals experiencing a provider shortage in the nation’s 6,433 pcHPSAs is currently around 77 million people–or roughly 24% of the U.S. population–and reflects a shortage of 13,939 primary care providers. [6] Further, HRSA is projecting that, under current patterns of utilization and care delivery, shortages of primary care providers will worsen over the coming years, driven largely by an increasing demand for health care services associated with demographic shifts in the nation’s population together with ongoing challenges in recruiting and retaining primary care providers. [8, 64] This is not solely a problem of primary care workforce magnitude, but also one of geography and distribution. For instance, in 2013, there were approximately 68 primary care physicians per 100,000 residents living in rural areas, which is approximately 19% lower than what the physicians available in urban areas. [64]

A number of other complex factors also limit health care accessibility, including access to primary care services. Such factors include care affordability, physical accessibility, relevance, effectiveness, and acceptability–as well as the presence of cultural barriers and issues of health equity. [36, 63, 65] The term “social determinants of health” aims to capture this collection of factors, which describe the circumstances into which people are born and live, and that are shaped by a wider set of factors such as economics, social policies, and politics. [19] The premise of this investigation was that certain social determinants of health–assessed through defined measures of medical, economic, and geographic vulnerability for which data were readily available–are likely to geographically co-occur with primary care workforce shortages, since access to quality health care, education, housing, transportation, employment, and the system of justice all relate to the societal distribution of resources. These factors, taken all together, in turn impact the prevalence of disease and injury, and overall population health. [19, 20]

Our study shows that nearly all (96 percent) counties in the United States are whole county or partial county pcHPSAs or have one or more risk markers. Perhaps unsurprisingly, this suggests that counties low in certain resources are likely to also be low in other resources. For example, areas with lower average educational attainment levels may also have higher degrees of poverty and unemployment. However, different SDOH markers appear to be variably pervasive in different regions of the country–meaning that the medical, economic, social, and political factors affecting access to primary care, the foundation for maintaining health and wellness across health care systems, are not uniformly contributory. As such, unique subnational factors may need to be considered and addressed while planning and developing public health policies and when making investments in initiatives–including workforce initiatives–intended to drive improvements in primary care health access.

The inherent intertwining of primary care physician shortages with markers of medical, economic, social, and political factors in the nation’s most vulnerable areas makes for great challenges with policy planning and investments, however. Recruitment and retention of primary care providers into rural and other areas with higher degrees of health workforce shortages, evident by lower per-resident numbers of primary care physicians in rural areas, has been a longstanding challenge for the U.S. health care system as well as for the health care systems in other nations, whether they are low- or high-income countries. [63, 66] The largely urban-centralized private medical sector often preferentially attracts health workers over competing rural areas, which also have disadvantages such as weaker labor markets and working and living conditions that may be less desirable to new medical training graduates as potential recruits to such high-need areas. [15, 16, 66] Rural medical practices may have more demanding schedules due to having fewer providers available for service coverage, and potentially due to federal level restrictions regarding reimbursement as well as the state-specific influence of scope of practice limitations for non-physician providers, such as nurse practitioners. [66, 67] Rural practices may also offer lower salaries, which may be an important factor for career choice among new medical school graduates, given the high and rising cost of medical education. [16] In contrast, urban areas are often more attractive locales for health care professionals to practice and live with their families, due to their relative advantages in employment and salary opportunities, social and cultural amenities, higher spousal satisfaction levels and work prospects, and increased access to education opportunities for their children. [56, 67]

Understanding the challenges in addressing provider workforce shortages and geographical imbalances in providers provides a path forward for strengthening health care access and equity in high-need areas. [63] A strategic approach to recruitment and retention of providers into high need areas that is inclusive of multiple interventions seems to be more effective than individual interventions. [63] Strategies can include exposing trainees to health care delivery in high-need areas through rotations or curricular content tailored to the particular needs of the populations served; targeted recruitment of health professionals with ties to and experience with the populations in a specific high-need area; providing personal and professional support; and offering financial incentives that require or entice service in an underserved area in return for educational financial aid. [68, 69]

HRSA is the principal federal agency in the U.S. that is charged with increasing access to effective and efficient basic health care for those who are medically underserved due to barriers in obtaining appropriate and quality care–be those barriers economic, geographic, linguistic, cultural, or attributable to other causes. [70] HRSA administers numerous programs through its $10.7 billion in fiscal year 2020 appropriations that are designed to expand services, increase supplies of health care workers, and improve the distribution of the workforce. For example, the National Health Service Corps supports primary care physicians and other health care professionals who have dedicated themselves to working in underserved, rural, and tribal areas that are located in HPSAs. This program significantly improves the distribution of a much needed primary care health workforce in areas that otherwise would have no access to these health care professionals. Additionally, the Teaching Health Center Graduate Medical Education program supports the resident training of future primary care physicians and dentists in community-based ambulatory care settings in rural and underserved areas. Empirical evidence indicates that physicians trained in these settings are more likely to practice in these same settings, such as health centers, when they complete their training. [69] HRSA’s new Rural Residency Planning and Development program similarly aims to develop new, rural-based residency training programs for physicians in the specialties of family medicine, internal medicine and psychiatry. Lastly, HRSA has awarded cooperative agreement funding to two new Health Workforce Research Centers that will focus their research on issues related to health equity in health workforce education, training, and practice. These centers will strengthen the evidence base for effective education and training programs that incorporate social determinants of health so that the future workforce will be better prepared to ensure health equity for all populations.

In a nationally representative sample of HRSA-funded federally qualified health centers, which disproportionately serve patients whose health is affected by socioeconomic disadvantages, patients reported a particularly high prevalence of social risk factors–including unstable housing and employment, lower education levels, and poverty. [71] HRSA is investing $5.6 billion total in health center program funding in FY2020 in order to support the nearly 1,400 health centers that provide care to approximately 27.2 million underserved patients. [70] HRSA’s primary care workforce programs, together with an understanding of population risk factors and disparities, are essential for addressing the nation’s mal-distribution of primary care providers and in particular ensuring sufficient staffing at health centers in medically underserved areas. [72]

Strengths and limitations

Our analyses offer a national view of the distribution of vulnerable populations in pcHPSAs across U.S. states and in the District of Columbia. By selecting a range of SDOH measures, we highlight not only the ubiquity but also the complicated needs of populations who may be geographically, economically, or medically vulnerable. Moreover, by overlaying SDOH measures with pcHPSA counties, our analyses provide important context for the consideration of policy implications specific to primary care and related workforce programs, including HRSA’s programs.

As with any analysis, there are limitations. First, it must be emphasized that HRSA’s designated pcHPSAs are based in regulation and so may not fully capture all primary care provider shortages at a particular point in time. While pcHPSAs offer a useful lens for our national analyses, we caution that more granular analyses at state and sub-state levels may require greater detail in identifying provider shortages than that afforded by looking at pcHPSAs.

We also did not attempt to estimate the number of people affected by either individual or collective vulnerabilities. While our use of proportions and categorical variables permitted us to evaluate characteristics across U.S. counties using a consistent suite of scales, our analyses cannot fully measure problem magnitude with respect to the size of vulnerable populations. Particularly in densely populated areas, more detailed analyses, to include estimates of population size, will be important in guiding health services planning at state and sub-state levels. In fact, replicating our analyses at the state and county levels may be beneficial to inform policy-makers about local decisions and to help find efficiencies in resource allocation and inter-related health policies.

Another limitation, commonly termed the ecological fallacy, involves assigning a single estimate or attributing a particular characteristic to every individual in a population. This is a well-recognized limitation of aggregated data analysis [73–75], and considerable caution is needed when inferring information about individuals from group-level measures. For example, it cannot be assumed that, in a county identified as a persistent poverty county, all individuals in the county experience persistent poverty to the same extent. However, our focus here is not on individuals, nor, as noted above, is it on population size. Rather, our aim is to elucidate regional complexities. Indeed, an important element of our analyses involves description of these regional or system-level patterns. As Galea and colleagues have recently discussed [76–78], understanding system level effects is essential to sound planning and policy development. In addition, capturing the complex effects of social determinants of health holistically is analytically challenging, and reliable data sources for some determinants are limited, particularly at the county level. Measures that accurately and completely capture the essence of certain determinants of health equity, such as social justice, do not exist.

Similarly, as a number of researchers have discussed, assessing rurality for the purposes of health care analyses is a complex undertaking [17, 79], and requires consideration of population size, population density, proximity to health care and other services, local and regional infrastructure, and other factors. While our blending of the RUCC and UIC classifications attempts to capture some of this complexity, it can reflect only a portion of the full nuance. Although beyond the scope of our analyses, we are intrigued by recent research from Doogan et al. [80] that proposes a “geographic isolation scale” for examining rural health disparities, and we expect to incorporate more rigorous analyses of rurality in future work.

We would also note the cross-sectional nature of our analyses. They are snapshots in time and, with the exception of the CRS persistent poverty indicator, our SDOH markers do not account for trends over time in population characteristics. Such trend analyses are indeed important, and represent an important next step in refining this work.

Finally, we acknowledge our use of “convenience” measures available in AHRF, rather than conducting primary data collection ourselves or retrieving data directly from the Census Bureau, ERS, and other resources. However, AHRF provides a comprehensive, county-level picture, and affords an opportunity to efficiently examine multiple markers at a common point in time and with consistent data validation checks applied. Again, an important next step will be to augment these analyses with additional variables directly from the U.S. Census Bureau and other primary data sources, as well as possible primary data collection. However, for the purposes of providing an initial county-level view, the AHRF affords an opportunity to examine pcHPSA counties across a range of medical, economic, and geographic measures.

Conclusion

Nationally, 89 percent of U.S. counties are classified as either whole county or partial county pcHPSAs and 96 percent of U.S. counties have one or more population risk markers, suggesting that medically, economically, and geographical vulnerable groups are nearly ubiquitous throughout the nation. Across all U.S. pcHPSA counties, the most common markers are rurality, having multiple pcHPSAs in a single county, education, low income, poverty, and sparse population. pcHPSA populations in HHS Region 4 and Region 6 have the most pervasive and complex risk patterns, and reflect the recognized social, economic, and health disparities across Appalachia and in the South. Collectively, these ecological measures of county-level population characteristics may help inform health workforce and health care planning at all levels, and, by shining a light on the risk complexities affecting U.S. populations, our findings may also help to guide efforts to strengthen the delivery of primary care in areas characterized by provider shortages.