The Growing Divide in the Composition of Public Health Delivery Systems in US Rural and Urban Communities, 2014–2018

Abstract

Objectives. To examine changes in the scope of activity and organizational composition of public health delivery systems serving rural and urban US communities between 2014 and 2018.

Methods. We used data from the National Longitudinal Survey of Public Health Systems to measure the implementation of recommended public health activities and the network of organizations contributing to these activities in a nationally representative cohort of US communities. We used multivariable regression models to test for rural–urban differences between 2014 and 2018.

Results. The scope of recommended activities implemented in rural areas declined by 3.4 percentage points between 2014 and 2018, whereas it increased by 1.4 percentage points in urban areas. The rural–urban disparity in scope of activities grew by a total of 4.8 percentage points (P < .05) over this time. The disparity in network density grew by 2.3 percentage points (P < .05).

Conclusions. Urban public health systems have enhanced their scope of activities and organizational networks since 2014, whereas rural systems have lost capacity. These trends suggest that system improvement initiatives have had uneven success, and they may contribute to growing rural–urban disparities in population health status.

Significant disparities in health care access and outcomes exist for the approximately 59 million people living in rural communities in the United States.^1,2 Overall life expectancy, infant mortality, chronic disease, and cancer outcomes are all worse for rural populations than for their urban and suburban counterparts.² The Affordable Care Act (ACA) coupled with calls to action for greater attention to social determinants of health have made progress in improving access to care and addressing preventable health concerns.³ However, whereas urban areas have seen significant improvements in some health indicators, rural areas continue to lag, which has widened rural–urban health disparities.^2,4 From 2007 to 2017, rural–urban mortality disparities increased for 5 of 7 major causes of death tracked by Healthy People 2020: coronary heart disease, cancer, diabetes, chronic obstructive pulmonary disease, and suicide.⁵

Important demographic differences partially explain rural–urban health care disparities. People living in rural communities are more likely to be older and less affluent than are their urban peers.⁶ Rural populations are more likely to experience barriers to health care, including provider shortages, longer travel times, hospital closures, and high health care costs.⁷ Furthermore, health risks such as tobacco use, poor diet, inadequate physical activity, and adiposity are more prevalent among rural populations.^8,9 A strong public health infrastructure is needed to change the policies and the social and environmental structures that impede good health in rural communities.

The federal government’s Public Health 3.0 framework calls for public health systems to play stronger roles in addressing social determinants of health by forging community partnerships with the medical and social service sectors.¹⁰ Greater collaboration among diverse partners may increase the efficiency of the allocation of community resources and bring a variety of voices and approaches to bear in solving community problems.¹¹ Additionally, multisectoral partnerships in public health systems have been shown to reduce health care utilization¹² and preventable deaths.¹³ Specifically, cardiovascular disease, diabetes, and influenza deaths declined significantly over time in communities that expanded multisector networks supporting population health activities.¹³ Rural communities may realize even greater benefits of strong public health networks if they reduce duplicative services and maximize limited resources.

Historically, rural health departments provided fewer health services and had fewer multisectoral partnerships than did urban health departments.¹⁴ However, service delivery and partnerships have not been studied in the context of recent initiatives like Public Health 3.0 and the ACA. Policies such as the ACA’s provision to require hospitals to conduct a community health needs assessment and the insurer mandate to cover individuals with preexisting conditions may have incentivized collaboration with local public health systems.³ On the other hand, funding for federal, state, and local public health departments has decreased in recent years. From 2015 to 2016, 31 states reduced their public health budgets.¹⁵ Moreover, 25% of local public health departments reported decreasing their budget for fiscal year 2016.¹⁵ This may limit the capacity of public health systems to provide additional services and develop community partnerships.

We used information from a nationally representative cohort of US communities to examine (1) the extent and nature of differences between rural and urban public health systems regarding the scope of activities they implement and the network of organizations they engage, and (2) the extent to which these differences changed from 2014 to 2018.

METHODS

We used data from the National Longitudinal Survey of Public Health Systems, which follows a nationally representative cohort of US communities with survey data originally collected in 1998 and in 5 subsequent waves through 2018. The survey expanded in 2014 to include a cohort of communities with less than 100 000 residents. The survey used a validated questionnaire administered to collect information about 20 public health activities recommended by national guidelines and federal consensus panels for use in improving community health status. These activities were derived from research-tested models of community health improvement and include periodic assessments of community health needs and risks, multisector priority setting and planning, community engagement in selecting and implementing health improvement strategies, resource allocation to support implementation of priority strategies, and monitoring and evaluation activities to track progress (see supplemental materials for a complete list of survey questions [available as a supplement to the online version of this article at http://www.ajph.org]). The activities correspond to activities recommended by the Institute of Medicine’s periodic studies of public health systems, the US Department of Health and Human Service’s Public Health 3.0 framework, and the Public Health Accreditation Board’s national accreditation standards for state and local public health agencies.^10,16,17

We administered the survey to local public health officials in each community who were instructed to report information about public health activities implemented for residents of their community, regardless of whether activities were implemented by public health agencies or by other organizations. To facilitate reporting accuracy, we instructed respondents to define their community based on the geopolitical jurisdiction served by their public health agency (county, city, town, or district). For each of the public health activities on the survey, local public health officials reported (1) whether the activity was implemented in the community during the past 3 years; and (2) if so, the network of organizations involved in activity implementation, including categories for public health agencies, hospitals, primary care providers, health insurers, employers, schools, community- and faith-based organizations, and other governmental agencies.

Our analysis included 2014, 2016, and 2018 survey data. Overall, 82% of the cohort communities were single-county jurisdictions, 12% were multicounty districts, and 6% were cities or towns. In 2014, response rates were 74.6% and 50.9% for urban and rural cohorts, respectively. We excluded 12 observations with missing data.

Measures

Public health system composition.

We focused on 3 primary measures of public health system composition. First, we measured the proportion of recommended public health activities that were implemented in each community (scope of activity). We also examined each individual activity. Second, we measured the network of community organizations that work together in implementing these activities (network density). We constructed a measure of public health network density for each community using the 13 types of organizations listed on the survey instrument. For each pair of organization type (dyads), we counted the number of public health activities in which both types of organizations participated, divided this count by the total number of public health activities listed on the survey instrument, and then computed the mean of these ratios across all possible dyads. The resulting measure had a theoretical range of 0 to 100, with larger values indicating higher levels of connectivity between organizations through joint participation in public health activities. Because the survey instrument collected information on organization types (sectors) rather than individual organizations, the network density measure can be interpreted as an indicator of cross-sector cooperation in public health. We also performed analyses on the proportion of activity participation by organization type.

Third, we constructed a composite measure of public health system capability that combined the 2 measures of scope of activity and network density into a single categorical measure. This measure, based on a previously developed and validated typology identified through latent class analysis,¹⁸ classified communities into ordinal categories: (1) comprehensive systems, which implemented the broadest scope of activities and engaged the most dense networks of organizations contributing to these activities; (2) conventional systems, which had moderate to high levels of activity scope and intermediate network density; and (3) limited systems, which had relatively low levels of activity scope and network density.¹⁸

Rurality and other community characteristics.

We linked our survey data with existing county-level community characteristics from the Area Health Resources Files.¹⁹ We identified rural and urban communities using the US Health and Human Services Office of Rural Health Policy’s definition of rurality based on rural–urban commuting area codes.²⁰ We classified counties that we did not include as part of metropolitan areas as rural. For communities spanning multiple counties, we used the county with the lowest (i.e., most urban) rural–urban commuting area code for rural–urban classification. Additionally, we used Area Health Resources Files to construct county-level measures indicating racial and age composition, poverty, household income, unemployment, and educational attainment (Table 1). To estimate the population size of each public health jurisdiction, we used data from the National Association of County and City Health Officials periodic national surveys of local health departments,²¹ adjusting for population growth using county growth rates estimated by the US Census Bureau.²²

TABLE 1—

Characteristics of Public Health Systems in Rural and Urban Community Cohorts: United States, 2014–2018

Characteristic	2014			2018
	Rural (n = 151), Mean (SD)	Urban (n = 340), Mean (SD)	Pa	Rural (n = 176), Mean (SD)	Urban (n = 374), Mean (SD)	Pa
Scope of activityb	62.3 (22.5)	69.3 (20.8)	< .001	58.0 (23.9)	68.9 (22.4)	< .001
Organization network densityc	8.2 (7.9)	12.4 (10.7)	< .001	7.8 (8.2)	14.4 (12.1)	< .001
Composite system ratingd (%)
Limited	58.3 (49.5)	44.4 (49.8)	.005	64.2 (48.1)	44.7 (49.8)	< .001
Conventional	12.6 (33.3)	15.6 (36.3)	.39	9.1 (28.8)	16.0 (36.8)	.03
Comprehensive	29.1 (45.6)	40.0 (49.1)	.02	26.7 (44.4)	39.3 (48.9)	.004
Below poverty (%)	16.4 (5.6)	14.1 (4.9)	< .001	14.8 (5.2)	12.8 (4.3)	< .001
Non-White (%)	16.7 (16.2)	29.7 (17.7)	< .001	15.5 (12.9)	31.7 (18.3)	< .001
Population older than 65 y (%)	18.6 (4.5)	15.0 (3.8)	< .001	20.7 (4.3)	16.1 (4.3)	< .001
Uninsured, (%)	14.0 (4.3)	11.9 (5.0)	< .001	10.2 (4.0)	9.2 (4.4)	.009
Income per capita ($10 000)	4.1 (1.7)	4.7 (1.4)	< .001	4.2 (1.8)	5.1 (1.4)	< .001
Unemployment (%)	6.2 (2.4)	6.1 (1.5)	.79	4.5 (1.5)	4.4 (1.3)	.29
4-y college degree (%)	19.5 (8.2)	30.2 (10.8)	< .001	19.8 (7.7)	31.3 (10.4)	< .001
Population (10 000)	4.1 (5.0)	34.6 (44.5)	< .001	3.5 (3.8)	44.3 (82.2)	< .001
Multicounty jurisdiction (%)	13.9 (34.7)	10.6 (30.8)	.29	10.2 (30.4)	12.8 (33.5)	.38
City health department (%)	0.00 (0.00)	7.4 (26.1)	.001	0.00 (0.00)	7.8 (26.8)	< .001

^aWe used the t test and χ² test to test the difference between characteristics of rural and urban communities. The P values from the results are reported.

^bScope of activity indicates the proportion of 20 recommended public health activities implemented in the community.

^cNetwork density indicates the extent to which community organizations work together in implementing the recommended public health activities.

^dComposite system rating is a composite measure that classifies systems into 1 of 3 ordinal categories based on the combination of scope of activity and network density.

Statistical Analysis

We used generalized linear models with a Gaussian family and identity link to estimate rural–urban differences over time for measures of scope of activity, network density, and activity participation by specific organizations. We estimated the binary measures for individual activity participation with logistic regression. To estimate rural–urban differences in trends for the composite measure of public health system capability, we used ordered logistic regression. The ordered logistic model satisfied tests for the proportional odds assumption. All models included the same set of county-level controls for community demographic and socioeconomic characteristics (Table 1). We controlled for within-survey unit correlation over time by including a community-specific random effect and used clustered SEs at the community level.

To determine whether rural–urban differences in public health systems changed over time, we estimated models using year by rurality indicator terms. We used urban communities in 2014 as the reference group. We estimated additional model specifications using year, rural, and rural–year interaction terms. The interaction terms between year and rural indicators allowed us to test differential trends between rural and urban communities. The coefficient on the interaction term tested whether the time trend in rural communities was significantly different in direction and magnitude compared with the trend in urban communities, after controlling for covariates. For the logistic regression models, we report marginal effects, which represent the average change in probability of the outcome for a small change or category change in the predictor variables. We display estimated models using rurality–year group for ease of interpretability. Model specifications testing the differences in trends are included in Tables A and B (available as a supplement to the online version of this article at http://www.ajph.org).

We conducted a sensitivity analysis using a balanced panel to ensure that our findings were not a result of attrition. The results were robust to balancing the sample; however, balancing resulted in a significant loss of precision because of dropping almost half of the sample size (unbalanced n = 1661; balanced n = 816). Thus, we used the unbalanced panel to preserve sample size.

We completed all analyses using Stata version 15.1 (StataCorp LP, College Station, TX).

RESULTS

Table 1 summarizes the public health system scope of activity and network density as well as demographic and socioeconomic characteristics of the rural and urban community cohorts in 2014 and 2018. Rural communities had a lower scope of activity and network density than did urban communities and were less likely to be categorized as a comprehensive system. Rural communities had higher poverty and uninsurance rates, and lower per capita income and percentage of residents with a 4-year college degree. Rural communities also had significantly higher proportions of older adults (older than 65 years) and non-Hispanic White residents than did their urban counterparts. Rural and urban communities did not significantly differ in percentage unemployed or whether the local public health department’s jurisdiction covered multiple counties. From 2014 to 2018, rural population sizes decreased, whereas urban populations grew. These changes over time confirmed our choice to include time-varying demographic variables as covariates.

Table 2 displays differences in public health system trends between rural and urban communities, adjusting for the socioeconomic and demographic covariates from Table 1. The parameter estimate for the rural 2014 variable indicates the baseline difference (i.e., difference in 2014) between rural and urban communities. In 2014, public health systems in rural communities on average implemented 5.2 percentage points fewer public health activities (confidence intervals [CI] = –9.7, –0.7; P < .05) and had a 3.2 percentage point lower network density (CI = –5.0, –1.4; P = .001) compared with urban communities. Similarly, rural communities were significantly less likely than were urban communities to achieve higher ratings of public health system composition (odds ratio [OR]  = 0.50; CI = 0.27, 0.91; P < .05).

TABLE 2—

Changes Over Time in Rural and Urban Public Health Systems: United States, 2014–2018

Variable	Scope of Activity,a b (95% CI)	Organization Network Density,b b (95% CI)	Composite System Rating,c OR (95% CI)
Rurality and period
Urban 2014 (Ref)	0	0	1
Urban 2016	1.6 (–1.3, 4.4)	0.5 (–0.9, 1.8)	1.22 (0.82, 1.82)
Urban 2018	1.4 (–1.8, 4.7)	1.7 (0.2, 3.2)	1.22 (0.77, 1.94)
Rural 2014	−5.2 (–9.7, –0.7)	−3.2 (–5.0, –1.4)	0.50 (0.27, 0.91)
Rural 2016	−4.2 (–9.0, 0.6)	−3.5 (–5.4, –1.5)	0.69 (0.38, 1.28)
Rural 2018	−8.6 (–13.7, –3.4)	−3.8 (–5.9, –1.7)	0.46 (0.24, 0.89)
Poverty rate (%)	−0.2 (–0.6, 0.2)	0.0 (–0.2, 0.2)	0.97 (0.92, 1.02)
Non-White (%)	0.0 (–0.1, 0.1)	0.0 (0.0, 0.1)	0.99 (0.98, 1.01)
Population older than 65 y (%)	−0.4 (–0.8, 0.1)	0.0 (–0.2, 0.2)	0.97 (0.92, 1.01)
Uninsured rate (%)	−0.2 (–0.6, 0.3)	−0.1 (–0.2, 0.1)	1.00 (0.96, 1.05)
Income per capita ($10 000)	−1.1 (–2.5, 0.2)	−0.7 (–1.5, 0.1)	0.90 (0.77, 1.05)
Unemployment rate (%)	0.5 (–0.6, 1.5)	−0.1 (–0.5, 0.4)	1.09 (0.96, 1.24)
4-y college degree (%)	0.0 (–0.2, 0.3)	0.1 (0.0, 0.1)	1.01 (0.98, 1.03)
Multicounty	0.8 (–3.0, 4.5)	1.1 (–0.7, 3.0)	1.07 (0.66, 1.73)
Constant	79.3 (66.0, 92.6)	13.8 (7.9, 19.7)	. . .

Note. CI = confidence interval; OR = odds ratio. We calculated all coefficients using generalized linear regression models except for the composite system rating model, for which we used an ordinal logit regression model. We controlled for within-survey unit correlation over time by including a community-specific random effect and used clustered SEs. All measures are reported by the local public health official in each jurisdiction (n = 1661).

Source. Authors’ analysis of data from the National Longitudinal Survey of Public Health Systems, 2014–2018.

^aScope of activity indicates the proportion of 20 recommended public health activities implemented in the community.

^bNetwork density indicates the extent to which community organizations work together in implementing the recommended public health activities.

^cComposite system rating is a composite measure that classifies systems into 1 of 3 ordinal categories based on the combination of scope of activity and network density.

From 2014 to 2016, the results suggest that there were limited changes in public health infrastructure. Urban and rural communities experienced a small and statistically insignificant increase in both scope of activity and organization density relative to 2014. However, it appears that from 2014 to 2018 rural public health systems were worsening compared with their urban counterparts. Scope of public health activities declined by an adjusted average of 3.4 percentage points in rural communities and increased by 1.4 percentage points in urban areas, resulting in a statistically significant difference in trend of –4.8 percentage points (P < .05). Similarly, network density declined by –0.6 percentage points in rural areas and increased in urban areas by 1.7 percentage points, resulting in a –2.3 percentage point difference in trend (P < .05). The composite measure of public health system capability also declined for rural areas, but the rural–urban differences in trend were not statistically significant (P > .05). Community characteristics were not independently associated with public health system characteristics after controlling for rurality.

Looking individually at organizations, we found that hospitals, higher education (e.g., universities and colleges), other nonprofits, and insurance groups were responsible for the largest rural–urban differences in organization density (Table 3). From 2014 to 2018, the proportion of activities contributed by hospitals declined by an adjusted average of 4.1 percentage points in rural areas and increased by 1.4 percentage points in urban areas, resulting in a –5.5 percentage point difference in trend (P < .05). Higher education and insurance groups declined by a –5.5 and –4.3 percentage point difference in trend, respectively (P < .001). Other nonprofits declined by an average of 4.7 percentage points difference in trend (P < .05). Significant rural–urban differences in scope were also found for other (nonhealth) state agencies and federal agencies (P < .05; Table A [available as a supplement to the online version of this article at http://www.ajph.org]).

TABLE 3—

Changes Over Time in Organization Participation in Public Health Activities by Rurality: United States, 2014–2018

Variable	Hospitals, b (95% CI)	Universities/Colleges, b (95% CI)	Other Nonprofits, b (95% CI)	Insurance Groups, b (95% CI)
Rurality and period
Urban 2014 (Ref)	0	0	0	0
Urban 2016	1.4 (–1.6, 4.5)	1.6 (–1.0, 4.3)	2.8 (–0.2, 5.8)	0.2 (–1.8, 2.2)
Urban 2018	1.4 (–2.0, 4.8)	4.0 (1.2, 6.9)	2.1 (–1.2, 5.4)	2.6 (0.3, 4.9)
Rural 2014	−4.6 (–9.4, 0.2)	−7.7 (–11.2, –4.2)	−5.1 (–9.4, –0.8)	−3.2 (–5.9, –0.5)
Rural 2016	−4.9 (–9.6, –0.1)	−8.4 (–12.0, –4.8)	−6.0 (–10.5, –1.6)	−3.6 (–6.7, –0.6)
Rural 2018	−8.7 (–13.7, –3.6)	−9.2 (–13.0, –5.3)	−7.6 (–12.3, –3.0)	−4.9 (–7.9, –1.8)
Poverty rate (%)	−0.3 (–0.8, 0.1)	0.5 (0.1, 0.8)	−0.4 (–0.7, 0.0)	0.2 (0.0, 0.5)
Non-White (%)	−0.1 (–0.2, 0.1)	0.0 (–0.1, 0.1)	0.1 (0.0, 0.2)	0.1 (0.0, 0.2)
Population older than 65 y (%)	−0.4 (–0.8, 0.0)	−0.2 (–0.5, 0.0)	0.0 (–0.3, 0.4)	0.0 (–0.2, 0.2)
Uninsured rate (%)	−0.3 (–0.7, 0.1)	0.0 (–0.3, 0.3)	−0.3 (–0.7, 0.1)	−0.5 (–0.7, –0.3)
Income per capita ($10 000)	−1.2 (–2.6, 0.3)	−1.6 (–2.8, –0.3)	−1.5 (–3.0, 0.0)	−0.8 (–1.6, 0.0)
Unemployment rate (%)	0.5 (–0.5, 1.6)	−0.2 (–1.0, 0.5)	1.0 (0.0, 1.9)	−0.3 (–0.8, 0.3)
4-y college degree (%)	0.0 (–0.2, 0.2)	0.2 (0.1, 0.4)	0.2 (0.0, 0.4)	0.0 (–0.1, 0.1)
Multicounty	3.2 (–0.7, 7.1)	1.7 (–1.9, 5.2)	2.7 (–1.3, 6.6)	−0.2 (–2.6, 2.3)
Constant	60.9 (46.4, 75.4)	16.0 (5.4, 26.6)	29.9 (17.0, 42.8)	13.5 (6.1, 21.0)

Note. CI = confidence interval. The outcomes indicate the proportion of recommended activities contributed by each type of organization. We calculated all coefficients using generalized linear regression models. We controlled for within-survey unit correlation over time by including a community-specific random effect and clustered SEs at the community level. All measures are reported by the local public health official in each jurisdiction (n = 1661).

Source. Authors’ analysis of data from the National Longitudinal Survey of Public Health Systems, 2014–2018.

Table 4 displays results for individual public health activities. Decreases in the total scope of activity measure were driven by changes in the following individual activities: whether a community health needs assessment was conducted in the past 3 years, whether the public health jurisdiction identified community health priorities in the past 3 years, and whether a community health action plan was developed to address community health needs in the past 3 years. These 3 activities had ORs suggesting a decline in activity implementation in rural communities from 2014 to 2018. The results for all organization types and recommended activities can be found in Tables A and B.

TABLE 4—

Changes Over Time in Public Health Activities by Rurality: United States, 2014–2018

Variable	Conducted Community Needs Assessment (Activity 1; n = 1658), OR (95% CI)	Identified Community Health Priorities (Activity 9; n = 1658), OR (95% CI)	Developed Community Health Action Plan (Activity 11; n = 1654), OR (95% CI)
Rurality and period
Urban 2014 (Ref)	1	1	1
Urban 2016	1.79 (0.95, 3.34)	1.33 (0.75, 2.38)	1.00 (0.63, 1.58)
Urban 2018	2.04 (1.03, 4.05)	1.42 (0.75, 2.71)	1.38 (0.83, 2.30)
Rural 2014	0.76 (0.33, 1.74)	0.57 (0.26, 1.24)	0.83 (0.43, 1.60)
Rural 2016	0.74 (0.32, 1.68)	0.54 (0.24, 1.20)	0.56 (0.29, 1.10)
Rural 2018	0.55 (0.23, 1.34)	0.28 (0.12, 0.65)	0.56 (0.28, 1.15)
Poverty rate (%)	0.91 (0.86, 0.98)	0.95 (0.89, 1.01)	0.96 (0.91, 1.02)
Non-White (%)	0.99 (0.97, 1.00)	0.99 (0.97, 1.01)	1.00 (0.98, 1.01)
Population older than 65 y (%)	0.94 (0.88, 1.00)	0.96 (0.91, 1.02)	0.98 (0.93, 1.03)
Uninsured rate (%)	1.03 (0.96, 1.10)	0.99 (0.93, 1.06)	1.00 (0.95, 1.05)
Income per capita ($10 000)	0.91 (0.72, 1.15)	0.91 (0.75, 1.11)	0.94 (0.79, 1.11)
Unemployment rate (%)	1.23 (1.01, 1.50)	1.14 (0.95, 1.37)	1.08 (0.94, 1.24)
4-y college degree (%)	0.98 (0.95, 1.02)	0.99 (0.96, 1.02)	1.00 (0.97, 1.02)
Multicounty	1.47 (0.70, 3.05)	1.55 (0.79, 3.03)	0.90 (0.51, 1.58)

Note. CI = confidence interval; OR = odds ratio. The outcomes are binary variables that indicate whether individual recommended public health activities were implemented. We calculated all coefficients using logistic regression models. We controlled for within-survey unit correlation over time by including a community-specific random effect and clustered SEs. All measures are reported by the local public health official in each jurisdiction.

Source. Authors’ analysis of data from the National Longitudinal Survey of Public Health Systems, 2014–2018.

DISCUSSION

These results provide the first, to our knowledge, nationally representative comparison of the structure and function of public health systems in rural and urban US communities. We found that systems serving rural areas implement fewer recommended public health activities and engage narrower networks of partners compared with their urban counterparts. These differences have grown larger since 2014, with systems growing stronger in urban areas and weaker in rural areas. In 2018, rural public health systems implemented fewer activities and engaged fewer partners in these activities than they had in 2014, even as the systems in urban communities gained strength. Because rural public health systems had significantly lower baseline levels of activity and network density in 2014, these negative trends over the ensuing 4 years widened an existing disparity.

Despite the constellation of initiatives to strengthen public health systems in recent years, urban systems have gained significant strength, whereas rural systems have yet to realize similar gains. These initiatives include enhanced community benefit requirements for nonprofit hospitals, national accreditation standards for public health agencies, expanded federal funding opportunities through the Prevention and Public Health Fund, and multisector demonstration programs such as the Accountable Health Communities Model.^3,16,23,24 Previous research provides evidence that policies such as the Medicare Hospital Readmissions Reduction Program²⁵ and the federal 340B Drug Discount Program²⁶ unintentionally benefit affluent communities and potentially exacerbate health disparities. Further research should explore whether these policies have similar unintended effects.

As a possible explanation for the diverging trends we found, rural public health systems operate with lower baseline levels of funding and staffing, which likely places them at a significant disadvantage in participating in new initiatives, particularly when participation is selective and competitive. For example, urban systems may be better positioned to advocate effectively to receive hospital community benefit investments, to demonstrate compliance with national accreditation standards, to compete for discretionary grants, and to negotiate for inclusion in demonstration programs. In these ways, initiatives designed to improve systems may unintentionally exacerbate preexisting rural–urban differences in system strength, consistent with the trends we observed.

Additionally, rural communities are vulnerable to marketplace and policy dynamics that constrain public health systems. Hospital closures and health insurer consolidations have disproportionately affected rural communities in recent years, reducing potential collaborators for public health activities.^27,28 During our study period, there were 70 rural hospital closures²⁷ and an increasing number of rural hospitals in financial distress.²⁹ At the same time, gains in health insurance coverage under the ACA have been less pronounced in rural areas, in part owing to the rural composition of the 14 states that have not yet expanded their Medicaid programs.^30,31 Lower insurance coverage results in higher demand for charity care provision, potentially at the expense of other types of public health activities. Structurally, rural communities have smaller tax bases available to support government services of any type, reducing the fiscal capacity to finance activities requiring high fixed costs, such as water and sewer systems, and activities requiring high-cost staff, such as physicians and epidemiologists.^14,32 Collectively, these dynamics leave rural areas with fewer resources and more competing unmet needs, relative to their size, than their urban counterparts.

Limitations

Several study limitations must be kept in mind when interpreting these results. The data on public health system characteristics we used in this analysis derive from information reported by local public health officials and, therefore, are limited by the knowledge and perspectives of these respondents. Survey respondents may not have full knowledge of all public health activities occurring in their communities, particularly of activities that do not involve the local public health agency. Changes in respondents over time because of agency leadership turnover may increase measurement error and limit the study’s ability to detect meaningful trends over time. The difference in survey response rates between rural and urban communities may also bias our study findings. However, we believe that rural health departments who have greater capability would be more likely to respond to the survey. If rural health departments that were limited in capability were less likely to respond to the survey, our findings may underestimate the disparity between urban and rural health departments.

Furthermore, the scope of activity measured only the quantity of activities performed and not the quality of implementation. Thus, it is possible that rural public health systems choose to concentrate their efforts on fewer activities and may perform those activities with higher quality. We believe this concern is mitigated by the fact that these activities are recommended by multiple governing agencies and viewed as basic activities for a functioning public health system.^10,33 Additionally, these measures have been linked to a reduction in preventable deaths in urban communities¹³; however, further research should explore whether findings can be generalized to rural communities. Finally, the observational research design we used did not allow us to identify causal mechanisms that generate rural–urban differences in public health systems. Our analysis controlled for an array of demographic and socioeconomic characteristics beyond rurality, and these adjustments had little impact on the estimated direction and magnitude of rural–urban differences. Nevertheless, unobserved factors may have confounded our estimated relationships between rurality and system composition.

Public Health Implications

The US health care system faces mounting pressure to reduce persistent and widening disparities in health status between rural and urban communities. Local public health systems are often overlooked in scientific and public policy discussions as potential solutions to these rural disparities, despite evidence of their effectiveness in improving population health. To our knowledge, our study is the first to estimate the magnitude of rural–urban disparities in public health system composition using nationally representative data and the first to document increases in these disparities since the enactment of the ACA. The results indicate that current efforts to strengthen public health systems may primarily benefit urban communities while leaving rural communities behind.

These findings demonstrate a need for greater targeting and tailoring of public health system improvement initiatives to the needs and resources of rural communities. Because state and federal governments are significant sources of funding for local public health activities, these entities should explore new mechanisms for allocating resources to communities in ways that mitigate rural–urban differences in fiscal capacity and opportunities for collaboration. At the same time, public officials at all levels should accelerate the implementation of cross-jurisdictional sharing arrangements that allow small and rural communities to combine systems and pool their resources and collaborators across neighboring jurisdictional boundaries.³⁴

These arrangements may include consolidated regional public health districts, joint purchasing and staffing models, multijurisdictional governance and advisory bodies, and other forms of pooling that allow systems to support a broader scope of activities and engage larger networks of collaborators. These arrangements have the additional potential to strengthen collaboration by achieving greater congruence among the geographic areas served by public health agencies, hospitals, social service agencies, schools, and other community organizations operating in rural areas. By recognizing and working together to reduce rural–urban disparities in public health systems, leaders in government and the private sector can make meaningful progress toward the promise of equality of opportunity in health.