Variation in Local Health Department Primary Care Services as a Function of Health Center Availability Brad Wright, PhD; Andrew J. Nice, MHA rrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr
Objectives: We aim to gain a clearer understanding of the effect of colocated federally qualified health centers (FQHCs) on the delivery of primary care and preventive services by local health departments (LHDs). Design: We collected data from the 2010 National Association of County & City Health Officials Profile of Local Health Departments, the 2010 Uniform Data System, and the 2011 Area Resource File. Setting: Forty-eight states, excluding Rhode Island and Hawaii. Participants: The analytic sample contained data on 2107 LHDs across the country. Main Outcome Measures: We modeled 4 measures of primary care activity for each LHD: provision of screening for (1) heart disease, (2) diabetes, and (3) hypertension, and (4) provision of comprehensive primary care. Results: Local health departments in counties with an FQHC grantee have, on average, 32% lower odds of providing hypertension screening, but having an FQHC grantee in the county does not influence LHDs’ primary care provision or screening for heart disease or diabetes. However, in an alternate model examining FQHC delivery sites per capita, each site is associated with a decrease in the odds of LHDs providing primary care (4% lower odds) or screening for heart disease (2% lower odds), diabetes, or hypertension (both 1% lower odds). Conclusion: Local health departments are more involved in disease screening than the provision of primary care. However, larger LHDs are more likely to provide both screening and primary care. Our current study provides evidence that the availability of an FQHC in the same county as an LHD partially explains the variation in LHDs’ provision of primary care services. Local health departments with colocated FQHCs can focus on more traditional public health activities, while LHDs in areas without an FQHC may find that ensuring access to care is more
challenging, possibly leading LHDs to directly provide primary care services. KEY WORDS: federally qualified health center, local health
department, primary care
Comprehensive, patient-centered primary care is associated with improved health outcomes.1-3 Yet, many vulnerable populations have difficulty obtaining access to primary care and must rely on health care safety-net providers to offer these services. While ensuring access to care is a core public health function, there is disagreement in the public health community about the role that local health departments (LHDs) should play in directly providing primary care services as part of the health care safety-net.4,5 Typically financed by state and local governments, LHDs are most often organized at the county level where they are tasked with performing a wide array of public health functions including, but not limited to, disease surveillance and prevention, regulation and oversight of environmental health risks, and direct care delivery. Local health departments are sometimes primary care providers of last resort for the uninsured and indigent, but the proportion of LHDs engaged in the direct provision of clinical services is decreasing, and the Institute of Medicine has called for LHDs to further minimize such activities while increasing their engagement in traditional (ie, nonclinical) public health activities.6 For example, between 1997 and 2008, the proportion of LHDs providing primary care declined by 31%, and the proportion providing screening for
Author Affiliation: Health Management and Policy, The University of Iowa, Iowa City, Iowa (Dr Wright and Mr Nice). The authors declare no conflicts of interest.
J Public Health Management Practice, 2015, 21(1), E1–E9 C 2015 Wolters Kluwer Health | Lippincott Williams & Wilkins Copyright
Correspondence: Brad Wright, PhD, Health Management and Policy, The University of Iowa, 145 N. Riverside Dr, N240 CPHB, Iowa City, IA ([email protected]). DOI: 10.1097/PHH.0000000000000112
E1 Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E2 ❘ Journal of Public Health Management and Practice diabetes and hypertension declined by 22% and 27%, respectively.5 There is substantial variation in the scope of primary care services provided by LHDs, but the explanation for this variation is not well understood. While 1 survey study found that LHD directors feel that whether LHDs should directly provide clinical services is at least partially dependent on the availability of other safety-net providers,7 it is unclear how the availability of other local safety-net primary care providers may actually influence LHD service provision.5,8-10 As the Affordable Care Act (ACA) is implemented, LHDs that provide comprehensive primary care services are being faced with an important decision about whether or not to continue providing those services.11 On the one hand, the increase in insurance coverage may mean that LHDs are able to seek reimbursement for providing primary care services that were previously supported by federal, state, and local public health funds.11 This opportunity can be expected to materialize predominantly in states that have opted to participate in the Medicaid expansion. On the other hand, LHDs may find that the increase in insurance coverage allows individuals to go elsewhere for their care. In that case, LHDs may simply decide to stop providing primary care and redirect primary care resources to more traditional public health activities.11 However, this decision may be conditional on the availability and capacity of other health care safety-net providers in the local area. Our study uses nationally representative quantitative data to explore variation in LHD primary care services as a function of federally qualified health center (FQHC) availability by county. Federally qualified health centers are federally funded facilities that are located in underserved areas and/or serve underserved populations and are legally mandated to provide primary care to all without regard to their ability to pay using an income-sensitive sliding fee scale. Consequently, FQHCs serve a low-income, often uninsured population with higher rates of chronic disease than the general population.12 Currently, there are 1200 FQHC grantees nationwide (a grantee may operate more than 1 delivery site) providing care for approximately 20 million patients annually, and their capacity is expected to double to 40 million patients under the ACA.13 Like LHDs, FQHCs also face opportunities and challenges under the ACA. For example, while their dedicated program funding has been increased, and they stand to receive increased reimbursement through Medicaid expansion and increases in private insurance coverage, they are also being expected to meet much of the increased demand resulting from those coverage increases, and it is unclear whether they will have adequate capacity to do so.14
Guided by the public health function of ensuring access to care, the direct provision of clinical services by LHDs seems likely to be limited to cases in which other provider options are few. Therefore, we hypothesize that LHDs in counties with an FQHC will be less likely to provide primary care services. However, a recent study by Luo et al10 finds no association between the presence of an FQHC and an LHD’s provision of medical care, and another study by Hsuan and Rodriguez5 found no association between the supply of FQHCs per 100 000 persons in poverty and the choice of LHDs to discontinue clinical services, or the extent of services discontinued. However, both of these studies relied on small samples (516 and 198 LHDs, respectively) and used a more limited measure of FQHC availability, while more broadly defining their outcomes. By contrast, we use a nationally representative sample of more than 2000 LHDs, 2 measures of FQHC availability, and more specific outcomes limited to primary care provision. By doing so, we aim to gain a clearer understanding of the relationship between FQHC availability and the provision of primary care services by LHDs.
● Methods To investigate primary care provision as a function of FQHC availability, we used data from the 2010 National Association of County & City Health Officials Profile of Local Health Departments, the 2010 Uniform Data System (UDS), and the 2011 Area Resource File (ARF). The National Association of County & City Health Officials survey data include information on the types of services provided by LHDs, as well as information on their staffing and finances. Hawaii and Rhode Island are excluded from the National Association of County & City Health Officials data because they do not operate LHDs. We merged these data by county (using a ZIP to FIPS crosswalk file) with the UDS data, which provided the location of FQHCs nationally. Finally, we merged data from the ARF by county to control for other aspects of local health care supply and population demographics. Together, these 3 data sets constituted a single cross-sectional analytic file. The analytic sample contained data on 2107 LHDs across the country. Pairwise correlation among all variables indicated only mild collinearity, with only 2 associations above 0.5, and all associations were as expected. We generated summary statistics and noted that 254 observations (12.06%) were missing data on whether the LHD employed a public health physician, and 136 observations (6.45%) were missing data for the LHD staffing variable. To address this, we used multiple imputation methods to replace missing values with
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
predicted values of the staffing and public health physician variables generated as a function of all covariates with nonmissing data in the data set. Using the analytic file, we estimated 4 logistic regression models that explain variation in the scope of primary care services provided by LHDs as a function of FQHC availability in the county where the LHD is located and other factors. Specifically, because we are primarily interested in investigating the delivery of care related to chronic disease, we modeled the provision of screening for heart disease, diabetes, and hypertension, as well as the provision of comprehensive primary care. For each of these outcomes, some observations (< 5%) were missing data. Given the limited number of missing cases, we simply excluded these observations from our analyses. For each outcome measure, we operationalized our key independent variable, FQHC availability, in 1 of 2 ways. First, using UDS data, we defined FQHC availability as the presence or absence of an FQHC grantee in the county. Alternately, using UDS data, we defined FQHC availability as the number of FQHC delivery sites (not grantees) per 100 000 persons residing in the county. This was done to determine the relative importance of having any FQHC in the county versus the importance of FQHC capacity in the county. We also controlled for several other LHD-specific and countyspecific characteristics. At the LHD-level, we controlled for staffing per 10 000 persons in the LHD’s jurisdiction, because adequate staffing is essential to the ability to provide a variety of services. We also controlled for whether or not the LHD employed a public health physician, as this is likely to be an important determinant of providing primary care services. In that spirit, we also controlled for whether or not the LHD was headed by an executive with a clinical degree (defined as MD or DO) or an executive with a public health degree (MPH or DrPH), as we expected such individuals might take a more comprehensive approach to primary care and/or public health delivery. At the LHD level, we also controlled for the jurisdiction of the LHD, which we categorized into 3 groups: county, city, and other (which included joint city-county jurisdiction, and both multicity and multicounty jurisdictions). Using data from the ARF, we controlled for the county supply of primary care physicians per capita, midlevel providers per capita (including nurse practitioners, physician assistants, and advanced practice nurses), short-term general hospitals, and rural health clinics. These variables were included to control for supply-side factors that are likely to influence LHD scope of services by providing additional resources to area residents. We also used the ARF to classify counties as metropolitan or nonmetropolitan using rural-urban continuum codes (metro = 1, 2, or 3).
❘ E3
Finally, we used clustered standard errors to adjust for data clustering within states. We opted to do this instead of using state dummy variables, both to avoid perfect prediction problems and because we lacked a strong conceptual framework to suggest that differences between states (beyond those measures explicitly controlled for in our model) would explain LHD activity at the local level.
● Results Sample summary statistics are shown in Table 1. The average health department in our study is responsible for a population of 133 606 persons, although this ranged from a low of 339 to a high of more than 1 million persons. Of the 2107 LHDs in our sample, nearly 38% are in counties that also have at least 1 FQHC grantee. On average, these counties have 4.8 FQHC delivery sites for every 100 000 persons. Our sample is nearly evenly split between metro and nonmetro areas. Nearly 3-quarters of LHDs in our sample serve single counties (72.9%), while the rest serve single cities (14.2%) or multiple cities or counties (12.9%). Chronic disease–screening activity varies significantly across LHDs. While a majority (66.0%) provide screening for hypertension, a minority provide screening for diabetes (43.7%) and heart disease (33.7%). Local health departments that provide 1 screening service are also more likely to provide others. By contrast, far fewer LHDs provide primary care services. In fact, only 12.9% report doing so. The exponentiated results of the logistic regressions to predict LHD screening activities appear in Tables 2 through 4. These results can be interpreted as odds ratios. The column on the left presents the results of the model using the binary measure of FQHC availability, while the column on the right uses the continuous measure. Table 2 presents the results of the heart disease screening model. We find that LHDs in counties with an FQHC have, on average, 26% lower odds of providing heart disease screening, but this figure is not statistically significant. In the alternate model, we find that each FQHC delivery site per 100 000 persons in the county is associated with 2% lower odds of an LHD providing screening for heart disease, and this figure is statistically significant. Across the 2 different model specifications, the odds ratios for the other covariates are remarkably consistent and several are of interest. Local health departments headed by an executive director with an MD or DO degree had 42% lower odds of providing heart disease screening. By contrast, the remaining significant covariates were all associated with an increase in the odds of an LHD providing heart disease screening.
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E4 ❘ Journal of Public Health Management and Practice TABLE 1 ● Sample Summary Statistics
qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable
Mean (SD)
County health care supply factors FQHC present in county, % 37.8 FQHC delivery sites per 100 000 4.8 (15.8) population General hospitals per 100 000 3.8 (6.2) population Rural health clinics per 100 000 5.6 (12.5) population Primary care physicians per 6.3 (3.2) 10 000 population Midlevels (PAs, NPs, APRNs) per 10.1 (6.8) 10 000 population County demographic factors % Elderly 15.3 (3.8) % White race 83.3 (15.1) % Non-white race 16.7 (15.1) % Hispanic ethnicity 8.0 (10.7) % In poverty 15.4 (5.7) % Uninsured 16.1 (5.6) LHD service provision % LHDs providing diabetes 43.7 screening % LHDs providing blood pressure 66.0 screening % LHDs providing cardiovascular 33.7 screening % LHDs providing primary care 12.9 LHD staffing and leadership LHD staff per 10 000 LHD 7.1 (7.3) population % LHDs employing public health 38.8 physician % LHDs with clinician (MD, DO) 12.5 executive director % LHDs with public health (MPH, 21.9 DrPH) executive director LHDs by location % Metro 50.7 % Nonmetro 49.3 LHDs by Jurisdiction % County 72.9 % City 14.2 % Other (city-county, multicity, 12.9 multicounty) N = 2107 local health departments/counties
Range
TABLE 2 ● Results of Logistic Regression of LHD Cardiac Screening Activitya qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable 0-100 0-346.9 0-71.5 0-140.4 0-26.3 0-72.9
5.6-43.4 10.7-99.2 0.8-89.3 0.2-95.7 3.7-40.4 3.6-37.6 0-100 0-100 0-100 0-100 0-92.8 0-100 0-100 0-100
0-100 0-100 0-100 0-100 0-100
Abbreviations: FQHC, federally qualified health center; LHD, local health department.
Specifically, each additional staff person hired per 10 000 persons served by the LHD is associated with a 5% increase in the odds of providing heart disease screening. If one of those staff persons happens to be a public health physician, it is associated with an ad-
Odds Ratio (SE)
FQHC present in county 0.74 (0.13) FQHC delivery sites per ... 100 000 population General hospitals per 0.98 (0.02) 100 000 population Rural health clinics per 1.01 (0.01) 100 000 population Primary care physicians 1.02 (0.03) per 10 000 population Midlevels (PAs, NPS, 1.00 (0.01) APRNs) per 10 000 population % Elderly 1.00 (0.03) % Non-white race 0.99 (0.01) % Hispanic ethnicity 0.99 (0.01) % in Poverty 0.98 (0.02) % Uninsured 1.04 (0.03) LHD staff per 10 000 LHD 1.05c (0.01) population LHD employs public 1.99c (0.39) health physician LHD has clinician (MD, 0.58d (0.13) DO) executive director LHD has public health 1.22 (0.19) (MPH, DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.90 (0.15) LHD jurisdiction (county is omitted reference group) City 0.68 (0.22) Other (city-county, 1.85d (0.46) multicity, multicounty) N = 2008 0.062 Pseudo R2 =
Odds Ratio (SE) ... 0.98b (0.01) 0.98 (0.02) 1.01 (0.01) 1.01 (0.03) 1.00 (0.01)
1.00 (0.03) 0.99 (0.01) 0.99 (0.01) 0.99 (0.02) 1.03 (0.03) 1.05c (0.01) 1.96b (0.38) 0.58d (0.13) 1.21 (0.19)
0.95 (0.16) 0.69 (0.21) 1.92b (0.47)
0.066
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown within parentheses. b P < .01. c P < .001. d P < .05.
ditional 96% to 99% increase in the odds of providing heart disease screening. Finally, LHDs with a multicity/county jurisdiction have, on average, between 85% and 92% higher odds of providing heart disease screening than LHDs with a single county jurisdiction. The results of the diabetes-screening model are shown in Table 3 and are remarkably similar to the heart disease–screening model. We find that LHDs in counties with an FQHC have, on average, 17% lower odds of providing diabetes screening, but this figure is not statistically significant. In the alternate model, we find that each FQHC delivery site per 100 000 persons
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
❘ E5
TABLE 3 ● Results of Logistic Regression of LHD Diabetes
Screening Activitya qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
TABLE 4 ● Results of Logistic Regression of LHD Hypertension Screening Activitya qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable
Variable
Odds Ratio (SE)
FQHC present in county 0.83 (0.12) FQHC delivery sites per ... 100 000 population General hospitals per 0.98 (0.01) 100 000 population Rural health clinics per 1.02c (0.01) 100 000 population Primary care physicians per 0.99 (0.03) 10 000 population Midlevels (PAs, NPS, APRNs) 1.01 (0.02) per 10 000 population % Elderly 0.99 (0.03) % Non-white race 0.99 (0.01) % Hispanic ethnicity 0.99 (0.01) % In poverty 0.99 (0.02) % Uninsured 1.05 (0.03) LHD staff per 10 000 LHD 1.04c (0.02) population LHD employs public health 1.49c (0.27) physician LHD has clinician (MD, DO) 0.63c (0.13) executive director LHD has public health (MPH, 0.89 (0.13) DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.81 (0.13) LHD jurisdiction (county is omitted reference group) City 0.79 (0.23) Other (city-county, 1.77c (0.45) multicity, multicounty) N = 2028 0.056 Pseudo R 2 =
Odds Ratio (SE) ... 0.99b (0.004) 0.98 (0.01) 1.02 (0.01) 0.99 (0.03) 1.01 (0.02) 0.99 (0.03) 0.99 (0.01) 0.99 (0.01) 1.00 (0.02) 1.05 (0.03) 1.04c (0.02) 1.46c (0.27) 0.63c (0.14) 0.88 (0.13)
0.83 (0.13) 0.82 (0.24) 1.83c (0.46)
0.059
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown within parentheses. b P < .01. c P < .05.
in the county is associated with 1% lower odds of an LHD providing screening for diabetes, and this figure is statistically significant. Local health departments headed by an executive director with an MD or DO degree had 37% lower odds of providing diabetes screening. By contrast, the remaining significant covariates were all associated with an increase in the odds of an LHD providing diabetes screening. Specifically, each additional staff person hired per 10 000 persons served by the LHD is associated with a 4% increase in the odds of providing diabetes screening. If one of those staff persons happens to be a public health physician, it is associated with an additional 46% to 49% increase in the odds of providing diabetes
Odds Ratio (SE)
FQHC present in county 0.68b (0.11) FQHC delivery sites per ... 100 000 population General hospitals per 1.00 (0.02) 100 000 population Rural health clinics per 1.02 (0.01) 100 000 population Primary care physicians 0.97 (0.03) per 10 000 population Midlevels (PAs, NPS, 1.02 (0.01) APRNs) per 10 000 Population % Elderly 0.99 (0.02) % Nonwhite race 0.99 (0.01) % Hispanic ethnicity 0.99 (0.01) % In poverty 0.97 (0.02) % Uninsured 1.05b (0.02) LHD staff per 10 000 LHD 1.04b (0.02) population LHD employs public 1.25 (0.27) health physician LHD has clinician (MD, 0.84 (0.21) DO) executive director LHD has public health 0.91 (0.16) (MPH, DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.93 (0.15) LHD jurisdiction (county is omitted reference group) City 1.06 (0.29) Other (city-county, 1.18 (0.40) multicity, multicounty) N = 2061 0.051 Pseudo R 2 =
Odds Ratio (SE) ... 0.99b (0.004) 1.00 (0.02) 1.02 (0.01) 0.96 (0.03) 1.02 (0.01)
0.99 (0.02) 0.99 (0.01) 0.99 (0.01) 0.97 (0.02) 1.05b (0.02) 1.04b (0.02) 1.22 (0.27) 0.85 (0.21) 0.89 (0.16)
0.99 (0.16) 1.01 (0.28) 1.19 (0.41)
0.050
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown within parentheses. b P < .05.
screening. Finally, LHDs with a multicity/county jurisdiction have, on average, between 77% and 83% higher odds of providing diabetes screening than LHDs with a single county jurisdiction. The results of the hypertension screening model are shown in Table 4 and differ substantially from the results of the heart disease and diabetes screening models. We find that LHDs in counties with an FQHC have, on average, 32% lower odds of providing hypertension screening, and this figure is statistically significant. In the alternate model, we find that each FQHC delivery site per 100 000 persons in the county is associated with 1% lower odds of an LHD providing screening
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E6 ❘ Journal of Public Health Management and Practice for hypertension, and this figure is also statistically significant. While the staffing variable has a similar effect as that observed in the other 2 screening models (a 4% increase in the odds of providing hypertension screening), there is no statistically significant relationship between hypertension screening and either LHDs being headed by an executive director with an MD or DO degree or employing a public health physician. Also, whereas none of the county demographic variables were significant for either the heart disease or diabetes screening models, the percentage of the county population that is uninsured is significantly associated with a 5% increase in the odds of an LHD providing hypertension screening. The exponentiated results of the logistic regression models to predict LHD provision of primary care are shown in Table 5. These results can be interpreted as odds ratios. In the first model, we find that—contrary to our expectations—simply having an FQHC present in the county does not influence LHDs’ primary care provision. While the presence of an FQHC in the county is associated with an LHD having 28% lower odds of providing primary care, this figure is not statistically significant. However, in the alternate model, we find that as the number of FQHC delivery sites per 100 000 persons in a county increases, LHDs do become less likely to provide primary care. Specifically, for each additional FQHC delivery site per 100 000 population, an LHD has 4% lower odds of providing primary care. Obviously, as the number of delivery sites per capita increases, the magnitude of this effect can become rather considerable. Similarly, we find that the county supply of primary care physicians per capita is negatively associated with the likelihood of an LHD providing primary care in both models. For each additional primary care physician per 10 000 population, an LHD has between 7% and 8% lower odds of providing primary care. Similarly, each additional hospital per 100 000 persons in a county is associated with a 3% decrease in the odds of an LHD providing primary care. Local health department staffing is also strongly associated with the provision of primary care. Specifically, if a LHD employs a public health physician, its odds of providing primary care are approximately 6.7 times as high as those of other LHDs. And each additional LHD staff person per 10 000 population is associated with between a 5% and 6% increase in the odds of providing primary care. Local health departments headed by an executive director with a graduate degree in public health have between 48% and 53% higher odds of providing primary care. By contrast, compared with county-based LHDs, city-based LHDs are less likely to provide primary care. Specifically, the
TABLE 5 ● Results of Logistic Regression of LHD Primary Care Provisiona qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable
Odds Ratio (SE)
FQHC present in county 0.72 (0.19) FQHC delivery sites per ... 100 000 population General hospitals per 0.97c (0.02) 100 000 population Rural health clinics per 0.99 (0.01) 100 000 population Primary care physicians 0.93c (0.03) per 10 000 population Midlevels (PAs, NPS, 1.03 (0.02) APRNs) per 10 000 population % Elderly 1.01 (0.03) % Nonwhite race 1.00 (0.01) % Hispanic ethnicity 0.98 (0.01) % In poverty 1.00 (0.03) % Uninsured 1.07 (0.04) LHD staff per 10 000 LHD 1.05d (0.02) population LHD employs public 6.65b (1.56) health physician LHD has clinician (MD, 0.81 (0.20) DO) executive director LHD has public health 1.53d (0.23) (MPH, DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.78 (0.14) LHD jurisdiction (county is omitted reference group) City 0.44c (0.17) Other (city-county, 0.50 (0.19) multicity, multicounty) N = 2036 0.191 Pseudo R2 =
Odds Ratio (SE) ... 0.96b (0.01) 0.97c (0.02) 0.99 (0.01) 0.92c (0.03) 1.03c (0.015)
1.01 (0.03) 0.99 (0.01) 0.98 (0.01) 1.01 (0.03) 1.07 (0.04) 1.06d (0.02) 6.72b (1.57) 0.82 (0.21) 1.48d (0.23)
0.85 (0.16) 0.48c (0.18) 0.53 (0.19)
0.201
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown in parentheses. b P < .001. c P < .05. d P < .01.
odds of providing primary care are between 52% and 56% lower for city-based LHDs.
● Discussion Not surprisingly, the most important factor by far in whether or not an LHD provides primary care or screening for heart disease or diabetes is whether or not it employs a public health physician. However, the results of our current study also provide evidence
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
that the availability of an FQHC in the same county as an LHD at least partially explains the variation in LHDs’ provision of primary care services. The relationship is much stronger for primary care provision and hypertension screening than screening for heart disease or diabetes and appears to be tied more closely to the extent of FQHC availability in the county, rather than the mere presence of a single FQHC grantee. However, the presence of a single FQHC grantee did have a larger association with hypertension screening than did the per capita supply of FQHC delivery sites. Turning our attention first to the provision of screening tests for chronic conditions, we find that LHDs are simply more likely at baseline to provide screening tests than to provide primary care, and that—with the exception of hypertension—the availability of FQHCs has little relationship to LHDs’ provision of screening services. One possible explanation for this is that offering screening tests is less resource-intensive than providing comprehensive primary care and thus is less dependent on the extent of the local FQHC network. Yet, in some counties where there is an FQHC to provide chronic disease screening tests, some LHDs clearly opt not to do so themselves. It is important to note that the 3 types of screening services we modeled are quite different from a clinical perspective. That is, screening for heart disease or diabetes requires a greater degree of clinical skill and resources than does screening for hypertension. Our results confirm this, as there is a strong positive relationship between an LHD employing a public health physician and choosing to provide screening for heart disease and/or diabetes, while no such relationship exists for hypertension screening. The flipside of this is that the low barriers to offering hypertension screening also become low barriers to withdrawing the service, which may explain why the presence of an FQHC grantee in the county is able to have a negative impact on LHDs offering hypertension screening, while having no effect on the other more intensive screening services or primary care delivery. Looking next at primary care provision, we find that the direct provision of primary care services by LHDs depends solely on the number of FQHC delivery sites per 100 000 county population, while the mere presence of an FQHC grantee has no effect. This is similar to the results for heart disease and diabetes screening but differs from the results for hypertension screening. This result suggests that where there is adequate safety-net capacity, LHDs are less likely to be involved in the direct provision of primary care, precisely because it is a resource-intensive enterprise and falls outside of the scope of traditional public health functions, as long as other provider options are available.
❘ E7
Likewise, a greater supply of primary care physicians and short-term general hospitals is also associated with a decreased likelihood of an LHD providing primary care directly. Thus, unlike their decision to provide screening services, the decision of LHDs to provide primary care is not limited to FQHC availability, but rather takes into account community primary care capacity as a whole, of which FQHCs are often an integral part. Yet, to the extent that this is the case, it is striking that the amount of rural health clinics has no effect on LHD primary care services. It may be that while rural health clinics are primary care providers serving vulnerable populations, they are not mandated to treat the uninsured or indigent in the way that FQHCs are, leaving some LHDs to feel a continued obligation to ensure access to care for these populations. As the ACA is implemented, both LHDs and FQHCs must consider their response to the opportunities and challenges facing them. Central to the ACA is the increase in insurance coverage through the establishment of subsidized health insurance marketplaces nationwide and the expansion of Medicaid in participating states. This increase in coverage is expected to increase demand for health care, and while many mistakenly assume that this will lessen demands on health care safety-net providers, research in Massachusetts has found that—at least for FQHCs—the opposite is true.11,15 In that state, FQHCs actually saw an increase in their patient volume, as the capacity of non-safetynet providers was overwhelmed by the influx of newly insured patients seeking care. Given the ACA’s focus on—and funding for—preventive care, FQHCs may also decide to increase the amount of preventive care and public health-oriented activities they provide, meaning that LHDs may find that they are competing among a growing pool of providers for limited funding.11 There is also some risk that state and local governments will further reduce public health funding with the expectation that the ACA reduces the need for LHDs to delivery primary care.11 Overall, our data indicate that, consistent with traditional public health activities, LHDs are more involved in disease screening than in the provision of primary care. However, LHDs with larger staffs per capita are more likely to provide both screening (for each of the 3 services modeled) and primary care. The direction of this relationship is unclear, however. It could be that LHDs opt to provide these services because of a perceived community need and must then employ the requisite staff, or it could be that better resourced LHDs subsequently opt to expand their scope of services, regardless of community need. The positive association we observe between LHD staffing and the provision of both chronic disease screening and direct provision of primary care services is consistent with existing
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E8 ❘ Journal of Public Health Management and Practice evidence that LHD inputs, including staffing, are positively associated with certain health outcomes.16 Indeed, clinical service provision and screening activity may be one of the mechanisms producing such improved outcomes. Finally, we observe an interesting set of findings regarding LHD leadership. While we find no relationship between leadership and hypertension screening, we find a positive relationship between having leaders trained in public health and the LHD’s provision of primary care, while we find a negative relationship between having leaders trained in medicine and the LHD’s provision of heart disease and diabetes screening. We are not entirely sure how to interpret these results but speculate that medical doctors may be biased against LHDs engaging in the provision of services that they feel are more appropriately within physicians’ scope of practice, while public health practitioners may be inclined to envision a broader scope of practice for LHDs that includes primary care provision. Why consistent effects of each covariate are not observed in both the screening and primary care models is unclear. We must acknowledge that our study has several limitations. First, our use of a cross-sectional data set does not permit us to make causal claims about the relationship between the presence of an FQHC and the primary care and preventive services provided by LHDs. Second, our analysis was conducted at the county level, and both FQHCs and LHDs may serve more than 1 county. While we control for LHD jurisdiction, we do not control directly for FQHC service area. Rather, we capture the presence of at least 1 grantee in the county as well as the presence of delivery sites per capita in the county (which may belong to >1 FQHC grantee). Finally, our analysis relies heavily on self-reported survey data that may suffer from a combination of recall and response biases. However, these remain the best available data on the activities of LHDs. In conclusion, as the ACA increases funding for FQHCs, they may be able to expand their capacity, and, consequently, LHDs may be freed up to increase their focus on more traditional public health activities. While avoiding the duplication of services or overlapping of roles is laudable, there is a risk that this may further silo medical care and public health.17 Conversely, the increased demand expected to result from the coverage provisions of the ACA may overwhelm the capacity of some FQHCs, and LHDs may find that ensuring access to care becomes a larger challenge, in some cases, leading LHDs to increasingly engage in the direct provision of primary care services. Some LHDs might even decide that pursuing FQHC designation themselves is a more attractive option than attempting to collaborate with an existing FQHC in the community.11 In either case, it is imperative to craft policies and make strate-
gic investments of resources that foster collaboration between LHDs and other health care safety-net and primary care providers, including FQHCs, at a time when both need to integrate to effectively improve population health.18 REFERENCES 1. Falik M, Needleman J, Wells BL, Korb J. Ambulatory care sensitive hospitalizations and emergency visits: experiences of Medicaid patients using federally qualified health centers. Med Care. 2001;39(6):551-561. 2. Starfield B, Shi L, Macinko J. Contribution of primary care to health systems and health. Milbank Q. 2005;83(3):457-502. 3. Probst JC, Laditka JN, Laditka SB. Association between community health center and rural health clinic presence and county-level hospitalization rates for ambulatory care sensitive conditions: an analysis across eight US states. BMC Health Serv Res. 2009;9:134-144. 4. Keane C, Marx J, Ricci E. Local health departments’ mission to the uninsured. J Public Health Pol. 2003;24(2):130-149. 5. Hsuan C, Rodriguez HP. The adoption and discontinuation of clinical services by local health departments. Am J Public Health. 2014;104(1):124-133. 6. Institute of Medicine. For the Public’s Health. Investing in a Healthier Future. Washington, DC: The National Academies Press; 2012. 7. Keane C, Marx J, Ricci E. Privatization and the scope of public health: a national survey of local health department directors. Am J Public Health. 2001;91(4):611-617. 8. Wall S. Transformations in public health systems. Health Aff. 1998;17(3):64-80. 9. Jacobson PD, Dalton VK, Berson-Grand J, Weisman CS. Survival strategies for Michigan’s health care safety net providers. Health Serv Res. 2005;40(3):923-940. 10. Luo H, Sotnikov S, Shah G. Local health department activities to ensure access to care. Am J Prev Med. 2013;45(6):720-727. 11. National Association of County & City Health Officials. Implementation of the Patient Protection and Affordable Care Act. http://www.naccho.org/advocacy/healthreform/ upload/ACA-white-paper-final.pdf. Published June 2011. 12. Calman N, Hauser D, Lurio J, Wu WY, Pichardo M. Strengthening public health and primary care collaboration through electronic health records. Am J Public Health. 2012;102(11):e13e18. 13. The Affordable Care Act and Health Centers. Bureau of Primary Health Care, Health Resources and Services Administration, U.S. Department of Health and Human Services. http://bphc.hrsa.gov/about/healthcenterfactsheet.pdf. Accessed May 5, 2014. 14. Taylor J. Changes in Latitudes, Changes in Attitudes: FQHCs and Community Clinics in a Reformed Health Care Market. National Health Policy Forum, Issue Brief No. 848. Washington, DC: The George Washington University; 2012. http://www .nhpf.org/library/issue-briefs/IB848_FQHCsandReform_ 12-18-12.pdf. Accessed June 2, 2014. 15. Ku L, Jones E, Finnegan B, Shin P, Rosenbaum S. How is the primary care safety net faring in Massachusetts? Community health centers in the midst of health reform. The Kaiser
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
Commission on Medicaid and the uninsured. https:// www.nachc.com/client/MA%20Health%20Reform%20Kaiser %20GW.pdf. Published March 2009. 16. Erwin PC, Greene SB, Mays GP, Ricketts TC, Davis MV. The association of changes in local health department resources with changes in state-level health outcomes. Am J Public Health. 2011;101(4):609-615.
❘ E9
17. Brandt AM, Gardner M. Antagonism and accommodation: interpreting the relationship between public health and medicine in the United States during the 20th century. Am J Public Health. 2000;90(5):707-715. 18. Institute of Medicine. Primary Care and Public Health: Exploring Integration to Improve Population Health. Washington, DC: The National Academies Press; 2012.
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Objectives: We aim to gain a clearer understanding of the effect of colocated federally qualified health centers (FQHCs) on the delivery of primary care and preventive services by local health departments (LHDs). Design: We collected data from the 2010 National Association of County & City Health Officials Profile of Local Health Departments, the 2010 Uniform Data System, and the 2011 Area Resource File. Setting: Forty-eight states, excluding Rhode Island and Hawaii. Participants: The analytic sample contained data on 2107 LHDs across the country. Main Outcome Measures: We modeled 4 measures of primary care activity for each LHD: provision of screening for (1) heart disease, (2) diabetes, and (3) hypertension, and (4) provision of comprehensive primary care. Results: Local health departments in counties with an FQHC grantee have, on average, 32% lower odds of providing hypertension screening, but having an FQHC grantee in the county does not influence LHDs’ primary care provision or screening for heart disease or diabetes. However, in an alternate model examining FQHC delivery sites per capita, each site is associated with a decrease in the odds of LHDs providing primary care (4% lower odds) or screening for heart disease (2% lower odds), diabetes, or hypertension (both 1% lower odds). Conclusion: Local health departments are more involved in disease screening than the provision of primary care. However, larger LHDs are more likely to provide both screening and primary care. Our current study provides evidence that the availability of an FQHC in the same county as an LHD partially explains the variation in LHDs’ provision of primary care services. Local health departments with colocated FQHCs can focus on more traditional public health activities, while LHDs in areas without an FQHC may find that ensuring access to care is more
challenging, possibly leading LHDs to directly provide primary care services. KEY WORDS: federally qualified health center, local health
department, primary care
Comprehensive, patient-centered primary care is associated with improved health outcomes.1-3 Yet, many vulnerable populations have difficulty obtaining access to primary care and must rely on health care safety-net providers to offer these services. While ensuring access to care is a core public health function, there is disagreement in the public health community about the role that local health departments (LHDs) should play in directly providing primary care services as part of the health care safety-net.4,5 Typically financed by state and local governments, LHDs are most often organized at the county level where they are tasked with performing a wide array of public health functions including, but not limited to, disease surveillance and prevention, regulation and oversight of environmental health risks, and direct care delivery. Local health departments are sometimes primary care providers of last resort for the uninsured and indigent, but the proportion of LHDs engaged in the direct provision of clinical services is decreasing, and the Institute of Medicine has called for LHDs to further minimize such activities while increasing their engagement in traditional (ie, nonclinical) public health activities.6 For example, between 1997 and 2008, the proportion of LHDs providing primary care declined by 31%, and the proportion providing screening for
Author Affiliation: Health Management and Policy, The University of Iowa, Iowa City, Iowa (Dr Wright and Mr Nice). The authors declare no conflicts of interest.
J Public Health Management Practice, 2015, 21(1), E1–E9 C 2015 Wolters Kluwer Health | Lippincott Williams & Wilkins Copyright
Correspondence: Brad Wright, PhD, Health Management and Policy, The University of Iowa, 145 N. Riverside Dr, N240 CPHB, Iowa City, IA ([email protected]). DOI: 10.1097/PHH.0000000000000112
E1 Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E2 ❘ Journal of Public Health Management and Practice diabetes and hypertension declined by 22% and 27%, respectively.5 There is substantial variation in the scope of primary care services provided by LHDs, but the explanation for this variation is not well understood. While 1 survey study found that LHD directors feel that whether LHDs should directly provide clinical services is at least partially dependent on the availability of other safety-net providers,7 it is unclear how the availability of other local safety-net primary care providers may actually influence LHD service provision.5,8-10 As the Affordable Care Act (ACA) is implemented, LHDs that provide comprehensive primary care services are being faced with an important decision about whether or not to continue providing those services.11 On the one hand, the increase in insurance coverage may mean that LHDs are able to seek reimbursement for providing primary care services that were previously supported by federal, state, and local public health funds.11 This opportunity can be expected to materialize predominantly in states that have opted to participate in the Medicaid expansion. On the other hand, LHDs may find that the increase in insurance coverage allows individuals to go elsewhere for their care. In that case, LHDs may simply decide to stop providing primary care and redirect primary care resources to more traditional public health activities.11 However, this decision may be conditional on the availability and capacity of other health care safety-net providers in the local area. Our study uses nationally representative quantitative data to explore variation in LHD primary care services as a function of federally qualified health center (FQHC) availability by county. Federally qualified health centers are federally funded facilities that are located in underserved areas and/or serve underserved populations and are legally mandated to provide primary care to all without regard to their ability to pay using an income-sensitive sliding fee scale. Consequently, FQHCs serve a low-income, often uninsured population with higher rates of chronic disease than the general population.12 Currently, there are 1200 FQHC grantees nationwide (a grantee may operate more than 1 delivery site) providing care for approximately 20 million patients annually, and their capacity is expected to double to 40 million patients under the ACA.13 Like LHDs, FQHCs also face opportunities and challenges under the ACA. For example, while their dedicated program funding has been increased, and they stand to receive increased reimbursement through Medicaid expansion and increases in private insurance coverage, they are also being expected to meet much of the increased demand resulting from those coverage increases, and it is unclear whether they will have adequate capacity to do so.14
Guided by the public health function of ensuring access to care, the direct provision of clinical services by LHDs seems likely to be limited to cases in which other provider options are few. Therefore, we hypothesize that LHDs in counties with an FQHC will be less likely to provide primary care services. However, a recent study by Luo et al10 finds no association between the presence of an FQHC and an LHD’s provision of medical care, and another study by Hsuan and Rodriguez5 found no association between the supply of FQHCs per 100 000 persons in poverty and the choice of LHDs to discontinue clinical services, or the extent of services discontinued. However, both of these studies relied on small samples (516 and 198 LHDs, respectively) and used a more limited measure of FQHC availability, while more broadly defining their outcomes. By contrast, we use a nationally representative sample of more than 2000 LHDs, 2 measures of FQHC availability, and more specific outcomes limited to primary care provision. By doing so, we aim to gain a clearer understanding of the relationship between FQHC availability and the provision of primary care services by LHDs.
● Methods To investigate primary care provision as a function of FQHC availability, we used data from the 2010 National Association of County & City Health Officials Profile of Local Health Departments, the 2010 Uniform Data System (UDS), and the 2011 Area Resource File (ARF). The National Association of County & City Health Officials survey data include information on the types of services provided by LHDs, as well as information on their staffing and finances. Hawaii and Rhode Island are excluded from the National Association of County & City Health Officials data because they do not operate LHDs. We merged these data by county (using a ZIP to FIPS crosswalk file) with the UDS data, which provided the location of FQHCs nationally. Finally, we merged data from the ARF by county to control for other aspects of local health care supply and population demographics. Together, these 3 data sets constituted a single cross-sectional analytic file. The analytic sample contained data on 2107 LHDs across the country. Pairwise correlation among all variables indicated only mild collinearity, with only 2 associations above 0.5, and all associations were as expected. We generated summary statistics and noted that 254 observations (12.06%) were missing data on whether the LHD employed a public health physician, and 136 observations (6.45%) were missing data for the LHD staffing variable. To address this, we used multiple imputation methods to replace missing values with
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
predicted values of the staffing and public health physician variables generated as a function of all covariates with nonmissing data in the data set. Using the analytic file, we estimated 4 logistic regression models that explain variation in the scope of primary care services provided by LHDs as a function of FQHC availability in the county where the LHD is located and other factors. Specifically, because we are primarily interested in investigating the delivery of care related to chronic disease, we modeled the provision of screening for heart disease, diabetes, and hypertension, as well as the provision of comprehensive primary care. For each of these outcomes, some observations (< 5%) were missing data. Given the limited number of missing cases, we simply excluded these observations from our analyses. For each outcome measure, we operationalized our key independent variable, FQHC availability, in 1 of 2 ways. First, using UDS data, we defined FQHC availability as the presence or absence of an FQHC grantee in the county. Alternately, using UDS data, we defined FQHC availability as the number of FQHC delivery sites (not grantees) per 100 000 persons residing in the county. This was done to determine the relative importance of having any FQHC in the county versus the importance of FQHC capacity in the county. We also controlled for several other LHD-specific and countyspecific characteristics. At the LHD-level, we controlled for staffing per 10 000 persons in the LHD’s jurisdiction, because adequate staffing is essential to the ability to provide a variety of services. We also controlled for whether or not the LHD employed a public health physician, as this is likely to be an important determinant of providing primary care services. In that spirit, we also controlled for whether or not the LHD was headed by an executive with a clinical degree (defined as MD or DO) or an executive with a public health degree (MPH or DrPH), as we expected such individuals might take a more comprehensive approach to primary care and/or public health delivery. At the LHD level, we also controlled for the jurisdiction of the LHD, which we categorized into 3 groups: county, city, and other (which included joint city-county jurisdiction, and both multicity and multicounty jurisdictions). Using data from the ARF, we controlled for the county supply of primary care physicians per capita, midlevel providers per capita (including nurse practitioners, physician assistants, and advanced practice nurses), short-term general hospitals, and rural health clinics. These variables were included to control for supply-side factors that are likely to influence LHD scope of services by providing additional resources to area residents. We also used the ARF to classify counties as metropolitan or nonmetropolitan using rural-urban continuum codes (metro = 1, 2, or 3).
❘ E3
Finally, we used clustered standard errors to adjust for data clustering within states. We opted to do this instead of using state dummy variables, both to avoid perfect prediction problems and because we lacked a strong conceptual framework to suggest that differences between states (beyond those measures explicitly controlled for in our model) would explain LHD activity at the local level.
● Results Sample summary statistics are shown in Table 1. The average health department in our study is responsible for a population of 133 606 persons, although this ranged from a low of 339 to a high of more than 1 million persons. Of the 2107 LHDs in our sample, nearly 38% are in counties that also have at least 1 FQHC grantee. On average, these counties have 4.8 FQHC delivery sites for every 100 000 persons. Our sample is nearly evenly split between metro and nonmetro areas. Nearly 3-quarters of LHDs in our sample serve single counties (72.9%), while the rest serve single cities (14.2%) or multiple cities or counties (12.9%). Chronic disease–screening activity varies significantly across LHDs. While a majority (66.0%) provide screening for hypertension, a minority provide screening for diabetes (43.7%) and heart disease (33.7%). Local health departments that provide 1 screening service are also more likely to provide others. By contrast, far fewer LHDs provide primary care services. In fact, only 12.9% report doing so. The exponentiated results of the logistic regressions to predict LHD screening activities appear in Tables 2 through 4. These results can be interpreted as odds ratios. The column on the left presents the results of the model using the binary measure of FQHC availability, while the column on the right uses the continuous measure. Table 2 presents the results of the heart disease screening model. We find that LHDs in counties with an FQHC have, on average, 26% lower odds of providing heart disease screening, but this figure is not statistically significant. In the alternate model, we find that each FQHC delivery site per 100 000 persons in the county is associated with 2% lower odds of an LHD providing screening for heart disease, and this figure is statistically significant. Across the 2 different model specifications, the odds ratios for the other covariates are remarkably consistent and several are of interest. Local health departments headed by an executive director with an MD or DO degree had 42% lower odds of providing heart disease screening. By contrast, the remaining significant covariates were all associated with an increase in the odds of an LHD providing heart disease screening.
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E4 ❘ Journal of Public Health Management and Practice TABLE 1 ● Sample Summary Statistics
qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable
Mean (SD)
County health care supply factors FQHC present in county, % 37.8 FQHC delivery sites per 100 000 4.8 (15.8) population General hospitals per 100 000 3.8 (6.2) population Rural health clinics per 100 000 5.6 (12.5) population Primary care physicians per 6.3 (3.2) 10 000 population Midlevels (PAs, NPs, APRNs) per 10.1 (6.8) 10 000 population County demographic factors % Elderly 15.3 (3.8) % White race 83.3 (15.1) % Non-white race 16.7 (15.1) % Hispanic ethnicity 8.0 (10.7) % In poverty 15.4 (5.7) % Uninsured 16.1 (5.6) LHD service provision % LHDs providing diabetes 43.7 screening % LHDs providing blood pressure 66.0 screening % LHDs providing cardiovascular 33.7 screening % LHDs providing primary care 12.9 LHD staffing and leadership LHD staff per 10 000 LHD 7.1 (7.3) population % LHDs employing public health 38.8 physician % LHDs with clinician (MD, DO) 12.5 executive director % LHDs with public health (MPH, 21.9 DrPH) executive director LHDs by location % Metro 50.7 % Nonmetro 49.3 LHDs by Jurisdiction % County 72.9 % City 14.2 % Other (city-county, multicity, 12.9 multicounty) N = 2107 local health departments/counties
Range
TABLE 2 ● Results of Logistic Regression of LHD Cardiac Screening Activitya qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable 0-100 0-346.9 0-71.5 0-140.4 0-26.3 0-72.9
5.6-43.4 10.7-99.2 0.8-89.3 0.2-95.7 3.7-40.4 3.6-37.6 0-100 0-100 0-100 0-100 0-92.8 0-100 0-100 0-100
0-100 0-100 0-100 0-100 0-100
Abbreviations: FQHC, federally qualified health center; LHD, local health department.
Specifically, each additional staff person hired per 10 000 persons served by the LHD is associated with a 5% increase in the odds of providing heart disease screening. If one of those staff persons happens to be a public health physician, it is associated with an ad-
Odds Ratio (SE)
FQHC present in county 0.74 (0.13) FQHC delivery sites per ... 100 000 population General hospitals per 0.98 (0.02) 100 000 population Rural health clinics per 1.01 (0.01) 100 000 population Primary care physicians 1.02 (0.03) per 10 000 population Midlevels (PAs, NPS, 1.00 (0.01) APRNs) per 10 000 population % Elderly 1.00 (0.03) % Non-white race 0.99 (0.01) % Hispanic ethnicity 0.99 (0.01) % in Poverty 0.98 (0.02) % Uninsured 1.04 (0.03) LHD staff per 10 000 LHD 1.05c (0.01) population LHD employs public 1.99c (0.39) health physician LHD has clinician (MD, 0.58d (0.13) DO) executive director LHD has public health 1.22 (0.19) (MPH, DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.90 (0.15) LHD jurisdiction (county is omitted reference group) City 0.68 (0.22) Other (city-county, 1.85d (0.46) multicity, multicounty) N = 2008 0.062 Pseudo R2 =
Odds Ratio (SE) ... 0.98b (0.01) 0.98 (0.02) 1.01 (0.01) 1.01 (0.03) 1.00 (0.01)
1.00 (0.03) 0.99 (0.01) 0.99 (0.01) 0.99 (0.02) 1.03 (0.03) 1.05c (0.01) 1.96b (0.38) 0.58d (0.13) 1.21 (0.19)
0.95 (0.16) 0.69 (0.21) 1.92b (0.47)
0.066
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown within parentheses. b P < .01. c P < .001. d P < .05.
ditional 96% to 99% increase in the odds of providing heart disease screening. Finally, LHDs with a multicity/county jurisdiction have, on average, between 85% and 92% higher odds of providing heart disease screening than LHDs with a single county jurisdiction. The results of the diabetes-screening model are shown in Table 3 and are remarkably similar to the heart disease–screening model. We find that LHDs in counties with an FQHC have, on average, 17% lower odds of providing diabetes screening, but this figure is not statistically significant. In the alternate model, we find that each FQHC delivery site per 100 000 persons
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
❘ E5
TABLE 3 ● Results of Logistic Regression of LHD Diabetes
Screening Activitya qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
TABLE 4 ● Results of Logistic Regression of LHD Hypertension Screening Activitya qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable
Variable
Odds Ratio (SE)
FQHC present in county 0.83 (0.12) FQHC delivery sites per ... 100 000 population General hospitals per 0.98 (0.01) 100 000 population Rural health clinics per 1.02c (0.01) 100 000 population Primary care physicians per 0.99 (0.03) 10 000 population Midlevels (PAs, NPS, APRNs) 1.01 (0.02) per 10 000 population % Elderly 0.99 (0.03) % Non-white race 0.99 (0.01) % Hispanic ethnicity 0.99 (0.01) % In poverty 0.99 (0.02) % Uninsured 1.05 (0.03) LHD staff per 10 000 LHD 1.04c (0.02) population LHD employs public health 1.49c (0.27) physician LHD has clinician (MD, DO) 0.63c (0.13) executive director LHD has public health (MPH, 0.89 (0.13) DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.81 (0.13) LHD jurisdiction (county is omitted reference group) City 0.79 (0.23) Other (city-county, 1.77c (0.45) multicity, multicounty) N = 2028 0.056 Pseudo R 2 =
Odds Ratio (SE) ... 0.99b (0.004) 0.98 (0.01) 1.02 (0.01) 0.99 (0.03) 1.01 (0.02) 0.99 (0.03) 0.99 (0.01) 0.99 (0.01) 1.00 (0.02) 1.05 (0.03) 1.04c (0.02) 1.46c (0.27) 0.63c (0.14) 0.88 (0.13)
0.83 (0.13) 0.82 (0.24) 1.83c (0.46)
0.059
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown within parentheses. b P < .01. c P < .05.
in the county is associated with 1% lower odds of an LHD providing screening for diabetes, and this figure is statistically significant. Local health departments headed by an executive director with an MD or DO degree had 37% lower odds of providing diabetes screening. By contrast, the remaining significant covariates were all associated with an increase in the odds of an LHD providing diabetes screening. Specifically, each additional staff person hired per 10 000 persons served by the LHD is associated with a 4% increase in the odds of providing diabetes screening. If one of those staff persons happens to be a public health physician, it is associated with an additional 46% to 49% increase in the odds of providing diabetes
Odds Ratio (SE)
FQHC present in county 0.68b (0.11) FQHC delivery sites per ... 100 000 population General hospitals per 1.00 (0.02) 100 000 population Rural health clinics per 1.02 (0.01) 100 000 population Primary care physicians 0.97 (0.03) per 10 000 population Midlevels (PAs, NPS, 1.02 (0.01) APRNs) per 10 000 Population % Elderly 0.99 (0.02) % Nonwhite race 0.99 (0.01) % Hispanic ethnicity 0.99 (0.01) % In poverty 0.97 (0.02) % Uninsured 1.05b (0.02) LHD staff per 10 000 LHD 1.04b (0.02) population LHD employs public 1.25 (0.27) health physician LHD has clinician (MD, 0.84 (0.21) DO) executive director LHD has public health 0.91 (0.16) (MPH, DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.93 (0.15) LHD jurisdiction (county is omitted reference group) City 1.06 (0.29) Other (city-county, 1.18 (0.40) multicity, multicounty) N = 2061 0.051 Pseudo R 2 =
Odds Ratio (SE) ... 0.99b (0.004) 1.00 (0.02) 1.02 (0.01) 0.96 (0.03) 1.02 (0.01)
0.99 (0.02) 0.99 (0.01) 0.99 (0.01) 0.97 (0.02) 1.05b (0.02) 1.04b (0.02) 1.22 (0.27) 0.85 (0.21) 0.89 (0.16)
0.99 (0.16) 1.01 (0.28) 1.19 (0.41)
0.050
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown within parentheses. b P < .05.
screening. Finally, LHDs with a multicity/county jurisdiction have, on average, between 77% and 83% higher odds of providing diabetes screening than LHDs with a single county jurisdiction. The results of the hypertension screening model are shown in Table 4 and differ substantially from the results of the heart disease and diabetes screening models. We find that LHDs in counties with an FQHC have, on average, 32% lower odds of providing hypertension screening, and this figure is statistically significant. In the alternate model, we find that each FQHC delivery site per 100 000 persons in the county is associated with 1% lower odds of an LHD providing screening
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E6 ❘ Journal of Public Health Management and Practice for hypertension, and this figure is also statistically significant. While the staffing variable has a similar effect as that observed in the other 2 screening models (a 4% increase in the odds of providing hypertension screening), there is no statistically significant relationship between hypertension screening and either LHDs being headed by an executive director with an MD or DO degree or employing a public health physician. Also, whereas none of the county demographic variables were significant for either the heart disease or diabetes screening models, the percentage of the county population that is uninsured is significantly associated with a 5% increase in the odds of an LHD providing hypertension screening. The exponentiated results of the logistic regression models to predict LHD provision of primary care are shown in Table 5. These results can be interpreted as odds ratios. In the first model, we find that—contrary to our expectations—simply having an FQHC present in the county does not influence LHDs’ primary care provision. While the presence of an FQHC in the county is associated with an LHD having 28% lower odds of providing primary care, this figure is not statistically significant. However, in the alternate model, we find that as the number of FQHC delivery sites per 100 000 persons in a county increases, LHDs do become less likely to provide primary care. Specifically, for each additional FQHC delivery site per 100 000 population, an LHD has 4% lower odds of providing primary care. Obviously, as the number of delivery sites per capita increases, the magnitude of this effect can become rather considerable. Similarly, we find that the county supply of primary care physicians per capita is negatively associated with the likelihood of an LHD providing primary care in both models. For each additional primary care physician per 10 000 population, an LHD has between 7% and 8% lower odds of providing primary care. Similarly, each additional hospital per 100 000 persons in a county is associated with a 3% decrease in the odds of an LHD providing primary care. Local health department staffing is also strongly associated with the provision of primary care. Specifically, if a LHD employs a public health physician, its odds of providing primary care are approximately 6.7 times as high as those of other LHDs. And each additional LHD staff person per 10 000 population is associated with between a 5% and 6% increase in the odds of providing primary care. Local health departments headed by an executive director with a graduate degree in public health have between 48% and 53% higher odds of providing primary care. By contrast, compared with county-based LHDs, city-based LHDs are less likely to provide primary care. Specifically, the
TABLE 5 ● Results of Logistic Regression of LHD Primary Care Provisiona qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq
Variable
Odds Ratio (SE)
FQHC present in county 0.72 (0.19) FQHC delivery sites per ... 100 000 population General hospitals per 0.97c (0.02) 100 000 population Rural health clinics per 0.99 (0.01) 100 000 population Primary care physicians 0.93c (0.03) per 10 000 population Midlevels (PAs, NPS, 1.03 (0.02) APRNs) per 10 000 population % Elderly 1.01 (0.03) % Nonwhite race 1.00 (0.01) % Hispanic ethnicity 0.98 (0.01) % In poverty 1.00 (0.03) % Uninsured 1.07 (0.04) LHD staff per 10 000 LHD 1.05d (0.02) population LHD employs public 6.65b (1.56) health physician LHD has clinician (MD, 0.81 (0.20) DO) executive director LHD has public health 1.53d (0.23) (MPH, DrPH) executive director LHD location (metro is omitted reference group) Nonmetro 0.78 (0.14) LHD jurisdiction (county is omitted reference group) City 0.44c (0.17) Other (city-county, 0.50 (0.19) multicity, multicounty) N = 2036 0.191 Pseudo R2 =
Odds Ratio (SE) ... 0.96b (0.01) 0.97c (0.02) 0.99 (0.01) 0.92c (0.03) 1.03c (0.015)
1.01 (0.03) 0.99 (0.01) 0.98 (0.01) 1.01 (0.03) 1.07 (0.04) 1.06d (0.02) 6.72b (1.57) 0.82 (0.21) 1.48d (0.23)
0.85 (0.16) 0.48c (0.18) 0.53 (0.19)
0.201
Abbreviations: FQHC, federally qualified health center; LHD, local health department. a State-clustered standard errors are shown in parentheses. b P < .001. c P < .05. d P < .01.
odds of providing primary care are between 52% and 56% lower for city-based LHDs.
● Discussion Not surprisingly, the most important factor by far in whether or not an LHD provides primary care or screening for heart disease or diabetes is whether or not it employs a public health physician. However, the results of our current study also provide evidence
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
that the availability of an FQHC in the same county as an LHD at least partially explains the variation in LHDs’ provision of primary care services. The relationship is much stronger for primary care provision and hypertension screening than screening for heart disease or diabetes and appears to be tied more closely to the extent of FQHC availability in the county, rather than the mere presence of a single FQHC grantee. However, the presence of a single FQHC grantee did have a larger association with hypertension screening than did the per capita supply of FQHC delivery sites. Turning our attention first to the provision of screening tests for chronic conditions, we find that LHDs are simply more likely at baseline to provide screening tests than to provide primary care, and that—with the exception of hypertension—the availability of FQHCs has little relationship to LHDs’ provision of screening services. One possible explanation for this is that offering screening tests is less resource-intensive than providing comprehensive primary care and thus is less dependent on the extent of the local FQHC network. Yet, in some counties where there is an FQHC to provide chronic disease screening tests, some LHDs clearly opt not to do so themselves. It is important to note that the 3 types of screening services we modeled are quite different from a clinical perspective. That is, screening for heart disease or diabetes requires a greater degree of clinical skill and resources than does screening for hypertension. Our results confirm this, as there is a strong positive relationship between an LHD employing a public health physician and choosing to provide screening for heart disease and/or diabetes, while no such relationship exists for hypertension screening. The flipside of this is that the low barriers to offering hypertension screening also become low barriers to withdrawing the service, which may explain why the presence of an FQHC grantee in the county is able to have a negative impact on LHDs offering hypertension screening, while having no effect on the other more intensive screening services or primary care delivery. Looking next at primary care provision, we find that the direct provision of primary care services by LHDs depends solely on the number of FQHC delivery sites per 100 000 county population, while the mere presence of an FQHC grantee has no effect. This is similar to the results for heart disease and diabetes screening but differs from the results for hypertension screening. This result suggests that where there is adequate safety-net capacity, LHDs are less likely to be involved in the direct provision of primary care, precisely because it is a resource-intensive enterprise and falls outside of the scope of traditional public health functions, as long as other provider options are available.
❘ E7
Likewise, a greater supply of primary care physicians and short-term general hospitals is also associated with a decreased likelihood of an LHD providing primary care directly. Thus, unlike their decision to provide screening services, the decision of LHDs to provide primary care is not limited to FQHC availability, but rather takes into account community primary care capacity as a whole, of which FQHCs are often an integral part. Yet, to the extent that this is the case, it is striking that the amount of rural health clinics has no effect on LHD primary care services. It may be that while rural health clinics are primary care providers serving vulnerable populations, they are not mandated to treat the uninsured or indigent in the way that FQHCs are, leaving some LHDs to feel a continued obligation to ensure access to care for these populations. As the ACA is implemented, both LHDs and FQHCs must consider their response to the opportunities and challenges facing them. Central to the ACA is the increase in insurance coverage through the establishment of subsidized health insurance marketplaces nationwide and the expansion of Medicaid in participating states. This increase in coverage is expected to increase demand for health care, and while many mistakenly assume that this will lessen demands on health care safety-net providers, research in Massachusetts has found that—at least for FQHCs—the opposite is true.11,15 In that state, FQHCs actually saw an increase in their patient volume, as the capacity of non-safetynet providers was overwhelmed by the influx of newly insured patients seeking care. Given the ACA’s focus on—and funding for—preventive care, FQHCs may also decide to increase the amount of preventive care and public health-oriented activities they provide, meaning that LHDs may find that they are competing among a growing pool of providers for limited funding.11 There is also some risk that state and local governments will further reduce public health funding with the expectation that the ACA reduces the need for LHDs to delivery primary care.11 Overall, our data indicate that, consistent with traditional public health activities, LHDs are more involved in disease screening than in the provision of primary care. However, LHDs with larger staffs per capita are more likely to provide both screening (for each of the 3 services modeled) and primary care. The direction of this relationship is unclear, however. It could be that LHDs opt to provide these services because of a perceived community need and must then employ the requisite staff, or it could be that better resourced LHDs subsequently opt to expand their scope of services, regardless of community need. The positive association we observe between LHD staffing and the provision of both chronic disease screening and direct provision of primary care services is consistent with existing
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
E8 ❘ Journal of Public Health Management and Practice evidence that LHD inputs, including staffing, are positively associated with certain health outcomes.16 Indeed, clinical service provision and screening activity may be one of the mechanisms producing such improved outcomes. Finally, we observe an interesting set of findings regarding LHD leadership. While we find no relationship between leadership and hypertension screening, we find a positive relationship between having leaders trained in public health and the LHD’s provision of primary care, while we find a negative relationship between having leaders trained in medicine and the LHD’s provision of heart disease and diabetes screening. We are not entirely sure how to interpret these results but speculate that medical doctors may be biased against LHDs engaging in the provision of services that they feel are more appropriately within physicians’ scope of practice, while public health practitioners may be inclined to envision a broader scope of practice for LHDs that includes primary care provision. Why consistent effects of each covariate are not observed in both the screening and primary care models is unclear. We must acknowledge that our study has several limitations. First, our use of a cross-sectional data set does not permit us to make causal claims about the relationship between the presence of an FQHC and the primary care and preventive services provided by LHDs. Second, our analysis was conducted at the county level, and both FQHCs and LHDs may serve more than 1 county. While we control for LHD jurisdiction, we do not control directly for FQHC service area. Rather, we capture the presence of at least 1 grantee in the county as well as the presence of delivery sites per capita in the county (which may belong to >1 FQHC grantee). Finally, our analysis relies heavily on self-reported survey data that may suffer from a combination of recall and response biases. However, these remain the best available data on the activities of LHDs. In conclusion, as the ACA increases funding for FQHCs, they may be able to expand their capacity, and, consequently, LHDs may be freed up to increase their focus on more traditional public health activities. While avoiding the duplication of services or overlapping of roles is laudable, there is a risk that this may further silo medical care and public health.17 Conversely, the increased demand expected to result from the coverage provisions of the ACA may overwhelm the capacity of some FQHCs, and LHDs may find that ensuring access to care becomes a larger challenge, in some cases, leading LHDs to increasingly engage in the direct provision of primary care services. Some LHDs might even decide that pursuing FQHC designation themselves is a more attractive option than attempting to collaborate with an existing FQHC in the community.11 In either case, it is imperative to craft policies and make strate-
gic investments of resources that foster collaboration between LHDs and other health care safety-net and primary care providers, including FQHCs, at a time when both need to integrate to effectively improve population health.18 REFERENCES 1. Falik M, Needleman J, Wells BL, Korb J. Ambulatory care sensitive hospitalizations and emergency visits: experiences of Medicaid patients using federally qualified health centers. Med Care. 2001;39(6):551-561. 2. Starfield B, Shi L, Macinko J. Contribution of primary care to health systems and health. Milbank Q. 2005;83(3):457-502. 3. Probst JC, Laditka JN, Laditka SB. Association between community health center and rural health clinic presence and county-level hospitalization rates for ambulatory care sensitive conditions: an analysis across eight US states. BMC Health Serv Res. 2009;9:134-144. 4. Keane C, Marx J, Ricci E. Local health departments’ mission to the uninsured. J Public Health Pol. 2003;24(2):130-149. 5. Hsuan C, Rodriguez HP. The adoption and discontinuation of clinical services by local health departments. Am J Public Health. 2014;104(1):124-133. 6. Institute of Medicine. For the Public’s Health. Investing in a Healthier Future. Washington, DC: The National Academies Press; 2012. 7. Keane C, Marx J, Ricci E. Privatization and the scope of public health: a national survey of local health department directors. Am J Public Health. 2001;91(4):611-617. 8. Wall S. Transformations in public health systems. Health Aff. 1998;17(3):64-80. 9. Jacobson PD, Dalton VK, Berson-Grand J, Weisman CS. Survival strategies for Michigan’s health care safety net providers. Health Serv Res. 2005;40(3):923-940. 10. Luo H, Sotnikov S, Shah G. Local health department activities to ensure access to care. Am J Prev Med. 2013;45(6):720-727. 11. National Association of County & City Health Officials. Implementation of the Patient Protection and Affordable Care Act. http://www.naccho.org/advocacy/healthreform/ upload/ACA-white-paper-final.pdf. Published June 2011. 12. Calman N, Hauser D, Lurio J, Wu WY, Pichardo M. Strengthening public health and primary care collaboration through electronic health records. Am J Public Health. 2012;102(11):e13e18. 13. The Affordable Care Act and Health Centers. Bureau of Primary Health Care, Health Resources and Services Administration, U.S. Department of Health and Human Services. http://bphc.hrsa.gov/about/healthcenterfactsheet.pdf. Accessed May 5, 2014. 14. Taylor J. Changes in Latitudes, Changes in Attitudes: FQHCs and Community Clinics in a Reformed Health Care Market. National Health Policy Forum, Issue Brief No. 848. Washington, DC: The George Washington University; 2012. http://www .nhpf.org/library/issue-briefs/IB848_FQHCsandReform_ 12-18-12.pdf. Accessed June 2, 2014. 15. Ku L, Jones E, Finnegan B, Shin P, Rosenbaum S. How is the primary care safety net faring in Massachusetts? Community health centers in the midst of health reform. The Kaiser
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FQHCs and Primary Care at Health Departments
Commission on Medicaid and the uninsured. https:// www.nachc.com/client/MA%20Health%20Reform%20Kaiser %20GW.pdf. Published March 2009. 16. Erwin PC, Greene SB, Mays GP, Ricketts TC, Davis MV. The association of changes in local health department resources with changes in state-level health outcomes. Am J Public Health. 2011;101(4):609-615.
❘ E9
17. Brandt AM, Gardner M. Antagonism and accommodation: interpreting the relationship between public health and medicine in the United States during the 20th century. Am J Public Health. 2000;90(5):707-715. 18. Institute of Medicine. Primary Care and Public Health: Exploring Integration to Improve Population Health. Washington, DC: The National Academies Press; 2012.
Copyright © 2015 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
