Assessing clinical quality performance and staffing capacity differences between urban and rural Health Resources and Services Administration-funded health centers in the United States: A cross sectional study.

BACKGROUND: In the United States, there are nearly 1,400 Health Resources and Services Administration-funded health centers (HCs) serving low-income and underserved populations and more than 600 of these HCs are located in rural areas. Disparities in quality of medical care in urban vs. rural areas exist but data on such differences between urban and rural HCs is limited in the literature. We examined whether urban and rural HCs differed in their performance on clinical quality measures before and after controlling for patient, organizational, and contextual characteristics. METHODS AND
FINDINGS: We used the 2017 Uniform Data System to examine performance on clinical quality measures between urban and rural HCs (n = 1,373). We used generalized linear regression models with the logit link function and binomial distribution, controlling for confounding factors. After adjusting for potential confounders, we found on par performance between urban and rural HCs in all but one clinical quality measure. Rural HCs had lower rates of linking patients newly diagnosed with HIV to care (74% [95% CI: 69%, 80%] vs. 83% [95% CI: 80%, 86%]). We identified control variables that systematically accounted for eliminating urban vs. rural differences in performance on clinical quality measures. We also found that both urban and rural HCs had some clinical quality performance measures that were lower than available national benchmarks. Main limitations included potential discrepancy of urban or rural designation across all HC sites within a HC organization.
CONCLUSIONS: Findings highlight HCs' contributions in addressing rural disparities in quality of care and identify opportunities for improvement. Performance in both rural and urban HCs may be improved by supporting programs that increase the availability of providers, training, and provision of technical resources.

Introduction

An estimated 72 percent of the land area in the United States is considered rural and 14 percent of the population or 46.1 million people live in rural areas [1]. Evidence indicates significant and persistent rural disparities in quality of care. Rural areas (micropolitan and noncore) were found to have worse quality of care based on performance on 250 quality indicators examined compared to large suburban areas [2]. In addition, these disparities have remained the same for over one third or worsened for about one in ten of these indicators from 2000 to 2015 [2]. The indicators with worse quality of care constituted about a third of the 26 effective treatment indicators, a third to half of the care coordination indicators, and a quarter to one third of the 18 access to care indicators in rural areas [2]. These disparities co-occur with variations in sociodemographics and disparities in health status [3-6]. Nationally, rural populations are more socioeconomically vulnerable than urban areas. This includes more often being without high school education (16% vs. 13%), low-income (38% vs. 33% under federal poverty guidelines), with Medicaid coverage (24% vs. 22%), and in poorer health (13% vs. 9%). It also includes being less often employed (56% vs. 58%), and female (48% vs 50%) [7-9]. In addition, 85% of rural counties in the United States were persistently considered primary care Health Professional Shortage Areas at least once from 1996 to 2005, which impacts access to care and might impact clinical quality measure performance [10, 11].

Health centers funded by the Health Resources and Services Administration (HRSA) (referred to as HCs hereon) have a significant presence in rural areas and may be the only providers in some rural locations [3]. In some areas, more than 20% of the low-income population are served by HCs [12]. HCs are also mandated as part of their mission to provide care to those who are geographically isolated, including those in rural settings. In 2017, about 44% of the 1,373 HCs served rural populations, operated more than 4,400 sites, and collectively provided care through 35 million visits to more than 27 million total patients [13]. Earlier studies of HCs indicated that rural HC patients were more often older, female, White, poor, uninsured, obese, in poor health, and with activity limitations compared to the general rural population [3]. In addition, rural HCs had lower staffing supply for primary, mental health, and dental care providers than urban settings [3, 4, 6, 14, 15]. Strategic quality improvement initiatives over the years have improved provider recruitment and retention in rural areas and may have reduced disparities in access to care and quality of care [16, 17]. In fact, past research assessing performance on clinical quality at HCs has shown that rural HCs performed well in prenatal care outcomes, cervical cancer screening, and childhood immunization rates [3, 18, 19].

Existing data on urban/rural disparities in quality of care among HCs is limited, and a gradual decline in the size of rural populations may have influenced previous understanding of this issue [1]. Recent literature has found that despite adjusting for individual and community-level factors, several differences in quality of care measures were observed between urban and rural general populations [20, 21]. These findings suggest other unknown mechanisms may be affecting quality of care among urban and rural areas. A comprehensive assessment of the clinical quality of HCs in rural communities is necessary to identify areas in need of improvement and reduce missed opportunities in addressing rural disparities in quality of care. To address this gap, we examined differences in performance on clinical quality measures between rural and urban HCs, controlling for differences in characteristics of these organizations as well as other contextual factors that might impact performance on these measures. We hypothesized that rural HCs performed as well as urban HCs because of the emphasis HRSA places on improving access to quality health care services for all HCs. HRSA has supported quality improvement by requiring reporting on performance, incentivizing improvement and meeting national benchmarks of performance and providing financial and technical support to improve performance [22-24]. Our study aimed to highlight disparities in performance between rural and urban HCs and to identify factors that may mediate urban/rural performance variations.

Methods

Data and sample

For this cross-sectional study, we used data from the 2017 Uniform Data System (UDS) reported by the entire HRSA-funded HC patient population on organizational characteristics and clinical quality measures. UDS is an administrative data source maintained by HRSA to monitor the Health Center Program and provide information to stakeholders. HRSA-funded HCs and other entities receiving federal funding authorized under Section 330 of the Public Health Service Act are required to report UDS data. UDS captures aggregate information at the organization level rather than individual delivery sites that operate within the organization [22].

We merged the UDS data with the latest available relevant data from the 2016 Area Health Resource File (AHRF). AHRF is a publicly available dataset maintained by HRSA. It compiles information from over 50 sources to provide county-level data on population characteristics, health workforce availability, health care utilization, and health facilities [25]. We merged these data using the Federal Information Process Standards (FIPS) code associated with the address of the HC organization. If HCs were present in multiple counties, we merged the data for the county where the largest share of HC patients lived. We included all HCs that reported serving patients in 2017 for a total analytic sample of 1,373.

Dependent variables

We studied 15 clinical quality measures that HCs are required to report. The majority of these measures were standard quality metrics and concordant with Centers for Medicare and Medicaid Services guidelines and electronically specified for automated reporting by HCs (electronic clinical quality measures) [26]. These measures had national benchmarks included in the 2017 Healthcare Effectiveness Data and Information Set (HEDIS) for Medicaid Managed Care patients [27, 28]. All 15 measures were included in our analyses, with seven measures examining prevention-related performance, five measures associated with care management, and three measures assessing clinical outcomes [29, 30]. The seven prevention measures included (1) up-to-date childhood immunization completion, (2) receipt of recommended cervical cancer screening, (3) receipt of colorectal cancer screening, (4) tobacco use and cessation counseling and intervention, (5) depression screening and receipt of a follow-up plan, (6) weight assessment and counseling for nutrition and physical activity for children and adolescents, and (7) body mass index (BMI) screening and follow-up plan for adults. The five measures examining aspects of care management included: (1) patients with asthma receiving appropriate medications, (2) patients with coronary artery disease that were prescribed lipid-lowering therapy, (3) patients with ischemic vascular disease who used aspirin or another antithrombotic drug, (4) patients seen for follow-up care within 90 days of initial HIV diagnosis, and (5) pregnant women who received early prenatal care. Three measures examined outcomes of care, including (1) patients with diabetes whose hemoglobin A1c level was greater than 9% (poorly controlled), (2) patients with diagnosed hypertension whose blood pressure was below 140/90 (controlled), and (3) patients born whose birthweight was below normal (2,500 grams). The full measure definitions are described in S1 Table.

We created a dichotomous indicator variable that measured the proportion of urban and rural HCs that met or exceeded (vs. did not) the 2017 HEDIS national benchmarks for each clinical quality measure.

Independent variables

The primary variable of interest was urban/rural status. HCs self-reported this status for the organization and used the same designation for all delivery sites. In some instances, HCs had several service delivery sites across both urban and rural areas, which resulted in some misclassification. We controlled for several HC organizational characteristics, patient characteristics, and contextual variables to account for any potential confounding among HC and local area factors. The HC organizational characteristics controlled for included the organizational size indicated by the number of sites and number of patients seen in 2017. We further controlled for patient demographic and health characteristics including percent of patients ages 0–17 and ages 65 or older, patients who were racial/ethnic minorities, patients who communicated with the provider in a language other than English, patients with heart related disease, patients with diabetes or endocrine disease, patients with respiratory disease, patients with HIV, and prenatal care patients who delivered during the year. These variables controlled for HC case mix and challenges to care outcomes.

We next controlled for primary care and other care capacity and service availability using several indicators. These included the ratio of HC patients per each full-time equivalent (FTE) primary care provider (PCPs include physicians, nurse practitioners, and physician assistants) and ratio of FTE nurses per PCP, the ratio of mental health providers (psychiatrists, psychologists, licensed clinical social workers, other licensed mental health providers) per 5,000 patients, dental providers (dentists and hygienists) per 2,500 patients, enabling service staff (case managers, transportation, and translation staff) per 5,000 patients, and an index of number of services available in addition to medical care. We also included financial resource indicators, including per capita total revenues to measure success in generating revenues and proportion of grant revenues from the Section 330 grants to measure success in fundraising. The contextual control variables, extracted from AHRF, included the ratio of PCP per 5,000 individuals in the county, the proportion of individuals below the federal poverty guideline, and the proportion of minorities in the county.

Statistical analysis

We compared the independent and control variables by urban/rural status using t-tests. We then developed fractional outcome regression models using the fracreg command and logit distribution [31, 32]. These models were used to compare clinical quality measures after adjusting for confounding impact of HC patient and organizational characteristics and county-level contextual factors. We further compared the proportion of urban vs. rural HCs that met or exceeded 2017 HEDIS benchmarks using logistic regression models and adjusting for control variables. We included only complete data for all analyses presented in this paper. All HCs were treated with equal analytical weight. All analyses were conducted using Stata v. 15, and we used the Margins command to report predicted probabilities for ease of interpretation. We discussed all statistically significant results with probability values of 0.05 or smaller.

Ethics statement

Secondary data on HCs were de-identified, and as such, the study was granted written exemption from review by the University of California Los Angeles Institutional Review Board (study number 16–001528).

Results

Table 1 indicated about 44% of HCs were rural. On average, rural HCs were smaller than non-rural HCs as indicated by fewer sites (7.4 [SD 8.1] vs. 8.6 [SD 9.6]) and patients (14,673 [SD 18,068] vs. 23,861[SD 26,623]). Rural HCs had more older patients (13.2% [SD 6.6%]vs 7.2% [SD 4.2%]), fewer racial/ethnic minorities (38.2% [SD 31.1%] vs 70.7% [SD 23.5%]), and less Medicaid patients (35.1% [SD 17.1%] vs. 50.7% [SD 18.6%]) than urban HCs. Rural HCs also had more patients with heart and respiratory diseases but fewer patients with HIV or prenatal patients who delivered than urban HCs. Rural HCs also differed from urban HCs in capacity of with a higher ratio of dental providers per 2,500 patients (0.8 [SD 0.8] vs 0.7 [SD 0.7]), and nurses to PCPs (0.9 [SD 0.6] vs. 0.7 [SD 0.5]), but lower PCP panel size (1,092 [SD 514] vs. 1,168 [SD 501]) and ratio of mental health providers to 5,000 patients (1.6 [SD 2.1] vs. 2.2 [SD 3.9]). Section 330 grants represented a higher percentage of total revenue in rural HCs than urban HCs (34.0% [SD 17.9%] vs. 25.9% [SD 17.9%]). Rural HCs also had lower PCP capacity in the county overall and less racial/ethnic diversity but more patients living in poverty than urban HCs.

Table 1

Contextual characteristics.

	Total	Urban	Rural	P-value (Urban vs. Rural)
Sample Size n (%)	1,373	765 (56%)	608 (44%)
	Mean (SD)	Mean (SD)	Mean (SD)
Organization Size
Average number of sites	8.1 (9.0)	8.6 (9.6)	7.4 (8.1)	0.010
Average number of patients seen during the year	19,792 (23,663)	23,861 (26,623)	14,673 (18,068)	0.000
Patient Characteristics/Complexity
Percent of patients 65 years and older	9.8% (6.2%)	7.2% (4.2%)	13.2% (6.6%)	0.000
Percent of patients between 0 to 17 years old	26.4% (12.9%)	26.9% (14.3%)	25.8% (10.7%)	0.130
Percent of patients that were racial/ethnic minorities	56.3% (31.6%)	70.7% (23.5%)	38.2% (31.1%)	0.000
Percent of patients that spoke with primary care provider (PCP) in a language other than English	19.1% (22.8%)	24.3% (22.0%)	12.5% (22.2%)	0.000
Percent of Medicaid patients	43.8% (19.5%)	50.7% (18.6%)	35.1% (17.1%)	0.000
Percent of patients with heart related disease	3.1% (2.1%)	2.5% (1.6%)	3.9% (2.4%)	0.000
Percent of patients with diabetes or endocrine diseases	9.5% (4.2%)	9.3% (4.1%)	9.6% (4.2%)	0.280
Percent of patients with respiratory diseases	3.3% (2.4%)	2.5% (1.8%)	4.3% (2.7%)	0.000
Percent of patients with HIV	0.8% (3.2%)	1.3% (4.1%)	0.2% (1.3%)	0.000
Percent of prenatal care patients who delivered during the year	0.8% (0.9%)	1.0% (1.0%)	0.6% (0.8%)	0.000
PCP Staffing and Capacity
PCP Panel Size (Patients per PCP)	1134.8 (508.1)	1168.6 (501.0)	1092.0 (514.1)	0.006
Ratio of nurses to PCP	0.8 (0.6)	0.7 (0.5)	0.9 (0.6)	0.000
Additional Staffing and Capacity
Ratio of mental health provider per 5,000 patients	1.9 (3.2)	2.2 (3.9)	1.6 (2.1)	0.000
Ratio of dental provider per 2,500 patients	0.8 (0.7)	0.7 (0.7)	0.8 (0.8)	0.004
Ratio of enabling service staff per 5,000 patients	5.3 (7.4)	6.0 (8.5)	4.5 (5.6)	0.000
Average number of services provided in addition to medical care	3.5 (1.6)	3.6 (1.6)	3.3 (1.6)	0.000
Financial Resources
Per capita total revenues	$1,084 ($928)	$1,112 ($890)	$1,048 ($973)	0.207
Percent of total revenues that are from 330 grants	29.5% (18.7%)	25.9% (17.9%)	34.0% (18.6%)	0.000
Contextual characteristics
Ratio of PCP per 5,000 population in county	3.8 (1.7)	4.3 (1.4)	3.2 (1.8)	0.000
Percent below federal poverty guideline in county	16.1% (5.6%)	15.3% (4.8%)	17.2% (6.3%)	0.000
Percent of minority in county	38.8% (23.7%)	46.8% (20.8%)	28.5% (23.2%)	0.000

Standard deviation in parentheses. Analyses involved comparing independent and control variables by urban and rural health center status using t-tests.

PCP, Primary care provider; SD, standard deviation.

Unadjusted clinical quality measures showed multiple differences between rural and urban HCs including lower rates of up-to-date child immunizations (30% [SD 23%] vs 38% [SD 23%]), recommended cervical cancer screening (47% [SD 17%] vs. 53% [SD 18%]), and weight assessment and counseling for nutrition and physical activity for children and adolescents (55% [SD 26%] vs. 62% [SD 26%], Fig 1). After adjusting for patient, organizational and county-level characteristics, there were no differences between rural and urban HCs among prevention measures. Among unadjusted care management measures, rural HCs had lower rates of appropriate pharmacological therapy for patients with persistent asthma (83% [SD 17%] vs. 86% [SD 13%]), lipid lowering therapy for patients with coronary artery disease (78% [SD 15%] vs. 80% [SD 13%]), use of antithrombotic drugs for patients with ischemic vascular disease (76% [SD 16%] vs. 78% [SD 14%]), and linkage to care for newly diagnosed HIV patients (71% [SD 40%] vs. 84% [SD 27%]) but higher rates of early prenatal care for pregnant patients (81% [SD 16%] vs. 74% [SD 15%]) compared to urban HCs. After adjustment, the differences in clinical quality measures remained statistically significant only in care management of newly diagnosed HIV patients. Rural HCs had a predicted probability of 75% [95% CI: 69%, 80%] of newly diagnosed HIV patients being linked to care in 90 days compared to 83% [95% CI: 80%, 86%] in urban HCs. Among unadjusted outcome quality measures, rural HCs had different performance rates, with 32% (SD 12%) of rural HCs reporting patients with diabetes had uncontrolled hemoglobin A1c levels (vs. 35% [SD 12%] of urban HCs) and 63% reporting patients with hypertension had their blood pressure controlled (vs. 61% of urban HCs). All the unadjusted and adjusted clinical quality measure outcomes are presented in S2 Table.

Fig 1

Unadjusted and adjusted predicted probabilities for health center quality indicators by urban and rural status.

Unadjusted analyses involved comparing urban and rural health center status using t-tests. Adjusted analyses were conducted using fractional outcome regression models using the logit distribution. Statistically significant at *p<0.05; **p<0.01; ***p<0.001 comparing urban vs. rural.

Differences in all clinical quality measures, with the exception of 90-day follow-up care for newly diagnosed HIV patients, were explained by underlying differences in patient demographics and health status, organizational characteristics, and contextual factors to varying degrees and depending on the performance measure. For example, the difference between urban and rural HC performance on childhood immunization completion was explained by the higher number of non-English speaking patients, higher rate of children at the HC, and lower rate of patients with respiratory diseases (S3 Table). We also compared the predicted probabilities of each measure for urban and rural HCs with HEDIS Medicaid Managed Care national benchmarks and found that on average rural HCs met or exceeded the benchmarks for the preventive measures of tobacco use and cessation counseling and intervention, care management measures of patients with asthmas receiving appropriate medications and lipid lowering therapy for patients with coronary artery disease, and outcome measures of patients with diabetes with hemoglobin A1c greater than 9% and patients with hypertension with blood pressure below 140/90mmHg (Table 2).

Table 2

Predicted probabilities of proportion of urban and rural health centers that met or exceed quality benchmarks.

	2017 HEDIS Medicaid Managed Care Benchmarks1	Predicted probabilities 2
Measure Definition		Urban		Rural
Sample Size n (%)		765 (56%)		608 (44%)
		Predicted Probability	95% CI	Predicted Probability	95% CI
Prevention
Childhood Immunization	35%	48%	[44%,53%]	41%	[36%,46%]
Cervical Cancer Screening	59%	32%	[28%,36%]	33%	[28%,37%]
Tobacco Use Counseling	77%	84%	[81%,87%]	80%	[76%,84%]
Child Weight Counseling	73%	38%	[34%,42%]	35%	[29%,40%]
Adult Body Mass Index (BMI) Screening	85%	19%	[15%,23%]	17%	[13%,21%]
Care Management
Asthma Treatment	61%	94%	[92%,97%]	93%	[91%,96%]
Lipid Therapy	76%	69%	[65%,73%]	67%	[63%,72%]
Early Prenatal Care	81%	44%	[40%,48%]	45%	[40%,50%]
Outcomes
Uncontrolled Diabetes	41%	79%	[76%,82%]	80%	[76%,84%]
Hypertension Control	57%	71%	[67%,74%]	73%	[68%,77%]

Notes:

1 Benchmarks are based on national benchmarks in the 2017 Healthcare Effectiveness Data and Information Set for Medicaid Managed Care patients.

2 Analyses were conducted using logistic regression models and creating an indicator outcome variable of whether the health center met or exceeded the 2017 HEDIS Medicaid Managed Care Benchmarks.

3 Several clinical quality measures did not have an associated 2017 HEDIS Medicaid Managed Care Benchmark. These included colorectal cancer screening, depression screening and follow-up, aspirin therapy, HIV linkage to care, and low birth weight.

Standard deviation in parentheses.

BMI, body mass index; CAD, coronary artery disease; IVD, ischemic vascular disease; HIV, human immunodeficiency virus; HbA1c, Hemoglobin A1c; SD, standard deviation; HC, health center; HEDIS, Healthcare Effectiveness Data and Information Set; CI, confidence interval.

We further examined the predicted probability of proportion of urban and rural HCs that met or exceeded these HEDIS benchmarks, when benchmarks were available. These data showed that predicted probabilities of proportion of urban and rural HCs that met or exceeded the preventive, care management, and outcome measures were statistically similar, despite apparent differences. For example, 48% of urban and 41% of rural HCs met or exceeded the national benchmark of 35% for up-to-date childhood immunization completion rate and this apparent difference was not statistically significant. The full logistic regression models are displayed in S6–S8 Tables.

The control variables that explained differences in urban/rural HC performance are displayed in Table 3. These results showed that performance differences in preventive measures were explained by proportion of non-English speaking patients, percentage of patients with diabetes, percentage of patients 0–17 years of age, and ratio of PCPs per 5,000 persons in the county. In contrast, contextual factors such as percentage of poor or minority patients in the county did not predict differences in preventive measures. Urban/rural differences in care management measures were most frequently explained by percentage of non-English speaking patients, but other characteristics did not play a major or any role. Finally, urban/rural differences in outcome measures were most frequently explained by percentage of patients who were racial/ethnic minorities and percentage who were non-English speaking patients. The remaining variables played a role less frequently or did not have a role. The full regression models with the coefficients for each control variable are noted in S3–S5 Tables.

Table 3

Select significant health center characteristics and organizational factors associated with clinical performance measures.

	Prevention							Care Management					Outcomes
	Up-to-Date Childhood Immunization Completion	Receipt of Recommended Cervical Cancer Screening	Receipt of Colorectal Cancer Screening	Tobacco Use and Cessation Counseling and Intervention	Depression Screening and Receipt of a Follow-up Plan	Weight Assessment and Counseling for Nutrition and Physical Activity for Children and Adolescents	Body Mass Index (BMI) Screening and Follow-up Plan for adults	Patients with Asthma Receiving Appropriate Medications	Patients with Coronary Artery Diseases That Were Prescribed Lipid-Lowering Therapy	Patients with Ischemic Vascular Disease Who Used Aspirin or Another Antithrombotic Drug	Patient Seen for Follow-up Care within 90 Days of initial HIV Diagnosis	Pregnant Women Who Received Early Prenatal Care	Patients with Diabetes with Hemoglobin A1c Greater Than 9%	Patients with Hypertension with Blood Pressure below 140/90	Patients Born Whose Birthweight Was Below Normal
Urban (vs. rural)											↑
Organization Size
Average number of sites						↑						↓			↓
Average number of patients seen during the year		↑	↑		↑							↑	↑
Patient Characteristics
Percent of patients that were racial/ethnic minorities			↑									↑	↑
Percent of patients that spoke with primary care provider (PCP) in a language other than English	↑	↑	↑			↑	↑								↓
Percent of patients 65 years and older			↓	↓						↓		↓	↓	↓	↑
Percent of patients between 0–17 years	↑	↑	↑	↑		↑		↑	↑	↑	↑	↑	↑	↑	↓
Percent of patients with heart related disease														↑
Percent of patients with diabetes or endocrine diseases			↑
Percent of patients with respiratory diseases		↑	↑	↑	↑	↑	↑		↑				↑	↑
Percent of patients with HIV	↓	↓	↓										↓
Percent of prenatal care patients who delivered during the year							↑		↑		↑		↑
Percent of Medicaid Patients		↑	↑			↑			↑			↓			↓
PCP Staffing and Capacity
PCP Panel Size (Patients Per Provider)					↑	↑	↓		↑
Ratio of nurses to PCP										↑		↓
Additional Staffing and Capacity
Ratio of mental health provider per 5,000 patients					↑
Ratio of dental provider per 2,500 patients		↑	↑		↑								↑
Ratio of enabling service staff per 5,000 patients									↓
Average number of services provided in addition to medical care		↓			↓				↑
Financial Resources
Per capita total revenues			↑		↓		↓		↓			↑		↑
Proportion of total revenues that are from 330 grants		↓	↓										↓	↓
Contextual Characteristics
Ratio of PCP per 5,000 population in county	↑	↑	↑					↑		↑
Proportion below federal poverty guideline in county												↑
Proportion of minority in county													↓		↓

Notes:

denotes positive statistically significant association at p<0.05.

denotes negative statistically significant association at p<0.05.

Analyses were conducted using fractional outcome regression models using the logit distribution.

BMI, body mass index; CAD, coronary artery disease; IVD, ischemic vascular disease; HIV, human immunodeficiency virus; HbA1c, Hemoglobin A1c.

Discussion

Our findings show that rural HCs had significant differences in patient, organizational, and contextual characteristics compared with urban HCs. Rural HCs also had lower performance on preventive and care management measures but better performance on outcome measures than urban HCs, though these differences were small. We found that nearly all urban-rural clinical quality measure differences could be attributed to patient, organizational, and contextual differences, with varying characteristics as the explanatory factors for performance differences on specific clinical quality measures.

For example, the higher percentage of patients who preferred care provided in a language other than English was associated with better performance measures across the board, with the exception of routine depression screening. This finding may indicate the value of delivering linguistically and culturally competent care. Enabling services staff in HCs provide translation services that can improve care outcomes because patients may better understand provider instructions [33, 34].

Similarly, we found a positive association for six of the seven measures between higher percentage of patients with diabetes and better preventive performance [35]. HCs focusing on improving diabetes outcomes may target diabetes patients for additional opportunities for comprehensive preventive care services (i.e., earlier pneumonia vaccines, weight screening, diet counseling) and patients with diabetes may visit HCs more frequently and therefore have more opportunities to receive preventive care [34, 36, 37]. Similarly, the positive association of higher percentages of younger patients with better child and adult preventive performance measures (five of seven) is likely because HCs with younger patients focused on provision of such services to children and their parents [38, 39]. In some instances, such a positive relationship between more diabetes patients with lower rates of poorly controlled diabetes and better hypertension control may be because HCs with a higher concentration of these patients spent more intensive effort on improving these outcomes, or more diabetes patients sought care from these organizations if they offered diabetes specific services such as lifestyle or exercise classes [40, 41].

Among clinical quality measures, few control variables systematically explained urban/rural differences. Among outcome measures, the negative relationship of higher rates of racial/ethnic minority patients at the HC with poorer outcomes may have been because the racial/ethnic case mix captured social determinants of health that were not separately controlled for in our models. A number of studies have found similar results among the patient minority case mix and its effect on clinical performance, particularly on outcome measures [42, 43].

The only difference that remained significantly lower among rural HCs after adjusting for patient and HC characteristics was follow-up care among newly diagnosed HIV patients. This lower rate has been observed among low-income rural populations nationally and highlights a more pervasive challenge in rural areas [44, 45]. Other data indicate that these lower rates may be due to inadequately trained providers in rural areas to treat persons with HIV and distance or lack of readily available transportation to obtain services in rural areas [45, 46]. With the exception of the rural disparity in HIV performance measure, our data indicated that after controlling for patient and HC factors, there were no statistically significant differences in rural and urban HCs in their performance of clinical quality measures, including similar proportions that met or exceeded national HEDIS benchmarks. Both urban and rural HCs reported high achievement rates in meeting or exceeding national performance benchmarks in tobacco screening and cessation counseling, asthma treatment, lipid lowering therapy, poorly controlled diabetes, and hypertension control, and these performance rates, particularly outcome measures, are consistent with previous findings [42, 47, 48]. However, the performance achievement rates of both rural and urban HCs were low in other national benchmarks. Thus, patients of rural HCs may still experience disparities in quality of care [3, 42].

Our study had limitations including a single urban or rural designation for HC organizations even if some delivery sites may not be in rural areas. However, we used delivery site addresses to determine that 11% of sites among HCs that self-designated as rural may be urban and 13% of sites among HCs that self-designated as urban may be rural. This potential discrepancy is likely to be a consequence of variations in definitions of rural designation and the lack of UDS data on individual HC sites, which requires HCs to make an overall determination even if there is an urban and rural mix among the organization’s sites. Given that our assessment found that potential misclassification is fairly uniform (11 and 13%), the bias that results is likely to weaken the associations between our outcomes of interest and urban and rural status. Additionally, because UDS data lacks information on individual HC sites, there is a potential masking of differences at site-level or patient-level. Our study is cross-sectional in nature and causal relationships between our independent and dependent variables cannot be readily determined. In addition, we examined the missing rate for the patients seen for follow-up care within 90 days of initial HIV diagnosis by urban/rural status and found that there is a positive association between this measure and rurality. This might be because HCs in rural areas have a low prevalence of patients with HIV and that these numbers were too small or sensitive to report, leading to a potential overestimate of the variations between urban and rural HCs for this outcome. Furthermore, it is possible that performance among clinical measures are independently correlated and are overestimated. Our national benchmarks are based on performance measures for Medicaid managed care organizations, which are a subset of HC patients and may limit national generalizability. However, because the majority of HC patients are Medicaid beneficiaries, these national benchmarks are likely to be the most relevant. Despite this limitation, both rural and urban HCs performed well in several preventive and outcome measures. Future research should include several years of data to assess the role HC characteristics have in eliminating differences among urban and rural clinical performance over time.

Policy implications

The number of rural populations has decreased over time and their demographics have shifted [1]. Our findings highlighted comparable clinical performance between urban and rural HCs, even with the cited challenges of providing care in rural geographies. These findings stress the integral role of rural HCs in alleviating disparities in quality of care and the potential negative impact of any reductions in resources to these crucial safety net providers in rural areas. Urban/rural disparities in HIV screening and follow-up requires further attention by assessing availability of trained providers in rural areas to treat persons with HIV, identifying procedures that improve confidentiality, or providing community health education to better inform the resident and provider communities about HIV, its epidemiology, and its implication for care and treatment [45, 49]. Improving availability of providers trained in HIV care in rural areas can be achieved by federal policies that are being implemented to improve access to care in rural areas with programs leveraging HCs to diagnose, treat, prevent, and respond to HIV in communities with substantial HIV burden [50]. HCs have received funding to provide medical care services to patients living with HIV through the Ryan White HIV/AIDS Program and in 2019, HRSA provided more than $2 billion to increase access to care for people living with HIV, including in rural areas [51]. Other programs including loan repayment programs under the National Health Service Corps for providers working in shortage areas, state-based loan repayment programs, and the Teaching Health Center Graduate Medical Education Program, which allow HCs to operate medical residency training programs, help address emerging public health priorities [17].

Other research indicates that lower clinical performance of HCs is linked to geographic disparities that could be alleviated by increasing the availability of resources and technical assistance [52]. HRSA has supported HC infrastructure development and provided funding to bolster the ability of these organizations to improve quality of care for low-income and uninsured patients [16, 53]. The Federal Office of Rural Health Policy has implemented programs to address access to quality health care and health professional capacity impacting rural communities. In addition, HRSA support of Health Center Controlled Networks and Primary Care Associations also provide technical resources to improve quality of care in rural and urban HCs [52, 54].

Our findings provide support for the continuation of these programs and the identification and implementation of new programs that address performance gaps among rural HCs. Promoting quality of care among rural and urban HCs could be achieved by providing technical assistance to develop skills and resources to conduct quality improvement activities [55]. Ultimately, our findings suggest that many urban/rural disparities in quality of care are concentrated among non-HC providers and further research is need to identify reasons for such disparities.

STROBE statement—Checklist of items that should be included in reports of cross-sectional studies.

(DOC)

Click here for additional data file.

(XLSX)

Click here for additional data file.

Performance measures and full clinical quality performance measure definition.

(DOCX)

Click here for additional data file.

Unadjusted and adjusted predicted probabilities of health center quality indicators by urban and rural status.

(DOCX)

Click here for additional data file.

Regression models of prevention quality performance indicators.

(DOCX)

Click here for additional data file.

Regression models of care management quality indicators.

(DOCX)

Click here for additional data file.

Regression models of outcome quality indicators.

(DOCX)

Click here for additional data file.

Logistic regression models of health centers that met prevention quality performance indicator benchmarks.

(DOCX)

Click here for additional data file.

Logistic regression models of health centers that met care management quality performance indicator benchmarks.

(DOCX)

Click here for additional data file.

Logistic regression models of health centers that met outcome quality performance indicator benchmarks.

(DOCX)

Click here for additional data file.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

24 Jan 2020

PONE-D-19-27498

Assessing clinical quality performance and staffing capacity differences between urban and rural Health Resources and Services Administration-funded health centers in the United States: A cross sectional study

PLOS ONE

Dear Dr. Pourat,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

While both authors had positive comments about the topic and the appropriateness of the data used for the analysis, one reviewer had concerns about missing data that was present in the data file that was included with the submission. Please carefully review Reviewer #1's concerns about data management and respond. The same reviewer also had major concerns about the appropriateness of the statistical tests that were conducted, please review, re-do analyses as necessary and respond.

We would appreciate receiving your revised manuscript by Feb 24 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

PLOS ONE Academic Editor

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Summary: The authors first merged the data collected from 1,373 HRSA-funded Health Centers (HC) and county level data of the 2016 AHRF, and then proceeded to tabulate the various health quality indicators broken down by HC-reported urban/rural status. To discern the adjusted differences, outcome-specific regression models were built (adjusted for patient-level, organizational level, and county-level variables.) The authors found that among adjusted model, HIV linkage persisted to be lower in the rural HCs, while other urban/rural differences were explained away by the various independent variables, with proportions of non-English speakers, people living with diabetes, and people aged 17 or below, as well as PCP/population being the frequent candidates. The authors provided policy-level suggestions on such findings.

Major comments:

Overall, I really enjoyed reading the paper. The data choice was proper and the idea that urban/rural engulfs many different factors was well displayed. The policy-level recommendations were also moderately realistic that they could be referenced by interested policy stakeholders. However, I have some major concerns on result presentation and statistical analyses. I am listing them here for the authors’ reference and I hope addressing them may further strengthen their work.

The results in the supplementary tables should be in the text: First, nearly every statistically significant difference in Table 1 was verbatim mentioned in text, rendering Table 1 unnecessary; Second, Tables 2 and Table 3 are better combined so that the unadjusted and adjusted differences can be more readily compared and contrasted; Third, urban/rural differences in healthcare quality is well-reported but the factors contributing to that in a systematic analysis are not as well-known. If the authors can creatively visualize the significance levels and directions of the regression coefficients without resorting to showing large amount of numbers, then readers will be able to appreciate important interpretations like the one spanning from line 236 through 241.

The use of GLM deserves some clarification: “Generalized Linear Model” includes a lot of different statistical models, so please specify that in the Abstract. From the Methods (Line 189 through 196) it seems the authors set it as a logistic regression, if so, please just state that. Assuming the analysis was indeed logistics regression, why is it the best choice given the dependent variables were continuous percentages? Since there are also beta regression and probit regression, etc. which could be more suitable for these kinds of outcomes, it would be helpful if the authors could provide a method-related paper to justify the use of logistic regression in this fashion. Along the same line, if dichotomization was performed, then please provide the scheme.

How did the model address the difference sizes of HC? The overall analysis seems to assume equal weight for every HC, are they comparable in size (e.g. in terms of patients served)? If not, should the summary statistics and regression models be weighted?

Concerns on data management: The attached Excel data set shows that HIV linkage was missing in almost half of the HCs. Given HIV linkage was the only characteristic still found to be different in urban/rural settings, the prevalence of missing outcomes as well as some speculation are merited. In addition, some of the percentage data were shown in percentage, and yet some were shown in fraction (e.g. percent of patient with HIV.) This may explain why the regression coefficients for those covariates in fraction were much larger. I would suggest a round of audit to verify the data, software syntax, and output results.

Minor comments/suggestions:

[Abstract] The starting sentence created a false impression that HCs are exclusive to rural area, creating some confusion later. Please revise.

[Line 115 through 117] The study design, data, and analysis do not support this objective. First, it is cross-sectional so “contribute in reducing disparities” could be over-reaching; second, the analysis adjusted for many causal downstream variables of urban/rural, while I would agree that the work unpacked what urban/rural entails, the fact that the urban/rural became largely statistically non-significant is not indicative of reducing disparity, but perhaps mediation adjustment.

[Line 137 through 139] Are there only 15 in HEDIS? If there were more than 15, how did the authors decide on the final list?

[Line 194] The brand name should be written as Stata.

[Line 202] Add SD to the acronym list.

[Table 1 and others, including Excel file] Please check the label “Percent of patients of patients 65 years and older.”

[Table 3] HEDIS benchmarks should be accompanied by “percent of HCs that exceed the benchmark” rather than using only the sample mean to determine adherence.

Reviewer #2: This article uses the Uniform Data Set to compare HRSA funded clinics operating in urban and rural areas using a number of standard quality metrics. The authors find that after controlling for confounders, there are no statistically significant differences between urban and rural clinics for most outcome measures. The one exception is that Rural clinics "had lower rates of linking patients newly diagnosed with HIV to care."

Overall, this is a nice article that makes a useful contribution to the literature. As the authors point out, there are relatively few comparisons of this sort, which is surprising given the well know differences in urban-rural health outcomes. Despite this, I think the paper could benefit from a bit more reflection on both the premise of the study and the implications of these findings. Most studies documenting urban-rural differences in health status point to social determinants as an explanation. To the degree that there is a focus on health care, the usual emphasis is on the availability of care, not quality. Did the authors have reason to believe there might be differences in the quality (as opposed to the quantity) of care available to people living in urban and rural areas? This is an underlying assumption of the analysis that the authors do not set up well.

Second, I think the quality of care available in HRSA funded clinics is important, but given the relatively small role that such clinics play in the overall health system, would it be reasonable to suggest that differences in quality would be sufficient to explain urban-rural differences in the first place? Or are the authors focused more narrowly on urban-rural differences in health care for populations who are likely to seek care in HRSA clinics and other safety-net organizations?

Third, what should policy makers do with this information? If there is a concern about quality differences among HRSA clinics in urban and rural areas, the authors have offered comforting evidence -- but the urban rural health differences remain. So if it is not a quality difference, what's driving the problem?

Finally, I think the authors should say a bit more about their HIV finding. There was an article published in the NYT recently arguing that HIV is increasing in rural areas, but that these parts of the country are not ready for it. The findings in this paper are one small part of that, but the findings are certainly consistent with the concerns expressed in the article. I think the authors should put the HIV finding into the larger context of rural public health and health care capacity to address a growing HIV problem in these communities.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

4 Mar 2020

Reviewer Comment

Reviewer #1:

Summary: The authors first merged the data collected from 1,373 HRSA-funded Health Centers (HC) and county level data of the 2016 AHRF, and then proceeded to tabulate the various health quality indicators broken down by HC-reported urban/rural status. To discern the adjusted differences, outcome-specific regression models were built (adjusted for patient-level, organizational level, and county-level variables.) The authors found that among adjusted model, HIV linkage persisted to be lower in the rural HCs, while other urban/rural differences were explained away by the various independent variables, with proportions of non-English speakers, people living with diabetes, and people aged 17 or below, as well as PCP/population being the frequent candidates. The authors provided policy-level suggestions on such findings.

Major comments:

Overall, I really enjoyed reading the paper. The data choice was proper and the idea that urban/rural engulfs many different factors was well displayed. The policy-level recommendations were also moderately realistic that they could be referenced by interested policy stakeholders. However, I have some major concerns on result presentation and statistical analyses. I am listing them here for the authors’ reference and I hope addressing them may further strengthen their work.

Reviewer Comment

The results in the supplementary tables should be in the text: First, nearly every statistically significant difference in Table 1 was verbatim mentioned in text, rendering Table 1 unnecessary; Second, Tables 2 and Table 3 are better combined so that the unadjusted and adjusted differences can be more readily compared and contrasted; Third, urban/rural differences in healthcare quality is well-reported but the factors contributing to that in a systematic analysis are not as well-known. If the authors can creatively visualize the significance levels and directions of the regression coefficients without resorting to showing large amount of numbers, then readers will be able to appreciate important interpretations like the one spanning from line 236 through 241.

Response to Reviewers

We have edited description of Table 1 in the Results to reduce length and only emphasize important findings.

We have combined Tables 2 and 3 and reformatted to visualize the significant findings as recommended.

We did not include the coefficients from the supplemental tables in the text. Instead we reported the overall predicted probabilities that show the degree to which a given outcome differs between urban and rural HCs. We then supplemented that discussion with what indicators in models contributed to the outcomes. We believe that it was a more efficient way of describing the results of the extensive models and less confusing to the readers. We added a table that highlighted the significant coefficients for each quality indicator and whether they were positive or negative in response to this comment.

Location in Manuscript

Results, Page 12-13, line 254-263

Revised Table 2 (previously Tables 2 and 3)

Table 3 (new)

Reviewer Comment

The use of GLM deserves some clarification: “Generalized Linear Model” includes a lot of different statistical models, so please specify that in the Abstract. From the Methods (Line 189 through 196) it seems the authors set it as a logistic regression, if so, please just state that. Assuming the analysis was indeed logistics regression, why is it the best choice given the dependent variables were continuous percentages? Since there are also beta regression and probit regression, etc. which could be more suitable for these kinds of outcomes, it would be helpful if the authors could provide a method-related paper to justify the use of logistic regression in this fashion. Along the same line, if dichotomization was performed, then please provide the scheme.

Response to Reviewers

The underlying data of the dependent variables are presented as a proportion. For ease of interpretation, the dependent variables are transformed to proportions ranging between zero and one. We used a generalized linear model, not a logistic regression, and have made edits to the Methods to make this clear. We have revised the Methods to make clear the dependent variables are proportions and added detail to the Statistical Analysis section on how we performed this transformation.

Location in Manuscript

Abstract, Page 4, line 69

Methods, Dependent variables, page 9, line 167

Methods, Statistical analysis, Page 11, line 231

Reviewer Comment

How did the model address the difference sizes of HC? The overall analysis seems to assume equal weight for every HC, are they comparable in size (e.g. in terms of patients served)? If not, should the summary statistics and regression models be weighted?

Response to Reviewers

Health centers are not comparable in size and as a result, we have controlled for the average number of sites and patients seen during the year in each model. We have also controlled for a number of patient characteristics. The full regression models with control variables are presented in the Supplemental Appendices. We are using the entire population of health centers in 2017 (N=1,373), not a sample.

Location in Manuscript

S2-S4 Tables

Reviewer Comment

Concerns on data management: The attached Excel data set shows that HIV linkage was missing in almost half of the HCs. Given HIV linkage was the only characteristic still found to be different in urban/rural settings, the prevalence of missing outcomes as well as some speculation are merited. In addition, some of the percentage data were shown in percentage, and yet some were shown in fraction (e.g. percent of patient with HIV.) This may explain why the regression coefficients for those covariates in fraction were much larger. I would suggest a round of audit to verify the data, software syntax, and output results.

Response to Reviewers

All clinical quality measure data are displayed as proportions, including HIV linkage to care. We have assessed the missing HIV linkage to care outcomes and found there was a relationship with urban/rural settings. We have added a sentence in the Limitations to address this. We further explain that missingness among rural health centers may be due to low prevalence of HIV and could be too sensitive to report and suggest a conservative interpretation.

We have also confirmed in the software syntax the percentage data displayed in the outcome variables were rescaled as a fraction.

Location in Manuscript

Discussion, page 31, lines 429-435

Methods, Page 9-10, lines 166-167

Reviewer Comment

Minor comments/suggestions:

[Abstract] The starting sentence created a false impression that HCs are exclusive to rural area, creating some confusion later. Please revise.

Response to Reviewers

We have revised this sentence to indicate that health centers are located nationally.

Location in Manuscript

Abstract, Page 4, line 59-62

Reviewer Comment

[Line 115 through 117] The study design, data, and analysis do not support this objective. First, it is cross-sectional so “contribute in reducing disparities” could be over-reaching; second, the analysis adjusted for many causal downstream variables of urban/rural, while I would agree that the work unpacked what urban/rural entails, the fact that the urban/rural became largely statistically non-significant is not indicative of reducing disparity, but perhaps mediation adjustment.

Response to Reviewers

We have revised this sentence to be more conservative and consistent with our study design and analysis.

Location in Manuscript

Introduction, Page 6, line 95-100

Reviewer Comment

[Line 137 through 139] Are there only 15 in HEDIS? If there were more than 15, how did the authors decide on the final list?

Response to Reviewers

There are a total of 15 clinical quality measures that health centers are required to report and we decided to use all 15 measures. These 15 measures have benchmarks in HEDIS. We have revised for clarity.

Location in Manuscript

Methods, Page 8, line 158-163

Reviewer Comment

[Line 194] The brand name should be written as Stata.

Response to Reviewers

Revised as suggested.

Location in Manuscript

Methods, Page 12, line 240

Reviewer Comment

[Line 202] Add SD to the acronym list.

Response to Reviewers

We have added to the acronym list as suggested.

Location in Manuscript

Title Page, Page 3, line 57

Reviewer Comment

[Table 1 and others, including Excel file] Please check the label “Percent of patients of patients 65 years and older.”

Response to Reviewers

Revised for correctness.

Location in Manuscript

Table 1 and S2 Table

Reviewer Comment

[Table 3] HEDIS benchmarks should be accompanied by “percent of HCs that exceed the benchmark” rather than using only the sample mean to determine adherence.

Response to Reviewers

We have revised Table 2 to include the proportion of health centers that achieved HEDIS benchmarks.

Location in Manuscript

Revised Table 2 (previously Table 2 and 3)

Methods, Statistical Analysis

Reviewer Comment

Reviewer #2: This article uses the Uniform Data Set to compare HRSA funded clinics operating in urban and rural areas using a number of standard quality metrics. The authors find that after controlling for confounders, there are no statistically significant differences between urban and rural clinics for most outcome measures. The one exception is that Rural clinics "had lower rates of linking patients newly diagnosed with HIV to care."

Overall, this is a nice article that makes a useful contribution to the literature. As the authors point out, there are relatively few comparisons of this sort, which is surprising given the well know differences in urban-rural health outcomes. Despite this, I think the paper could benefit from a bit more reflection on both the premise of the study and the implications of these findings. Most studies documenting urban-rural differences in health status point to social determinants as an explanation. To the degree that there is a focus on health care, the usual emphasis is on the availability of care, not quality. Did the authors have reason to believe there might be differences in the quality (as opposed to the quantity) of care available to people living in urban and rural areas? This is an underlying assumption of the analysis that the authors do not set up well.

Response to Reviewers

There is literature that shows differences in quality of care outcomes in the general urban/rural population, as well as other providers. Due to this evidence, we believe in examining urban/rural differences in health center settings in particular. We have added additional sentences in the Introduction in order to address differences in quality

Location in Manuscript

Introduction, Page 6, line 106-108

Reviewer Comment

Second, I think the quality of care available in HRSA funded clinics is important, but given the relatively small role that such clinics play in the overall health system, would it be reasonable to suggest that differences in quality would be sufficient to explain urban-rural differences in the first place? Or are the authors focused more narrowly on urban-rural differences in health care for populations who are likely to seek care in HRSA clinics and other safety-net organizations?

Response to Reviewers

HCs are the cornerstone of the safety net system and the only providers that provide care to low-income and uninsured patients regardless of income or any other factor. Other providers including public systems organized by counties have various restrictions. In addition, HCs are often the only comprehensive provider of care in rural areas. Therefore, the role of HCs in providing access to high quality care in general and in rural areas in particular is very important.

We are focused on urban/rural differences among populations that are served by HRSA-funded health centers, but our results have implications for quality of care in urban and rural areas more broadly, particularly when HCs are the primary or only providers of care

Location in Manuscript

Introduction, Page 7, line 123-125

Reviewer Comment

Third, what should policy makers do with this information? If there is a concern about quality differences among HRSA clinics in urban and rural areas, the authors have offered comforting evidence -- but the urban rural health differences remain. So if it is not a quality difference, what's driving the problem?

Response to Reviewers

We cannot explain what explain urban/rural differences in quality from our data. However, we have shown that a subset of rural populations have quality of care on par with urban areas. Therefore, policy efforts should be focused on other providers who are likely to be the source of urban/rural disparities. We have revised the implications to address this point.

Location in Manuscript

Discussion, Page 34, line 479-481

Reviewer Comment

Finally, I think the authors should say a bit more about their HIV finding. There was an article published in the NYT recently arguing that HIV is increasing in rural areas, but that these parts of the country are not ready for it. The findings in this paper are one small part of that, but the findings are certainly consistent with the concerns expressed in the article. I think the authors should put the HIV finding into the larger context of rural public health and health care capacity to address a growing HIV problem in these communities.

Response to Reviewers

We have added additional information to support how health centers have provided additional funding support to address HIV.

Location in Manuscript

Discussion, Page 33, line 458-460

Submitted filename: Response to Reviewers-final.docx

Click here for additional data file.

31 Jul 2020

PONE-D-19-27498R1

Assessing clinical quality performance and staffing capacity differences between urban and rural Health Resources and Services Administration-funded health centers in the United States: A cross sectional study

PLOS ONE

Dear Dr. Pourat,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

*****************************

Editor's Comments:

Given that the concerns of Reviewer 1 remain significant, and Reviewer 2 was unable to provide a second review, we have worked with the Editorial office to find another reviewer with a strong quantitative background. In your revision, please be sure to address the areas of overlap of both reviewers (1 & 3), as they share similar concerns about the statistical methodology.

Thank you.

*****************************

Please submit your revised manuscript by Sep 14 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Major comments: Thank you for addressing most of my comments concerning the presentation of the results. Tables 2 & 3 appear informative and I hope readers will find the results more readily digestible.

There are a few remaining issues that the authors did briefly responded to and I would like to raise them again:

It appears that the GLM analysis was based on binomial family and logit link, which makes this a logistic regression. After reading the authors’ response, my curiosity remains unsated and I wish the authors will address this explicitly: why was logistic regression used on a continuous variable that is bound between 0 and 1? It would be great if the author can cite a technical document that this is indeed a correct approach. The fractional dependent variables (like pct_diabete) look very much like a bell curve and I couldn’t wrap my head around using binomial. If we tried to use the Stata command “logit” or “logistic” to repeat this analysis we may also find that Stata would in fact display a warning and refuse to proceed. Please also consider if fracreg (Fractional response regression) function in Stata would be a more appropriate candidate.

With the Excel data, I was able to check some statistics in Table 1. It seems the data remained unweighted. Given the mean (SD) patient of 19,792 (23,663), there is a huge variability among the sizes. If a HC serving 1,000 patients has and indicator at 0%, and a HC serving 20,000 patients has it at 50%, would the mean be closer to 25.0% or 47.6%? It’d be great if the authors can justify why they favored the unweighted approach.

Reviewer #3: The analyses in this manuscript appear to be done well. I had some suggestions below to strengthen the descriptions and, potentially, improve the results. I think the biggest problem is the manuscript is so dense. Table 2 and the results section have quite a bit of information. I have a couple suggestions for this in my comments which may or may not help.

1. (line 172) This statement is a little confusing since an indicator variable is 0 or 1 and not a proportion. I'm guessing created indicator variables for each of the 15 DVs and then created a proportion from those? That would fit with your methods.

2. (lines 205-206) I think this choice of method is good. I suggest including a methodological citation for the method, probably McCullagh and Nelder's book will be fine. I think the main question is what your residuals look like and whether there is potential for systematic over- or under-fitting. My guess is that your covariates have eliminated this, but it would be good for you to check this. It may be that adding random effects by state to account for any state-to-state differences could be helpful. Another thought were spatial random effects, but my guess is that the urban/rural variable will be good enough for that.

3. (line 210) How many or what percentage of HCs were dropped due to missing data? Rules of thumb vary on when complete case analyses are still valid. My suggestion is if > 5% are dropped, then start exploring the missing data. If > 15% are missing, then you'll probably need to do something about it, e.g., multiple imputation.

4. (line 211) Sorry, I'm not a Stata user. Does the margins command produce confidence intervals or prediction intervals?

5. Did you perform any variable selection or were all the IVs included in all analyses?

6. (Table 2) This table is quite dense and I wonder if you thought about moving some of these results to a figure, especially the predicted probabilities. It might make for a nice visual and could make table 2 more readable. I also wondered if there was any way to abbreviate the measures in the first column.

7. (Table S2) I am not sure I understand the coefficients in this table. Are they the raw coefficient values from the binomial models? If so, they should be exponentiated to become odds ratios and reported as such.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

29 Sep 2020

Reviewer #1

Reviewer Comment

Major comments: Thank you for addressing most of my comments concerning the presentation of the results. Tables 2 & 3 appear informative and I hope readers will find the results more readily digestible.

There are a few remaining issues that the authors did briefly responded to and I would like to raise them again:

It appears that the GLM analysis was based on binomial family and logit link, which makes this a logistic regression. After reading the authors’ response, my curiosity remains unsated and I wish the authors will address this explicitly: why was logistic regression used on a continuous variable that is bound between 0 and 1? It would be great if the author can cite a technical document that this is indeed a correct approach. The fractional dependent variables (like pct_diabete) look very much like a bell curve and I couldn’t wrap my head around using binomial. If we tried to use the Stata command “logit” or “logistic” to repeat this analysis we may also find that Stata would in fact display a warning and refuse to proceed. Please also consider if fracreg (Fractional response regression) function in Stata would be a more appropriate candidate.

Response to Reviewers

We used GLM regressions for the analyses because the dependent variables are a proportion bound between 0 and 1 and therefore had both a lower and upper bound ceiling. The following citation supports our rationale for using the GLM analysis (Baum CF. Stata Tip 63: Modeling Proportions. The Stata Journal. 2008;8(2):299-303. doi:10.1177/1536867X0800800212). In response to the reviewer’s comment we changed the regressions to fracreg even though our tests indicated the two models yielded the exact same results. We included a technical citation to support use of fracreg in the manuscript.

Location in Manuscript

Methods, Page 11, line 209-210

Reviewer Comment

With the Excel data, I was able to check some statistics in Table 1. It seems the data remained unweighted. Given the mean (SD) patient of 19,792 (23,663), there is a huge variability among the sizes. If a HC serving 1,000 patients has and indicator at 0%, and a HC serving 20,000 patients has it at 50%, would the mean be closer to 25.0% or 47.6%? It’d be great if the authors can justify why they favored the unweighted approach.

Response to Reviewers

We did not weight the data because the goal of this manuscript was to compare the performance of HC in urban and rural locations. For this goal, every HC would have to exert the same amount of influence on the results. As the reviewer points out, the weighted data would allow larger HCs to exert more of an influence on the results. That analyses would be more appropriate if our aim was to compare the population-level impact.

To illustrate the difference between weighted and unweighted data, we conducted weighted regressions to assess how the results change. We found the majority of coefficients had similar effects in both types of analyses, with the exception of HIV linkage to care and low birthweight. In weighted analyses, HIV linkage to care was no longer significant, but the direction of association did not change. In the weighted model, low birthweight became significant but the direction of the associations did not change.

Location in Manuscript

No change.

Reviewer #3

Reviewer Comment

The analyses in this manuscript appear to be done well. I had some suggestions below to strengthen the descriptions and, potentially, improve the results. I think the biggest problem is the manuscript is so dense. Table 2 and the results section have quite a bit of information. I have a couple suggestions for this in my comments which may or may not help.

Response to Reviewers

We thank the reviewer for their comments. We have added Figure 1 to visually display the significant differences in performance measures between urban and rural HCs. We did not include results that were not significant to reduce the size of this figure. We moved the original Table 2 into the appendix to all performance measures .

Location in Manuscript

Figure 1 (new)

Table 2

Reviewer Comment

1. (line 172) This statement is a little confusing since an indicator variable is 0 or 1 and not a proportion. I'm guessing created indicator variables for each of the 15 DVs and then created a proportion from those? That would fit with your methods.

Response to Reviewers

We have removed this sentence to reduce confusion. The dependent variables are indeed proportions bound between 0 and 1 and therefore had both a lower and upper bound ceiling. They indicate the proportion of patients that had received specific services or had a specific outcome. The methods sections are revised in response to both reviewers’ comments.

Location in Manuscript

Methods, Page 8, line 158

Reviewer Comment

2. (lines 205-206) I think this choice of method is good. I suggest including a methodological citation for the method, probably McCullagh and Nelder's book will be fine. I think the main question is what your residuals look like and whether there is potential for systematic over- or under-fitting. My guess is that your covariates have eliminated this, but it would be good for you to check this. It may be that adding random effects by state to account for any state-to-state differences could be helpful. Another thought were spatial random effects, but my guess is that the urban/rural variable will be good enough for that.

Response to Reviewers

As indicated in the previous responses to similar comments above, we are now using fracreg or fractional outcome regression, which produced the same results as GLM and have added a citation to support the use of this regression model.

We have also checked the residual plots for all our models and found that there was no evidence of model under- or overfitting.

We have controlled for county-level variations using county-level characteristics, as well as the urban/rural variable as the reviewer mentioned. We have not included random effects by state to avoid any potential overfitting/overcontrolling.

Location in Manuscript

Methods, Page 11, lines 209-210

Reviewer Comment

3. (line 210) How many or what percentage of HCs were dropped due to missing data? Rules of thumb vary on when complete case analyses are still valid. My suggestion is if > 5% are dropped, then start exploring the missing data. If > 15% are missing, then you'll probably need to do something about it, e.g., multiple imputation.

Response to Reviewers

We examined the proportion of missing for each dependent variable and found that no variables were missing by greater than 5%. We did not impute the dependent variables with missing values following recommendation from Hippel (Von Hippel, P.T. (2007), REGRESSION WITH MISSING YS: AN IMPROVED STRATEGY FOR ANALYZING MULTIPLY IMPUTED DATA. Sociological Methodology, 37: 83-117. doi:10.1111/j.1467-9531.2007.00180.x).

Location in Manuscript

No change

Reviewer Comment

4. (line 211) Sorry, I'm not a Stata user. Does the margins command produce confidence intervals or prediction intervals?

Response to Reviewers

The margins command produces predicted probabilities with confidence intervals.

Location in Manuscript

S2 Table

Reviewer Comment

5. Did you perform any variable selection or were all the IVs included in all analyses?

Response to Reviewers

We had selected independent variables that could influence HC performance conceptually and included all of them in the models to avoid omitted variable bias.

Location in Manuscript

No change

Reviewer Comment

6. (Table 2) This table is quite dense and I wonder if you thought about moving some of these results to a figure, especially the predicted probabilities. It might make for a nice visual and could make table 2 more readable. I also wondered if there was any way to abbreviate the measures in the first column.

Response to Reviewers

We agree that the table was dense. This was the result of requests from previous reviewers to present the data in that way. We have now presented some of the data in the original Table 2 into Figure 1 as suggested and abbreviated the name of the measures for better readability. We then adjusted the other tables including S2 Table and Table 2 still reports probabilities of met or exceeding benchmarks.

Location in Manuscript

Figure 1 (new)

S2 Table

Table 2

Reviewer Comment

7. (Table S2) I am not sure I understand the coefficients in this table. Are they the raw coefficient values from the binomial models? If so, they should be exponentiated to become odds ratios and reported as such.

Response to Reviewers

Yes, the supplemental data displayed beta coefficients. We have now revised them to report odds ratios as suggested.

Location in Manuscript

S3 Table- S5 Table

Submitted filename: Urban Rural CQM Response to Reviewers RR2-final.docx

Click here for additional data file.

2 Nov 2020

PONE-D-19-27498R2

Assessing clinical quality performance and staffing capacity differences between urban and rural Health Resources and Services Administration-funded health centers in the United States: A cross sectional study

PLOS ONE

Dear Dr. Pourat,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

Thank you for addressing the comments of our reviewers. While both reviewers indicated that they are ready to accept the manuscript, Reviewer One requested that an additional statement be added to the Methods section indicating that all HC's were treated with equal analytical weight. Please add this statement where appropriate and resubmit the manuscript with this revision.

==============================

Please submit your revised manuscript by Dec 17 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Candace C. Nelson

Academic Editor

PLOS ONE

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for addressing the previous comments by adding the technical reference and explaining the paradigm between weighted and unweighted analysis. I just have one minor suggested change: in the Methods section please indicate that all HCs were treated with equal analytical weight.

Reviewer #3: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

9 Nov 2020

Reviewer #1

Reviewer Comment

Thank you for addressing the previous comments by adding the technical reference and explaining the paradigm between weighted and unweighted analysis. I just have one minor suggested change: in the Methods section please indicate that all HCs were treated with equal analytical weight.

Response to Reviewers

The authors have included a sentence in the Methods as suggested.

Location in Manuscript

Methods, Line 211, page 11

11 Nov 2020

Assessing clinical quality performance and staffing capacity differences between urban and rural Health Resources and Services Administration-funded health centers in the United States: A cross sectional study

PONE-D-19-27498R3

Dear Dr. Pourat,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Candace C. Nelson, ScD

Academic Editor

PLOS ONE

24 Nov 2020

PONE-D-19-27498R3

Assessing clinical quality performance and staffing capacity differences between urban and rural Health Resources and Services Administration-funded health centers in the United States: A cross sectional study

Dear Dr. Pourat:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Candace C. Nelson

Academic Editor

PLOS ONE