Racial and Ethnic Minority Concentration in Veterans Affairs Facilities and Delivery of Patient-Centered Primary Care

Abstract

Despite patient-centered medical home implementation by the Veterans Health Administration (VHA), delivery of patient-centered care varies across VHA facilities. Facility characteristics underlying this variation are not fully understood. This study used administrative data from 908 VHA outpatient facilities to examine the association of racial and ethnic minority patient concentration and other facility characteristics with facility ratings of patient-centered care. The primary finding was that patient-centered ratings were lower for facilities with medium or high concentrations of Hispanic patients (medium: estimate [Est] = −0.40, standard error [SE] = 0.20, P = 0.046; high: Est = −0.99, SE = 0.23, P < 0.001). In addition, patient-centered ratings decreased as patient panel sizes increased, especially among facilities with higher concentrations of black patients. This study indicates that efforts to improve patient-centered care may be needed at VHA facilities that serve Hispanic and VHA facilities with large panel sizes of high concentrations of black patients.

Introduction

There has been a nationwide push to increase delivery of patient-centered primary care by widespread adoption of the Patient-Centered Medical Home (PCMH) model.^1

–4 The PCMH model includes structural, organizational, and reimbursement changes that are designed to personalize and prioritize care to improve the health of whole populations.⁵ The Veterans Health Administration (VHA) began implementing a PCMH model in 2010 by organizing primary care-based providers and staff into Patient Aligned Care Teams (PACTs).^6,7 Progress toward PACT implementation has been monitored using the PACT Implementation Progress Index (Pi²), a measure developed to assess achievement on 8 core PCMH objectives: access, continuity, care coordination, team-based care, comprehensiveness, self-management support, patient-centered communication, and shared decision making. Facilities that score higher on this measure also have higher patient satisfaction, better clinical quality, and less staff burnout.⁶ However, VHA facilities vary widely in the extent to which they have achieved PCMH goals.⁶ Given the potential benefits, understanding why certain VHA facilities lag behind others in achieving PCMH goals is essential.

Of particular interest is whether facilities serving high concentrations of racial and ethnic minorities are achieving PCMH goals. Within and outside of VHA, health care facilities with higher concentrations of racial and ethnic minority patients tend to deliver lower quality care on multiple measures.^8

–15 Pertaining to the PCMH goal of delivering patient-centered care, studies conducted before VHA PACT implementation reported significant racial and ethnic differences in patient-reported experiences with VHA care that were driven largely by interfacility variation.^16,17 Patterns of racial/ethnic differences in patient experiences were unique for black or African American (subsequently referred to as black), Hispanic, and other racial/ethnic minorities, underscoring the need to focus on the patient-reported components of Pi² and to disaggregate racial/ethnic groups when possible.

Focusing specifically on the patient-reported components of Pi², the primary objective of this study was to assess the unique associations of VHA facilities’ concentrations of black, Hispanic, and other racial and ethnic minority patients with facility ratings of patient-reported patient-centered care. In an effort to identify potential ways to ensure equitable delivery of patient-centered care across facilities, this study also examined the extent to which a broad range of facility characteristics explain differences in patient-centered care between facilities with high and low concentrations of black, Hispanic, and other racial/ethnic minority patients. This study represents a critical step in this area.

Methods

A cross-sectional analysis was conducted using facility-level administrative data on VHA outpatient facilities from fiscal year 2013 (FY13: October 1, 2012–September 30, 2013). All evaluation activities were reviewed and approved as quality improvement by the Corporal Michael J. Crescenz Veterans Affairs Medical Center Institutional Review Board.

Sample

The sample included VHA outpatient facilities that (1) were located in the 50 states and Washington, DC, (2) offered primary care services during FY13, (3) had data available on the study outcome of interest, and (4) remained open through FY13.

Study measures and data sources

Study outcome

The study outcome was facility-level delivery of patient-centered care measured using the patient-centered care domains of the Pi². The full Pi² was developed by the National Evaluation Team at the VHA PACT Demonstration Laboratory Initiative Coordinating Center to assess the VHA transition to a PCMH model of primary care.⁶ Pi² is calculated from 53 components from patient and provider surveys and administrative data sources that pertain to key PACT initiative goals, including delivery of care that is accessible, continuous, coordinated, team based, and patient centered. The patient-centered component includes care comprehensiveness (ie, attention to mental health issues), self-management support, patient-centered communication, and shared decision making.

As described elsewhere,⁶ Pi² scores are constructed by first calculating a facility score for each of the 53 components. Component scores are standardized across facilities and averaged within 8 domains of care. Facilities are rank ordered for each domain, then assigned a 1 or −1 for every domain for which they are in the top or bottom quartile, respectively, and assigned 0 otherwise. A facility's Pi² value is the sum of its ranked scores across all domains.

For this analysis, a patient-centered Pi² score was constructed based on the 4 patient-centered Pi² domains (Table 1). Data for these domains are from the FY13 PCMH Survey of Healthcare Experiences of Patients, an ongoing survey conducted by the VHA Office of Analytics and Business Intelligence to assess veteran experiences with the VA Healthcare System using the Consumer Assessment of Healthcare Providers and Systems (CAHPS) PCMH survey.^18,19 Previous work has demonstrated adequate internal reliability for CAHPS multi-item scales assessing comprehensiveness (3 items, α = 0.80), communication with providers (6 items, α = 0.93), self-management support (2 items, α = 0.61), and shared decision making about medications (3 items, α = 0.67).²⁰ The mail-based survey was administered using CAHPS guidelines to a random sample of veterans from all VHA outpatient facilities who received outpatient care from an assigned PACT provider from October 1, 2012, through September 30, 2013. Patient-centered Pi² was calculated by applying specifications for calculating the full Pi² to the 4 patient-centered domains.⁶ The resulting 9-point scale ranged from −4 to 4, with higher scores reflecting more patient-centered care based on patient surveys at that facility.

Table 1.

Patient-Centered Domains and Relevant Survey Items Used to Calculate Facility-Level Patient-Centered Pi ² Scores

Patient-centered domains	Survey items	Response options
Comprehensiveness	In the last 12 months
	…did anyone in this provider's office ask you if there was a period of time when you felt sad, empty, or depressed?	Yes, No
	…did you and anyone in this provider's office talk about things in your life that worry you or cause you stress?	Yes, No
	…did you and anyone in this provider's office talk about a personal problem, alcohol use, drug use, or a mental or emotional illness?	Yes, No
Self-management support	In the last 12 months
	…did anyone in this provider's office talk with you about specific goals for your health?	Yes, No
	…did anyone in this provider's office ask you if there are things that make it hard for you to take care of your health?	Yes, No
Communication	In the last 12 months
	…how often did this provider explain things in a way that was easy to understand?	Never, sometimes, usually, always
	…how often did this provider listen carefully to you?	Never, sometimes, usually, always
	…how often did this provider give you easy to understand information about these health questions or concerns?	Never, sometimes, usually, always
	…how often did this provider seem to know the important information about your medical history?	Never, sometimes, usually, always
	…how often did this provider show respect for what you had to say?	Never, sometimes, usually, always
	…how often did this provider spend enough time with you?	Never, sometimes, usually, always
Shared decision making	In the last 12 months when you talked about starting or stopping a prescription medicine
	…how much did this provider talk about the reasons you might want to take a medicine?	Never, sometimes, usually, always
	…how much did this provider talk about the reasons you might not want to take a medicine?	Never, sometimes, usually, always
	…did this provider ask you what you thought was best for you?	Yes, No

Pi², Patient Aligned Care Team implementation progress index.

Primary predictors

The predictor variables were the concentration of outpatients from non-Hispanic black, Hispanic, and non-Hispanic other racial/ethnic minority groups at each facility. The VHA MedSAS outpatient file from the National Patient Care Database was used to calculate the percent of patients in each group at each facility with at least 1 outpatient visit during FY13 and race/ethnicity data on file. Because of the positively skewed distribution of each minority group, facilities were categorized into tertiles based on their percent of black, Hispanic, and other racial/ethnic patients. The lowest tertile was used as the reference group for each minority group in analyses, comparing facilities with relatively medium or high percentages of each minority group with those with a relatively low percentage.

Additional facility characteristics

This study examined several facility characteristics that could be associated with patient-centered care ratings and/or explain the association between patient racial/ethnic concentration and delivery of patient-centered care. Following Glasgow et al's analytic framework of organizational factors,²¹ indicators of 2 broad dimensions of organizational context were selected: facility structure and staffing/culture (Table 2). Facility structure variables included basic, unmodifiable structural characteristics, whereas staffing/culture variables included more modifiable characteristics that may affect the success of a facility's quality improvement efforts. All facility characteristics were drawn from VHA administrative data sources (Table 2).

Table 2.

Definitions of Facility Characteristics, Variable Type, Level of Disaggregation, and Classification

Facility characteristic ^a	Definition (data source)	Variable type (level of disaggregation ^b )
Facility structure
% Female	Percentage of patients receiving outpatient care who were female (FY13 Medical SAS Outpatient file)	Continuous (substation)
Number of annual primary care visits	Total number of outpatient visits in primary care clinics (FY13 Medical SAS Outpatient file)	Quartiles (substation)
Patient severity (DCG)	Average DCG across patients, with DCG representing the cost of care for a patient, relative to the average VHA patient, based on his or her age, sex, and recorded diagnoses from VHA inpatient, VHA outpatient, and non-VHA fee-based care records over the prior 12-month period. Scores greater than 0.5 indicate greater than average cost and clinical complexity. (VHA PACT Compass^c)	Continuous, scaled by 0.10 (substation)
Facility type	Type of outpatient facility (VHA Site Tracking System)	4 categories collapsed into 2: VAMC/IOC or CBOC/MOC (substation)
Staffing/culture
Patient panel size	Average panel size of primary care providers (VHA PACT Compass^c)	Continuous, scaled by 100 (substation)
% Non-Hispanic white MD	Percentage of medical doctors employed at a facility who are non-Hispanic white (VSSC Human Resources Turnover Cube)	Continuous (parent station)
% Non-Hispanic white nurse	Percentage of nurses employed at a facility who are non-Hispanic white (VSSC Human Resources Turnover Cube)	Continuous (parent station)
Employee satisfaction	Aggregate responses to the following question: Considering everything, how satisfied are you with your job? (All Employee Survey, a survey of job satisfaction administered annually to all VHA employees)	5-point Likert scale treated as continuous, with possible values ranging from 1 (not at all satisfied) to 5 (very satisfied). (parent station)
Primary care staff ratio	Number of primary care support staff divided by the number of direct care primary care providers (MDs or other direct care providers such as nurse practitioners). Higher staff ratios indicate more support staff per direct care provider, with 3 being the recommended minimum ratio for successful PACT implementation. (VHA PACT Compass^c)	Continuous (substation)
MD turnover	Number of MDs lost (related to retirement, death, termination, or voluntary separation) divided by the number of medical doctors employed across the year (VSSC Human Resources Turnover Cube)	Continuous (parent station)
Nurse turnover	Number of nurses lost (related to retirement, death, termination, or voluntary separation) divided by the number of nurses employed across the year (VSSC Human Resources Turnover Cube)	Continuous (parent station)

Facility characteristics were classified according to an analytical framework of factors that determine organizational context.²¹ Facility structure includes basic structural characteristics that are generally unmodifiable. Staffing and culture includes more modifiable structural characteristics that may affect the success of quality improvement efforts.

Characteristics were aggregated at the level of parent stations or substations. A parent station is an administrative unit that usually includes a medical center and 2 or more CBOCs. A substation is a component of a station and can be a medical center or a CBOC. Characteristics were aggregated at the substation level whenever possible. For characteristics only available at the parent station level, the parent station value was assigned to all substations within that parent station.

VHA PACT Compass is an electronic dashboard of metrics that reflect the extent to which facilities are successful in providing team-based, continuous, and coordinated care.

CBOC, community-based outpatient clinic; DCG, diagnostic cost group; FY, fiscal year; IOC, independent outpatient clinic; IQR, interquartile range; MD, medical doctor; MOC, mobile outpatient clinic; PACT, Patient Aligned Care Team; SD, standard deviation; VAMC, Veterans Affairs Medical Center; VHA, Veterans Health Administration; VSSC, Veterans Support Service Center.

Statistical analyses

Preliminary analyses were conducted to inform the analytic approach. Specifically, variables were examined for out-of-range values, sparse categories, and high correlations that would indicate multicollinearity. As a result, average patient panel size across providers for each facility was calculated using months with panel sizes within the expected range (500–2000). In addition, because of small numbers, independent outpatient clinics (IOCs, n = 5) were combined with Veterans Affairs Medical Centers (VAMCs, n = 148), and mobile outpatient clinics (MOCs, n = 4) were combined with community-based outpatient clinics (CBOCs, n = 750). For pairs of variables with correlations ≥0.50, only the variable that was more highly correlated with patient-centered Pi² was included. To aid interpretation of model coefficients, panel size was scaled by 100, patient severity was scaled by 0.10, and all variables were centered before running multivariable models. A likelihood ratio test was conducted to determine whether the study needed to account for the fact that individual VHA facilities are organized into administrative units called parent stations. This test indicated that 21.6% of the variance of patient-centered Pi² was accounted for by variation between parent stations. Therefore, mixed-effects linear regression models with a random effect for parent station were used for subsequent analyses.

The multivariable analyses proceeded in 3 steps. First, separate mixed-effect models were run to examine associations of each facility characteristic with patient-centered Pi² scores. This began with a base model containing only the minority concentration variables and random effect for parent station, then other facility characteristics were added that were associated with patient-centered Pi² scores at P < 0.10 when tested individually. Second, a final multivariable model was constructed that retained the minority concentration variables and other facility characteristics that were associated with patient-centered Pi² scores at P < 0.05 in the fully adjusted model. The final step was to test for interactions between all minority concentration variables and facility characteristics retained in the final model.

Results

Facility characteristics

Of the 1049 VHA facilities with outpatient activity in FY13, the following were excluded: 11 (1.0%) in outlying areas (San Juan, Manila, Guam, and American Samoa), 96 (9.2%) that did not offer primary care services, 32 (3.1%) without Pi² scores, and 2 (0.2%) that closed before the end of FY13. Analyses included 908 facilities clustered within 139 administrative parent stations.

Median percentages of black, Hispanic, and other minority patients across facilities were 5.1, 1.6, and 3.4, respectively. Facilities were categorized as having low, medium, or high concentrations of patients from each minority group using the 33rd and 66th percentiles of black (low: <2.3% and high: >10%), Hispanic (low: <0.9% and high: >3%), and other minority (low: <2.6% and high: >4.5%) patient concentration. As shown in Table 3, higher concentrations of black, Hispanic, or other minority patients were significantly associated with higher percentages of female patients and lower percentages of white nurses. Patterns for all other facility characteristics depended on which racial/ethnic group was being considered. For example, employee satisfaction was significantly higher at facilities with greater concentrations of black or other minority patients, but was significantly lower at facilities with greater concentrations of Hispanic patients. Percentage of VAMC and IOC facilities (vs. CBOC and MOC) increased across the low, medium, and high black tertiles, but peaked in the medium tertiles of Hispanic and other minority concentrations. Patient clinical severity increased across the low, medium, and high black tertiles, but was not associated with concentration of Hispanic or other minority patients.

Table 3.

Characteristics of VHA Facilities That Provide Outpatient Primary Care Services (n = 908), Compared by Percentage of Black, Hispanic, and Other Racial/Ethnic Minority Patient Concentrations

	Black patient concentration
Facility characteristics	Low (<2.3%)	Medium (2.3%–10.0%)	High (>10.0%)
% Non-Hispanic white patients, median (IQR)	94.3 (91.2–96.0)	88.0 (82.4–91.5)	66.8 (53.7–76.1)^a
% Non-Hispanic black patients, median (IQR)	0.9 (0.5–1.4)	5.0 (3.4–7.1)	22.1 (14.6–34.2)^a
% Hispanic patients, median (IQR)	0.9 (0.5–2.3)	2.1 (0.8–5.3)	2.0 (1.1–6.0)^a
% Other patients, median (IQR)	3.3 (2.1–4.9)	3.4 (2.2–4.9)	3.5 (2.4–5.8)^b
% Female, mean (SD)	4.6 (1.4)	5.1 (1.6)	7.6 (3.5)^a
Number of annual primary care visits, n (%)
Q1	124 (41.1)	54 (17.8)	49 (16.2)^a
Q2	99 (32.8)	74 (24.4)	54 (17.8)
Q3	55 (18.2)	96 (31.7)	76 (25.1)
Q4	24 (8.0)	79 (26.1)	124 (40.9)
Patient panel size, mean (SD)^b	1277 (1688)	1309 (1609)	1157 (397)
Patient severity (DCG), mean (SD)^b	0.5 (0.1)	0.6 (0.2)	0.6 (0.2)^a
% Non-Hispanic white MD, mean (SD)	74.0 (14.5)	62.1 (13.7)	59.4 (11.4)^a
% Non-Hispanic white nurse, mean (SD)	83.8 (17.2)	68.5 (21.7)	45.0 (20.3)^a
Employee satisfaction, mean (SD)	3.66 (0.13)	3.69 (0.10)	3.69 (0.08)^c
Primary care staff ratio, mean (SD)^c	3.10 (0.99)	3.11 (0.70)	3.11 (0.76)
MD turnover, mean (SD)	0.11 (0.05)	0.10 (0.05)	0.10 (0.04)^c
Nurse turnover, mean (SD)	0.10 (0.03)	0.09 (0.03)	0.08 (0.02)^a
Facility type
VAMC or IOC, n (%)	21 (7.0)	48 (15.8)	84 (27.7)^a
CBOC or MOC, n (%)	281 (93.1)	255 (84.2)	219 (72.3)
Patient-centered Pi², mean (SD)	0.32 (2.41)	−0.17 (2.19)	−0.18 (2.23)^c

	Hispanic patient concentration
	Low (<0.9%)	Medium (0.9%–3.0%)	High (>3.0%)
% Non-Hispanic white patients, median (IQR)	93.6 (87.4–95.9)	87.1 (74.2–92.4)	73.3 (59.2–85.4)^a
% Non-Hispanic black patients, median (IQR)	2.3 (0.7–7.4)	6.3 (1.7–19.0)	8.0 (2.9–15.1)^a
% Hispanic patients, median (IQR)	0.5 (0.4–0.7)	1.6 (1.2–2.2)	6.7 (4.1–15.2)^a
% Other patients, median (IQR)	2.6 (1.7–4.1)	3.6 (2.3–5.0)	4.3 (2.8–7.4)^a
% Female, mean (SD)	4.5 (1.3)	6.1 (2.6)	6.7 (3.3)^a
Number of annual primary care visits, n (%)
Q1	88 (29.1)	72 (23.8)	67 (22.1)^c
Q2	89 (29.5)	73 (24.1)	65 (21.5)
Q3	75 (24.8)	71 (23.4)	81 (26.7)
Q4	50 (16.6)	87 (28.7)	90 (29.7)
Patient panel size, mean (SD)^b	1262 (1630)	1168 (329)	1313 (1682)
Patient severity (DCG), mean (SD)^b	0.6 (0.1)	0.6 (0.2)	0.6 (0.2)
% Non-Hispanic white MD, mean (SD)	67.5 (13.7)	67.2 (14.7)	60.7 (14.6)^a
% Non-Hispanic white nurse, mean (SD)	78.8 (21.2)	66.9 (24.4)	51.6 (22.9)^a
Employee satisfaction, mean (SD)	3.69 (0.11)	3.67 (0.11)	3.67 (0.09)^b
Primary care staff ratio, mean (SD)^b	3.09 (0.89)	3.11 (0.74)	3.12 (0.84)
MD turnover, mean (SD)	0.10 (0.06)	0.10 (0.04)	0.10 (0.04)
Nurse turnover, mean (SD)	0.09 (0.02)	0.09 (0.03)	0.09 (0.02)
Facility type
VAMC or IOC, n (%)	31 (10.3)	65 (21.5)	57 (18.8)^a
CBOC or MOC, n (%)	271 (89.7)	238 (78.6)	246 (81.2)
Patient-centered Pi², mean (SD)	0.57 (2.19)	0.06 (2.25)	−0.66 (2.25)^a

	Other racial/ethnic minority patient concentration
	Low (<2.6%)	Medium (2.6%–4.5%)	High (>4.5%)
% Non-Hispanic white patients, median (IQR)	91.9 (82.4–95.9)	86.5 (71.3–93.3)	77.9 (61.5–88.5)^a
% Non-Hispanic black patients, median (IQR)	3.9 (1.2–11.8)	5.9 (1.3–18.2)	6.0 (1.6–14.5)
% Hispanic patients, median (IQR)	1.0 (0.5–2.5)	1.6 (0.7–3.9)	2.8 (1.1–7.5)^a
% Other patients, median (IQR)	1.9 (1.4–2.2)	3.4 (3.0–3.9)	6.7 (5.2–9.3)^a
% Female, mean (SD)	5.0 (1.8)	5.9 (2.3)	6.5 (3.5)^a
Number of annual primary care visits, n (%)
Q1	85 (28.2)	69 (22.8)	73 (24.1)
Q2	78 (25.8)	67 (22.1)	82 (27.1)
Q3	73 (24.2)	82 (27.1)	72 (23.8)
Q4	66 (21.9)	85 (28.1)	76 (25.1)
Patient panel size, mean (SD)^b	1267 (1619)	1201 (453)	1275 (1667)
Patient severity (DCG), mean (SD)^b	0.6 (0.2)	0.6 (0.1)	0.6 (0. 2)
% Non-Hispanic white MD, mean (SD)	66.2 (14.9)	64.3 (15.0)	64.9 (14.0)
% Non-Hispanic white nurse, mean (SD)	70.3 (25.7)	65.6 (25.6)	61.3 (24.4)^a
Employee satisfaction, mean (SD)	3.70 (0.10)	3.67 (0.11)	3.66 (0.11)^a
Primary care staff ratio, mean (SD)^b	3.13 (0.80)	3.09 (0.77)	3.11 (0.90)
MD turnover, mean (SD)	0.10 (0.05)	0.11 (0.05)	0.10 (0.04)^b
Nurse turnover, mean (SD)	0.09 (0.03)	0.09 (0.03)	0.10 (0.03)
Facility type
VAMC or IOC, n (%)	39 (12.9)	64 (21.1)	50 (16.5)^b
CBOC or MOC, n (%)	263 (87.1)	239 (78.9)	253 (83.5)
Patient-centered Pi², mean (SD)	0.42 (2.18)	−0.15 (2.33)	−0.29 (2.29)^a

There were 3 variables with missing cases: panel size (n = 11, 1.2%), patient severity (n = 18, 2.0%), and primary care staffing ratio (n = 15, 1.7%). P values are based on comparisons across facilities with low, medium, and high concentrations of each minority group. Analysis of Variance was used to compare means, Kruskal–Wallis tests were used to compare medians, and chi-square tests were used to compare categorical variables.

P < 0.001.

P < 0.05.

P < 0.01.

CBOC, community-based outpatient clinic; DCG, diagnostic cost group; IOC, independent outpatient clinic; IQR, interquartile range; MD, medical doctor; MOC, mobile outpatient clinic; Pi², patient aligned care team implementation progress; Q1, first quartile (<5990.5 visits); Q2, second quartile (≥5990.5 and <11051visits); Q3, third quartile (≥11051 and <24170.5 visits); Q4, fourth quartile (≥24170.5 visits); SD, standard deviation; VAMC, Veterans Affairs Medical Center; VHA, Veterans Health Administration.

Patient-centered Pi²

Patient-centered Pi² ranged from −4 to 4, with a mean of −0.01 (standard deviation = 2.29). Patient-centered Pi² was significantly different across facilities in the low, medium, and high tertiles for each racial/ethnic minority concentration, with facilities in the lowest tertile of each minority group having the highest patient-centered Pi² (Table 3). The largest difference was observed when facilities were categorized based on Hispanic patient concentration, with mean patient-centered Pi² scores of 0.57, 0.06, and −0.66 (P < 0.001) in facilities with low, medium, and high Hispanic patient concentrations, respectively. Somewhat smaller differences were observed when facilities were categorized based on concentration of other minority patients (means = 0.42, −0.15, and −0.29 for low, medium, and high facilities, respectively; P < 0.001) or black patients (means = 0.32, −0.17, and −0.18 for low, medium, and high facilities, respectively; P = 0.009).

Modeled associations of black, Hispanic, and other minority patient concentration with patient-centered Pi²

In the multilevel base model, the medium and high tertiles of Hispanic patient concentration were significantly associated with lower patient-centered Pi² scores (medium: estimate [Est] = −0.40, standard error [SE] = 0.20, P = 0.046; high: Est = −0.99, SE = 0.23, P < 0.001). Black and other minority patient indicators were not associated with patient-centered Pi² scores in this model (Table 4, Base Model).

Table 4.

Mixed-Effects Linear Regression Models Examining the Association of Patient Racial/Ethnic Minority Concentration and Other Facility Characteristics with Patient-Centered Pi ²

	Base model		Final model
Predictors	Estimate	SE	Estimate	SE
Black patient concentration
Low: <2.3%	Ref		Ref
Medium: 2.3%–10.0%	−0.24	0.19	−0.23	0.19
High: >10.0%	−0.15	0.22	−0.28	0.22
Hispanic patient concentration
Low: <0.9%	Ref		Ref
Medium: 0.9%–3.0%	−0.40	0.20^a	−0.40	0.19^a
High: >3.0%	−0.99	0.23^b	−0.95	0.23^b
Other racial/ethnic minority patient concentration
Low: <2.6%	Ref		Ref
Medium: 2.6%–4.5%	−0.17	0.19	−0.16	0.18
High: >4.5%	−0.27	0.21	−0.22	0.21
Patient panel size (scaled by 100)			−0.12	0.03^b
Patient severity (scaled by 0.10)			0.26	0.06^b
Annual primary care visits
Q1			Ref
Q2			0.54	0.20^c
Q3			0.12	0.21
Q4			−0.35	0.23

Models included a random effect for parent station to account for the clustering of facilities.

P < 0.05.

P < 0.001.

P < 0.01.

Q1, first quartile (<5990.5 visits); Q2, second quartile (≥5990.5 and <11051visits); Q3, third quartile (≥11051 and <24170.5 visits); Q4, fourth quartile (≥24170.5 visits); Pi², Patient Aligned Care Team implementation progress index; SE, standard error.

Based on preliminary analyses, the following variables were added to the base model to determine whether they explained the association of Hispanic patient concentration with patient-centered Pi² scores: percentage of female patients, number of primary care visits, patient panel size, patient severity, percentage of white medical doctors, and employee job satisfaction (data not shown). Of these, patient panel size, patient severity, and number of annual primary care visits were retained in the final model as these remained significantly (P < 0.05) associated with patient-centered Pi² scores in the fully adjusted model.

In the final model (Table 4, Final Model), the medium and high Hispanic tertiles remained significant predictors of patient-centered Pi² scores (medium: Est = −0.40, SE = 0.19, P = 0.040; high: Est = −0.95, SE = 0.23, P < 0.001). In other words, facilities in the medium and high tertiles of Hispanic patient concentration had patient-centered Pi² scores that were 0.40 and 0.95 points lower than facilities with the lowest concentration of Hispanic patients. Facilities with larger panel sizes also exhibited worse patient-centered Pi² scores than facilities with lower panel sizes, such that patient-centered Pi² was reduced by 0.12 points for every increase of 100 patients in average patient panel size among primary care providers. However, patient-centered Pi² scores increased by 0.26 for every 0.10 increase in the average clinical severity of patients. Finally, facilities in the second smallest volume quartile had patient-centered Pi² scores that were 0.54 points higher than facilities in the lowest quartile of primary care visits.

Interactions between minority patient concentration and other facility characteristics

This study explored all potential 2-way interactions between each minority patient concentration indicator and patient panel size, patient severity, and annual number of primary care visits (data not shown). There were significant interactions between panel size and the medium and high black tertiles (medium: Est = −0.15, SE = 0.07, P = 0.046; high: Est = −0.19, SE = 0.08, P = 0.015). To interpret these interactions, the predicted patient-centered Pi² scores were plotted for facilities with low, medium, and high concentrations of black patients at the mean patient panel size and at 1 standard deviation above and below the mean (Fig. 1). The figure illustrates that among facilities with low percentages of black patients, patient-centered Pi² scores are slightly lower (−0.15 points) in facilities with patient panel sizes 1 standard deviation below (vs. above) the mean. The difference is much larger among facilities with medium (−0.85 points) or high percentages (−1.03 points) of black patients.

FIG. 1.

Patient-centered Pi² scores for facilities with low, medium, and high concentrations of black patients at the mean patient panel size (1138 patients), as well as at 1 standard deviation above (1373 patients) and below (903 patients) the mean. Pi², Patient Aligned Care Team implementation progress index.

Discussion

In this study of VHA facilities that have undergone primary care transformations to implement the PCMH model, ratings of patient-centered care were ∼0.5–1 point lower on a 9-point scale at facilities serving higher concentrations of Hispanic patients and at facilities with a combination of higher concentrations of black patients and larger panel sizes. Although many factors contribute to racial and ethnic disparities in health care,^22,23 this study contributes to a body of work showing that segregation of racial and ethnic minority patients in relatively low-performing facilities may underlie and perpetuate disparities.^{8

–17,24,25}

Most prior studies showing that racial and ethnic disparities in health care occur because of differences in facilities where minority and nonminority patients get care have focused on quality of care metrics in inpatient settings.^{8

–15,24,25} Fewer studies have examined the association of patient racial and ethnic concentration with patient-reported experiences in the outpatient setting, as was the focus of the present study.^16,26
–28 Addressing this gap is important given that access to high-quality outpatient primary care is recognized as essential for maintaining health and health equity²⁹ and there continue to be racial and ethnic disparities in primary care access and utilization.³⁰

The PCMH model is thought to be a potential mechanism to reduce racial and ethnic disparities.^31
–33 For the PCMH model to meet the needs of racial and ethnic minority patients, however, it must be well implemented in facilities serving high concentrations of minority patients. Outside the VHA, some racial and ethnic minority groups remain less likely than whites to receive care from providers with PCMH characteristics.^34,35 Within VHA, one study using Pi² scores from 2012 found that facilities serving a higher percentage of non-white veterans were struggling with PACT implementation overall.³⁶

Focusing on the patient-reported components of Pi² from 2013, and disaggregating the racial/ethnic groups, the present study shows that a facility's concentration of Hispanic, but not black or other minority, patients is associated with patient-centered Pi². This study was unable to identify any facility characteristics that explained differences in patient-centered care between high, medium, and low-Hispanic VHA facilities, despite examining a broad range of variables pertaining to facility structure, staffing, and culture. These results suggest that targeted efforts are likely needed to improve patient-centered care at VHA facilities with relatively high proportions of Hispanic patients. To identify specific strategies to improve patient-centered care at Hispanic-serving facilities, it may be necessary to go beyond what can be ascertained from administrative data by conducting site visits or interviewing patients and providers at those facilities. One factor that could be explored is the presence and impact of language barriers. On the same survey used to calculate this study's facility-level rating of patient-centered care, it was found that 15% of Hispanic veteran respondents reported that they spoke a language other than English at home (vs. <2% of veterans in the sample overall). The higher prevalence of non-English speakers among the Hispanic veteran population calls for additional research on whether language barriers interfere with the delivery of patient-centered care at facilities that serve Hispanics.

This study identified additional facility characteristics that were independently associated with patient-centered care, one of which may provide a modifiable target to improve patient-centered care overall. In particular, patient-centered scores were lower at facilities with larger patient panel sizes, especially at facilities with higher concentrations of black patients. Taking steps to decrease the average patient panel size of providers (eg, by increasing staffing levels) may increase patient-centered care, with the greatest potential benefits likely being at facilities serving more black patients. Other facility characteristics associated with patient-centered care in this study (ie, clinical severity of patients, overall primary care volume) would be more difficult to modify.

This study has limitations. The main predictor was the racial and ethnic concentration of patients at facilities as documented in electronic medical records. This may have undercounted Hispanic and other minority patients.^37,38 By relying on administrative data, the study also was limited in the facility characteristics that could be included in the analyses. Finally, no conclusions could be drawn about causal relationships because of the cross-sectional nature of this analysis.

Conclusion

Although the PCMH model has been seen as having potential to reduce disparities, a reduction of disparities has not been an explicit target or a routinely monitored outcome in PCMH initiatives.³⁹ This study demonstrates that patient-centered care in VHA facilities where the PCMH model has been implemented is lower among facilities that serve more Hispanic patients and in facilities with large panel sizes that serve large numbers of black patients. Moving forward, steps are needed to ensure that the PCMH goal of delivering patient-centered care is being achieved successfully across all VHA facilities, regardless of the racial and ethnic composition of their patient populations.

Footnotes

Acknowledgments

This work was undertaken as part of the Veterans Administration's PACT Demonstration Laboratory initiative, supporting and evaluating Veterans Affairs transition to a patient-centered medical home. The views expressed here are those of the authors and do not represent those of the Department of Veterans Affairs or the US Government. Preliminary findings from this work were presented at the VA HSR&D/QUERI National Conference in 2016.

Author Disclosure Statement

Drs. Hausmann, Gao, Jones, Long, and Werner, Ms. Canamucio, and Ms. Keddern declared no conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received financial support from the VA Office of Patient Care Services (XVA 72-065; Coprincipal Investigators Drs. Werner and Long).

References

Takach

. About half of the states are implementing patient-centered medical homes for their Medicaid populations. Health Aff (Millwood), 2012; 31:2432–2440.

Cunningham

, Felland

, Stark

. Safety-net providers in some US communities have increasingly embraced coordinated care models. Health Aff (Millwood), 2012; 31:1698–1707.

Werner

, Canamucio

, Shea

, True

. The medical home transformation in the Veterans Health Administration: An evaluation of early changes in primary care delivery. Health Serv Res, 2014; 49:1329–1347.

Edwards

, Bitton

, Hong

, Landon

. Patient-centered medical home initiatives expanded in 2009–2013: providers, patients, and payment incentives increased. Health Aff, 2014; 33:1823–1831.

Stange

, Nutting

, Miller

, et al. Defining and measuring the patient-centered medical home. J Gen Intern Med, 2010; 25:601–612.

Nelson

, Helfrich

, Sun

, et al. Implementation of the patient-centered medical home in the Veterans Health Administration: associations with patient satisfaction, quality of care, staff burnout, and hospital and emergency department use. JAMA Intern Med, 2014; 174:1350–1358.

Rosland

, Nelson

, Sun

, et al. The patient-centered medical home in the Veterans Health Administration. Am J Manag Care, 2013; 19:e263–e272.

Hasnain-Wynia

, Baker

, Nerenz

, et al. Disparities in health care are driven by where minority patients seek care: examination of the hospital quality alliance measures. Arch Intern Med, 2007; 167:1233–1239.

Hasnain-Wynia

, Kang

, Landrum

, Vogeli

, Baker

, Weissman

. Racial and ethnic disparities within and between hospitals for inpatient quality of care: an examination of patient-level Hospital Quality Alliance measures. J Health Care Poor Underserved, 2010; 21:629–648.

10.

Jha

, Orav

, Li

, Epstein

. Concentration and quality of hospitals that care for elderly black patients. Arch Intern Med, 2007; 167:1177–1182.

11.

Jha

, Orav

, Zheng

, Epstein

. The characteristics and performance of hospitals that care for elderly Hispanic Americans. Health Aff (Millwood), 2008; 27:528–537.

12.

Barnato

, Alexander

, Linde-Zwirble

, Angus

. Racial variation in the incidence, care, and outcomes of severe sepsis: analysis of population, patient, and hospital characteristics. Am J Respir Crit Care Med, 2008; 177:279–284.

13.

Barnato

, Lucas

, Staiger

, Wennberg

, Chandra

. Hospital-level racial disparities in acute myocardial infarction treatment and outcomes. Med Care, 2005; 43:308–319.

14.

Skinner

, Chandra

, Staiger

, Lee

, McClellan

. Mortality after acute myocardial infarction in hospitals that disproportionately treat black patients. Circulation, 2005; 112:2634–2641.

15.

Merchant

, Becker

, Yang

, Groeneveld

. Hospital racial composition: a neglected factor in cardiac arrest survival disparities. Am Heart J, 2011; 161:705–711.

16.

Hausmann

, Gao

, Mor

, Schaefer

Jr. , Fine

. Understanding racial and ethnic differences in patient experiences with outpatient health care in Veterans Affairs Medical Centers. Med Care, 2013; 51:532–539.

17.

Hausmann

LRM

, Gao

, Mor

, Schaefer

JHJ

, Fine

. Patterns of sex and racial/ethnic differences in patient health care experiences in US Veterans Affairs hospitals. Med Care, 2014; 52:328–335.

18.

Agency for Healthcare Research and Quality. Patient Experience Measures from the CAHPS Clinician & Group Surveys. Document No. 1309 Updated September 26, 2014. www.ahrq.gov/sites/default/files/wysiwyg/cahps/surveys-guidance/cg/about/measures_cg.pdf Accessed June 29, 2016 .

19.

Scholle

, Vuong

, Ding

, et al. Development of and field test results for the CAHPS PCMH survey. Med Care, 2012; 50 suppl:S2–S10.

20.

Hays

, Berman

, Kanter

, et al. Evaluating the psychometric properties of the CAHPS Patient-Centered Medical Home survey. Clin Ther, 2014; 36:689–696.e1.

21.

Glasgow

, Yano

, Kaboli

. Impacts of organizational context on quality improvement. Am J Med Qual, 2013; 28:196–205.

22.

Smedley

, Stith

, Nelson

, eds. Unequal treatment: confronting racial and ethnic disparities in health care. Washington, DC: National Academies Press, 2003.

23.

Kilbourne

, Switzer

, Hyman

, Crowley-Matoka

, Fine

. Advancing health disparities research within the health care system: a conceptual framework. Am J Public Health, 2006; 96:2113–2121.

24.

Howell

, Egorova

, Balbierz

, Zeitlin

, Hebert

. Black-white differences in severe maternal morbidity and site of care. Am J Obstet Gynecol, 2016; 214:122-e1–122-e7.

25.

Brooks-Carthon

, Kutney-Lee

, Sloane

, Cimiotti

, Aiken

. Quality of care and patient satisfaction in hospitals with high concentrations of black patients. J Nurs Scholarsh, 2011; 43:301–310.

26.

Weech-Maldonado

, Elliott

, Morales

, Spritzer

, Marshall

, Hays

. Health plan effects on patient assessments of Medicaid managed care among racial/ethnic minorities. J Gen Intern Med, 2004; 19:136–145.

27.

Weech-Maldonado

, Morales

, Elliott

, Spritzer

, Marshall

, Hays

. Race/ethnicity, language, and patients’ assessments of care in Medicaid managed care. Health Serv Res, 2003; 38:789–808.

28.

Price

, Haviland

, Hambarsoomian

, et al. Do experiences with Medicare managed care vary according to the proportion of same-race/ethnicity/language individuals enrolled in one's contract?. Health Serv Res, 2015; 50:1649–1687.

29.

Starfield

, Shi

, Macinko

. Contribution of primary care to health systems and health. Milbank Q, 2005; 83:457–502.

30.

Agency for Healthcare Research and Quality. 2014. National Healthcare Quality and Disparities Report. Pub. No. 15-0007. Rockville, MD: Agency for Healthcare Research and Quality, 2015.

31.

Berenson

, Doty

, Abrams

, Shih

. Achieving better quality of care for low-income populations: The roles of health insurance and the medical home in reducing health inequities. The Commonwealth Fund. May 2012. www.healthreformgps.org/wp-content/uploads/1600_Berenson_achieving_better_quality_care_low_income_v2.pdf Accessed June 29, 2016 .

32.

Nocon

, Gao

, Gunter

, et al. Associations between medical home characteristics and support for patient activation in the safety net: understanding differences by race, ethnicity, and health status. Med Care, 2014; 52(11 suppl 4):S48–S55.

33.

Garg

, Jack

, Zuckerman

. Addressing the social determinants of health within the patient-centered medical home: lessons from pediatrics. JAMA, 2013; 309:2001–2002.

34.

Beal

, Doty

, Hernandez

, Shea

, Davis

. Closing the divide: How medical homes promote equity in health Care. The Commonwealth Fund. June 2007. www.commonwealthfund.org/usr_doc/1035_Beal_closing_divide_medical_homes.pdf?section=4039 Accessed June 29, 2016 .

35.

Beal

, Hernandez

, Doty

. Latino access to the patient-centered medical home. J Gen Intern Med, 2009; 24 suppl 3:514–520.

36.

Hernandez

, Taylor

, Grembowski

, et al. A first look at PCMH implementation for minority veterans: room for improvement. Med Care, 2016; 54:253–261.

37.

Kressin

, Chang

, Hendricks

, Kazis

. Agreement between administrative data and patients’ self-reports of race/ethnicity. Am J Public Health, 2003; 93:1734–1739.

38.

Stroupe

, Tarlov

, Zhang

, Haywood

, Owens

, Hynes

. Use of Medicare and DOD data for improving VA race data quality. J Rehabil Res Dev, 2010; 47:781–795.

39.

Reibling

, Rosenthal

. The (missed) potential of the patient-centered medical home for disparities. Med Care, 2016; 54:9–16.

Racial and Ethnic Minority Concentration in Veterans Affairs Facilities and Delivery of Patient-Centered Primary Care

Abstract

Introduction

Methods

Sample

Study measures and data sources

Study outcome

Primary predictors

Additional facility characteristics

Statistical analyses

Results

Facility characteristics

Patient-centered Pi2

Modeled associations of black, Hispanic, and other minority patient concentration with patient-centered Pi2

Interactions between minority patient concentration and other facility characteristics

Discussion

Conclusion

Footnotes

Acknowledgments

Author Disclosure Statement

References

Patient-centered Pi²

Modeled associations of black, Hispanic, and other minority patient concentration with patient-centered Pi²