Concordance and Patient-Centered Care in Medicaid Enrollees' Care Experience With Providers.

Patient-centered care is at the nexus of several overlapping institutional reforms to improve health care system performance. However, we know little regarding Medicaid patients' experience with their doctors along several key dimensions of patient-centered care, and how their experience compares with Medicare and privately insured patients. We studied 4 outcomes using the 2017 National Health Interview Survey: patient-provider concordance on racial/sexual/cultural identity, respectful provider attitude, solicitation of patient opinion/beliefs during the care encounter, and patient-centered communication (PCC). The primary independent variable was Medicaid enrollee status. We dichotomized responses and ran multivariate logistic regressions for each type of care experience outcome, controlling for sociodemographic factors, health care access, and health care utilization of respondents. Compared to Medicare and privately insured enrollees, Medicaid enrollees reported much lower odds of seeing providers who treated them with respect (OR = 1.91, P < .001; OR = 1.62, P < .01) and who offered PCC (OR = 1.35, P < .05; OR = 1.35, P < .01), but similar odds of seeing concordant providers (OR = 0.78, P = .96; OR = 0.96, P = .72). Importantly, Medicaid enrollees reported higher odds of seeing providers who solicited their opinion/beliefs/preferences than their Medicare or privately insured counterparts (OR = 0.82, P < .05; OR = 0.87 P < .10). Medicaid enrollees report less patient-centered experiences in some important facets of their provider interaction than their Medicare or privately insured counterparts. Federal, state, and local policies and practices directed at improving these facets of patient-provider interaction are needed and should be aimed squarely at Medicaid providers, especially those working in geographic areas and settings with a disproportionate number of racial, gender, cultural, and linguistic minorities.

Introduction

Medicaid enrollees are notably more likely to need health care services than the uninsured and privately insured, yet they are faced with several barriers to primary care, specialist care, and associated emergency department (ED) utilization.(1 –3) Access to primary care and ED utilization are important indicators of the health care system’s performance (2 –4). Recent studies show that the expansion of insurance coverage alone is not sufficient to ensure Medicaid enrollees’ access to primary, specialty, and emergency services(5). Although the Patient Protection and Affordable Care Act reduced barriers to access to care by increasing Medicaid eligibility, other barriers persist (6 –8).

Literature examining the care experience of Medicaid enrollees has heavily focused on identifying system-level barriers to access, such as the clinic not being open during the time of access, limited numbers of primary care physicians, and transportation issues as the reason for poor access to care for Medicaid enrollees (2,7). Other studies identified stigma, discrimination, and cost-related barriers to access for Medicaid enrollees (9,10). Of the few studies that focus on Medicaid enrollees’ subjective experiences of care, most used qualitative approaches. These studies suggest Medicaid enrollees face several challenges when accessing care, poor communication with the provider, stigma, and the lack of respect from providers (5,11).

Previous studies are useful but limited in 2 important ways. Qualitative approaches can yield powerful insights into underlying phenomena but lack strong reliability and external validity available with large-scale representative surveys of the US population. Moreover, existing literature provides little information on several key aspects of care experience considered vital to care quality. For example, it is increasingly recognized that providers who share some form of identity (eg, racial, gender, or cultural) with their patients may provide higher quality care, especially for minority patients. Concordance on race/ethnicity and/or sex seems to be correlated with higher satisfaction with care experience, possibly due to heightened odds of physicians’ familiarity with patients’ cultural norms, language, and unique needs (12). A desire for higher concordance may also simply reflect heightened distrust of the care system by some minority patients, an outgrowth of past historical practices grounded in racial/ethnic discrimination (13,14). This lack of trust often translates into lower compliance with recommendations and subpar clinical outcomes (13). More behavioral aspects of provider–patient interaction may also yield significant benefits in terms of both subjective and objective measures of care quality. Doctors who communicate their recommendations in simpler terms include patients’ voices in decisions regarding key aspects of care and assume and maintain respectful attitudes while interacting with patients tend to elicit higher satisfaction and greater compliance (15,16) and, in some studies, measurable improvements in clinical metrics of disease progression (17). Patient-centered care is at the nexus of several overlapping institutional reform efforts designed to alter delivery, payment, and evaluation of medical care to improve health care system performance (18). The concept is defined by IOM as “care that is respectful of and responsive to individual patient preferences, needs, and values and ensuring that patient’s values guide all clinical decisions” (6). A commonly used framework categorizes key dimensions of patient-centeredness into a network of interrelated principals (characteristics of physicians, physician–patient relationship, recognition of patient as a unique individual under a biopsychosocial perspective), enabling factors (nature of provider communication, integration of medical and nonmedical care, teamwork, access to care, and coordination and continuity in care), and activities (patient information, involvement of patients and family in care, and empowerment and emotional support activities) (19). We draw on this framework to identify important aspects of patient-centeredness in Medicaid enrollees’ experience of provider–patient relationship and care encounters. Specifically, we explore Medicaid enrollees’ perception of being treated with respect during the care encounter, the degree to which providers solicited patients’ opinions/beliefs/preferences during the encounters, and the overall quality of provider communication. It is surprising how little we know regarding Medicaid patients’ experience with their doctors along these several key dimensions of patient-centeredness, and how their experience compares with Medicare and privately insured patients. We address both of these limitations in the present study using new survey data from the National Health Interview Survey (NHIS). Therefore, we use the NHIS to characterize participants’ experience of patient-centered care and identify differences in the experience of patient-centered care for Medicaid enrollees compared with other insurers.

Methods

Data

The study utilized data from the NHIS, which is a cross-sectional face-to-face interview survey conducted by the National Center for Health Statistics and representative of the noninstitutionalized US civilian population. The NHIS uses a stratified multistage probability design for the selection of subjects. The survey tool includes a core questionnaire with basic demographic and health questions. Certain subgroups including racial and ethnic minorities are oversampled in the NHIS. These subgroups are oversampled related to the need for better data on minority groups in the United States and an acknowledgment of the inadequacy of race data classified solely by observation. Oversampling certain subgroups of households helps to improve the precision of estimates for these respondents. Our baseline sample comprised 26 742 adults who were interviewed in 2017 (aged 18 years and older; Figure 1).

Figure 1.

Cohort selection table for inclusion in the study.

Study Variables

Outcome variables

We defined 5 outcome variables for this study. To assess their care experience, respondents were asked a set of questions in a distinct order (Figure 1). Respondents who acknowledged seeing a physician in past 12 months (n = 22 864) were first asked, “Some people think it is important for their providers to understand or share their race or ethnicity or gender or religion or beliefs or native language. How important is it to you that your health care providers understand or are similar to you in any of these ways?” Responses were dichotomized so that 1 = very important or somewhat important and 0 = slightly or not important (Table 1). These individuals were then asked the following 3 questions: “How often were you treated with respect by your health care providers?” (1 = always or most of the time, 0 = some or none of the time), “How often did the provider ask about your opinions and beliefs?” (1 = always or most of the time, 0 = some or none of the time), and “How often did the provider give you information that was easy to understand?” (1 = always or most of the time, 0 = some or none of the time). Finally, individuals who thought sharing of identity with the provider was at least slightly important (n = 10 526) were asked the following question: “How often were you able to see health care providers who were similar to you in any of these ways? (1 = always or most of the time, 0 = some or none of the time). We generated 4 additional binary outcome variables from these items, respectful provider attitude, solicited patient opinions/beliefs/preferences, patient-centered communication (PCC), and patient–provider concordance, respectively, by dichotomizing responses as indicated.

Table 1.

Measurement of Patient-Centered Care Dimensions.

Number	Dimension	Question	How response options were dichotomized
1	Acknowledged importance of shared identity between provider and patient	“Some people think it is important for their providers to understand or share their race or ethnicity or gender or religion or beliefs or native language. How important is it to you that your health care providers understand or are similar to you in any of these ways?”	Very important	1
			Somewhat important
			Slightly important	0
			Not important at all
2	Patient–provider concordance	How often were you able to see health care providers who were similar to you in any of these ways?	Always	1
			Most of the time
			Some of the time	0
			None of the time
3	Treated patients with respect	How often were you treated with respect by your health care providers?	Always	1
			Most of the time
			Some of the time	0
			None of the time
4	Solicited patient opinions/beliefs/preferences	How often did your health care providers ask for your opinions or beliefs about your medical care or treatment? For example, what kind of tests, procedures, or medications you prefer.	Always	1
			Most of the time
			Some of the time	0
			None of the time
5	Patient-centered communication	How often did your health care providers tell or give you information about your health and health care that was easy to understand?	Always	1
			Most of the time
			Some of the time	0
			None of the time

Independent and control variables

Our primary independent variable was Medicaid enrollee status. Since we were primarily interested in how Medicaid beneficiaries compared to individuals with other types of insurance, we generated a single multicategory variable where Medicaid status was coded as a reference category, while privately insured, Medicare, dual-eligible, and uninsured patients were the comparison groups. Patients who had both Medicare and private insurance were coded as Medicare enrollees, while patients having both private insurance and Medicaid were considered Medicaid enrollees. We considered adults with both Medicare and Medicaid as dual eligible. We evaluated Medicaid enrollees’ experience of clinical care after controlling for a variety of patient-level demographic, clinical, access, and health care utilization-related factors. Specifically, we controlled for sex, race/ethnicity, marital status, employment status, functional disability, self-reported health, delay or denial in getting needed care, and high utilization of ED, hospital inpatient services, and office visits. Adjusted demographics also permitted comparison among various sociodemographic subgroups that may have different age structures.

Data Analysis

We generated summary statistics to compare care experience, sociodemographics, health care access, and health care utilization of Medicaid, Medicare, and privately ensured enrollees separately. We ran multivariate logistic regressions for each type of care experience outcome, controlling for sociodemographic, health care access, and health care utilization of respondents. The models yielded adjusted odds (with 95% CI) of reporting a specific type of care experience by respondents who did not have Medicaid as their primary insurance (ie, Medicare, private insurance, dual eligible, and uninsured), relative to that of the Medicaid enrollees. All statistical analysis was performed with Stata (version 16). We adjusted all summary statistics and regressions for the complex, multistage sampling design, using the svyset command in Stata version 15 along with the design variables and population weights provided by the NHIS.

Results

Compared to privately insured and Medicare, Medicaid enrollees were more likely to be younger, female, belong to a racial minority group, and unmarried (Table 2). They also reported poorer health status, more functional disability, lack of steady employment, more frequent ED use, and poorer access to needed care compared to privately insured and Medicare enrollees. In uncontrolled analyses, Medicaid enrollees were more likely to acknowledge the importance of a shared identity between patients and providers (mean = 46%, 95% CI, 42-49). At the same time, fewer Medicaid enrollees reported seeing providers who were concordant with them on some form of identity (mean = 64%, 95% CI, 60-68), providers who treated them with respect (mean = 93%, 95% CI, 92-95), or providers who offered PCC (mean = 87%, 95% CI, 85-89) than those who had private insurance or Medicare. The incidence of seeing providers who solicited patient opinions/beliefs was higher for Medicaid compared to Medicare and private insurance (mean = 60%, 95% CI, 57-63).

Table 2.

Summary Statistics by Payer Status.

	Medicaid, N = 2133	Medicare, N = 7028	Private insurance, N = 13 444	Dual enrollee, N = 905	Not insured, N = 2382
	Mean (95% CI)	Mean (95% CI)	Mean (95% CI)	Mean (95% CI)	Mean (95% CI)
Acknowledged importance of shared identity between provider and patient	46% (42-49)	37% (35–38)	35% (33–35)	70% (65-76)	48% (44-52)
Patient–provider concordance	64% (60-68)	77% (75-79)	71% (69-72)	71% (65-76)	58% (–)a
Provider treated patient with respect	93% (92-95)	97% (97-98)	97% (97-98)	94% (92-96)	94% (92-95)
Provider shared decision-making	60% (57-63)	57% (56-59)	58% (57-59)	64% (59-68)	60% (57-64)
Provider communication was patient centered	87% (85-89)	92% (92-93)	93% (92-94)	89% (86-91)	86% (84-89)
Age	37.62 (36.8-38.4)	71.75 (71-72)	41.72 (41–42)	64.77 (63-66)	38.39 (37–39)
Male	37% (35–39)	45% (44-47)	50% (51-42)	38% (33–42)	53% (50-56)
Race/ethnicity
Non-Hispanic white	43% (39-47)	79% (78-81)	68% (66-69)	47% (42-52)	41% (37-44)
Non-Hispanic black	21% (19–24)	89% (08-10)	11% (09-12)	21% (17–25)	14% (12–17)
Non-Hispanic Indian American	1% (0-24)	0.05% (00-01)	0.01% (0-01)	2% (01-03)	2% (01-03)
Non-Hispanic Asian	7% (05-09)	3% (03-04)	7% (06-08)	6% (05-11)	4% (03-05)
Hispanic	27% (23–31)	7% (06-08)	12% (12–14)	22% (17-11)	38% (35–42)
Married	42% (38-44)	59% (58-61)	66% (65-67)	29% (25–34)	55% (52-57)
Employment status	46% (43-49)	16% (16–18)	83% (82-84)	7% (05-09)	68% (65-70)
Health care access
Health care delayed due to cost	8% (07-09)	5% (45-58)	7% (06-07)	8% (05-09)	27% (24–29)
Health care denied foregone to cost	7% (06-08)	4% (37-05)	3% (03-04)	5% (04-07)	22% (20–24)
Health care utilization
Frequent emergency department visits, past 12 months	5% (04-06)	2% (02-03)	1% (01-01)	8% (06-11)	2% (01.03)
Frequent overnight hospital admission, past 12 months	0.7% (00-01)	1% (01-01)	0.2% (00-00)	2% (01-03)	0.4% (00-01)
Frequent office visits, past 12 M	29% (27–31)	36% (35–38)	18% (18–19)	51% (26-54)	9% (07-10)
Any functional limitation, all conditions	41% (39-45)	69% (68-71)	25% (34-26)	86% (82-88)	26% (24–29)
Self-reported health status	77% (75-79)	77% (76-79)	94% (94-95)	43% (39-47)	87% (86-89)

a Missing SE because of stratum with single sampling unit.

After controlling for sociodemographic, health care utilization, and access to needed care, Medicaid enrollees were much more likely to acknowledge the importance of sharing some form of identity with their providers than respondents who had Medicare or private insurance (Medicare OR 0.76, P < .001; private insurance OR 0.79, P < .001; Table 3). At the same time, odds of seeing a concordant provider did not significantly differ between respondents on different types of insurance (Medicare OR 0.78, P = .96; private insurance OR 0.96, P = .72; dual enrollee OR 1.03, P = 0.85). Medicaid enrollees reported much lower odds of seeing providers who treated them with respect compared to Medicare and privately insured enrollees (OR 1.91, P < .001; OR 1.62, P < .01). Similarly, Medicare and privately insured respondents had a much higher likelihood of seeing providers who communicated in patient-centered ways (OR = 1.35, P < .05) and (OR = 1.35, P < .01). On the other hand, Medicaid enrollees had higher odds of seeing providers who solicited patient opinions/beliefs regarding medical and diagnostic procedures (Medicare OR 0.82, P < .05, private insurance OR 0.87 P < .10).

Table 3.

Attitudes Toward Concordance and Care Experience of Medicaid Enrollees Relative to Respondents Enrolled in Other Types of Health Care Plans.

	Acknowledged importance of shared identity between provider and patient, N = 22 864	Patient–provider concordance, N = 10 526	Provider treated with respect, N = 22 864	Solicited patient’s opinion/beliefs, N = 22 864	Provider communication was patient centered, N = 22 864
	OR (P value)	OR (P value)	OR (P value)	OR (P value)	OR (P value)
Payer status
Medicaid	Reference	Reference	Reference	Reference	Reference
Medicare	0.76 (.00)	0.96 (.78)	1.91 (.00)	0.82 (.02)	1.35 (.04)
Private insurance	0.79 (.00)	0.96 (.72)	1.62 (.01)	0.87 (.06)	1.35 (.01)
Dual enrollee	1.25 (.07)	1.03 (.85)	1.19 (.45)	1.09 (.48)	1.27 (.18)
Not insured	1.10 (.33)	0.74 (.02)	1.09 (.68)	1.03 (.76)	0.77 (.15)
Age	1.01 (.00)	1.01 (.00)	1.01 (.00)	1.00 (.00)	1.01 (.00)
Sex	0.85 (.00)	0.93 (.17)	0.92 (.31)	0.90 (.00)	0.96 (.81)
Race/ethnicity
Non-Hispanic white	0.53 (.02)	2.23 (.03)	0.56 (.32)	0.80 (.37)	1.16 (.69)
Non-Hispanic black	1.29 (.31)	0.95 (.89)	0.37 (.77)	0.90 (.37)	0.89 (.79)
Non-Hispanic Indian American	1.26 (.46)	1.25 (.59)	0.29 (.07)	1.38 (.78)	1.10 (.84)
Non-Hispanic Asian	1.29 (.32)	0.82 (.63)	0.31 (.05)	0.99 (.95)	0.62 (.24)
Hispanic	1.30 (.31)	0.95 (.91)	0.41 (.13)	0.89 (.67)	0.79 (.57)
Married	1.04 (.30)	1.09 (.17)	0.95 (.66)	1.05 (.18)	1.07 (.37)
Employment status	0.85 (.00)	0.94 (.44)	1.15 (.26)	0.98 (.67)	0.94 (.48)
Health care access
Health care delayed due to cost	1.27 (.03)	0.84 (.18)	0.54 (.00)	0.87 (.10)	0.91 (.52)
Health care denied foregone to cost	1.21 (.02)	0.87 (.35)	0.72 (.09)	0.76 (.01)	0.49 (.04)
Health care utilization
Frequent emergency department visits, past 12 months	1.18 (.23)	0.63 (.17)	0.62 (.04)	0.93 (.59)	0.67 (.11)
Frequent overnight hospital admission, past 12 months	1.08 (.75)	1.89 (.10)	2.22 (.29)	1.97 (.01)	1.35 (.46)
Frequent office visits, past 12 months	0.89 (.02)	0.89 (.11)	0.96 (.75)	1.02 (.69)	1.04 (.61)
Any functional limitation, all conditions	1.01 (.79)	0.80 (.00)	0.67 (.00)	0.81 (.00)	0.63 (.00)
Self-reported health status	0.86 (.01)	0.99 (.99)	1.88 (.00)	0.97 (.65)	1.54 (.00)

Non-Hispanic whites were much less likely to acknowledge the importance of shared provider identity (OR 0.53, P < .05) but had greater odds of seeing a concordant provider (OR 2.23, P < .05) than other racial/ethnic groups. There were no significant differences in care experience between racial groups on other patient-centered outcomes.

Discussion

It is well established that Medicaid enrollees’ are less likely to have access to high-quality care relative to Medicare or private insurance (20,21). While some studies report no significant effects of concordance in race and the quality of patient–physician communication, more studies found racial discordance had a negative effect on the quality of communication (22). There is mixed but growing evidence that providers who share racial or sexual identity with their patients provide care of higher quality than nonconcordant providers (23,24). Given this reality, empirical data on Medicaid enrollees’ attitudes toward concordance could yield important insights and policy implications. Our study provides new empirical evidence that Medicaid enrollees are more likely to prefer seeing providers who share some form of identity with them. This finding may reflect the demographic composition of Medicaid, which is strongly skewed toward groups known to have a preference towards seeing concordant providers, such as racial minorities and other disadvantaged groups (12,25). Such preferences may often be less about sharing a specific core identity than a desire to have fulfilling interactions with providers that are respectful, comprehensible, linguistically unchallenging, and imbued with trust, for which a concordant identity may simply act as a rough proxy (26 –29). While these findings support policies that seek to expand access to concordant providers, it is important to note that once we controlled for demographic and other confounders, Medicaid enrollees did not have a lesser chance of seeing a concordant provider compared to Medicare or privately insured enrollees. There are several possible explanations for this finding. First, it is possible that providers from racial minority groups participate in Medicaid disproportionately, limiting race-based discordance between Medicaid minority enrollees and providers. However, there are very little data on the demographic composition of providers who participate in Medicaid. Second, Medicaid enrollees may have powerful informal social networks that help them find providers that they want (26,30,31). State Medicaid administrators have recognized this fact by providing network directories and provider lists and outreach programs designed to match enrollees with providers (32 –34).

Previous literature suggests Medicaid enrollees report that providers often treat them with disrespect and our study confirms these findings (35,36). Perceived lack of respect often stems from perceived discrimination and strongly correlates with suboptimal provider–patient relationships that translate into compromised access to care, poor quality of care, and limited compliance with recommendations (7). Our findings reinforce the need for policy solutions targeted at removing the stigma associated with Medicaid status. Provider education and training protocols that emphasize special care and sensitivity during interactions with Medicaid enrollees, state outreach programs that de-stigmatize utilization of welfare programs, and institutional efforts at dispelling negative stereotypes around the consumption of these benefits may be helpful.

Medicaid enrollees in our study reported that their providers solicited their opinions/beliefs/preferences at rates higher than those with Medicare and private insurance. This is important and reassuring: Centering patients’ opinions/preferences has multiple benefits including fostering a healthy patient–provider relationship, increasing patient satisfaction, improving clinical outcomes, and better compliance (37 –39). Moreover, actively soliciting patients’ preferences to inform diagnostic and therapeutic strategies is a small but key component of the broader push toward shared decision-making (SDM), making it of concern that barely more than half of the respondents reported seeing providers who asked them about their opinions/beliefs. Our finding underscores the continuing need for policies directed at improving SDM across the board, such as revamping training and education protocols for medical professionals, incentivizing SDM through provider reimbursement models and organizational best practices, and fortifying accreditation requirements with clear expectations regarding SDM goals.

Even though Medicaid providers were good at soliciting patients’ preferences in practice encounters, Medicaid enrollees were much less satisfied with their providers’ communication skills than their Medicare or privately insured counterparts. This is unsurprising since racial and linguistic minorities (eg, recent immigrants) face substantial cultural, linguistic, financial, and social barriers in communicating their preferences or even comprehending the basic care recommendations, which may frustrate doctors’ efforts to involve them in care delivery (40,41). Doctors are seldom good at accurately identifying patient preferences, leaving critical gaps between provider perceptions and patient preferences (42,43). Federal and state policies directed toward training and accreditation with particular stress on effective communication of primary care Medicaid providers may be needed. State Medicaid administrators and Medicaid-managed care programs could initiate periodic reviews of doctors’ communication skills and generate and implement quality metrics that incorporate less tangible intercommunication skills in payment reimbursement mechanisms.

A significant strength of our study is that NHIS oversampled minority groups, which are disproportionately likely to use Medicaid benefits, adding precision to our estimates. However, our study has a few limitations. First, since we relied on a special NHIS data supplement that was fielded for just 1 year (2017), this may have introduced some measurement error in our estimates. However, NHIS design is considered the gold standard among large survey programs and provides some reassurance that the topic area can be estimated reliably with only 1 year of data. Second, our study is correlational and cannot be used to infer casual relationships. Third, self-reported data are subject to recall, social desirability, and other biases. Finally, our measure of concordance is derived from a survey item that is too broadly worded to distinguish patient preference for or experience of different types of concordance between providers and patients.

In conclusion, we find that Medicaid enrollees report more mixed experiences with their provider interactions than their counterparts in Medicare or privately insured plans in ways that go beyond the earlier documented issues of limited access to needed care or providers. Medicaid enrollees do get matched to concordant providers at roughly the same rates as other patients, even though Medicaid enrollees are more likely to prefer their providers to be concordant with them on some form of identity. Moreover, once you account for demographic and other confounding factors, Medicaid enrollees report seeing providers who solicit their opinions/beliefs at higher rates than patients on other plans. However, they report strikingly lower odds of seeing providers that communicate their care recommendations in simple understandable terms and distressingly higher perceptions of being disrespected by providers. Federal, state, and local policies and practices directed at improving these facets of patient–provider interaction are sorely needed and should be aimed squarely at Medicaid providers, especially those working in geographic areas and settings with a disproportionate number of racial, gender, cultural, and linguistic minorities.