An Early Look at the Association Between State Medicaid Expansion and Disparities in Cardiovascular Diseases: A Comprehensive Population Health Management Approach

Abstract

Cardiovascular disease (CVD) is one of the most prevalent chronic diseases nationally and disproportionately affects low-income individuals. There are substantial disparities on CVD outcomes that stem from the lack of health insurance among low-income populations. The Affordable Care Act expands Medicaid health insurance to low-income populations, and aims to increase the utilization of health, social, and economic preventive services to reduce health disparities and prevent chronic diseases. The authors analyzed data from the 2014 Behavioral Risk Factor Surveillance System to understand the potential impact of Medicaid expansion on disparities in CVD among low-income populations. Logistic regression models examined the association between CVD self-reported outcomes among low-income adults with incomes at or below 138% of the federal poverty level in states that have chosen to expand Medicaid and those states choosing not to expand, controlling for socioeconomic, demographic, behavioral, social, and health variables that affect CVD. Overall, the results show that adults in Medicaid expansion states have significantly lower odds of experiencing poor heart health compared to those in non-Medicaid expansion states (odds ratio = 0.767, 95% confidence interval 0.667–0.882). Additionally, significant findings were found between the association of CVD and demographic, socioeconomic, health, and health behavioral covariates. Policy makers should consider policies, systems, and interventions that increase access to a comprehensive set of preventive, population health, and socioeconomic services targeting the key determinants of CVD and other outcomes when expanding Medicaid and designing state plans and waivers.

Introduction

Cardiovascular disease (CVD), including stroke, angina, coronary heart disease, and myocardial infarction (heart attack), is the leading cause of death for both men and women in the United States.¹ More than 17.3 million people die from CVD in a given year.² In the United States a new case of heart disease is diagnosed approximately every 42 seconds.² CVD disproportionately affects ethnic minorities including African Americans, Hispanics, and American Indians.¹ Additionally, people with lower incomes and/or less education are disproportionately affected by and are much more likely to develop CVD.³ From 2011 to 2012, CVD and strokes cost the US health care system an estimated $316.6 billion.²

Despite increased efforts over the years to reduce the economic, mortal, and morbid burden of CVD, preventable risk factors remain the leading cause of newly diagnosed cases.⁴ The increased risk of CVD is associated with physical inactivity, smoking/tobacco use, obesity, diabetes mellitus, and lack of health insurance.² Additionally, social determinants of CVD include unemployment, low-income, and low educational attainment.^5
–7

Access to health insurance is a critical focal point for efforts to prevent CVD among low-income populations.⁸ Medicaid is the single largest supplier of health insurance for low-income people in the United States, covering nearly 72 million Americans,⁹ and can serve as an important safety net in CVD prevention.⁸ The Patient Protection and Affordable Care Act (ACA) aims to expand Medicaid coverage to adults younger than 65 years of age with incomes at or below 138% of the federal poverty level (FPL) and increase the number of low-income persons states must cover.¹⁰ However, many states have not expanded Medicaid as a result of a recent US Supreme Court decision that made the ACA's Medicaid expansion optional for states.¹¹

Furthermore, some of the other major goals of the ACA are to increase utilization of preventive services to reduce health disparities and prevent chronic diseases such as heart disease, obesity, and diabetes for Medicaid eligible adults, and to provide enhanced support for community preventive services related to social and economic environments among special populations.^12

–16 As a result of the Supreme Court's ruling and the various policy initiatives sponsored by the US federal government to address CVD, the ability of Medicaid and the ACA to address differences in a range of medical and community services to prevent CVD across groups remains uncertain, especially in low-income areas of the United States.

Aday and Andersen's Behavioral Model of Health Services Use¹⁷ provides a useful framework for this study as it posits that enabling factors such as health insurance and the enactment of laws can remove barriers to care and enable improved access to care. This increased access to care will then lead to improved health outcomes. This framework assesses the impact of improvements in human health resulting from predisposing, need, and enabling characteristics.

Predisposing factors include sociodemographic factors (eg, age, sex, race, ethnicity, education, employment status) that describe the use of health care services and, subsequently, health status. Need factors seek to explain how patients view their own general health or the reason(s) for seeking health care services.

Finally, of vital importance to this study and the basis for this analysis are the enabling factors, which are the community and personal resources an individual has available to obtain the health care services they need to optimize their health. Enabling factors include health insurance, transportation, wait times, regular access to care, income, and enactment of rules or laws. With respect to this study, Medicaid expansion can be viewed as an enabling factor that can influence CVD outcomes because of its ability to serve as an important means of access to health and social care for safety net individuals by affecting demographic, behavioral, social, and health variables.

This study utilized a recent national study–the 2014 Behavioral Risk Factor Surveillance System (BRFSS) survey¹⁸–to examine whether low-income individuals living in Medicaid expansion states have better self-reported CVD health than low-income people living in non-Medicaid expansion states. An understanding of the variation of self-reported CVD rates and associated determinants across states currently expanding Medicaid and those not expanding Medicaid based on this national data will contribute new knowledge to the population health literature and will inform policy discussions concerning health care reform.

Methods

Study design and variables

This is a cross-sectional study that focused on a sample of adults between the ages of 18 and 64 years with incomes at or below 138% of the FPL who responded to the 2014 BRFSS survey. The BRFSS does not provide income as a percentage of FPL. Therefore, the authors calculated the respondent's median household income in the reported category and applied 2014 Census Bureau federal poverty guidelines for household size. The FPL for a household size of 1 in 2014 was $11,670, and $23,850 for a family of 4. A total of 50,713 low-income adults were included in the analysis. Respondents who answered “don't know/not sure” or did not supply information for income and household size variables were excluded.

The BRFSS survey contains 3 items of particular interest to this study related to CVD. The prologue to each of the 3 questions asked, “Has a doctor, nurse, or other health professional ever told you that you had any of the following?” The first item asked, “(Ever told) you that you had a heart attack, also called a myocardial infarction?” The next item asked, “(Ever told) you that you had angina or coronary heart disease?” Lastly, the BRFSS asked, “(Ever told) you that you had a stroke?” There were 4 possible responses: yes, no, don't know/not sure, and refused. Respondents who refused to answer the question or answered don't know/not sure were excluded. As a result, 3 dependent variables were coded dichotomously (yes or no) for analysis: heart attack, angina or coronary heart disease (CHD), and stroke.

A summary variable–poor heart health–was created to assess the overall impact of CVD on Medicaid expansion and confounding variables. Poor heart health was coded dichotomously (yes or no). Respondents who answered “yes” to ever being diagnosed with a stroke, heart attack, or angina or CHD, or any combinations of the diagnoses were coded as “yes.” All other respondents who were not diagnosed with a stroke, heart attack, or angina or CHD, or any combinations of the diagnoses were coded as “no.”

The predictor variable of interest was an indicator of residence in a Medicaid expansion state or non–Medicaid expansion state on CVD outcomes. Medicaid expansion states were determined using data from the Kaiser Family Foundation.¹⁹ By 2014, 27 states including the District of Columbia had expanded Medicaid, and 24 states had not. All but 2 of the 27 states, Michigan (April 1, 2014) and New Hampshire (August 15, 2014), had expanded Medicaid as of January 1, 2014. Because Michigan and New Hampshire did not expand Medicaid until later in 2014, this analysis included Michigan and New Hampshire as non–Medicaid expansion states. In summary, 25 states were included in the predictor group and 26 states were included in the reference group. Confounding demographic, behavioral, social, and health variables and their measures are listed in Table 1.

Table 1.

Descriptive Statistics and Bivariate Associations Between Demographic, Behavioral, Social, and Health Variables and Heart Health-Related Characteristics for Adults with Incomes at or Below 138% of the Federal Poverty Level

Variable name	N (Wt%)	Expansion states 23,429 (50.4), Wt%	Non-expansion states 27,284 (49.6), Wt%	P
Poor heart health				<0.0001
Yes	5387 (7.9)	3.45	4.42
No	45,326 (92.1)	46.95	45.19
Stroke				<0.001
Yes	2352 (3.5)	1.50	1.95
No	48,361 (96.5)	48.90	47.65
Heart attack				<0.01
Yes	2679 (3.9)	1.76	2.15
No	48,034 (96.1)	48.64	47.45
Angina or coronary heart disease				<0.0001
Yes	2487 (3.5)	1.49	2.02
No	48,226 (96.5)	48.91	47.58
Age				0.824
18–24	6124 (19.3)	9.93	9.35
25–34	10,191 (26.2)	13.13	13.06
35–44	9918 (20.7)	10.39	10.30
45–54	11,499 (18.5)	9.27	9.26
55–64	12,981 (15.3)	7.68	7.63
Sex				<0.0001
Male	19,080 (43.6)	22.84	20.80
Female	31,633 (56.4)	27.56	28.80
Race				<0.0001
White	29,134 (43.0)	20.05	22.96
Black	7318 (17.9)	6.31	11.55
Hispanic	8985 (30.9)	18.53	12.38
Multiracial^a	2218 (2.4)	1.33	1.08
Other race^b	3058 (5.8)	4.17	1.63
Education				<0.0001
Did not graduate high school	9280 (29.5)	15.47	14.03
Graduated high school	19,060 (33.3)	15.84	17.47
Attended college or technical school	15,117 (28.9)	14.62	14.29
Graduated college or technical school	7256 (8.3)	4.46	3.81
Employment				0.361
Employed	22,241 (48.5)	24.64	23.87
Unemployed	28,472 (51.5)	25.76	25.74
Health care coverage				<0.0001
Yes	37,102 (67.5)	37.24	30.24
No	13,611 (32.5)	13.16	19.36
Diabetes				<0.01
Yes	6968 (10.7)	5.27	5.42
No	42,021 (85.8)	43.07	42.75
No, prediabetes or borderline diabetes	936 (1.7)	1.01	0.73
Overweight or obese				<0.0001
Yes	32,585 (66.2)	32.54	33.64
No	15,123 (33.8)	17.97	15.84
Smoking status				<0.0001
Current smoker–smokes every day	11,462 (19.7)	8.90	10.79
Current smoker–smokes some days	4529 (8.5)	3.86	4.66
Former smoker	9977 (16.5)	8.27	8.24
Never smoked	24,745 (55.3)	29.37	25.91
Exercise				<0.0001
Yes	33,876 (67.7)	35.23	32.47
No	16,837 (32.3)	15.17	17.13
Delay getting medical care				<0.05
Yes, delay getting medical care	11,699 (33.5)	18.96	14.56
No, delay getting medical care/did not need care	23,254 (66.5)	35.94	30.53

Data are from the Behavioral Risk Factor Surveillance System survey 2014.¹⁸

Multiracial includes not of Hispanic, Latino/a, or Spanish origin and don't know/not sure.

Other race includes Asian, American Indian, Alaskan Native, Native Hawaiian, or Other Pacific Islander.

Wt, weighted percentage.

Statistical analysis

Demographic characteristics were reported for Medicaid expansion and non–expansion states separately, and these differences were tested using Wald chi-square tests. Separate binary logistic regression was performed for each dependent variable of interest–poor heart health, stroke, angina or CHD, and heart attack. The regression models estimated the self-reported CVD outcomes as a function of state Medicaid expansion status, controlling for age, race/ethnicity, and the other demographic, behavioral, social, and health variables. All analysis was computed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Observations with missing values on the dependent variable and any independent variable were excluded as the reasons for missing were not certain and the percentages of missing were not appropriate for imputations. Missing covariates included the variables overweight or obese (5.9%) and delay getting medical care (31.1%). In order to avoid bias and adjust for the complex sampling design of BRFSS, PROC SURVEYLOGISTIC was used to conduct logistic regression. PROC SURVEYFREQ was used to produce frequencies and percentages. All models used BRFSS raking weighting methodology.²⁰

Models were estimated for all individuals younger than 65 years of age with incomes at or below 138% FPL. Odds ratio (ORs) and their 95% confidence intervals (CIs) were reported to estimate the impact of variables. Predictor variables with a P < .05 were considered statistically significant.

Results

Table 1 presents descriptive statistics and bivariate associations for adults living in Medicaid expansion and non–expansion states with incomes at or below 138% of the FPL across all dependent and predictor variables.

As seen in Table 1, low-income adults living in Medicaid expansion states were less likely to report CVD across all dependent variables compared to low-income adults in non–expansion states. Specifically, this study found that low-income adults living in Medicaid expansion states were less likely to report poor heart health compared to low-income adults in non–expansion states (3.45% vs. 4.42%; χ² = 26, P < 0.0001).

This study found that more than 90% of low-income respondents had not graduated from college or technical school. More than half of the 50,713 respondents included in the analyses were unemployed. Roughly one third of the respondents had some form of delay in receiving medical care in the past 12 months. Nearly one third of respondents had not exercised or participated in physical activity in the past 30 days. Most respondents reported a health risk factor, with almost two thirds indicating they were overweight or obese, and approximately 11% diagnosed with diabetes. Just less than half of the respondents reported smoking at some point in their life.

Table 2 presents binary logistic regression analyses results for regression coefficients and ORs to determine the association between poor heart health and Medicaid expansion status, when controlling for demographic, behavioral, social, and health variables.

Table 2.

Results from Binary Logistic Regression Analyses Showing the Effect of Medicaid Expansion Status on Poor Heart Health, Controlling for Confounding Variables

	Poor heart health
Variable	B	OR
Medicaid expansion status
Expansion state	−0.13^***	0.767 (0.667–0.882)
Age group
18–24	−1.34^***	0.476 (0.271–0.838)
35–44	−0.02	1.786 (1.329–2.400)
45–54	0.81^***	4.082 (3.157–5.278)
55–64	1.16^***	5.828 (4.535–7.490)
Sex
Female	0.16^***	0.732 (0.640–0.838)
Race
Black	−0.00	1.065 (0.901–1.257)
Hispanic	−0.28^**	0.804 (0.622–1.039)
Multiracial^a	0.39^**	1.579 (1.127–2.211)
Other race^b	−0.05	1.017 (0.715–1.447)
Education
Did not graduate high school	0.17^**	1.352 (1.075–1.700)
Graduated high school	−0.01	1.124 (0.911–1.387)
Attended college or technical school	−0.04	1.093 (0.874–1.367)
Employment
Employed	−0.44^***	0.415 (0.352–0.489)
Health care coverage
No	−0.17^***	0.717 (0.610–0.842)
Diabetes
Yes	0.52^***	2.192 (1.868–2.571)
No, prediabetes or borderline diabetes	0.02	1.331 (0.903–1.963)
Overweight or obese
Yes	0.07	1.143 (0.975–1.339)
Smoking status
Current smoker–smokes every day	0.14^*	1.752 (1.463–2.098)
Current smoker–smokes some days	0.20^**	1.867 (1.488–2.341)
Former smoker	0.08	1.648 (1.384–1.962)
Exercise
No	0.14^***	1.330 (1.167–1.515)
Delay getting medical care
Yes, delay getting medical care	0.21^***	1.515 (1.325–1.732)

Data are from the Behavioral Risk Factor Surveillance System survey 2014.¹⁸

Multiracial includes not of Hispanic, Latino/a, or Spanish origin and don't know/not sure.

Other race includes Asian, American Indian, Alaskan Native, Native Hawaiian, or Other Pacific Islander.

P < 0.05.

P < 0.01.

***

P < 0.001.

B, estimated ordered log-odds (logit) regression coefficients; OR, odds ratio.

Results in Table 2 show that when controlling for confounding variables, low-income individuals in Medicaid expansion states were 23% significantly less likely to have poor heart health compared to low-income individuals in non–Medicaid expansion states (OR = 0.767, 95% CI = 0.667, 0.882). This finding suggests that Medicaid expansion is associated with a lower likelihood that a person will have poor heart health, given the other variables in the model are held constant.

Additionally, the association between demographic variables and poor heart health also is important to report. The odds of experiencing poor heart health significantly increases with age. Adults ages 45–54 were 4.1 times as likely to have poor heart health compared to adults ages 25–34. In addition, adults age 55–64 have 5.8 times the rate of poor heart health as those in the 25–34 age group. These results for age were statistically significant. Multiracial individuals were approximately 1.58 times significantly more likely to have poor heart health compared to white low-income individuals.

Low-income women were approximately 27% significantly less likely to experience poor heart health compared to low-income males. Socioeconomic factors had an impact on poor heart health with individuals who did not graduate from high school being 35% significantly more likely to have poor heart health compared to low-income individuals who graduated from college or technical school. Respondents who were employed were significantly less likely to experience poor heart health (OR = 0.415, 95% CI = 0.352, 0.489) compared to unemployed low-income adults.

Furthermore, comorbidities had an impact on CVD, with low-income individuals diagnosed with diabetes 2.19 times significantly more likely to be diagnosed with poor heart health. Also, risky health behaviors remain an important determinant of CVD, with individuals who smoke every day 75% significantly more likely to have poor heart health compared to individuals who never smoked.

The lack of exercise or physical activity within the past 30 days had a consistent impact on poor heart health with low-income adults 33% significantly more likely to experience poor heart health compared to individuals who exercised or engaged in physical activity within the past 30 days. Lastly, the OR for those who reported they had a delay in receiving medical care in the past 12 months is approximately 52% significantly higher than those who did not have a delay in the past 12 months.

Table 3 presents results for binary logistic regression analyses including regression coefficients and ORs to determine the association between stroke, heart attack, and angina or CHD and Medicaid expansion status, and other confounding variables, respectively.

Table 3.

Results from Binary Logistic Regression Analyses Showing the Effect of Medicaid Expansion Status on Stroke, Heart Attack, and Coronary Heart Disease, Controlling for Confounding Variables

	Stroke		Heart attack		Angina or CHD
Variable	B	OR	B	OR	B	OR
Medicaid expansion status
Expansion state	−0.13^*	0.773 (0.635–0.942)	−0.06	0.880 (0.731–1.060)	−0.15^**	0.735 (0.606–0.892)
Age group
18–24	−1.05^**	0.547 (0.255–1.172)	−1.78^***	0.385 (0.140–1.055)	−1.15^***	0.641 (0.262–1.566)
35–44	−0.19	1.294 (0.857–1.952)	0.21	2.811 (1.801–4.388)	−0.04	(1.941 (1.152–3.269)
45–54	0.74^***	3.281 (2.292–4.698)	0.97^***	6.007 (4.086–8.830)	0.74^***	4.209 (2.637–6.717)
55–64	0.95^***	4.041 (2.851–5.727)	1.41^***	9.312 (6.413–13.520)	1.15^***	6.344 (4.005–10.049)
Sex
Female	−0.03	0.943 (0.782–1.137)	−0.27^***	0.583 (0.485–0.701)	−0.21^***	0.660 (0.549–0.795)
Race
Black	0.16	1.474 (1.174–1.851)	−0.14	0.958 (0.762–1.206)	−0.18	0.783 (0.621–0.988)
Hispanic	−0.34^*	0.888 (0.601–1.314)	−0.38^**	0.750 (0.523–1.076)	−0.13	0.826 (0.581–1.173)
Multiracial^a	0.38^*	1.827 (1.164–2.865)	0.43^**	1.689 (1.126–2.533)	0.07	1.002 (0.657–1.530)
Other race^b	0.03	1.291 (0.800–2.084)	0.18	1.314 (0.828–2.084)	0.18	1.120 (0.667–1.878)
Education
Did not graduate high school	0.09	1.101 (0.805–1.508)	0.26^**	1.666 (1.206–2.300)	0.13	1.279 (0.917–1.786)
Graduated high school	0.02	1.023 (0.767–1.363)	−0.05	1.226 (0.910–1.652)	0.00	1.129 (0.826–1.543)
Attended college or technical school	−0.10	0.907 (0.677–1.214)	0.05	1.351 (0.979–1.865)	−0.02	1.107 (0.804–1.524)
Employment
Employed	−0.51^***	0.364 (0.280–0.473)	−0.35^***	0.501 (0.397–0.631)	−0.45^***	0.411 (0.321–0.524)
Health care coverage
No	−0.21^**	0.663 (0.524–0.839)	−0.12^*	0.791 (0.638–0.982)	−0.14^*	0.751 (0.596–0.947)
Diabetes
Yes	0.39^**	1.930 (1.537–2.423)	0.64^***	2.073 (1.673–2.570)	0.55^**	2.461 (1.984–3.053)
No, prediabetes or borderline diabetes	0.12	1.462 (0.885–2.415)	0.20	1.335 (0.784–2.273)	0.08	1.545 (0.904–2.639)
Overweight or obese
Yes	0.01	1.024 (0.828–1.268)	0.05	1.098 (0.881–1.368)	0.14^*	1.334 (1.062–1.676)
Smoking status
Current smoker–smokes every day	0.19^*	1.775 (1.375–2.292)	0.21^**	2.105 (1.629–2.720)	0.12	1.736 (1.356–2.223)
Current smoker–smokes some days	0.23^*	1.841 (1.346–2.519)	0.19^*	2.072 (1.532–2.801)	0.17	1.829 (1.347–2.483)
Former smoker	−0.05	1.396 (1.162–1.676)	0.14	1.971 (1.528–2.542)	0.13	1.759 (1.396–2.215)
Exercise
No	0.17^**	1.396 (1.082–1.802)	0.11^*	1.243 (1.040–1.486)	0.12^*	1.270 (1.054–1.529)
Delay getting medical care
Yes, delay getting medical care	0.17^**	1.403 (1.168–1.684)	0.20^***	1.495 (1.247–1.791)	0.16^**	1.390 (1.153–1.676)

Data are from the Behavioral Risk Factor Surveillance System survey 2014.¹⁸

Multiracial includes not of Hispanic, Latino/a, or Spanish origin and don't know/not sure.

Other race includes Asian, American Indian, Alaskan Native, Native Hawaiian, or Other Pacific Islander.

P < 0.05.

P < 0.01.

***

P < 0.001.

B, estimated ordered log-odds (logit) regression coefficients; CHD, coronary heart disease; OR, odds ratio.

Table 3 regression analysis reveals that when controlling for confounding variables, low-income adults in Medicaid expansion states were nearly 23% less likely to be diagnosed with a stroke compared to low-income adults in non–expansion states. Hence, this analysis suggests that Medicaid expansion is associated with a lower likelihood of stroke, given the other variables in the model are held constant. This result was statistically significant. Additionally, there was no difference in the odds of heart attack in adults in Medicaid expansion states versus non–Medicaid expansion states (OR = 0.880, 95% CI = 0.731, 1.060), when controlling for confounding variables.

Finally, results show that low-income adults in Medicaid expansion states were roughly 27% less likely to be diagnosed with angina or CHD compared to low-income adults in non–expansion states when controlling for confounding variables. Thus, Medicaid expansion is associated with a lower likelihood that a person will be diagnosed with angina or CHD given the other variables in the model are held constant. This result was statistically significant.

Discussion

The major goals of the ACA with respect to this study are to expand and increase Medicaid health insurance to low-income populations; to increase the utilization of preventive services to reduce health disparities and prevent chronic diseases such as heart disease, obesity, and diabetes for Medicaid eligible adults; and to provide enhanced support for community preventive services related to social and economic environments among special populations.^{10,12

–16} This is the first study to the authors' knowledge to rely on nationally representative federal survey data to evaluate the relationship of the ACA Medicaid expansions on CVD.

The results of this study show low-income adults living in Medicaid expansion states have significantly lower odds of poor heart health, specifically stroke and angina or CHD. In contrast, although individuals living in Medicaid expansion states were less likely to be diagnosed with a heart attack because the CI contains the null hypothesis of 1.0, the results are consistent with the hypothesis of no association, and these results were not statistically significant. Yet, one cannot exclude the possibility of small changes in the odds of heart attack in Medicaid expansion states versus non-Medicaid expansion states.

The finding modeling the association of Medicaid expansion on CVD, specifically relating to poor heart health, stroke, and angina or CHD, is consistent with previous studies that used simulation models to estimate reductions in the incidence of CVD diagnoses resulting from ACA health insurance expansions.⁸ However, the OR is not statistically significant in the model depicting the relationship between heart attack and Medicaid expansion.

In addition, significant but decreasing evidence of self-reported CVD outcomes and health insurance status among low-income adults was found. One potential explanation is that, as a result of an increase in CVD risk and diagnoses, the need for health insurance prior to or after a major CVD event may be related to one's motivation, attitude, and knowledge, or other social and demographic factors.^21,22

Moreover, socioeconomic and demographic factors had a consistent statistically significant effect on the odds of CVD across all dependent variables, consistent with previous literature.^2,23 Specifically, across all CVD outcome measures, the odds of experiencing a CVD event increased with age. There are several possible reasons for this finding, one being that with all other variables held constant, as age increases the decline in the natural physiologic changes of the heart and blood vessels results in declining health and to some degree reduced cardiovascular function.²⁴ Further significant demographic results emerged across some models with women less likely to experience a heart attack or angina or CHD event compared to men. This finding underscores the importance of sex-specific care to ensure primary CVD prevention is established with Medicaid recipients.

Statistically significant socioeconomic results emerged from this analysis across all outcome categories. Low-income individuals who are employed were significantly less likely to suffer poor heart health, angina or CHD, heart attack, or stroke. Numerous studies have found associations of lower CVD risk among individuals who are employed,^25,26 which suggests that initiatives to expand employment opportunities among low-income individuals is warranted.

Consistent with prior literature,^3,25 this study also found significant evidence between the association of CVD and lower levels of education. Specifically, low-income adults who did not graduate from high school were more likely to suffer poor heart health or myocardial infarction. This suggests that differences in Medicaid coverage across states may widen CVD disparities among low-income adults if efforts to provide supportive adult education programs are not taken into consideration. Additionally, states deciding to expand Medicaid could benefit from improved CVD outcomes if they offered programs that support job training, job readiness, and job placement in conjunction with Medicaid.

In this model of CVD and Medicaid expansion, consistent with prior literature,^27
–29 evidence was found of the association between CVD and several health and health behavior variables–namely obesity, diabetes, and smoking. Notably, a strong association was found between being overweight or obese and angina or CHD. It is likely that the percentage of missing observations in BRFSS for this covariate altered its relationship with the independent variables, or attenuated the magnitude of the associations.

A chi-square test of independence was calculated comparing the frequency of heart disease (poor heart health, stroke, heart attack, and angina or CHD) in non–missing and missing samples for the variable overweight or obese. A significant interaction (P < .05) was found for each test. Therefore, one can conclude that the reporting of overweight or obesity status significantly influences heart disease outcomes.

Furthermore, a chi-square test of independence was calculated comparing the frequency of Medicaid expansion status in non–missing and missing samples for the variable overweight or obese. No significant relationship was found (P = 0.102). Steps were taken in the analysis to control for these effects. Nevertheless, this study's finding suggests that any efforts to expand Medicaid should include utilization and investment of appropriate or necessary programs and resources to reduce obesity among low-income individuals.

Another important finding in this study is the strong and consistent association between diabetes and CVD across all models. The magnitude of the effect of Medicaid expansion on CVD must take into consideration the impact diabetes has on stroke, myocardial infarction, and angina or CHD. Policies that support evidenced-based diabetes prevention and management interventions are needed to reduce the proliferation of diabetes-associated CVD among low-income individuals.

Additionally, low-income adults who currently smoke, consistently or occasionally, have higher reported rates of poor heart health, specifically stroke and heart attack. Understanding health behavior factors associated with CVD, such as smoking, is vital to designing policy strategies aimed at reducing heart disease among the Medicaid population.

The findings related to the association between CVD and being overweight or obese, having diabetes, and smoking must spur increased efforts to detect chronic diseases and reduce poor health behaviors for any efforts to expand Medicaid to have important implications for population health and health care spending.

Consistent and strong evidence was observed in the relationship between access to care and CVD outcomes among low-income adults. The question related to access to care asked respondents “Have you delayed getting needed medical care…in the past 12 months.” Because this analysis assessed respondents' level of access to care in the first year of Medicaid expansion (2014) and data collection took place throughout 2014, it is likely that these results include some months before Medicaid expansion, which may weaken the results, especially if respondents did not sign up for Medicaid upon eligibility.

Additionally, it is likely that the high percentage of missing observations in BRFSS for this covariate altered its relationship with the independent variables, or attenuated the magnitude of the associations. A chi-square test of independence was calculated comparing the frequency of heart disease (poor heart health, stroke, heart attack, and angina or CHD) in non–missing and missing samples for the variable medical care delay. A nonsignificant interaction (P > .05) was found for each test. Therefore, one cannot conclude that the reporting of medical care delay status significantly influences heart disease outcomes.

Furthermore, a chi-square test of independence was calculated comparing the frequency of Medicaid expansion status in non–missing and missing samples for the variable medical care delay. A significant relationship was found (P < 0.001). Therefore, one can conclude that the reporting of medical care delay status significantly influences Medicaid expansion status. Steps were taken in the analysis to control for these effects. Nevertheless, significant results emerged from this examination of the relationship between access to care, which included variables such as lack of transportation, and poor heart health, stroke, heart attack, and angina or CHD, controlling for confounding variables.

Access to transportation services has long been one of the major barriers faced by low-income populations in accessing medical care.³⁰ Although nonemergency medical transportation (NEMT) is a benefit provided to Medicaid beneficiaries to access medical services, states have great latitude in making such services available. In an effort to control utilization of services, many states set limits on travel distance, require co-payments, require prior approval, or have limits on the number of service days per month.³¹ Additionally, it is not clear whether enrollees are aware of such transportation services and this benefit for increasing access to medical care. Previous studies have shown that the association between NEMT and the ability to pay is a barrier faced by Medicaid beneficiaries in accessing medical services.³²

Despite previous studies providing evidence that increased access to transportation services improves access to disease management for chronic diseases such as heart disease and diabetes,³³ the present study results show that low-income individuals are still having challenges with accessing medical care for reasons such as lack of transportation. In order to improve CVD outcomes, states that have already expanded Medicaid must ensure that they are aggressively educating beneficiaries about the availability of NEMT and taking steps to address access challenges. Additionally, states that have not expanded Medicaid must be certain to offer broad services to ensure access to needed medical care while limiting the utilization control measures.

Limitations

There are several limitations to this study. First, CVD is a progressive disease that evolves over years as a result of damage to the heart and blood vessels; using data from the first 12 months after the ACA Medicaid expansion to examine its early effects could cause the estimates to differ from longer term effects.

Another potential limitation inherent in the BRFSS survey data is that survey results are based on self-reports and not validated with medical records, which makes them vulnerable to recall bias. Because of the nature of the sampling method inherent to the BRFSS survey, this study also is prone to sampling bias although efforts were taken to adjust for this bias. Nonresponse bias is also a concern in this study because of the inability of BFRSS to obtain data for all questionnaire items from a respondent. BRFSS survey weights were employed to reduce nonresponse bias.

Additionally, given the cross-sectional design of the BRFSS survey data, these results do not allow one to infer a causal relationship between heart health and Medicaid expansion. Lastly, the validity of the results is also limited because of the quality of measurements used in the telephone survey, which can be affected by the structure of the questions used, coding and data entry errors, the length of the interview, the ergonomics of the respondent's environment when they answered survey questions, and the technique of the interviewer.

Despite these limitations, this study contributes to the literature concerning the factors predicting heart disease and provides unique and new evidence that ACA Medicaid expansion is associated with lower CVD outcomes among low-income adults. The results of this study must be interpreted bearing in mind these limitations.

Conclusions

This is the first study to the authors' knowledge that demonstrates the relationship between enabling factors such as Medicaid expansion (health insurance) and the ACA (enactment of laws) on lower odds of CVD, specifically stroke and angina or CHD.

This study further demonstrates the independent contribution of social determinants to CVD and, thus, the increasing contribution of social disparities to poor heart health. When controlling for confounding health and health behavior variables, the risks associated with obesity, diabetes, smoking, and lack of exercise are evident through the findings of this study.

Many state Medicaid programs have no or very limited programs that cover evidence-based chronic disease prevention services.³⁴ Additionally, many state Medicaid programs have no or very limited programs that provide access to services that improve social and economic factors that remove barriers to shaping healthy behaviors that promote chronic disease prevention.³⁵ Despite continued research that demonstrates the need for community-based prevention initiatives to address the impact of social determinants on health outcomes,³⁶ it remains a challenge for such initiatives to be included in the Medicaid framework for most states.³⁵

As such, this study provides useful evidence to support the policy assertion that Medicaid expansion has the potential to improve health outcomes, namely CVD, after controlling for confounding variables. This study also provides evidence that in order to address the unique needs of Medicaid beneficiaries, any effort to expand Medicaid must be amalgamated with policies that increase access to preventive and social care services, such as CVD management programs that promote management of chronic diseases such as diabetes and reduced tobacco use, employment and training programs, and access and linkage to transportation services. Although undecided states remain likely to expand Medicaid, some states are doing so at the expense of needed health and social care services.³⁷

Results from this study demonstrate that in order for Medicaid expansion and the ACA to accomplish the goals of improved CVD health and quality of care for low-income adults, policy makers must account for additional factors related to improved heart health, such as access to comprehensive health and socioeconomic supportive services.

States that have expanded Medicaid to date should conduct a thorough analysis of their existing state Medicaid plan and waivers to ensure that their programs offer a comprehensive set of health and social care services that address the range of health disparities and social determinants of health that affect underserved and vulnerable populations. States considering expanding Medicaid should ensure that their state plan includes access to a broad range of preventive and social care services that address chronic disease prevention and access to health-related supportive services that address the social determinants of health.

Using a national study that incorporates divergent populations across diverse states to conclude the positive effects of access to health insurance on certain types of heart disease makes this study credible and allows for a range of conclusions to be drawn.

Footnotes

Author Disclosure Statement

Mr. Rogers and Dr. Zhang declared no conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received no financial support for this article.

References

Centers for Disease Control and Prevention. Heart disease facts. 2015. www.cdc.gov/heartdisease/facts.htm Accessed August 4, 2016 .

Writing Group

Members

, Mozaffarian

, Benjamin

, et al. Heart disease and stroke statistics-2016 update: a report from the American Heart Association. Circulation, 2016; 133:e38–e60.

Franks

, Winters

, Tancredi

, Fiscella

. Do changes in traditional coronary health disease risk factors over time explain the association between socio-economic status and coronary heart disease?. BMC Cardiovasc Disord, 2011; 11:28.

World Health Organization. Cardiovascular diseases (CVDs). 2016. www.who.int/mediacentre/factsheets/fs317/en Accessed August 4, 2016 .

Noelke

, Avendano

. Who suffers during recessions? Economic downturns, job loss, and cardiovascular disease in older Americans. Am J Epidemiol, 2015; 182:873–882.

Levenson

, Skerrett

, Gaziano

. Reducing the global burden of cardiovascular disease: the role of risk factors. Prev Cardiol, 2002; 5:188–199.

Kilander

, Berglund

, Boberg

, Vessby

, Lithell

. Education, lifestyle factors and mortality from cardiovascular disease and cancer. A 25-year follow-up of Swedish 50-year-old men. Int J Epidemiol, 2001; 30:1119–1126.

, Bruen

, Lantz

, Mendez

. Impact of health insurance expansions on nonelderly adults with hypertension. Prev Chronic Dis, 2015; 12:150111.

Centers for Medicare & Medicaid Services. Medicaid & CHIP: May 2016 Monthly Applications, Eligibility Determination and Enrollment Report. 2016. www.medicaid.gov/medicaid/program-information/downloads/may-2016-enrollment-report.pdf Accessed August 4, 2016 .

10.

Patient Protection and Affordable Care Act. Title II–Role of public health programs. Section 2001 Medicaid coverage for the lowest income populations. Pub. L. No. 111–148, 42 USC 18001, 2010.

11.

National Federation of Independent Business et al. v. Sebelius, Secretary of Health and Human Services, et al. 2012; 11–393. www.supremecourt.gov/opinions/11pdf/11-393c3a2.pdf Accessed March 5, 2016 .

12.

Patient Protection and Affordable Care Act. Title II–Role of public health programs. Section 2703, State option to provide health homes for enrollees with chronic conditions. Pub. L. No. 111–148, 42 USC 18001, 2010.

13.

Patient Protection and Affordable Care Act. Title IV–Prevention of chronic disease and improving public health. Section 4004, Education and outreach campaign regarding preventive benefits. Pub. L. No. 111–148, 42 USC 18001, 2010.

14.

Patient Protection and Affordable Care Act. Title IV–Prevention of chronic disease and improving public health. Section 4106, Improving access to preventive services for eligible adults in Medicaid. Pub. L. No. 111–148, 42 USC 18001, 2010.

15.

Patient Protection and Affordable Care Act. Title IV–Prevention of chronic disease and improving public health. Section 4108, Incentives for prevention of chronic diseases in Medicaid. Pub. L. No. 111–148, 42 USC 18001, 2010.

16.

Patient Protection and Affordable Care Act. Title IV–Prevention of chronic disease and improving public health. Section 399 U, Community Preventive Services Task Force. Pub. L. No. 111–148, 42 USC 18001, 2010.

17.

Aday

, Andersen

. A framework for the study of access to medical care. Health Serv Res, 1974; 9:208–220.

18.

Centers for Disease Control and Prevention. 2014 BRFSS Survey Data and Documentation. 2015. www.cdc.gov/brfss/annual_data/annual_2014.html Accessed January 5, 2016 .

19.

Kaiser Family Foundation. Status of State Action on the Medicaid Expansion Decision. http://kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act Accessed March 5, 2016 .

20.

Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System. Weighting BRFSS data. 2014. www.cdc.gov/brfss/annual_data/2014/pdf/weighting-data.pdf Accessed August 4, 2016 .

21.

Remler

, Glied

. What other programs can teach us: increasing participation in health insurance programs. Am J Public Health, 2003; 93:67–74.

22.

Diehr

, Madden

, Cheadle

, Martin

, Patrick

, Skillman

. Will uninsured people volunteer for voluntary health insurance? Experience from Washington State. Am J Public Health, 1996; 86:529–532.

23.

Karlamangla

, Merkin

, Crimmins

, Seeman

. Socioeconomic and ethnic disparities in cardiovascular risk in the United States, 2001–2006. Ann Epidemiol, 2010; 20:617–628.

24.

Young

. Heart disease in the elderly. In: Zaret

, Moser

, Cohen

, eds. Yale University School of Medicine Heart Book. New York, NY: William Morrow and Company, 1992:263.

25.

Kaplan

, Keil

. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation, 1993; 88:1973–1998.

26.

Luckhaupt

, Calvert

, Centers for Disease Control and Prevention. Prevalence of coronary heart disease or stroke among workers aged <55 years—United States, 2008–2012. Morb Mortal Wkly Rep, 2014; 63:645–649.

27.

Bhupathiraju

, Hu

. Epidemiology of obesity and diabetes and their cardiovascular complications. Circ Res, 2016; 118:1723–1735.

28.

Rosenberg

, Palmer

, Rao

, Adams-Campbell

. Risk factors for coronary heart disease in African American women. Am J Epidemiol, 1999; 150:904–909.

29.

Ockene

, Miller

. Cigarette smoking, cardiovascular disease, and stroke: a statement for healthcare professionals from the American Heart Association. American Heart Association Task Force on Risk Reduction. Circulation, 1997; 96:3243–3247.

30.

Syed

, Gerber

, Sharp

. Traveling towards disease: transportation barriers to health care access. J Community Health, 2013; 38:976–993.

31.

Kaiser Family Foundation. Medicaid benefits: Non-emergency medical transportation services. 2012. http://kff.org/medicaid/state-indicator/non-emergency-medical-transportation-services Accessed September 30, 2016 .

32.

Bentler

, Damiano

, Momany

, Ingleshwar

, Robinson

. First look at Iowa's Medicaid expansion: how well did members transition to the Iowa Health and Wellness Plan from Iowa Care. 2014. http://ir.uiowa.edu/ppc_health/100 Accessed September 30, 2016 .

33.

Hughes-Cromwick

, Wallace

, Mull

, Bologna

. Cost benefit analysis of providing non-emergency medical transportation. 2005. http://altarum.org/sites/default/files/uploaded-publication-files/05_project_report_hsd_cost_benefit_analysis.pdf Accessed September 30, 2016 .

34.

Gates

, Ranji

, Snyder

. Coverage of preventive services for adults in Medicaid. 2014. http://files.kff.org/attachment/coverage-of-preventive-services-for-adults-in-medicaid-issue-brief Accessed September 30, 2016 .

35.

Crawford

, Houston

. State payment and financing models to promote health and social service integration. 2015. www.chcs.org/media/Medicaid_-Soc-Service-Financing_022515_2_Final.pdf Accessed September 30, 2016 .

36.

Williams

, Costa

, Odunlami

, Mohammed

. Moving upstream: how interventions that address the social determinants of health can improve health and reduce disparities. J Public Health Manag Pract, 2008; 14:S8–S18.

37.

Grant

. In Kentucky's new Medicaid plan evidence takes a back seat. 2016. http://healthaffairs.org/blog/2016/08/25/in-kentuckys-new-medicaid-plan-evidence-takes-a-back-seat Accessed September 30, 2016 .