The Impact of Medicaid Expansion on Black-White Disparities in Cardiovascular Disease Mortality

Abstract

Cardiovascular disease (CVD) is a leading cause of mortality among U.S. adults, especially low-income and uninsured adults. Non-Hispanic Black adults, who are overrepresented among low-income and uninsured populations, are disproportionately burdened by CVD mortality compared with non-Hispanic White adults. Medicaid expansion is associated with improved insurance coverage and access to care among low-income adults as well as reduced CVD mortality. It is unclear whether Medicaid expansion has reduced the Black-White disparity in CVD mortality. This study estimated a difference-in-differences model to compare changes in county-level CVD mortality ratios between expansion and non-expansion states. Findings indicate that Medicaid expansion is not associated with a statistically significant reduction in Black-White disparities in CVD mortality (β = −.039; p =.30). In conclusion, Medicaid expansion may be associated with improved health outcomes and access to care overall; however, it is insufficient to overcome other (i.e., social and economic) drivers of racial/ethnic disparities in CVD mortality.

Despite a steady decline in cardiovascular disease (CVD) mortality over the past several decades, it remains a leading cause of death among U.S. adults,¹ especially those in lower socioeconomic status (SES)² and uninsured groups.³ Not surprisingly, non-Hispanic Black adults, who are overrepresented in these at-risk populations, are disproportionately burdened by poor cardiovascular health,^4,5 higher CVD mortality,^1,5,6 and more CVD-related years of life lost.⁷ Medicaid expansion has been explored as a potential policy lever to address racial/ethnic health disparities by expanding Medicaid eligibility to nonelderly, single adults who meet certain income thresholds.⁸ In addition to improved insurance coverage⁹ and access to care,¹⁰ Medicaid expansion is associated with reduced CVD mortality.¹¹ Adults with CVD in expansion states have higher rates of insurance coverage and access to care compared with their non-expansion counterparts, potentially reducing their risk of preventable CVD mortality.¹² Yet, coverage and access gaps appear to persist among non-Hispanic Black adults.¹² Though overall uninsurance disparities appear to have declined,¹³ some research suggests that the benefits of Medicaid expansion, such as improved access to care and health outcomes, have been realized at a lower rate among non-Hispanic Black adults compared with non-Hispanic White adults.^14,15 It is unclear whether Medicaid expansion has affected the leading cause of death in this population—cardiovascular disease (CVD).

The goal of the current study was to address this knowledge gap by investigating the relationship between Medicaid expansion and Black-White disparities in CVD mortality. To accomplish this goal, we employed a quasi-experimental research design to compare changes in Black-White CVD mortality ratios between U.S. counties in expansion and non-expansion states. This contribution is particularly significant as states explore innovative ways to make use of Medicaid spending to improve chronic disease outcomes, including CVD mortality.

Methods

This study used publicly available data sources and a quasi-experimental research design to test whether Black-White CVD mortality ratios changed in expansion states relative to non-expansion states. This research was deemed not human subjects research under exemption 4 by the institutional review board at the University of Pittsburgh.

Data.

We examined nine years of CVD mortality data (2011–2019) from the Wide-Ranging Online Data for Epidemiologic Research (WONDER) system from the Centers for Disease Control and Prevention (CDC). Data in WONDER include annual mortality counts as well as crude and age-adjusted mortality rates for all U.S. counties based on death certificates. Underlying cause of death and demographic information (e.g., geographic location, age group, race/ethnicity, gender) for the decedents are available through WONDER’s publicly available database. This allows national data to be disaggregated to assess mortality rates among specific subpopulations by location (e.g., state, U.S. region, county), gender, age group, race/ethnicity, and/or year. These data were merged with data from the U.S. Census Bureau and the Health Resources and Services Administration (HRSA) Area Health Resource Files (AHRF) to adjust for county-level factors.

Sample.

The prevalence of CVD is highest in adults over 45 and Medicaid expansion is associated with a significant reduction in all-cause mortality in adults under 64.^16,17 Therefore, we restricted our sample to county-year mortality rates for non-Hispanic Black and White adults, ages 45–64. Additionally, by using an age cut-off of 64, we limited the inclusion of adults who may have been dually eligible for Medicare at the time of their death. Due to WONDER privacy constraints, county-year mortality rates are suppressed or coded as unreliable if fewer than 20 deaths occurred. Therefore, our final analytic sample consisted of 1,720 county-year observations. To be included in the treatment (i.e., Medicaid expansion) group, the policy change must have been enacted by January 1, 2019. Our final sample represented 246 counties in 22 expansion states (n=937 county-year observations in the treatment group) and 14 non-expansion states (n=783 county-year observations in the control group). Table 1 lists states included in the study sample and whether the state expanded Medicaid.

Table 1.

STATES INCLUDED IN THE STUDY SAMPLE BY STATUS OF MEDICAID EXPANSION

Medicaid Expansion		No Expansion
Arizonaa	Massachusettsa	Alabama	Tennessee
Californiaa	Michigan	Arkansas	Texas
Coloradoa	Minnesotaa	Florida	Wisconsin
Connecticuta	Nevada	Georgia
Delawarea	New Jerseya	Kansas
District of Columbiaa	New Yorka	Mississippi
Illinois	Ohio	Missouri
Indiana	Oregon	Nebraska
Kentucky	Pennsylvania	North Carolina
Louisiana	Virginia	Oklahoma
Maryland	Washingtona	South Carolina

Note:

^aState partially expanded Medicaid to offer coverage to certain childless adults prior to the Affordable Care Act.

Dependent variable.

Our outcome of interest was the county-year age-adjusted mortality ratio (AAMR). The AAMR is a relative measure of racial/ethnic health disparity calculated by dividing the county-year age-adjusted mortality rate for one racial/ethnic group by the county-year age-adjusted mortality rate for the reference group. Based on the most recent mortality report from National Vital Statistics,¹ we retrieved county-level, age-adjusted mortality rates using only deaths with ICD-10 codes I00–I09, I11, I13, I20–I51 (i.e., diseases of the heart) listed as cause of death. We used age-adjusted CVD mortality rates for non-Hispanic Black adults and non-Hispanic White adults to calculate the AAMR.

Independent variable.

Our determinant of interest was a binary county-year indicator of whether the county was in a state that had adopted Medicaid expansion under the Affordable Care Act (ACA). Though most expansion states included in our sample expanded in 2014, several states expanded in later years. Indiana and Pennsylvania expanded in 2015; and Louisiana expanded in 2016. Virginia’s expansion took effect on January 1, 2019.

Analysis.

To examine the impact of Medicaid expansion, we estimated a difference-in-difference regression model. This quasi-experimental design attempts to mimic randomized designs and allow for stronger assumptions of causality. The model controlled for county-level factors including uninsurance rate for adults (ages 40–64), health professional shortage area (HPSA) scores, primary care providers and cardiologists per 100,000 population, unemployment rates, median household income, and poverty rates for the population.^* Models also controlled for whether the county was in a state that had partially expanded Medicaid prior to 2014. Lastly, we included county and year fixed effects to adjust for within-county variation over time and used robust standard errors to account for state clusters. Analyses were performed using STATA 16 statistical software (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC.) and statistical significance was considered at the p<.05 level.

Results

From 2011–2019, on average, the age-adjusted CVD mortality rates for both non-Hispanic Black and non-Hispanic White adults were significantly lower (p<.001) among counties in expansion states than in counties located in non-expansion states (see Table 2). However, the Black-White disparity in CVD mortality or age-adjusted mortality ratio (AAMR)—the county-year ratio of age-adjusted CVD mortality rates among non-Hispanic Black adults over non-Hispanic White adults—was significantly higher among expansion counties compared with non-expansion counties (p<.001). With respect to control factors, the poverty rate was lower, while the median household income was higher in expansion counties compared with non-expansion counties(p<.001). There was no statistically significant difference in the unemployment rates between the two groups. As expected, a larger percentage of the population was uninsured in non-expansion than in expansion counties (15.89% vs. 10.57%; p<.001). Further, there were significantly more primary care providers and cardiologists per 100,000 in expansion counties than in non-expansion counties (p<.001).

Table 2.

COUNTY-LEVEL CHARACTERISTICS IN EXPANSION AND NON-EXPANSION STATES, 2011–2019

	Mean (Standard Error)
	Expansion	Non-Expansion	p-value
Cardiovascular Disease Mortality Rates
Non-Hispanic Black	214.70 (2.4)	231.85 (2.8)	<.001
Non-Hispanic White	117.84 (1.4)	150.60 (2.2)	<.001
Age-Adjusted Mortality Ratio	1.93 (.02)	1.65 (.02)	<.001
Population Characteristics
Percent Unemployed	6.13 (.08)	5.92 (.08)	.30
Percent in Poverty	14.88 (.18)	17.28 (.16)	<.001
Median Household Income	61,586.02 (575.40)	50,193.7 (363.59)	<.001
Health Care Access
Percent without Health Insurance	10.57 (.15)	15.89 (.15)	<.001
Primary Care Providers per 100,000	88.49 (.88)	77.62 (.85)	<.001
Cardiologists per 100,000	9.54 (.19)	8.32 (.18)	<.001

Figure 1 displays trends in the age-adjusted mortality ratio (AAMR) between counties in expansion and non-expansion states. Prior to the first wave of Medicaid expansion in 2014, the AAMR trends were roughly parallel. While the AAMR continued to trend upward in both groups after the first wave of expansion, the percent change over time was higher in counties located in expansion states. In expansion counties, the AAMR increased by 5.8% between 2015 and 2019, whereas the percent increase in non-expansion counties was 1.3% during the same period. Notably, between 2011 and 2015, the percent change was lower in expansion counties (4.0%) and higher in non-expansion counties (1.4%).

Figure 1.

Trends in age-adjusted mortality ratio in expansion and non-expansion states, 2011–2019.

Results from the difference-in-difference regression model are presented in a forest plot (see Figure 2). To interpret the percent change using the regression coefficients, we exponentiated the coefficients, subtracted 1, and then multiplied by 100. Controlling for other county-level factors, Medicaid expansion was associated with a 3.9% decrease in the disparity in CVD mortality between non-Hispanic Black and White adults (i.e., age-adjusted mortality ratio or AAMR), but this relationship was not statistically significant. Two factors were associated with a statistically significant increase in the AAMR. With each unit increase in the rate of cardiologists per 100,000, the county-level AAMR increased by 4.0%. Similarly, for each $1000 increase in the median household income, the AAMR increased by 17.8%.

Figure 2.

Factors associated with the age-adjusted mortality ratio for CVD, 2011–2019.

Discussion

Non-Hispanic Blacks account for approximately 13% of the total U.S. population¹⁸ but nearly one-fourth of all CVD deaths.¹ Medicaid expansion is looked to as a panacea for improving health outcomes among low-income populations and addressing racial/ethnic health disparities. However, our findings suggest that Medicaid expansion alone may not suffice to reduce Black-White disparities in CVD mortality. Specifically, our primary finding is that Medicaid expansion was not associated with a statistically significant decrease in our measure of the Black-White disparity—the age-adjusted mortality ratio (AAMR). It is possible that this disparity did not improve after expansion because Medicaid expansion has fewer impacts on access to care and health outcomes among non-Hispanic Black adults compared with non-Hispanic White adults.^14,15,18 Indeed, the CVD mortality trend (2011–2019) among non-Hispanic Black adults in expansion counties shows a marked increase in age-adjusted mortality rates over time compared with a slight decrease in mortality rates among non-Hispanic White adults over time (figure not shown but available upon request). These differences in mortality trends likely explain why the disparity did not significantly decrease. Additionally, our results suggest inequitable access to specialized cardiovascular care (i.e., availability of cardiologists) is a potential driver of the Black-White disparity in CVD mortality. The rate of cardiologists per 100,000 was associated with a 4.0% increase in the age-adjusted mortality ratio. Non-Hispanic White adults, who often reside in higher-income neighborhoods than non-Hispanic Black adults,¹⁹ may benefit from spatial access to specialized cardiovascular care,²⁰ thus driving the disparity.

Importantly, emerging research links Black-White disparities in risk of CVD mortality to inequalities related to socioeconomic status.²¹ This aligns with our regression results which suggest that higher median household income at the county-level (which can be attributed to non-Hispanic White populations)²² is associated with an increase in the age-adjusted mortality ratio (AAMR). Budget constraints are a known barrier to seeking medical care among individuals with CVD, especially for those who are underinsured or unisured.¹² Additionally, income and socioeconomic status influence other determinants of health and health behaviors that may be affecting CVD and CVD-related disparities. It is likely that Medicaid expansion alone is simply not sufficient to counteract a lifetime of inequitable social and economic policy exposures that affect social and economic conditions, and, thus, CVD risk factors among minoritized populations.

There are some limitations to this study. First, WONDER does not include individual details about the decedent. Therefore, we are unable to limit the analysis to low-income, Medicaid populations or adjust for health behaviors. Additionally, WONDER suppresses data for counties with fewer than 20 cases. This limits our sample size, options for stratified analyses (e.g., by gender), and the ability to conduct falsification tests using populations unaffected by Medicaid expansion. Moreover, Medicaid varies from state to state, including what services are covered, whether the beneficiary is responsible for premiums, co-insurance, and/or deductibles, and whether the state has participated in Medicaid waiver programs such as Delivery System Reform Incentive Payment (DSRIP). Our study does not explore such variation in how states administer Medicaid. Finally, the first wave of Medicaid expansion occurred in 2014. It is possible that the post-ACA implementation period is not enough to determine whether expansion has significantly improved the disparity in CVD mortality rates. In fact, when 1- and 2-year lags are applied, Medicaid expansion is associated with an increase in the disparity, but the relationship remains statistically insignificant.

Conclusion.

Premature mortality due to CVD is a known public health crisis in the U.S., particularly among low-income, uninsured, and minoritized populations. Not surprisingly, non-Hispanic Black adults, who are overrepresented in low-income and uninsured groups, are disparately burdened by CVD mortality. Our study provides empirical evidence that several years of access to care under the ACA’s expanded Medicaid did not narrow disparities in CVD mortality between non-Hispanic Black and non-Hispanic White adults. Future research is needed to explore the relationship between policies influencing socioeconomic status and Black-White disparities in CVD outcomes.