Medicaid Expansion and State-Level Differences in Premature Cardiovascular Mortality by Subtype, 2010–2017

For the first time in decades, a decline in life expectancy was noted in the United States in 2018 and attributed to increases in midlife mortality primarily due to cardiovascular disease (CVD) and secondarily due to the opioid epidemic.^{1, 2} However, significant disparities exist in trends of CVD mortality by socioeconomic status, which may in part be related to healthcare access and affordability. In fact, Medicaid expansion as part of the Affordable Care Act (ACA) has been linked with favorable improvements in cardiovascular mortality rates in counties in states that expanded compared with non-expansion states between 2014–2016.³ However, subtypes of CVD are heterogeneous and it is unclear whether Medicaid expansion was associated with similar improvements in CVD mortality across all CVD subtypes. Therefore, we examined whether changes in cause-specific premature CVD mortality occurring before the age of 65 years varied by state-level policy on Medicaid expansion.

Annual age-adjusted mortality rates (AAMR) by state for cardiovascular causes of death from 2010 through 2017 among adults aged 25–64 years were calculated using the Centers for Disease Control and Prevention Wide-ranging ONline Data for Epidemiologic Research (CDC WONDER), which includes all death certificates filed in the US. Specifically, we identified decedents with any mention of cause of death of the following: CVD (I00-I99); ischemic heart disease (I20-I25); hypertensive disease (I10-I15); heart failure (I50); and cerebrovascular disease (I60-I69) leveraging the multiple cause of death files. The date of expansion for most states was January 1, 2014; states that expanded Medicaid between 2014 and 2017 were excluded as the intervention occurred during the post-period for analysis as well as Wisconsin (already provides coverage up to 100% federal poverty level). We used a differences-in-differences quasi-experimental design to compare premature mortality rates (overall CVD and by subtypes) before and after Medicaid expansion. After confirming that the parallel trends assumption was met, difference-in-differences models were adjusted for a linear time trend and the following state-level variables: percent non-white race, percent female, percent living under 100% of the federal poverty line, percent living in a rural area, percent current smokers, percent obese, and number of primary care physicians per 100,000 population. State-level random intercepts accounted for clustering at the state-level, and analyses were weighted according to state population size.

CVD mortality declined significantly in expansion compared with non-expansion states with a differences-in-difference estimate of −3.8 (−5.6, −2.1, p <0.001), which represents 3.8 fewer CVD deaths per 100,000 population annually after 2014 in expansions compared with non-expansion states (Table). When examining subtypes of cardiovascular death, the change in ischemic heart disease and cerebrovascular-related deaths did not differ significantly by Medicaid expansion status. However, significant differences were observed in trends in deaths related to hypertensive heart disease and heart failure in expansion compared with non-expansion states (DID −2.2 (−3.6, −0.8, p=0.002) and −0.8 (−1.2, −0.3, p<0.001), respectively).

Table.

Difference-in-Differences Analysis of Changes in Age-Adjusted Mortality Rates Among US Adults Aged 25–64 Before and After Implementation of the Affordable Care Act Medicaid Expansion, by State Expansion Status and Cardiovascular Cause of Death, 2010–2017

Underlying Cause of Death*	Expansion States (N=24)†			Non-Expansion (N=18)†			Difference-in-Differences‡		Δ Absolute Deaths
	AAMR 2010–2013	AAMR 2014–2017	Change After Expansion (95% CI)	AAMR 2010–2013	AAMR 2014–2017	Change After Expansion (95% CI)	Adjusted DiD, per 100,000 (95% CI)	DiD p-value
All CVD	141.3	140.2	−1.1 (−3.0, 0.7)	128.2	130.9	2.7 (0.7, 4.7)	−3.8 (−5.6, −2.1)	p<0.001	−12,472 (−6,892, −18,379)
Ischemic Heart Disease	53.8	53.7	−0.1 (−0.9, 0.8)	47.3	47.9	0.6 (−0.4, 1.5)	−0.6 (−1.4, 0.2)	p=0.14	NS
Hypertensive Heart Disease	44.2	42.6	−1.6 (−3.1, −0.1)	37.7	38.4	0.6 (−1.0, 2.2)	−2.2 (−3.6, −0.8)	p=0.002	−7220 (−2626, −11,815)
Heart Failure	14.8	14.8	0.0 (−0.4, 0.5)	14.6	15.4	0.8 (0.3, 1.2)	−0.8 (−1.2, −0.3)	p<0.001	−2626 (−985, −3938)
Cerebrovascular Disease	16.5	16.5	0.0 (−0.4, 0.4)	16.3	16.7	0.4 (−0.1, 0.8)	−0.3 (−0.8, 0.1)	p=0.09	NS

Mortality rates age-adjustment, per 100,000 AAMR denotes age-adjusted mortality rate; CI, confidence interval; DiD, difference-in-differences.

^*The following ICD-10 codes represent the cause of death: Cardiovascular Disease (I00-I99); Ischemic Heart Disease (I20-I25); Hypertensive Disease (I10-I15); Heart Failure (I50); Cerebrovascular Disease (I60-I69).

^†Medicaid expansion states included 24 states: Arizona, Arkansas, California, Colorado, Connecticut, Delaware, District of Columbia, Hawaii, Illinois, Iowa, Kentucky, Maryland, Minnesota, Nevada, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, West Virginia. Non-Medicaid expansion states included 18 states: Alabama, Florida, Georgia, Idaho, Kansas, Maine, Mississippi, Missouri, Nebraska, North Carolina, Oklahoma, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wyoming. Maine and Virginia expanded Medicaid after the study period ended, as thus are included as non-expansion.

^‡Difference-in-differences models were adjusted for a linear time trend and the following state-level variables (data source): percent non-white race (U.S. Census Bureau, American Community Survey, 2010–2017), percent female ((U.S. Census Bureau, American Community Survey, 2010–2017), percent living under 100% of the federal poverty line (U.S. Census Bureau, American Community Survey, 2010–2017), percent living in a rural area (U.S. Census Bureau, 2010 Decennial Census), percent current smokers (Centers for Disease Control and Prevention, Behavioral Risk Factor Surveillance System, 2010–2017), percent obese (Centers for Disease Control and Prevention, Behavioral Risk Factor Surveillance System, 2010–2017), and number of primary care physicians per 100,000 population (Health Resources and Services Administration, Area Health Resource Files, 2014–2016). State-level random intercepts accounted for clustering at the state-level, and analyses were weighted according to state population size. All covariates were annually time-varying, except for rural population which was measured in 2010 only, and primary care density which was measured in 2010, 2015, 2016.

Our analysis confirm prior findings regarding the favorable impact of the ACA with the decreasing burden of CVD mortality in states that expanded Medicaid in stark contrast with increasing burden of premature CVD mortality in states that did not expand Medicaid.³ We extend these findings by identifying that these differences were primarily attributed to cardiovascular deaths related to hypertensive heart disease and heart failure. The contrasting state-level trends are observed in the context of national trends demonstrating increasing AAMR for both of these CVD subtypes since 2011⁴, and point to the importance of hypertension prevention and management as a key driver for mortality related to both hypertensive heart disease and heart failure. While we cannot assess the causal contribution of Medicaid expansion on improved hypertension management, the significant differences in Medicaid expansion states suggest that improving access to preventive health care may improve hypertension-related health outcomes.⁵ Further, the impact of these findings is especially relevant given the continued controversy of the constitutionality of the ACA and in the context of the current coronavirus pandemic that has led to unprecedented rates of unemployment in the US that is threatening health insurance coverage and health for millions of Americans.

Limitations of this analysis include the potential for miscoding cause of death, especially for heart failure given it is more challenge to attribute as a cause of death. We also note the potential for residual confounding due to incomplete adjustment for health (e.g. diet, physical activity) and environmental factors (e.g. air pollution). While using aggregated AAMR data on a state-level may introduce confounding from heterogeneity in risk of death within states, a smaller unit of analysis (such as counties) would have required excluding all units with <20 premature deaths for specific causes, which can result in significant bias. In summary, states that expanded Medicaid experienced significant reductions in deaths related to hypertensive heart disease and heart failure, in spite of national mean increases in deaths from these causes supporting the role of enhanced Medicaid access to reduce the burden of CVD mortality in the US.