Medicaid Expansion And Outpatient Cardiovascular Care Use Among Low-Income Nonelderly Adults, 2012–15

Abstract

Adults with lower socioeconomic status have a disproportionately higher burden of cardiovascular disease. Medicaid expansion under the Affordable Care Act, which went into effect January 1, 2014, in adopting states, led to an expansion of health insurance coverage for low-income adults. To understand whether Medicaid expansion was associated with increased access to outpatient cardiovascular care in expansion states, we examined Medicaid Analytic eXtract administrative claims data for nonelderly adult beneficiaries from the period 2012–15 for two states that expanded Medicaid eligibility (New Jersey and Minnesota) and two states that did not (Georgia and Tennessee) and calculated population-level rates of cardiovascular care use. There was a 38.1 percent greater increase in expansion states in the rate of beneficiaries with outpatient visits for cardiovascular disease management associated with Medicaid expansion relative to nonexpansion states. This was accompanied by a 42.9 percent greater increase in the prescription rate for cardiovascular disease management agents. These results suggest that expansion of Medicaid eligibility was associated with an increase in cardiovascular care use among low-income nonelderly adults in expansion states.

Introduction:

The Affordable Care Act (ACA) expanded Medicaid eligibility as of January 1, 2014, for nonelderly adults with incomes up to 138 percent of the federal poverty level, with certain states expanding eligibility to people with even higher income levels. Other states opted not to expand eligibility, and as of July 2023, ten states had not yet done so.(1) More than sixteen million newly eligible nonelderly adults had gained Medicaid coverage as of 2022.(2) As low-income people have a disproportionately higher prevalence of cardiovascular disease, Medicaid expansion has had the potential to play an important role in improving cardiovascular health.(3)

Expanded Medicaid eligibility has been associated with lower population-level cardiovascular mortality rates among nonelderly adults.(4) How that improvement occurred, however, is unclear. Prior studies found that outcomes for patients hospitalized for cardiovascular disease do not differ by state expansion status.(5, 6) This suggests that improved access to cardiovascular care, rather than the quality of care a patient receives once hospitalized, may play a role in improving cardiovascular health. This is particularly the case for people with established cardiovascular disease, who may require care from subspecialist providers and may benefit from several guideline-recommended medical therapies. Uninsured people are less likely to have a regular source of outpatient care and are more likely to delay or forgo necessary medical treatment.(7, 8) Whether outpatient cardiovascular care, especially for those with chronic cardiovascular disease, changed in the setting of Medicaid expansion is not well studied.

Medicaid administrative claims are a rich yet underused data source that can provide detailed information on health care use for low-income people. Unlike in studies of Medicaid that have relied on national survey data, the use of detailed administrative claims data sources such as the Medicaid Analytic eXtract (MAX) in this study allows for the potential analysis of use in subgroups of interest such as those based on age, sex, race, and ethnicity. We analyzed MAX data for beneficiaries ages 19–64 during the period 2012–15 for two states that expanded Medicaid income eligibility—Minnesota (up to 200 percent of the federal poverty level in 2014, followed by a decrease to 138 percent of poverty in 2015) and New Jersey (up to 138 percent of poverty)—and two states that did not (Georgia and Tennessee) to evaluate whether expansion was associated with changes in outpatient cardiovascular care use and the prescribing of cardiovascular disease management medications.

Study Data And Methods

This study was considered exempt from review by the Institutional Review Board of the University of Pennsylvania. Our analysis used MAX data from the first quarter of 2012 to the third quarter of 2015 from two expansion states (Minnesota and New Jersey) and two states that did not expand eligibility during the study period (Georgia and Tennessee). Using a difference-in-differences approach, we analyzed trends in outpatient cardiovascular care use (outpatient visits, prescription of cardiovascular disease–related medications, emergency department [ED] visits, and hospitalizations) and evaluated whether expansion of Medicaid eligibility was associated with a differential change in trends in the postexpansion period.

The study period of 2012–15 allowed for two years of data before and after expansion took effect. As a result of the transition of diagnostic coding from the International Classification of Diseases, Ninth Revision (ICD-9), to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), in October 2015, data from the fourth quarter of 2015 were not used. Also, as there may have been changes in beneficiary behavior in the quarters immediately before and after expansion, the fourth quarter of 2013 and the first quarter of 2014 were also excluded as transition periods.

Analysis States

The four states analyzed were chosen on the basis of the availability of suitable MAX data. These data are based on the Medicaid Statistical Information System. Between 2011 and 2015, states transitioned from the Medicaid Statistical Information System to the Transformed Medicaid Statistical Information System, which makes longitudinal analyses in states that switched systems challenging.(9) Therefore, we limited candidate states to the seventeen states that used the older Medicaid Statistical Information System throughout our study period. Among these states, we evaluated the availability of data for Medicaid managed care beneficiaries. To determine which states included encounter data for these beneficiaries, we used a combination of a Department of Health and Human Services Office of Inspector General report, which reviewed which states reported managed care encounter data,(10) and Medicaid managed care crosswalk files, which provide information on whether any managed care plans in a state are missing encounter data.(11) On the basis of these sources, the four states used in this analysis were the only states that provided encounter information for all managed care beneficiaries. For these four states, we then reviewed MAX validation reports produced by the Centers for Medicare and Medicaid Services, which confirmed that virtually all outpatient claims, which were used to determine utilization, had diagnosis and procedure codes available for all study period years.(12) Additional details about state selection and data validation are in online appendix exhibit 1.(13)

Medicaid Eligibility

The Medicaid income eligibility limit for childless, nondisabled, nonelderly adults in the two nonexpansion states was 0 percent of the federal poverty level throughout the study period. In Minnesota, there was an earlier, smaller expansion from 0 percent to 75 percent of poverty in January 2012; this increased further to 200 percent of poverty in January 2014 and then decreased to 138 percent of poverty in January 2015. In New Jersey, the limit increased from 0 percent of poverty to 138 percent of poverty in January 2014.(14) Income eligibility limits for parents and pregnant adults are in appendix exhibit 2.(13) Nonincome eligibility criteria include factors such as disability. For the two expansion states in this analysis, from the first quarter of 2012 to the third quarter of 2013 was considered the preexpansion period, and from the second quarter of 2014 to the third quarter of 2015 was considered the expansion period.

Study Population

The study population consisted of all Medicaid beneficiaries ages 19–64 enrolled during the study period in the four states. The primary analysis excluded people dually enrolled in Medicaid and Medicare. Beneficiaries with Medicaid eligibility listed as unknown or ineligible or with missing ZIP code information, out-of-state residents, or residents of ZIP codes with fewer than ten residents with income below the federal poverty level were excluded. The number of people excluded in each state is in appendix exhibit 3.(13)

We considered all members of the study population to be low income, inasmuch as each one is likely to have an income within the applicable upper limit for Medicaid eligibility in their state during any given calendar quarter of the study period.

To evaluate how Medicaid expansion could affect the population-level rate of cardiovascular care use, the ideal denominator would be all Medicaid-eligible or potentially eligible adults who had cardiovascular disease or risk factors (that is, were candidates for cardiovascular care). However, such data were unavailable. As a surrogate, we used the number of nonelderly adults with income below the federal poverty level in each ZIP code, obtained from the Census Bureau’s 2012–16 American Community Survey five-year estimates, as this population would be eligible for Medicaid after expansion (in expansion states). Such an approach of using census-derived estimates of the number of people in an area who may be eligible for Medicaid as a denominator has been used in previous analyses of the association of Medicaid expansion with health care use.(15, 16) We limited the analysis to ZIP codes with at least ten residents with income below the federal poverty level to provide more stable statistical estimates. In sensitivity analyses, we used alternative populations as the denominator: the number of nonelderly adults who were primarily insured through Medicaid or were uninsured, the total number of nonelderly adult residents, and the number of annual county-level nonelderly adult residents with income up to 138 percent of poverty.

Outcomes

The primary outcome of interest was the rate of Medicaid beneficiaries ages 19–64 with at least one outpatient visit for cardiovascular disease management: encounters with Current Procedural Terminology codes 99201–15 and a visit diagnosis for any of the atherosclerotic cardiovascular disease conditions (coronary artery disease, stroke, and peripheral arterial disease), heart failure, or atrial fibrillation in a quarter per 1,000 nonelderly adult residents with income below the federal poverty level in each ZIP code. Visit diagnoses were based on ICD-9 codes, as listed in appendix exhibit 4.(13)

Secondary outcomes were the rate of Medicaid beneficiaries ages 19–64 with at least one outpatient visit for any cause, at least two outpatient visits for cardiovascular disease management, and at least one outpatient visit with a cardiovascular disease–related specialist (cardiology, neurology, vascular surgery, or cardiothoracic surgery); the rate of beneficiaries prescribed cardiovascular disease management medications, as listed in appendix exhibit 5; and the rate of all-cause or cardiovascular disease–related ED visits and hospitalizations in a quarter per 1,000 nonelderly adult residents with income below the federal poverty level in each ZIP code.(13)

Data sources used other than MAX are in appendix exhibit 6.(13)

Missing Data

Multiple imputation (ten imputations) with fully conditional specification was used to account for missing race and ethnicity data. Covariates used in the multiple imputation model are in appendix exhibit 7.(13)

Statistical Analysis

Using census data, we first examined state-level insurance coverage in the included states for nonelderly adults with income up to 138 percent of poverty (and up to 100 percent and between 138 percent and 200 percent of poverty for Minnesota). We then calculated summary measures of demographic and clinical variables at the beneficiary level, using MAX data, and of health care availability, demographic characteristics, and components of the Centers for Disease Control and Prevention’s Social Vulnerability Index for each ZIP code in each state, using census data. We also compared demographic and clinical characteristics between the preexpansion and expansion periods.

We then calculated the rate of the outcomes listed above per 1,000 nonelderly adults with incomes below the federal poverty level for each ZIP code. Using a difference-in-differences approach, we assessed changes in the outcome measures associated with expansion. We fit a multivariable mixed effects model with a negative binomial distribution and log link, with the ZIP code population as the offset (appendix exhibit 8).(13) The model accounted for autocorrelation over time and clustering of ZIP codes in states. As cardiovascular disease prevalence and care use may be affected by area-level factors, including demographics and area social vulnerability, as well as health care availability, we included ZIP code–level demographic, economic, and health care–related covariates included in the model,(17, 18) as listed in appendix exhibit 8.(13) We confirmed that the slope of preexpansion trends for the primary outcome did not differ statistically significantly between the two groups, using an interaction between a linear time trend and expansion status in the preexpansion period.

Subgroup analyses examined the primary outcome among subgroups by age (younger and middle-aged adults; ages 19–44 and 45–64, respectively), sex, and race and ethnicity (non-Hispanic Black, non-Hispanic White, non-Hispanic other race, and Hispanic any race). We estimated “triple difference” models comparing these subgroups with each other in the expansion and nonexpansion states. We performed sensitivity analyses using alternative denominator populations as described previously. For the county-level sensitivity analysis, all covariates were aggregated at the county level. We also estimated the primary model with dually enrolled beneficiaries included.

All statistical tests were two-tailed and used a p value of 0.05 as a threshold for significance. Regression estimates are presented with 95% confidence intervals. All analyses were conducted using SAS, version 9.4.

Limitations

Our study had several limitations. Because it was an observational study, the associations noted cannot be assumed to be causal. The analysis did not capture care paid for by non-Medicaid payers or uncompensated care. However, because Medicaid is the largest source of health insurance for low-income, nonelderly adults, trends among Medicaid beneficiaries provide insights into how the ACA and Medicaid expansion influenced health care access among low-income people. Our analysis was limited to the four states that met all data requirements, including consistent use of the older Medicaid Statistical Information System, and might not be generalizable to other states. The income eligibility limit for Medicaid in Minnesota was expanded to 200 percent of poverty in 2014, which was higher than the limit required by the ACA and the limit set by most expansion states. This limit was decreased to 138 percent in 2015, but this may have affected utilization rates and the study population in the three included quarters in 2014. The nonexpansion states differed in meaningful ways from the expansion states. We did, however, note that before expansion, trends in the primary outcome did not differ statistically significantly between the two groups of states. The measures of cardiovascular care used might not reflect all aspects of appropriate or necessary care. However, we attempted to examine different aspects of care that have been shown to have a meaningful impact on cardiovascular outcomes in previous studies. Our denominator population (all nonelderly adults with income below the federal poverty level) might not fully capture all Medicaid-eligible people who could have been candidates for cardiovascular care or the variation in disease prevalence between areas. However, our model attempted to control for differences in area-level variables that were associated with cardiovascular disease or risk-factor prevalence.(17)

Study Results

Using census data, among adults ages 19–64 with income up to 138 percent of poverty in the included states, Medicaid coverage increased from 37.0 percent to 47.8 percent in the two expansion states and from 25.3 percent to 28.8 percent in nonexpansion states from 2012 to 2015 (appendix exhibit 9).(13) In Minnesota, where there was an earlier increase in the income eligibility limit (from 0 percent to 75 percent of poverty) in 2012 and an increase to 200 percent of poverty in 2014, followed by a decrease to 138 percent of poverty in 2015, the proportion of people with income below 100 percent of poverty and income between 138 percent and 200 percent of poverty with Medicaid coverage was relatively stable between 2011 and 2013, followed by a larger increase in 2014 (appendix exhibit 10).(13)

Between the first quarter of 2012 and the third quarter of 2015, a total of 3,936,638 unique Medicaid beneficiaries ages 19–64 were included in our analytical sample from MAX data (exhibit 1). The mean age was 37.6 years and was similar across all states. The proportion of women was 63.4 percent, with the highest proportion found in Georgia (73.3 percent) and the lowest in Minnesota (56.4 percent). Race and ethnicity information was missing for 9.9 percent of beneficiaries. After multiple imputation, the proportion of Hispanic beneficiaries was the highest in New Jersey (21.9 percent) and the lowest in Georgia (1.0 percent). The highest proportion of non-Hispanic Black beneficiaries was in Georgia (51.3 percent), and the lowest was in Minnesota (16.8 percent). The proportion with cardiovascular diagnoses was 4.4 percent for coronary artery disease, 3.2 percent for heart failure, 3.2 percent for stroke, 1.5 percent for peripheral arterial disease, and 1.2 percent for atrial fibrillation. In the 3,041 included ZIP codes, the median proportion of nonelderly adult residents with income below the federal poverty level was 11.5 percent (appendix exhibit 12).(13)

Exhibit 1.

Beneficiary-level characteristics of Medicaid beneficiaries ages 19–64 in 4 states, total and by state Medicaid expansion status, 2012–15

	Expansion states		Nonexpansion states
Characteristics	Minnesota	New Jersey	Georgia	Tennessee	Total
Unique beneficiaries (no.)	907,239	1,210,376	995,218	823,805	3,936,638
Mean age (years)	37.0	38.6	36.9	37.7	37.6
Female (%)	56.4	58.7	73.3	66.0	63.4
Race and ethnicity subgroups (%)
Before multiple imputation
Hispanic (any race)	5.7	21.8	0.9	1.8	8.2
Non-Hispanic Black	17.1	29.1	51.8	26.9	31.6
Non-Hispanic other race	10.7	4.8	2.2	7.6	6.1
Non-Hispanic White	66.5	44.3	45.1	63.8	54.1
Missing race and ethnicity	6.3	16.5	7.8	6.6	9.9
After multiple imputationa
Hispanic (any race)	5.5	21.9	1.0	1.7	8.6
Non-Hispanic Black	16.8	27.2	51.3	26.6	30.8
Non-Hispanic other race	10.8	5.7	2.2	7.4	6.3
Non-Hispanic White	66.8	45.1	45.5	64.3	54.2
Medicaid enrollment (%)
Dually enrolled in Medicare	12.1	16.1	22.4	25.8	18.8
Enrolled in managed care plan	80.9	85.8	52.4	89.7	77.0
Cardiovascular diagnosis (%)b
Coronary artery disease	2.7	5.1	4.7	4.7	4.4
Heart failure	1.8	3.0	4.5	3.2	3.2
Stroke	2.0	3.5	3.7	3.3	3.2
Peripheral arterial disease	0.6	1.9	1.6	1.6	1.5
Atrial fibrillation	1.0	1.3	1.3	1.1	1.2
Any cardiovascular disease	5.8	9.9	10.1	9.4	8.9

SOURCE Authors’ analysis of Medicaid Analytic eXtract data from the period 2012–15.

NOTE An unabridged version of this table is in appendix exhibit 11 (see note (13) in text).

^a.Multiple imputation (10 imputations) with fully conditional specification was used to account for missing race and ethnicity data. Covariates used in the multiple imputation model are in appendix exhibit 7.

^b.Diagnosis codes used to identify cardiovascular diagnoses are in appendix exhibit 4.

In the two expansion states, between the preexpansion and expansion periods, the mean age of beneficiaries changed from 37.3 to 38.0 years (p<0.001), and the proportion who were female declined from 60.7 percent to 57.9 percent (p<0.001) (appendix exhibit 13).(13) The proportion of beneficiaries with any cardiovascular diagnosis was 10.6 percent in the preexpansion period and 9.0 percent in the expansion period (p<0.001). In the two nonexpansion states, these measures were similar across the two periods.

During the preexpansion period, and for every 1,000 nonelderly residents with income below the federal poverty level, the adjusted mean quarterly number of beneficiaries with at least one outpatient visit for cardiovascular disease management was 8.4 (95% CI: 6.1, 11.6) in the expansion states and 4.5 (95% CI: 3.3, 6.1) in the nonexpansion states (appendix exhibit 14).(13) During the expansion period, the adjusted mean rates in expansion and nonexpansion states rose to 11.7 (95% CI: 8.5, 16.0) and 4.9 (95% CI: 3.6, 6.8), respectively. Quarterly rates are displayed in exhibit 2. Trends in the baseline period did not differ statistically significantly between the two groups (0.1 percent; 95% CI: −0.5 percent, 0.8; data not shown). In the adjusted negative binomial difference-in-differences model, there was a 38.1 percent (95% CI: 35.7, 40.6; p<0.001) increase in the rate of beneficiaries with outpatient visits for cardiovascular conditions (3.6 [95% CI: 2.3, 4.9; p<0.001] additional beneficiaries per quarter per 1,000 nonelderly adults with income below the federal poverty level per quarter) associated with Medicaid expansion relative to nonexpansion states (exhibit 3). The unadjusted difference-in-differences estimate for the primary outcome (without any ZIP code–level covariates included) was 44.4 percent (95% CI: 39.1, 49.9; p<0.001; data not shown).

Exhibit 2. Quarterly number of Medicaid beneficiaries ages 19–64 in 4 states with at least 1 outpatient visit for cardiovascular disease management per 1,000 nonelderly adult residents with income below the federal poverty level, before and after Medicaid expansion, by state expansion status, 2012–15.

SOURCE Authors’ analysis of Medicaid Analytic eXtract data from the period 2012–15.

NOTES Vertical dotted line indicates Medicaid expansion in 2 expansion states (Minnesota and New Jersey) in January 2014. The fourth quarter of 2013 and the first quarter of 2014 were excluded from the model as transition periods. Mean quarterly rates were estimated from a mixed-effects negative binomial model, with covariates listed in appendix exhibit 5 (see note (13) in text). The denominator for the rate is the number of nonelderly adult state residents with income below the federal poverty level. The nonexpansion states were Georgia and Tennessee.

Exhibit 3.

Quarterly rates of cardiovascular care use among Medicaid beneficiaries ages 19–64 in 4 states, by state expansion status and percent changes associated with Medicaid expansion versus nonexpansion, 2012–15

	Beneficiaries per 1,000 residentsa
Outcomes	Expansion states		Nonexpansion states
	Preexpansionb	Expansionc	Preexpansionb	Expansionc	DIDd (%)
Outpatient visitse
≥1 cardiovascular-related visits	8.4	11.7	4.5	4.9	38.1
≥2 cardiovascular-related visits	1.2	3.3	1.1	2.3	35.4
≥1 visits for any cause	472.8	646.2	206.0	225.0	34.5
≥1 cardiovascular specialty visitsf	6.5	10.2	11.3	12.9	57.8
Cardiovascular medications prescribedg
Cholesterol management agent	78.1	115.1	30.8	33.4	46.9
Anticoagulant	7.8	10.5	3.1	3.3	36.0
Heart failure management agent	140.1	200.1	64.2	69.6	42.3
Any cardiovascular disease–related agent	195.3	279.5	87.9	94.4	42.9
Hospitalizations
Cardiovascular disease related	1.8	2.5	0.6	0.8	8.6
All cause	42.9	58.2	27.5	37.1	4.6
Emergency department visits
Cardiovascular disease related	3.5	5.2	1.7	1.9	41.5
All cause	242.0	341.3	122.3	131.9	29.1

SOURCE Authors’ analysis of Medicaid Analytic eXtract data from the period 2012–15.

NOTES Expansion and nonexpansion states are in exhibit 1. An unabridged version of this table is in appendix exhibit 14 (see note (13) in text).

^a.The number of beneficiaries with visits or prescriptions per 1,000 nonelderly residents with income below the federal poverty level.

^b.The preexpansion period was from the first quarter of 2012 through the third quarter of 2013.

^c.The expansion period was from the second quarter of 2014 through the third quarter of 2015.

^d.Differences-in-differences (DID) were estimated using a mixed effects model with negative binomial distribution and log link (appendix exhibit 8). All estimates p<0.05.

^e.Encounters for cardiovascular disease management were based on diagnosis codes as listed in appendix exhibit 4.

^f.Outpatient visits with providers in any of the following specialties: cardiology, neurology, vascular surgery, cardiothoracic surgery.

^g.Medications in each category are listed in appendix exhibit 5.

In secondary analyses, there was a 35.4 percent (95% CI: 29.1, 41.9; p<0.001) increase in the rate of beneficiaries per quarter with at least two visits for cardiovascular conditions associated with Medicaid expansion relative to nonexpansion states (exhibit 3). There was a 57.8 percent (95% CI: 55.3, 60.3; p<0.001) increase in the rate of beneficiaries per quarter with at least one outpatient visit with a cardiovascular disease–related specialist. There was a 42.9 percent (95% CI: 38.7, 47.3; p<0.001) increase in the rate of beneficiaries per quarter per 1,000 nonelderly adult residents with income below the federal poverty level prescribed any cardiovascular medication, as well as a significant increase in the rate of beneficiaries per quarter for whom each category of cardiovascular medication was prescribed (p<0.001 for all categories). The rate of beneficiaries with cardiovascular hospitalizations, all-cause hospitalizations, cardiovascular ED visits, and all-cause ED visits also increased to a greater degree in expansion states (8.6 percent [95% CI: 2.6, 14.9; p=0.004], 4.6 percent [95% CI: 1.1, 8.3; p=0.010], 41.5 percent [95% CI: 36.5, 46.7; p<0.001], and 29.1 percent [95% CI: 24.1, 34.3; p<0.001], respectively) relative to nonexpansion states. Quarterly rates for specialist visits, medication prescriptions, ED visits, and hospitalizations are in appendix exhibits 15–18.(13)

As shown in exhibit 4, there was a statistically significant increase in the rate of Medicaid beneficiaries ages 19–64 with at least one outpatient visit for cardiovascular disease management among all subgroups by age, sex, race, and ethnicity (p=0.036 for non-Hispanic adults of other races and p<0.001 for all other subgroups). In “triple-difference” models (appendix exhibit 20),(13) expansion was associated with a 27.7 percent greater increase in middle-aged adults compared with younger adults (p<0.001), a 3.5 percent greater increase in men compared with women (p=0.041), and a smaller increase among non-Hispanic Black and non-Hispanic other race beneficiaries compared with non-Hispanic White beneficiaries (−9.9 percent [p<0.001] and −20.4 percent [p<0.001], respectively).

Exhibit 4.

Estimated changes in Medicaid beneficiaries ages 19–64 with at least 1 outpatient visit per quarter for cardiovascular disease management associated with Medicaid expansion versus nonexpansion in 4 states, by population subgroup, 2012–15

Population subgroups	DID (%), expansion versus nonexpansion states	95% CI
Age (years)
19–44	15.6	9.7, 21.9
45–64	43.5	40.7, 46.4
Sex
Men	40.2	36.6, 43.9
Women	34.5	31.4, 37.7
Race and ethnicitya
Hispanic (any race)	65.6	45.4, 88.6
Non-Hispanic Black	27.5	23.1, 32.0
Non-Hispanic other race	8.5	0.5, 17.0
Non-Hispanic White	45.1	39.9, 50.4

SOURCE Authors’ analysis of Medicaid Analytic eXtract data from the period 2012–15.

NOTES An unabridged version of this table is in appendix exhibit 19 (see note (13) in text). Expansion and nonexpansion states are in exhibit 1. Differences-in-differences (DID) were estimated using a mixed-effects model with negative binomial distribution and log link and adjusted for covariates. All estimates p<0.01.

^a.Race and ethnicity subgroups after applying multiple imputation for missing values (see the exhibit 1 notes).

The overall results for the change in the rate of beneficiaries ages 19–64 with at least one outpatient visit for cardiovascular disease management were similar in the sensitivity analyses performed: including dually enrolled beneficiaries in the analysis, using all nonelderly adults as the denominator, using nonelderly adults without health insurance or insured through Medicaid as the denominator, and aggregating at the county level with the annual county-level number of nonelderly adults with income up to 138 percent of poverty as the denominator (appendix exhibit 21).(13)

Discussion

In this analysis of low-income, nonelderly adult Medicaid beneficiaries in two states that expanded Medicaid eligibility (Minnesota and New Jersey) and two states that did not (Georgia and Tennessee), expansion was associated with an increase in the rate of Medicaid beneficiaries with outpatient visits for cardiovascular conditions. Expansion was also associated with increases in the rate of beneficiaries with visits to cardiovascular disease–related specialists and with prescriptions for cardiovascular conditions.

How Medicaid expansion has affected outpatient cardiovascular care access has not been well studied. Previous studies have noted that expansion was associated with an increase in medication prescriptions for certain cardiovascular diseases and risk factors in the overall population.(19, 20) Our analysis confirmed these findings among the Medicaid-insured population and expanded on them to also include treatments for therapies for serious cardiovascular conditions such as medications used to manage heart failure and anticoagulants that are associated with improved health outcomes.

We also noted an increase in the rate of Medicaid beneficiaries with outpatient visits for cardiovascular conditions. Outpatient visits for cardiovascular disease management, particularly by cardiology subspecialists, has been associated with improvements in cardiovascular outcomes, including death.(21, 22) Therefore, it is possible that greater continuity of cardiovascular care and an increase in the use of recommended medical therapies could lead to the improvements in cardiovascular outcomes associated with Medicaid expansion noted previously.(4) As uninsured people are less likely to access regular outpatient care, the increases noted are unlikely to solely be a result of a shift of care from uncompensated care to Medicaid-reimbursed care.(23, 24)

The analysis also noted an increase in the rate of hospitalizations and ED visits. These could have resulted from either increased contact with the health care system or reduced financial concerns when beneficiaries sought care after gaining health insurance. However, previous studies suggest that although the proportion of ED visits in which the primary payer was Medicaid increased after expansion, with a concomitant decrease in uncompensated care, the overall volume of ED visits stayed the same.(25) The Oregon Health Insurance Experiment noted that although a limited Medicaid expansion in Oregon was associated with more ED visits, there was no significant difference in emergent, nonpreventable, visits, which are likely to include those for acute cardiovascular conditions such as heart attack or stroke.(26) Therefore, it is possible that at least some of the increases in cardiovascular ED visits and hospitalizations noted in our analysis are the result of a shift from uncompensated to Medicaid-reimbursed care.

A significantly greater increase in the rate of beneficiaries with at least one outpatient visit for cardiovascular disease management per 1,000 low-income residents was noted among middle-aged compared with younger adults and among men compared with women. This was likely a result of the higher prevalence of cardiovascular disease in these segments of the study population. We also noted a significant increase among all subgroups of race and ethnicity studied. However, unlike in some previous studies that noted a potential narrowing of racial disparities in health care use among minority compared with White populations,(27, 28) our analysis noted a greater relative increase in use for non-Hispanic White adults compared with non-Hispanic Black and non-Hispanic other race adults. This may, in part, be related to the demographic profiles of the individual states being studied and might not be generalizable to other states. In addition, it points to potential non-insurance-related barriers to access experienced by minority populations, such as difficulty getting appointments and transportation issues.(29, 30) Analyses of care use from additional states, particularly those that expanded Medicaid eligibility more recently, are needed to determine whether the findings noted in this subgroup analysis are generalizable to other states.

Conclusion

In this analysis, Medicaid expansion was associated with a significant increase in the rates of beneficiaries with outpatient visits for cardiovascular disease management and with prescriptions for guideline-recommended cardiovascular medications among low-income adults ages 19–64. These findings suggest that Medicaid expansion was associated with an increase in access to and use of cardiovascular care among low-income nonelderly adults in the expansion states. This increased access may be a contributing factor in the previously noted population-level improvements in cardiovascular mortality among nonelderly adults associated with Medicaid expansion.