INTRODUCTION

Atherosclerotic cardiovascular disease (CVD) is the leading cause of death in the USA1 and over 40% of Americans are expected to have some form of CVD by 2030.1, 2 Access to medical care is essential for optimal treatment of CVD and management of cardiovascular risk factors (CVRFs)3,4,5 including high blood pressure, high cholesterol, diabetes, smoking, physical inactivity, and obesity.6 For example, people with financial barriers to care are less likely to be screened for CVRFs7 and, following acute myocardial infarction (AMI), are less likely to receive evidence-based CVD treatments.8 Similarly, a lack of health insurance has been associated with worse control of CVRFs9, 10 and failure to receive recommended outpatient medications for CVD.11 In addition, racial and ethnic minorities—groups with a high prevalence of CVRFs12,13,14,15—experience greater financial barriers to care, higher rates of uninsurance,16 and worse outcomes of CVD treatment.12, 17, 18

The Patient Protection and Affordable Care Act (ACA) was designed to reduce the number of uninsured Americans and improve access to and affordability of care.19, 20 The main coverage provisions of the ACA, implemented on January 1, 2014,21 offered premium and out-of-pocket subsidies to many lower income individuals, a mandate requiring most persons to have health coverage and, at states’ discretion, the expansion of Medicaid to adults with family incomes at or below 138% of the federal poverty level.22 Among non-elderly adults nationally, the ACA has resulted in an approximately 40% decrease in uninsurance23, 24 and improved access and affordability,23, 25 with greater gains among racial and ethnic minorities,26, 27 low-income individuals,24 and in states that expanded Medicaid.24, 28 Because people with CVD and CVRFs are at risk for cardiovascular mortality and have high health care needs, understanding the impact of the ACA on these populations has high health policy and public health relevance. However, little is known about the extent to which the ACA or states’ Medicaid expansions altered coverage or access to care for Americans with CVD or CVRFs.

We examined the effect of the ACA on non-elderly adults with CVD or CVRFs using a large nationally representative data set. We sought to assess (1) whether the ACA was associated with improvements in insurance coverage and access to care; (2) whether the ACA’s optional state Medicaid expansion resulted in increased coverage and access; (3) whether racial and ethnic disparities in these outcomes diminished post-ACA; and (4) the extent to which uninsurance and poor access persist in the post-ACA era.

METHODS

Data and Study Population

We analyzed data from the Behavioral Risk Factor Surveillance System (BRFSS), a state-based telephone survey of non-institutionalized civilians conducted every year by state health departments and the Centers for Disease Control and Prevention.29 Using a random-digit dialing survey methodology, the BRFSS asks United States (US) residents of each state questions about health-related risk behaviors, chronic health conditions, demographic characteristics, access to care, and health insurance status.29 We analyzed data from 2012 through 2016, the latest year for which BRFSS data was available at the time of the analysis; we excluded data from 2014, the first year of the ACA, as we considered this to be a “washout” period (see Supplement Methods for further details). For this study, the pre-reform period was defined as 2012 and 2013 while the post-reform period was defined as 2015 and 2016.

Our study population was adults aged 18–64 (the target population of the ACA) who reported having history of CVD or two or more CVRFs. We defined individuals as having CVD if they reported having had a history of heart attack, angina, coronary heart disease, or stroke. We defined individuals as having two or more CVRFs (risk factors associated with cardiovascular mortality)30 if they reported having two or more of the following: diabetes, being a current cigarette smoker, getting no physical exercise, having a BMI > 25 (calculated using self-reported weight and height), or being age > 45 for men or > 55 for women. We did not include hypertension and hyperlipidemia in the definition of CVRFs in our main analysis because BRFSS data on these conditions was not collected in 2012 or 2016; however, we did include them in a sensitivity analysis (described below).

Study Variables

We assessed four primary outcomes: having health insurance (at the time of the interview), having had to forgo a physician visit in the last year due to cost (to facilitate comparisons, we report this variable as the inverse, not having to forgo a physician visit due to cost; for brevity, we refer to this as “able to afford a physician visit”), having had a check-up in the last year and having a personal physician. We also used information on demographic characteristics including age, gender, race/ethnicity, marital status, education, employment, and household income. Race/ethnicity was classified as non-Hispanic white, non-Hispanic black, Hispanic, and other (see Methods in the Supplement for a full description of all variables).

Statistical Analysis

For all outcomes, we calculated adjusted and unadjusted percentage point changes on the absolute scale from the pre- to post-ACA periods. We used multivariable logistic regression models to obtain estimates adjusted for marital status, income, sex, employment status, education, and race. Percentage point changes were derived from logit models using the method of Korn and Graubard31 to calculate predictive marginal effects at representative values. We assessed pre- to post-ACA changes separately for the overall population, a subset of lower income individuals (family income < $35,000/year), and residents of each US state. To assess the effect of Medicaid expansion on each outcome, we employed a quasi-experimental difference-in-differences modeling approach32 that estimated the adjusted net change in expansion states relative to non-expansion states (the referent group). To assess changes in pre-existing racial and ethnic disparities, we estimated pre- to post-changes in white-black and white-Hispanic disparities in each outcome using the same modeling approach (except race/ethnicity was not included as a model covariate) with whites as the reference group.

We assessed the magnitude of gaps in coverage and access for people with CVD/CVRFs that remain in the post-ACA era for white, black, and Hispanic residents of expansion and non-expansion states in unadjusted analyses.

We conducted several analyses to assess whether pre-ACA trends were similar in our difference-in-differences analyses33 (see Supplement for details).

In addition, we conducted two sensitivity analyses. First, to determine whether inclusion of people with hypertension and hyperlipidemia would have changed our results, we repeated our main analyses using only data from study years in which these diagnoses were elicited, 2013 (pre-ACA) and 2015 (post-ACA). Second, we repeated our analyses separately for individuals with established CVD and those with multiple CVRFs to determine if results could vary between these subgroups.

In all analyses, we applied sampling weights provided by BRFSS to account for differences in probabilities of selection and non-response. All analyses were performed using Stata, version 15 (StatCorp). This study was granted exemption by the Cambridge Health Alliance Institutional Review Board because BRFSS is a publicly available data source.

RESULTS

Sample

For our main study cohort, we analyzed data on 1,014,450 individuals with CVD/CVRFs. Patient characteristics of the study population are shown in Supplement Table 1. Pre- and post-ACA demographics were similar.

Changes in Insurance Coverage and Access Among Adults with CVD/CVRFs, Overall

For the overall study sample, insurance coverage increased by 6.9 percentage points (95% CI, 6.6 to 7.2) in adjusted analysis, reducing the number of people with CVD/CVRFs who were uninsured by 9.8 million (Table 1). All three measures of access to care also improved: ability to afford a physician visit by 3.6 percentage points (95% CI, 3.3 to 3.9), having an annual check-up by 2.2 percentage points (95% CI, 1.8 to 2.6), and having a personal physician by 1 percentage point (95% CI, 0.6 to 1.3), in adjusted analyses. Among those with incomes < $35,000/year, these percentage point gains were approximately twice as great (Table 1).

Table 1 Effect of the ACA on Changes in Health Insurance Coverage and Access to Care Among US Adults with CVD or Multiple CVRFs
Full size table

Changes in Outcomes Among Adults with CVD/CVRFs by Medicaid Expansion Status and in Individual States

Residents of Medicaid expansion states had higher levels of coverage and access in the pre-ACA period and experienced larger increases than residents of non-expansion states in the post-ACA period in unadjusted analyses (Table 2). In adjusted difference-in-differences analyses, residents of Medicaid expansion states experienced net percentage point increases over non-expansion states of 1.7 (95% CI, 1.0 to 2.3) for insurance coverage, 1.3 (95% CI, 0.6 to 1.9) for ability to afford a physician visit, 1.3 (95% CI, 0.5 to 2.1) for having an annual check-up, and 2.8 (95% CI, 2.0 to 3.5) for having a personal physician. Gains in expansion states among those with incomes < $35,000/year were, again, approximately twice as great.

Table 2 Effect of the ACA’s Medicaid Expansion on Changes in Health Insurance Coverage and Access to Care Among Adults with CVD or Multiple CVRFs
Full size table

Health insurance coverage increased in all states but we observed substantial variation by state and Medicaid expansion status (Table 3). The proportion of the population of people with CVD/CVRF that remains without insurance in the post-ACA era also varied substantially by state from a low of 6.1% in Massachusetts to a high of 27.0% in Texas (Table 3).

Table 3 Effect of the ACA’s Medicaid Expansion on Changes in Health Insurance Coverage and Access to Care Among Adults with CVD or Multiple CVRFs Among US States
Full size table

Changes in Coverage and Access Among Adults with CVD/CVRFs by Race and Ethnicity

All measures of coverage and access improved for each racial and ethnic group in unadjusted analyses (Table 4). However, when we examined whether pre-existing racial and ethnic disparities were reduced after the ACA, we observed a mixed picture. In adjusted difference-in-differences analyses, we found a 2.7 percentage point (95% CI, 2.6 to 2.9) increase in coverage among blacks relative to whites and a 5.7 percentage point (95% CI, 5.4 to 6.0) increase among Hispanics relative to whites. We also found that the white-black and white-Hispanic gaps in ability to afford a physician visit narrowed slightly (1.0 percentage points (95% CI, 0.8 to 1.1) and 1.5 percentage points (95% CI, 1.3 to 1.6), respectively). For the remaining access measures, we found only small changes of borderline significance.

Table 4 Effect of the ACA on Changes in Health Insurance Coverage and Access to Care Among Adults with CVD or Multiple CVRFs According to Race/Ethnicity
Full size table

Remaining Gaps in Coverage and Access Among Adults with CVD/CVRFs, Overall and by Medicaid Expansion Status, and Race and Ethnicity

Despite gains in the post-ACA era, large proportions of people with CVD/CVRFs continue to lack coverage and access. For example, nationally, 14.5% (Table 1) or 20.6 million people with CVD/CVRFs continued to lack insurance coverage after ACA implementation. In addition, 12.9% of black and 28.7% of Hispanic people lacked coverage in the post-ACA period (Table 4). We found that the persistent lack of coverage and access was particularly great for minorities in non-expansion states (Fig. 1). For example, among Hispanic people with CVD/CVRFs in non-expansion states post-ACA, 42% continued to lack insurance coverage, 25% could not afford a physician visit, 40% did not have a check-up in the last year, and 48% did not have a personal physician.

Figure 1
figure 1

Unadjusted percentage of respondents reporting health insurance and access outcomes, showing the pre-ACA percentage, change after ACA implementation, and percentage remaining without coverage or access in Medicaid expansion and non-expansion states. Figure 1 contains poor-quality text inside the artwork. Please do not re-use the file that we have rejected or attempt to increase its resolution and re-save. It is originally poor; therefore, increasing the resolution will not solve the quality problem. We suggest that you provide us the original format. We prefer replacement figures containing vector/editable objects rather than embedded images. Preferred file formats are .eps, .ai, .tiff, and .pdf.Attached is the original excel sheet with the original format with the graphs. I also attach a PDF version of the just the figures.

Full size image

Pre-ACA Placebo Time Trend and Sensitivity Analyses

Our placebo analyses comparing pre-ACA trends (2012 to 2013) in outcomes between Medicaid expansion and non-expansion states demonstrated that trends were parallel (Supplement Table 2). The placebo analysis of pre-ACA trends according to race/ethnicity showed a mixed picture with pre-ACA increases in coverage and ability to afford a physician among black and Hispanic relative to white individuals and an increase in having a personal physician among white relative to black and Hispanic individuals; trends were parallel for having a check-up in the past year (Supplement Table 3). All noted pre-trend differences were small (0.2–0.7 percentage points), however. Sensitivity analyses that included diagnoses of hypertension and hyperlipidemia as CVRFs (using data only from 2013 and 2015) produced results largely similar to the main analysis (Supplement Table 4). In sensitivity analyses examining the CVD and CVRF populations separately, we found similar changes in insurance coverage for the CVD and CVRF cohorts overall and by Medicaid expansion status (Supplement Table 5).

DISCUSSION

In this nationally representative cohort of adults with CVD/CVRFs, the ACA was associated with improvements in insurance coverage and measures of access to care, particularly among those with lower incomes. Residents of Medicaid expansion states experienced greater improvements for each outcome than residents of non-expansion states, again, with much larger effects seen among those with lower incomes. The ACA was also associated with a narrowing of racial and ethnic disparities in insurance coverage and ability to afford a physician visit. Despite these overall improvements, the proportion of adults with CVD/CVRFs who lack coverage and experience barriers to care in the post-ACA era remains substantial, particularly among residents of Medicaid non-expansion states and among racial and ethnic minorities.

Prior studies have shown that insurance coverage and access to care improved among people with multiple chronic conditions,34 diabetes,35 and cigarette smokers.36 Only one prior study (which was not peer reviewed) has examined the effect of the ACA on people with CVRFs,37 but was limited to the first year of the post-ACA period. It found a 4.4% decline in uninsurance, an effect that is smaller than but consistent with our findings. No studies have assessed the effect of the ACA on people with established CVD.

Our finding that the ACA improved insurance coverage and access to care for people with CVD/CVRFs, particularly those with lower incomes, is important since individuals with CVD/CVRFs are at elevated risk of mortality and typically have high health care needs. When patients with CVD/CVRFs are uninsured or face access barriers, they are more likely to experience a wide range of adverse outcomes. For example, the uninsured and those reporting financial barriers to care are more likely to delay hospital presentation for AMI,38 and during the first year of recovery after AMI, those reporting financial barriers are more likely to have lower quality of life scores, higher cardiac rehospitalization rates, and more angina.39 In addition, a lack of insurance coverage11 and low income8 are also both associated with failure to receive guideline-recommended outpatient medications in patients with established coronary artery disease.

Our finding that coverage and access improved to a greater degree in Medicaid expansion states—with a particularly large increase in coverage of 14.4 percentage points among lower income adults in expansion states—indicates that the benefits of the Medicaid expansion observed in previous studies of all non-elderly adults23 extend to people with CVD/CVRFs and are substantial. Similarly, our finding that pre-existing racial and ethnic disparities among adults with CVD or CVRFs narrowed is also important. Socioeconomic and racial and ethnic disparities in cardiovascular disease prevalence and outcomes have been recognized for decades12, 13 and elimination of these disparities has been a key goal of clinicians, policy experts, and the federal government.40, 41

The gains we observed, however, should be viewed in context. While we found that 9.8 million people with CVD/CVRFs gained coverage, 20.6 million continue to lack insurance coverage, including strikingly large proportions of racial and ethnic minorities and residents of Medicaid non-expansion states. There are likely multiple reasons why the ACA did not close these gaps more than it did, including states’ decisions to not expand Medicaid, ineligibility of millions of Hispanic persons for coverage due to their immigration status, perceived unaffordability of plans offered through marketplaces,42 and persistence of out-of-pocket expenses that are high relative to income.25

Our study has several limitations. First, our analysis of the overall pre-post ACA changes in outcomes did not employ a control group, as all states were affected by the ACA. Thus, these findings could have been influenced by unmeasured secular trends in factors affecting our outcomes such as, potentially, improvements in the economy or employment-related insurance gains not captured by our controlling for unemployment in our models. However, our analyses of the effects of the Medicaid expansion and changes in racial and ethnic disparities used a more robust study design (quasi-experimental difference-in-differences analysis) and were thus less susceptible to the influence of secular trends. Second, for three outcomes, we found pre-ACA trend differences; however, these were small and thus likely did not have a substantial impact on our results. Third, while we were not able to include individuals with hypertension and hyperlipidemia in our main analysis, our sensitivity analysis that included these CVRFs showed effect sizes that were very similar to our main results. Fourth, our study population combines two subpopulations with higher (CVD) and lower (CVRFs) underlying risks of cardiovascular mortality and of needing health care services. However, while modifiable CVRFs such as tobacco use, overweight/obesity, and physical inactivity carry lower risks than established CVD, these CVRFs alone account for approximately 40% of all US deaths.6 In addition, our sensitivity analyses that examined CVD and CVRF populations separately found comparable effect sizes. Fifth, the BRFSS is only conducted in English and Spanish which may limit the applicability of our findings to populations speaking these languages. Lastly, the BRFSS is a telephone survey and not all people eligible to participate could be reached (response rate for the study years varied from 45.2 to 47.2%) potentially leading to nonresponse bias. However, the use of BRFSS-provided survey weights to account for non-response likely reduced this potential effect.

CONCLUSION

Our findings have several implications relating to the ongoing health insurance expansion debate. First, the gains in access and coverage we found represent significant progress and likely have substantial health ramifications for people with CVD/CVRFs. Second, states’ failure to expand Medicaid has likely resulted in a substantial missed opportunity to improve health outcomes in people with CVD/CVRFs, particularly because they comprise larger shares of the populations in most non-expansion states43 Third, repealing all or significant parts of the ACA without a comparable alternative would risk the loss of coverage and access to care for millions of Americans with CVD/CVRFs. This is a realistic concern since several changes that could increase uninsurance have already recently been adopted, including repeal of the ACA’s individual mandate, cuts in advertising and outreach during marketplace open enrollment periods, and the imposition of Medicaid work requirements in some states.44 Finally, even if the ACA remains in place, many patients with CVD/CVRFs will continue to be uninsured and without access.45 In order to substantially improve outcomes for Americans with CVD/CVRFs, a much more comprehensive health reform than the ACA would be required.