Effects of the 2014 Affordable Care Act Medicaid Expansions on Health Care Access and Health Status of Poor Adults Aged 60–64 Years: Evidence From the First 6 Years

Abstract

Background and Objectives

Little is known on the effects of the Affordable Care Act (ACA) Medicaid expansions on health care access and health status of adults closest to 65. This study examines the effects of ACA Medicaid expansion on access and health status of poor adults aged 60–64 years.

Research Design and Methods

The study employs a difference-in-differences design comparing states that expanded Medicaid in 2014 under the ACA and nonexpansion states over 6 years postexpansion. The data are from the 2011–2019 Behavioral Risk Factor Surveillance System for individuals aged 60–64 years below the Federal Poverty Level.

Results

Having any health care coverage rate increased by 8.5 percentage points (p < .01), while the rate of forgoing a needed doctor’s visit due to cost declined by 6.6 percentage points (p < .01). Similarly, rates of having a personal doctor/provider and completing a routine checkup increased by 9.1 (p < .01) and 4.8 (p < .1) percentage points, respectively. Moreover, days not in good physical health in the past 30 declined by 1.5 days (p < .05), with suggestive evidence for decline in days not in good mental health and improvement in self-rated health.

Discussion and Implications

The ACA Medicaid expansions have improved health care access and health status of poor adults aged 60–64 years. Expanding Medicaid in the states that have not yet done so would reduce barriers to care and address unmet health needs for this population. Bridging coverage for individuals aged 60–64 years by lowering Medicare eligibility age could have long-term effects on well-being and health services utilization.

Background

Poor adults within a few years from reaching 65 are a particularly vulnerable age group in the United States. Not yet eligible for Medicare, they have historically had high uninsured rates due to low eligibility levels for Medicaid and greater difficulty in accessing private coverage than higher-income adults (Smolka et al., 2012). They are also less likely to have employer-sponsored coverage due to higher rates of unemployment, part-time employment, and less skilled jobs (Mikelson et al., 2017) and less able to purchase coverage independently. Furthermore, this group has a greater prevalence of chronic conditions, needs more prescription drugs, and is at greater risk of experiencing major health events and incurring substantial out-of-pocket medical expenses than same-age, higher-income individuals (Smolka et al., 2012).

The Affordable Care Act (ACA) increased Medicaid eligibility to 138% Federal Poverty Level (FPL) for adults younger than 65 in states that expanded Medicaid under the ACA. By 2021, 38 states plus Washington, DC, have expanded Medicaid under the ACA (Kaiser Family Foundation [KFF], 2021d). Before this expansion, Medicaid income eligibility varied across states, but was overall much lower than the expansion limit. Across states that expanded in 2014, average preexpansion income eligibility in 2013 for parents and childless adults was 104% and 24% of FPL, respectively. In states that had not expanded under the ACA by 2019, average income eligibility for parents and childless adults in that year was 43% and 0% of FPL, respectively (KFF, 2021b, 2021c).

The Medicaid expansions led to significant increases in coverage and access to care for low-income individuals in expansion states (Courtemanche et al., 2018; Gruber & Sommers, 2019; Guth et al., 2020; Simon et al., 2017; Wehby & Lyu, 2018). Nearly 15 million in the newly eligible income levels were enrolled in 2020 (KFF, 2021a). In contrast, there were nearly four million individuals below 138% of FPL in nonexpansion states who were not income-eligible for Medicaid, 2.2 million of whom are below the FPL (Garfield, Orgera et al., 2021). The increase in Medicaid enrollment in expansion states was observed across demographic groups including by age, race/ethnicity, and gender (Lyu & Wehby, 2019). There is also emerging evidence of improvement in health status (Cawley et al., 2018; Semprini et al., 2020; Simon et al., 2017; Sommers et al., 2016, 2017). The evidence on the ACA Medicaid expansion effects on access and health status, however, is mostly for averaged effects across the entire age range of newly eligible adults or wide ranges of age subgroups. There is less evidence on those close to 65, the eligible age for Medicare. A few studies examined the ACA Medicaid expansion effects on low-income individuals aged 50–64 years (Courtemanche et al., 2017; McInerney et al., 2020; Miller et al., 2021; Semprini et al., 2020; Tipirneni et al., 2021; Van Houtven et al., 2020; Wehby & Lyu, 2018). For this age group, there is evidence of an increase in Medicaid coverage and a drop in uninsured rate (Courtemanche et al., 2017; McInerney et al., 2020; Miller et al., 2021; Tipirneni et al., 2021; Wehby & Lyu, 2018), better self-rated health status (Semprini et al., 2020), reduction in any work limitations from health (Tipirneni et al., 2021), improvement in activities of daily living (McInerney, 2020), and reduction in disease-related deaths (Miller et al., 2021). There is also an increase in use of long-term care (Van Houtven et al., 2020) and higher likelihood of hospitalizations (Tipirneni et al., 2021), suggesting previously unmet needs.

In this study, we focus on poor adults aged 60–64 years and examine how the ACA Medicaid expansions affected their access to health care and health status. There are several reasons to focus on this age group. First, this age group has a greater prevalence of chronic conditions and more unmet health care needs than younger adults combined in the same broader age group of 50–64 in prior studies. In 2011–2019, 68% of adults aged 60–64 years below the FPL reported having at least two chronic conditions, and 47% reported three or more conditions, compared to 53% and 34%, respectively, of those aged 50–59 of similar income (Behavioral Risk Factor Surveillance System [BRFSS], 2011–2019). Therefore, poor adults aged 60–64 years may benefit differently and possibly more in access and health gains from Medicaid expansions than younger adults. Second, individuals aged 60–64 years are closer to retirement decisions than younger individuals and their decisions are partly dependent on their health insurance coverage options. This difference could also modify the expansion effects on access and health status for this group. Third, reducing Medicare eligibility age to 60 has been a commonly debated policy agenda (Levitt, 2021) and was recently introduced in a congressional bill (Jayapal, 2021). The ACA Medicaid expansions provide an opportunity to understand the effects of coverage expansions for poor adults aged 60–64 years, and this evidence can inform the debate on lowering Medicare eligibility age.

One study found an increase in Medicaid coverage and decline in the uninsured rate from the ACA Medicaid expansions for individuals aged 60–64 years using data through 2018 (Duggan et al., 2020). However, that study did not examine changes in access (except coverage) or health status. To our knowledge, we provide the first evidence on changes in access and health status from the ACA Medicaid expansions specifically for the 60–64 age group. Another contribution of our study is including data over 6 years from the 2014 expansions (2014–2019), whereas most previous studies (combining individuals aged 50–64 years) only covered the first 2–3 years of the expansions. In 2019, 14.5% of adults aged 60–64 years were enrolled in Medicaid, but 7.8% were uninsured; 10% of the uninsured are below the FPL and ineligible for Medicaid in nonexpansion states (Garfield, Rae et al., 2021). Our study sheds light on the implications of this continuing disparity in coverage for access to care and health status for this age group.

Conceptual Framework and Hypotheses

An increase in income-based Medicaid eligibility can improve access to care and health status of adults aged 60–64 years who gain coverage. Before the ACA expansions, most states had low eligibility levels, particularly for childless adults. Many adults in this age group are unable to have private coverage due to unemployment or less generous employment-based health benefits (including out-of-pocket cost sharing) and not affording independent coverage. Without access to Medicaid, many go uninsured with constrained access to care including forgoing needed services and preventive care. Therefore, we hypothesize that gaining Medicaid from the ACA expansions will increase access to health care by reducing out-of-pocket expenditures, which would reduce avoiding necessary care due to cost and an increase in seeking care from a usual source and use of preventive services such as routine checkups. As noted above, these effects have been documented for broader age groups of nonolder adults, although not specifically for individuals aged 60–64 years.

Similarly, we hypothesize that the increased access would improve health status in both physical and mental domains. This improvement might happen gradually for some individuals with better management of chronic conditions (e.g., due to access to prescription drugs), or immediately for others if it addresses previously undiagnosed health problems and initiates effective treatments. Furthermore, the greater financial security associated with gaining health insurance can reduce stress and increase resources to meet other needs that affect health such as nutrition or housing. The reduction in stress can also improve mental health, and such effects might take shorter time to materialize compared to physical health changes. Prior evidence on the ACA Medicaid expansions suggests that effects on health status among the broader age range of 50–64 years are more prominent over time (Semprini et al., 2020).

Method

Data

We use data from the BRFSS from 2011 through 2019. BRFSS is a telephone-based large annual health survey of a nationally representative cross-sectional sample of more than 300,000 adults per year in the United States. The survey is coordinated by the Centers for Disease Control and Prevention and implemented by state health departments (all 50 states and the District of Columbia). BRFSS collects data on health status, health care access, health behavior, and socioeconomic and demographic characteristics.

Sample

We include adults aged 60–64 years with household income below 100% of the FPL as part of the study sample. We exclude individuals between 100% and 138% of the FPL from the main sample because they were eligible for subsidies in the ACA private insurance marketplace in nonexpansion states. In a sensitivity check, however, we include those individuals back to the sample. We use the midpoints of the household income categories reported in BRFSS to calculate household income as a percentage of FPL considering family size and annual changes in FPL (Lyu & Wehby, 2019; Semprini et al., 2020). Because BRFSS did not ask individuals in the cellphone sample about the number of adult household members, we approximate that number based on marital status (two adults for married or cohabiting, and one for single). Because BRFSS modified its sampling strategy in 2011 to include a cellphone sample, and to avoid any conflating effects during early recovery from the Great Recession, we start the sample period in 2011 and include data through 2019, the most recent year of data currently available. Depending on the outcome measure (described below), the analytical sample with complete data ranges between 18,180 and 25,761. Descriptive data on the study sample are given in Supplementary Table 1.

Data on Medicaid Expansions

We obtain data on state ACA Medicaid expansions from the KFF (2021d). We include in the expansion group states that expanded in 2014. We focus on those early expansions in 2014 so that we have six postexpansion years for all expansion states. For that reason, we exclude from the sample seven states (Alaska, Indiana, Louisiana, Maine, Montana, Pennsylvania, and Virginia) that expanded after 2014. Four of the states that announced expansion under the ACA in 2014 (Delaware, Massachusetts, New York, Vermont) and Washington, DC, had full or near full expansions before 2014. Therefore, we also exclude these four states and Washington, DC, from the main sample. As a sensitivity check, we include those four states and Washington, DC, back in the control (nonexpansion group) because they had expansions before 2014. Finally, we exclude Wisconsin from the main sample because it expanded eligibility to 100% of the FPL in 2014 but not under the ACA. As another sensitivity check, we include Wisconsin back in the expansion group. The details of state assignments into expansion or nonexpansion groups and excluded states are given in Supplementary Table 2.

Outcome Measures

We utilize four measures for coverage and access to care that have been commonly studied with the Medicaid expansions. These are binary (0/1) indicators for having any health care coverage, forgoing a needed doctor’s visit due to cost in the past 12 months, having a personal doctor/health care provider, and having a routine checkup with a doctor in the past 12 months.

We also include four measures of health status. The first measure is self-rated health on five categories from 1 (poor) to 5 (excellent). We use both the 5-point scale (1–5) and a binary (0/1) indicator for good/very good/excellent health versus fair/poor health. The second indicator is days not in good physical health in the past 30 days, and the third is a similar measure for mental health. The last measure is number of days limited by bad physical or mental health in the past 30 days (reported directly by survey respondents and not constructed from the two other days’ measures for physical and mental health).

Study Design and Estimation

The research design is a difference-in-differences model comparing changes in outcomes within expansion states before and after the 2014 expansions to outcome changes within the nonexpansion states over the same period. The difference-in-differences regression is specified as follows:

$${\displaystyle \begin{array}{ll}{\displaystyle Y_{ist}}& ={\mathrm{\beta }}_0+{\mathrm{\beta }}_1(\mathrm{M}\mathrm{E}\mathrm{D}\mathrm{I}\mathrm{C}\mathrm{A}\mathrm{I}{\mathrm{D}}_s\times \mathrm{P}\mathrm{O}\mathrm{S}{\mathrm{T}}_t)+{\mathrm{\beta }}_2{X}_{ist}\\ & {\displaystyle +{\mathrm{\theta }}_s+{\mathrm{\delta }}_t+{\mathrm{\epsilon }}_{ismt}}\end{array}}$$ (1)

In this model (Equation 1), Y_ist is one of the access or health outcomes mentioned above for individual i in state s in year t. MEDICAID_s is a binary indicator equal to 1 for states that expanded Medicaid in 2014 and to 0 for nonexpansion states. POST_t is a binary indicator equal to 1 for years 2014 through 2019 and 0 for years 2011–2013. X_ist includes the following individual-level covariates: sex, race/ethnicity, education, marital status, whether there are children at home, and month of survey fixed effects (0/1 indicators). θ _s represents state fixed effects (0/1 indicators), which capture time-invariant differences between states, and replace adding MEDICAID_s as a separate term in the regression. δ _t represents year fixed effects, which represent national trends shared between states, and replace adding POST_t as a separate variable. β ₁ is the difference-in-differences estimate of the Medicaid expansion effect on outcome Y.

The model (Equation 1) assumes a constant effect of the Medicaid expansions over time (i.e., the effect is the same in each year since enactment). To evaluate whether the Medicaid expansion effects are changing over time after the expansion, we estimate a second model that includes the Medicaid expansion effect separately for the first 3 years (2014–2016) and years four through six (2017–2019) after the expansion. This model is specified as follows:

$${\displaystyle \begin{array}{ll}{\displaystyle Y_{ist}=}& {\mathrm{\alpha }}_0+{\mathrm{\alpha }}_1(\mathrm{M}\mathrm{E}\mathrm{D}\mathrm{I}\mathrm{C}\mathrm{A}\mathrm{I}{\mathrm{D}}_s\times \mathrm{P}\mathrm{O}\mathrm{S}{\mathrm{T}}_{t}1)\\ & {\displaystyle +{\mathrm{\alpha }}_2(\mathrm{M}\mathrm{E}\mathrm{D}\mathrm{I}\mathrm{C}\mathrm{A}\mathrm{I}{\mathrm{D}}_s\times \mathrm{P}\mathrm{O}\mathrm{S}{\mathrm{T}}_{t}2)+{\mathrm{\alpha }}_3{X}_{ist}+{\mathrm{\theta }}_s}\\ & {\displaystyle +{\mathrm{\delta }}_t+r_{ist}}\end{array}}$$ (2)

In this model, POST_t1 is a binary (0/1) variable equal to one during 2014–2016 (0 otherwise), while POST_t2 is a binary (0/1) variable equal to one during 2017–2019 (0 otherwise). All other variables are as defined previously. α ₁ is the Medicaid expansion effect within the first 3 years after expansion, while α ₂ is the expansion effect in Years 4–6.

The difference-in-differences design assumes that outcomes would have changed similarly between the expansion and nonexpansion states in 2014–2019 had the expansion not happened. To check this assumption, we compare outcome trends between expansion and nonexpansion states in 2011–2013 using a regression model similar to the above model but limiting the data to 2011–2013 and interacting the Medicaid expansion group indicator (MEDICAID) with one binary indicator for 2011, and another for 2012 (with 2013 as the reference category). We then test these interactions, which if significant would indicate differential pretrends that might bias the difference-in-differences estimates.

We estimate the regression models using weighted least squares with the BRFSS individual sampling weights. Standard errors are clustered at the state level. All analyses are conducted using Stata 17.

Results

Medicaid Expansion Effects on Access to Care

Table 1 reports the estimated ACA Medicaid expansion effects over 2014–2019 on access to care from Equation 1. The estimates are for 60–64 years old adults below the FPL. Relative to nonexpansion states, expansion states had an increase in coverage compared to nonexpansion states by 8.5 points (t value = 7.87; df = 25,760; p < .01). Following Medicaid expansion, the likelihood of forgoing a needed doctor’s visit due to cost in the past 12 months declined by 6.6 percentage points in expansion states compared to nonexpansion states (t value = −2.73; df = 25,754; p < .01). Similarly, expansion states had an increase in the likelihood of having a personal doctor or health care provider by 9.1 percentage points (t value = 3.45; df = 25,739; p < .01). Also, there was a marginally significant increase in the likelihood of completing a routine checkup with a doctor in the past 12 months by 4.8 percentage points (t value = 1.95; df = 25,135; p < .1).

Table 1.

Difference-in-Differences Unstandardized Regression Coefficients Representing Estimates of the Affordable Care Act Medicaid Expansion Effects on Health Care Access of Adults Aged 60–64 Below 100% Federal Poverty Level in Behavioral Risk Factor Surveillance System 2011–2019

Access to care outcomes	Unstandardized regression coefficients (SE)	Outcome mean in 2011–2013 (preexpansion)
Having any health care coverage	0.085*** (0.011)	0.679
Forwent a needed doctor’s visit due to cost in the past 12 months	−0.066*** (0.024)	0.324
Having a personal doctor/health care provider	0.091*** (0.026)	0.808
Completed a routine checkup with a doctor in the past 12 months	0.048* (0.024)	0.710

Notes: The models adjust for race/ethnicity, gender, marital status, education, and children at home and include fixed effects for survey month, year, and state. Standard errors are clustered by state. Sample size ranges between 25,136 and 25,761 depending on outcome.

*p < .10, **p < .05, ***p < .01.

Table 2 reports the Medicaid expansion effect estimates separately for Years 1–3 and 4–6 after expansion (Equation 2). As a whole, effects were generally comparable between the two periods, especially for coverage and having a personal doctor or health care provider. Estimates were smaller and not statistically significant in the first 3 years for forgoing a needed doctor’s visit due to cost and routine checkups. However, estimates were not significantly different between the two periods.

Table 2.

Difference-in-Differences Unstandardized Regression Coefficients Representing Estimates of the Affordable Care Act Medicaid Expansion Effects on Health Care Access of Adults Aged 60–64 Below 100% Federal Poverty Level in Behavioral Risk Factor Surveillance System 2014–2016 and 2017–2019 Versus 2011–2013

	Unstandardized regression coefficients (SE)		Outcome mean in 2011–2013
Access to care outcomes	1–3 years postexpansion	4–6 years postexpansion
Having any health care coverage	0.084*** (0.019)	0.087*** (0.021)	0.679
Forwent a needed doctor’s visit due to cost in the past 12 months	−0.058 (0.039)	−0.074*** (0.019)	0.324
Having a personal doctor/health care provider	0.087*** (0.031)	0.095*** (0.027)	0.808
Completed a routine checkup with a doctor in the past 12 months	0.034* (0.020)	0.061* (0.032)	0.710

Notes: The models adjust for race/ethnicity, gender, marital status, education, and children at home and include fixed effects for survey month, year, and state. Standard errors are clustered by state. Sample size ranges between 25,136 and 25,761 depending on outcome.

*p < .10, **p < .05, ***p < .01.

Medicaid Expansion Effects on Health Status

Table 3 presents estimates of the Medicaid expansion effects in 2014–2019 on health status of adults aged 60–64 years below the FPL (Equation 1). Days not in good physical health in the past 30 days declined by 1.5 days in expansion states (t value = −2.36; df = 24,955; p < .05). This decline is notable because it represents nearly 13% of the mean number of days not in good physical health in the sample in 2011–2013. Also, there was a marginally significant (t value = 1.93; df = 25,710; p < .1) improvement in self-rated health (by 0.08 points on the 1–5 points scale) in expansion states relative to nonexpansion states. Estimates for the other outcomes were not statistically significant.

Table 3.

Difference-in-Differences Unstandardized Regression Coefficients Representing Estimates of the Affordable Care Act Medicaid Expansion Effects on Health Status of Adults Aged 60–64 Below 100% Federal Poverty Level in Behavioral Risk Factor Surveillance System 2011–2019

Health status outcomes	Unstandardized regression coefficients (SE)	Outcome mean in 2011–2013 (preexpansion)
Five-category health status ranking from poor (1) to excellent (5)	0.076* (0.039)	1.40
Good/very good/excellent health versus fair/poor health	0.012 (0.019)	0.44
Days not in good physical health in past 30 days	−1.534** (0.649)	11.81
Days not in good mental health in past 30 days	−1.135 (0.678)	7.83
Days limited by bad physical or mental health in past 30 days	−0.915 (0.921)	11.70

Notes: The models adjust for race/ethnicity, gender, marital status, education, and children at home and include fixed effects for survey month, year, and state. Standard errors are clustered by state. Sample size ranges between 18,130 and 25,711 depending on outcome.

*p < .10, **p < .05, ***p < .01.

Table 4 reports estimates of the Medicaid expansion effects on health status separately for Years 1–3 and 4–6 after expansion. The decline in days not in good physical health was larger and statistically significant (t value = −2.53; df = 24,955; p < .05) only in Years 4–6. In contrast, there was a larger and statistically significant (t value = −2.26; df = 25,063; p < .05) decline in days not in good mental health in Years 1–3. Improvement in self-rated health on the 5-point scale was marginally significant (t value = 1.78; df = 25,711; p < .1) in Years 4–6. However, differences in estimates between the two periods were not statistically significant.

Table 4.

Difference-in-Differences Unstandardized Regression Coefficients Representing Estimates of the Affordable Care Act Medicaid Expansion Effects on Health Status of Adults Aged 60–64 Below 100% Federal Poverty Level in Behavioral Risk Factor Surveillance System 2014–2016 and 2017–2019 Versus 2011–2013

Health status outcomes	Unstandardized regression coefficients (SE)		Outcome mean in 2011–2013
	1–3 years postexpansion	4–6 years postexpansion
Five-category health status ranking from poor (1) to excellent (5)	0.067 (0.049)	0.085* (0.048)	1.40
Good/very good/excellent health versus fair/poor healtha	−0.011 (0.022)	0.036 (0.022)	0.44
Days not in good physical health in past 30 days	−1.161 (0.795)	−1.910** (0.756)	11.81
Days not in good mental health in past 30 days	−1.240** (0.550)	−1.030 (0.913)	7.83
Days limited by bad physical or mental health in past 30 days	−1.373 (1.310)	−0.460 (0.717)	11.70

Notes: The models adjust for race/ethnicity, gender, marital status, education, and children at home and include fixed effects for survey month, year, and state. Standard errors are clustered by state. Sample size ranges between 18,130 and 25,711 depending on outcome.

^aCoefficients for 1–3 years postexpansion and 4–6 years postexpansion were different from each other at p < .05.

*p < .10, **p < .05, ***p < .01.

Preexpansion Trend Checks

As noted above, we compare outcome trends between expansion and nonexpansion states before 2014 as a check of the difference-in-differences design. Supplementary Table 3 reports the estimates of preexpansion trends by Medicaid expansion status. For all but one access outcome (health coverage), there are no statistically significant differences between expansion and nonexpansion states. There is a statistically significant difference in the 2012–2013 change in coverage rates. Overall, these results support the validity of the difference-in-differences estimates for the access measures (with the noted caveat for insurance coverage).

Supplementary Table 4 reports the pretrends of health status measures by Medicaid expansion status. There are no statistically significant differences in pretrends of days not in good health between expansion and nonexpansion states (based on joint year tests or individual year differences). There is a difference in the self-rated health pretrend (although it is opposite in direction from the postexpansion difference). As a whole, these results support the estimated effects of the Medicaid expansions on the health measures (with the noted caveat on self-rated health).

Sensitivity Checks

We perform four sensitivity checks. The first check adds Wisconsin as a treatment state because it expanded eligibility to 100% FPL in 2014 (although not under the ACA). The second check adds into the control group four states (Delaware, Massachusetts, New York, and Vermont) and Washington, DC, that announced expansion under the ACA in 2014 but already had full or near full expansions before 2014. The third check expands the income range in the sample from 100% FPL up to 138% of the FPL. Finally, the fourth check drops the survey sampling weights. The results for access from these analyses are generally similar to the main results (Supplementary Table 5). The results for health status are also generally comparable to the main results when adding Wisconsin or the states with full or near full expansions before 2014 with the estimates for days not in good health becoming slightly smaller and marginally significant (Supplementary Table 6). However, adding individuals 100%–138% of the FPL and not using sampling weights results in markedly smaller and mostly nonsignificant estimates (Supplementary Table 6), suggesting possible treatment effect heterogeneity across income, states, or time.

Discussion

This study provides evidence on the ACA Medicaid expansion effects on access and health status of adults aged 60–64 years below the FPL. Our study is among the first to examine these effects for this specific age group and over the first 6 years after the 2014 expansions. We find that the expansions have increased coverage and improved health care access by reducing forgoing a needed doctor’s visit due to cost in the past year and increasing the likelihood of having a personal doctor/health care provider. The changes represent about 20% and 11% of the 2011–2013 sample means of forgoing care due to cost and having a personal doctor or health care provider, respectively. There is also suggestive evidence of an increase in routine checkups with a doctor (the estimate is marginally significant). For health outcomes, there are declines in days not in good physical health concentrated in Years 4–6 and in days not in good mental health in Years 1–3 postexpansion representing nearly 16% of these outcome sample means in 2011–2013. There is also suggestive evidence for improvement in self-rated health (the estimate is marginally significant). Despite observing some estimates to be statistically significant only in Years 1–3 or 4–6 after the expansions, there is no statistical evidence of effects differing over time. Taken as a whole, these results suggest important benefits in access to medical care and health status of the poor adults aged 60–64 years from the recent ACA Medicaid expansions.

The study offers evidence that realized access gains among individuals aged 60–64 years have measurable effects on health status. This evidence suggests access and health benefits for this group if Medicare eligibility age is lowered to 60. This is particularly important for poor adults aged 60–64 years who have more constrained access to employer-sponsored coverage and cannot afford purchasing private coverage on their own. The effects we find on access and health status might have additional long-term effects on health and health care utilization including long-term care and hospitalizations. Examining these long-term effects in future research is important for a comprehensive cost–benefit assessment of policy proposals to change Medicare eligibility age or other policies to bridge coverage gaps for adults aged 60–64 years.

Our results are overall consistent with the broader literature on Medicaid expansions but offer new insights for effects on poor adults aged 60–64 years. For this group, we find improvement in both days not in good physical health and days not in good mental health, as well as suggestive evidence for improvement in self-rated health. In contrast, recent work examining the Medicaid expansion effects on similar health outcomes for poor adults aged 18–64 years using BRFSS data finds improvement in self-rated health but smaller and statistically insignificant effects on days not in good physical or mental health (Semprini et al., 2020). Similarly, another recent study using data from the Health and Retirement Study for low-educated individuals aged 51–64 years finds no effects of the ACA Medicaid expansions on several access measures including having a usual source of care and forgoing care due to cost, among others (Tipirneni et al., 2021). Furthermore, that study finds no effects on self-rated health and changes in depressive symptoms, although they find a decline in the likelihood of health limiting any work. Two other studies of individuals aged 50–64 years have reported improvement in activities of daily living (McInerney, 2020) and reduction in disease-related mortality (Miller et al., 2021) from the Medicaid expansions. Compared to these studies, the findings from our study suggest broad benefits for individuals aged 60–64 years based on general measures of access and health status beyond those reported in broader age ranges. The suggested improvement in health might be due to new diagnoses and treatment of previously undiagnosed chronic conditions or better management of previously diagnosed illnesses. Further research using other data on diagnoses and utilization of medical services and medication prescriptions such as Medicaid claims data would be useful to evaluate such potential mechanisms.

Our study has some limitations. BRFSS does not ask about the type of health provider or setting of care, so we cannot measure those as outcomes. Similarly, there are no data on timing of chronic condition diagnoses or medication use to explore changes due to the Medicaid expansions. However, BRFSS offers the advantage of a relatively large national sample of individuals aged 60–64 years in each year of the study period compared to some other data sets such as the Medical Expenditure Panel Survey or the Health and Retirement Study. Finally, understanding the long-run effects of the Medicaid expansions particularly on health status requires a longer timeframe.

In conclusion, this study provides evidence that the ACA Medicaid expansions have improved access to medical care and health status of poor adults aged 60–64 years. These effects indicate missed opportunities to address barriers to care and unmet health needs for this population in states that have not yet expanded Medicaid. The evidence also highlights the importance of policies that bridge coverage for this age group including lowering Medicare eligibility age.

Funding

The work was partly supported by the National Institutes of Health (grant number 1R03 DE02804101).

Conflict of Interest

The authors have declared no conflicts of interest for this work.