Changes in Insurance Coverage and Access to Care for Young Adults in 2017

Abstract

Purpose:

Previous work has demonstrated the Affordable Care Act (ACA) increased young adults’ health care access during its first years. However, it is unclear if these trends continued through 2017; recent policies enacted by the Trump administration may have decreased the ACA’s effectiveness. Our purpose was to determine changes in young adults’ health care access during the transition from Obama to Trump administrations.

Methods:

Data on noninstitutionalized U.S. young adults (18–24 years) was obtained from the Behavioral Risk Factor Surveillance System 2011–2017 (N = 173,848). We used interrupted time series and difference-in-differences analysis to quantify changes in self-reported insurance coverage, access to a primary care physician, and unmet care because of cost from 2013 to 2017.

Results:

Young adults’ health care access continued to improve through 2016; for instance, the percentage of respondents experiencing uninsurance declined by 8.7 points from 2013 to 2016 (95% confidence interval [CI] −9.4 to −8.0). However, these trends began to reverse and from 2016 to 2017, the percentage of young adults who experienced uninsurance increased by 1.4 points (95% CI .6–2.1), not having a personal doctor increased by 1.1 points (95% CI .2–2.0), and unmet care because of cost increased by 1.0 points (95% CI .3–1.7). The 2017 declines in access were concentrated in states which did not expand Medicaid and in households earning above 138% of federal poverty level.

Conclusions:

Health care access declined for young adults in 2017, after several years of improvements. These changes correspond with recent policy actions, which may have weakened the ACA’s reforms.

The Affordable Care Act (ACA) sought to increase insurance coverage through a variety of provisions, which positively impacted health insurance coverage for young adults. The ACA required all employer-sponsored and individual Marketplace plans that offer dependent child coverage to make that coverage available until the child reaches the age of 26 years. In addition, the ACA provided substantial federal funding for states to expand Medicaid household income eligibility up to 138% of Federal Poverty Level (FPL), created federal and state insurance exchanges with premiums capped on a sliding scale according to income, among other requirements [1]. Increasing insurance coverage for young adults is especially important, as poor health during young adulthood has important implications for future health and economic outcomes [2]. Access to care for young adults is also vital to well-functioning insurance markets because their participation helps to keep premiums more stable and affordable for all enrollees [3]. Previous work has demonstrated that passage of the ACA in 2010 was associated with increased overall insurance coverage for children and young adults during its first few years. For instance, Sommers [4] found that 3.1 million adults aged 19–25 years gained insurance coverage, and McMorrow et al. [1] found that the uninsurance rate for young adults fell by 11 percentage points from 2009 to 2014. These gains in coverage for young adults also occurred in certain vulnerable subpopulations, such as oncology and mental health patients [5,6].

It is unclear if young adults’ access has continued to improve, stagnated, or potentially reversed in recent years. Several states have implemented the Medicaid expansion since 2014, which could further improve access. However, policy actions under-taken by the Trump administration may have weakened the ACA’s reforms. These included repeal of the individual insurance mandate penalty (effective 2019) [7], expansion of short-term insurance options, which are exempt from most ACA consumer protections [8], cancellation of cost-sharing reduction payments [9], and a series of videos and statements produced by the Department of Health & Human Services arguing that the ACA has hurt consumers [10]. Misinformation may also have affected rates of insurance take-up; a 2017 Morning Consult/Politico survey found nearly one in four Americans incorrectly believed that the ACA was partially repealed, whereas 15% believed it was totally repealed [11]. The number of insurers participating in the ACA’s individual Marketplaces has also declined in 2016 and 2017, and insurance premiums continued their several-year climb [12,13]. Recent work has shown either no significant declines or slight declines in insurance coverage under the Trump administration, but these studies focus on the U.S. adult population more broadly [14–16].

Our goal was to quantify changes in health care access for young adults from 2013 through 2017, incorporating additional outcomes beyond insurance coverage. The study period allows us to observe changes through the final year of the Obama administration as well as transition to the Trump administration in 2017. Previous work has demonstrated all three outcomes have important implications for health services utilization, chronic disease management, and mortality [17–19]. We also stratified our analyses to determine whether household income group or state Medicaid expansion status mediate the observed changes in access.

Methods

Sample selection

Data for this study were provided by the 2011–2017 BRFSS. BRFSS is an annual telephone survey of noninstitutionalized U.S. residents regarding their health-related risk behaviors, chronic health conditions, and use of preventive services. Approximately 400,000 households respond to the survey each year, and their responses are weighted to ensure national representativeness. BRFSS weights its responses to match known proportions of age, sex, categories of ethnicity, geographic regions within states, marital status, education level, home ownership, and type of phone ownership; the BRFSS sampling and weighting methodologies have been described previously [20,21]. Response rates for each year ranged from 45% to 50%. We excluded data from U.S. territories and limited our sample to adults aged 18–24 years, the youngest age category recorded in the BRFSS. Our final sample included 173,848 young adults; the demographic characteristics of our analytic sample are summarized in Table 1.

Table 1.

Characteristics of the study sample (N = 173,848)

Variable	Frequency	Weighted (%)a
Female	85,445	48.6
Married	21,300	9.4
Children in household	38,026	36.4
Medicaid expansion state	104,363	61.5
Household income group
<138% FPL	77,712	45.3
138% FPL to 400% FPL	67,230	38.0
>400% FPL	28,906	16.6
Race/ethnicity
White	111,608	54.9
Black	17,193	13.2
Hispanic	25,179	20.9
Other	13,492	8.8
Multiracial	6,376	2.2
Educational attainment
Less than high school	14,814	13.4
High school grad	64,050	36.4
Some college	64,806	39.0
College graduate	30,178	11.2
Employment status
Unemployed	17,300	10.8
Employed	94,085	49.4
Not in labor force	62,463	39.8
Homeowner	46,429	36.4

Source: Authors’ analysis of data noninstitutionalized U.S. adults aged 18–24 years who responded to the 2011–2017 from the Behavioral Risk Factor Surveillance System (BRFSS). Household income group is defined as an imputed percentage of Federal Poverty Level (FPL); see the Appendix for more details.

^aBRFSS weights its responses to match known proportions of age, sex, categories of ethnicity, geographic regions within states, marital status, education level, home ownership, and type of phone ownership.

Study variables

Our analyses focused on three measures of health care access contained within BRFSS: whether the respondent (1) had any form of insurance coverage, (2) had a personal doctor, or (3) unmet care because of cost within the previous 12 months. Question texts for these outcomes are contained in Table A1 (Supplementary Data). We reverse coded the first two outcomes (yes = 0, no = 1) to indicate uninsurance and lack of a personal doctor.

We included a variety of demographic controls in adjusted models: respondent sex, educational attainment, household size, presence of children in the household, race/ethnicity, home-ownership, and veteran status. Following prior research in this area, we used an imputed percentage of the FPL to classify respondents into three household income groups: <138% (the income cutoff for eligibility under the Medicaid expansion), 138%–400% (corresponding to eligibility for subsidies on the ACA Marketplaces), and >400% (not eligible for Medicaid or subsidies). This process is described in more detail in Sommers et al. [22] and the Appendix. Models also included indicator variables for respondent state, state Medicaid expansion status, and survey year, with 2013 set as the reference period. To be counted as an expansion state in this analysis, a state must have implemented the Medicaid expansion with an effective date in 2017 or earlier [23]. Hot-deck imputation was used to replace missing answers to specific survey questions and reduce potential nonresponse bias [24].

Analytic approach

Our analysis proceeded in three steps. We first conducted simple linear regressions of our outcomes and indicators for each year 2011–2017, allowing us to explicate unadjusted trends in health care access during the study period.

Next, we constructed adjusted interrupted time series models controlling for the covariates described earlier. Our key coefficients of interest were the indicator variables for survey years 2016 and 2017, allowing us to quantify changes in access from the last year before implementation of the ACA Medicaid expansion (2013). Data for years 2011 and 2012 were used to establish baseline trends in access.

Finally, we used adjusted difference-in-differences (DID) models, which included a binary variable for state Medicaid expansion status, interacted with the aforementioned indicators for each year after full implementation of the ACA (2014–2017). Our key coefficients of interest for this step were the interactions between expansion status and indicator variables for survey years 2016 and 2017, allowing us to determine whether state Medicaid expansion status mediated the observed changes in access during the study period. Because income may be affected by Medicaid expansion, we also stratified our models by educational attainment in sensitivity analyses.

All regressions were estimated as linear probability models using BRFSS sampling weights; results were similar when using logistic models with predicted probabilities to describe absolute changes. Wald tests were used to determine whether the changes from 2016 to 2017 were statistically significant. Standard errors were clustered at the state level in DID models. For more details on the study methodology, see the Appendix. Analyses were conducted using Microsoft R Open version 3.3.

Results

Unadjusted changes in health care access by household income group

Figure 1 presents national unadjusted percentages of young adults who reported they did not have health insurance, stratified by household income group. In 2013, young adults from households earning <138% FPL had an uninsurance rate of 32.0% compared with 13.3% in households earning >400% FPL, an absolute difference of 18.7%. This difference narrowed over time as rates of uninsurance dropped, reaching a minimum of 12.1% in 2016. However, coverage gains began to reverse for young adults in all income groups in 2017. Similar trends were observed for having a personal doctor or having unmet care because of cost (Table A2, Supplementary Data).

Figure 1.

Unadjusted trends in uninsurance for young adults 2011–2017, by household income group. Source: Authors’ analysis of data for 2011–2017 from the Behavioral Risk Factor Surveillance System (BRFSS). The exhibit displays the percentage of noninstitutionalized U.S. adults aged 18–24 years who reported that they have insurance coverage, by household income group. Household income group is defined as an imputed percentage of Federal Poverty Level (FPL); see the Appendix for more details.

Adjusted changes in health care access, overall and by household income group

Table 2 presents adjusted changes in health insurance and access to care for young adults from 2013 to 2016, and then from 2016 to 2017. Young adults from all income groups benefitted during the first few years of the ACA. Overall rates of uninsurance declined by 8.7 percentage points from 2013 to 2016 (95% confidence interval [CI] −9.4 to −8.0), whereas the rates of not having a personal doctor fell by 4.3 points (95% CI −5.1 to −3.5) and avoided cost fell by 3.5 points (−4.1 to −2.9). These benefits disproportionately accrued to young adults in the <138% FPL group, especially for insurance coverage. For instance, uninsurance among low-income young adults declined by 10.1 percentage points from 2013 to 2016 (95% CI −11.3 to −9.0). During the same period, uninsurance among high-income young adults declined by 4.8 percentage points (95% CI −6.1 to −3.5).

Table 2.

Adjusted changes from 2013 to 2017 in health care access for young adults under the ACA, by household income group

Household income	Uninsured		No personal doctor		Avoided care because of cost
	Change 2013–2016	Change 2016–2017	Change 2013–2016	Change 2016–2017	Change 2013–2016	Change 2016–2017
<138% FPL	−10.1***	.9	−4.9***	1.0	−5.1***	.4
138%–400% FPL	−7.0***	1.6**	−2.6***	.6	−.9	1.3**
>400% FPL	−4.8***	1.3	−3.4***	2.4*	−1.5**	1.2
Overall	−8.7***	1.4***	−4.3***	1.1*	−3.5***	1.0**

Source: Authors’ analysis of data for 2011–2017 from the Behavioral Risk Factor Surveillance System (BRFSS). Data for years 2011–2012 were included to control for baseline trends.

The exhibit displays changes in the percentage of noninstitutionalized U.S. adults aged 18–24 years who reported that they had insurance coverage, had a primary care provider, or avoided care because of cost associated with the Affordable Care Act (ACA) rollout. Household income group is defined as an imputed percentage of Federal Poverty Level (FPL); see the Appendix for more details. All columns show regression estimates adjusted for covariates described in the text.

^*p < .05;

^**p < .01;

^***p < .001.

We observed deterioration in all three access outcomes for young adults in 2017. The percentage who experienced uninsurance increased by 1.4 percentage points from 2016 to 2017 (95% CI .6–2.1), not having a personal doctor increased by 1.1 percentage points (95% CI .2–2.0), and unmet care because of cost increased by 1.0 percentage points (.3–1.7). The observed declines in access were concentrated among households earning 138% FPL or higher; among this income group, the uninsured rate increased.

Changes in health care access by state Medicaid expansion status

Unadjusted trends in self-reported health care access for young adults 2013–2017 by state Medicaid expansion status are contained in Figure 2 (Table A2, Supplementary Data). Uninsurance rates declined by approximately the same amount from 2013 to 2016, regardless of state expansion status; from 28.9% to 20.3% in nonexpansion states; and from 21.2% to 12.9% in expansion states. We observed smaller improvements in other outcomes during the same period. Access generally worsened for all outcomes in both types of states from 2016 to 2017. However, the unadjusted insurance rate declined by .4% for expansion states in 2017 compared with an increase of 4.2% for nonexpansion states.

Figure 2.

Trends in self-reported health care access for young adults 2011–2017, by state Medicaid expansion status. Source: Authors’ analysis of data for 2011–2017 from the Behavioral Risk Factor Surveillance System (BRFSS). The exhibit displays the average responses for noninstitutionalized U.S. adults aged 18–24 years who reported that they did not have insurance coverage, had no personal doctor, or avoided care because of cost, stratified by whether their state expanded Medicaid.

Adjusted changes in health care access by state Medicaid expansion status are contained in Table 3. Improvements in access were observed for both categories of statues from 2013 to 2016. For instance, the uninsured rate declined by 8.7 percentage points in expansion states (95% CI −9.5 to −7.9) and by 8.5 percentage points in nonexpansion states (95% CI −9.7 to −7.4). In DID models, Medicaid expansion did not appear to mediate 2013–2016 changes in insurance coverage or having a personal doctor but was associated with greater declines in unmet care because of cost (−1.5 percentage points, 95% CI −2.9 to −.0).

Table 3.

Adjusted changes from 2013 to 2017 in health care access for young adults under the ACA, by state Medicaid expansion status

Medicaid expansion state	Uninsured		No personal doctor		Avoided care because of cost
	Change 2013–2016	Change 2016–2017	Change 2013–2016	Change 2016–2017	Change 2013–2016	Change 2016–2017
No	−8.5***	3.9***	−3.4***	1.3	−2.5***	.6
Yes	−8.7***	−.4	−4.9***	1.0	−4.1***	1.3**
Differencea	.0	4.4***	−1.2	−.4	−1.5*	.7

Source: Authors’ analysis of data for 2011–2017 from the Behavioral Risk Factor Surveillance System (BRFSS). Data for years 2011–2012 were included to control for baseline trends.

The exhibit displays changes in the percentage of noninstitutionalized U.S. adults aged 18–24 years who reported that they had insurance coverage, had a primary care provider, or avoided care because of cost associated with the Affordable Care Act (ACA) rollout. All columns show regression estimates adjusted for covariates described in the text.

^*p < .05;

^**p < .01

^***p < .001.

^aDifference between expansion and nonexpansion states in changes over time, adjusted for covariates.

From 2016 to 2017, the adjusted rate of uninsurance increased by 4.4 percentage points in nonexpansion states (95% CI 2.0–6.7) compared with expansion states. During this same period, uninsurance in expansion states showed an insignificant decline (−.4 percentage points, 95% CI −1.3 to .5). Both groups of states showed increases in the rate of not having a personal doctor, although these changes were also not statistically significant. Finally, unmet care because of cost increased in expansion states by 1.3 percentage points from 2016 to 2017 (95% CI .4–2.1), whereas nonexpansion states saw an increase of .6 percentage points, which were not statistically significant (95% CI −.6–1.7); these differences were not significant in adjusted DID models. Expanded versions of our results tables with 95% confidence intervals are available in Tabls A3 & A4. Similar results were found when stratifying by educational attainment instead of income (Table A5, Supplementary Data).

Discussion

Previous research has demonstrated large gains in insurance coverage for young adults during the years immediately after the initial passage of the ACA, although a substantial portion remained uninsured [1,4]. We examined whether these trends continued through the final year of the Obama administration (2016) and then during the first year of the Trump administration (2017). Although the observational nature of our study design limits our ability to make causal claims, the observed declines in access were accompanied by concomitant policy changes which some commentators believe were intended to undermine the ACA [25,26]. We found that both unadjusted and adjusted rates of young adults reporting insurance coverage, having a personal doctor and in unmet care continued to improve through 2016. However, health care access began to erode for young adults in 2017, especially for respondents in expansion states and those in households earning 138% FPL or greater. At a population level, a 1.4% decrease in insurance coverage implies that approximately 436,000 young adults aged 18–24 years lost coverage in 2017 [27]. Our results suggest a protective effect of Medicaid expansion and comport with Miller and Wherry [28] and Sommers et al. [14] who find an increasing effect of Medicaid expansion on health care access and related financial strain for low-income adults through 2017. Notably, the unadjusted BRFSS estimates of uninsurance for 18-to 24-year-olds (15.7% in 2016 vs. 17.1% in 2017) is higher than similar estimates from the U.S. Census Bureau’s Current Population Survey (11.9% in 2016 vs. 12.9% in 2017) [29]. However, CPS’ estimated changes in uninsurance 2016–2017 (+1.0%) is within the 95% CI of our adjusted estimate for changes during this period (+.6% to +2.1%).

Although we found significant reductions in health care access for young adults during 2017, it is unclear if these changes are temporary or part of longer term trends. Several ACA-related policies enacted by the Trump administration became effective after our study period such as repeal of the individual mandate penalty requirement [7], reductions in funding for Marketplace navigator programs and open enrollment advertising [30], shorter open enrollment periods, and CMS’ approval of state waivers to implement Medicaid work restrictions in certain states [31]. Although these policies may weaken the ACA’s effectiveness in terms of access, recent Medicaid expansions in Virginia, Maine, Idaho, Nebraska, and Utah may provide countervailing effects for young adults in those states [23]. Further monitoring of these trends will be essential to track the effects of this fluctuating policy environment.

Limitations

Our study has several important limitations. First, the BRFSS is a repeated cross-sectional survey, and our results should be interpreted as associations and not necessarily causal. Our outcomes are self-reported, although previous research has shown these measures have high validity and reliability [32]. Furthermore, our insurance coverage outcome does not allow us to determine the source of coverage (e.g., Medicaid, Medicare, or private insurance). Income measurement in the BRFSS is not precise, and our imputed household income measure is a proxy of eligibility for Medicaid and Marketplace subsidies. We are also limited by the age categories reported in the BRFSS. We focused our analyses on the youngest age category available (18- to 24-year-olds), which does not exactly map with the ACA’s dependent coverage expansion. Finally, the potential exists for nonresponse bias despite the relatively high BRFSS response rates for a telephone survey (~40%–50% per year). Young adults tend to have higher response rates than average and are underrepresented in the BRFSS; thus, we follow Schneider et al.’s [21] recommendation and use multiple imputations to replace missing values and reduce the potential for nonresponse bias.

IMPLICATIONS AND CONTRIBUTION.

The Affordable Care Act was associated with continued improvements in health care access for young adults from 2013 to 2016. However, these encouraging trends reversed, and access began to deteriorate in 2017. At a population level, a 1.4% increase in insurance coverage implies that approximately 436,000 young adults aged 18–24 years lost coverage in 2017.

Question text and answer format for selected BRFSS outcomes

Outcome	Question Text	Answer Format
Has insurance coverage (%)	Do you have any kind of health care coverage, including health insurance, prepaid plans such as HMOs, or government plans such as Medicare, or Indian Health Service?	Yes/No
Has a personal doctor (%)	Do you have one person you think of as your personal doctor or health care provider?	Yes/No
Unmet care due to cost (%)	Was there a time in the past 12 months when you needed to see a doctor but could not because of cost?	Yes/No

Source: 2017 Behavioral Risk Factor Surveillance System (BRFSS)

Unadjusted BRFSS outcomes 2011–2017

Outcome	Strata	2011	2012	2013	2014	2015	2016	2017
Uninsured	Overall	27.4	26.4	24.1	19.7	17.7	15.7	17.1
	<138% FPL	34.7	34.9	32	26.3	23.5	21.6	23.1
	138 to 400% FPL	20.5	20.4	18.5	15.7	14.9	12.7	14.3
	>400% FPL	18.9	14.2	13.3	10.7	10.5	9.5	10.5
	Non-Expanders	34.8	30.6	28.9	24.8	23.6	20.3	24.5
	Expanders	22.9	23.7	21.2	16.5	14	12.9	12.5
No personal doctor	Overall	38.7	38	40.4	40.2	37.9	38	39.1
	<138% FPL	44.8	43.3	46.3	45.4	42.8	42.4	43.6
	138 to 400% FPL	32.6	33.8	36.4	37.7	35.5	37.1	37.5
	>400% FPL	33.2	31.7	31.9	31.7	31.9	30.9	33.2
	Non-Expanders	44.9	43.7	45.6	46.5	44.7	43.6	44.8
	Expanders	35	34.4	37.2	36.2	33.6	34.5	35.5
Avoided care due to cost	Overall	19.5	18.2	16.6	14.3	14.3	13.5	14.5
	<138% FPL	25.5	25	23	19.9	18.7	18	18.7
	138 to 400% FPL	13.8	13.5	11.4	10.6	12.3	11.5	12.8
	>400% FPL	12.8	8.1	8.8	7.6	8.2	8.1	8.9
	Non-Expanders	22.6	20.8	18.8	16.3	16.8	16.5	17
	Expanders	17.6	16.6	15.2	13.1	12.6	11.7	12.9

Source:Authors' analysis of data for 2011–17 from the Behavioral Risk Factor Surveillance System (BRFSS).

Notes: The exhibit displays the percentage of noninstitutionalized US adults age 18–24 who reported that they have insurance coverage, had a primary care provider, or avoided care due to cost, stratified by household income group and residence in a Medicaid expansion state.

Adjusted changes from 2013–2017 in health care access for young adults under the ACA, by household income group

Household Income	Uninsured				No personal doctor				Avoided care due to cost
	Change 2013–2016		Change 2016–2017		Change 2013–2016		Change 2016–2017		Change 2013–2016		Change 2016–2017
	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI
<138% FPL	−10.1***	(−11.3, −9.0)	0.9	(−0.4, 2.2)	−4.9***	(−6.2, −3.7)	1.0	(−0.5, 2.4)	−5.1***	(−6.2, −4.1)	0.4	(−0.8, 1.6)
138% to 400% FPL	−7.0***	(−8.0, −6.0)	1.6**	(0.5, 2.7)	−2.6***	(−3.9, −1.3)	0.6	(−0.8, 2.0)	−0.9	(−1.8, 0.1)	1.3**	(0.4, 2.3)
>400% FPL	−4.8***	(−6.1, −3.5)	1.3	(−0.0, 2.7)	−3.4***	(−5.3, −1.6)	2.4*	(0.4, 4.3)	−1.5**	(−2.7, −0.4)	1.2	(−0.0, 2.4)
Overall	−8 7***	(−9.4, −8.0)	1.4***	(0.6, 2.1)	−4.3***	(−5.1, −3.5)	1.1*	(0.2, 2.0)	−3.5***	(−4.1, −2.9)	1.0**	(0.3, 1.7)

Source:Authors' analysis of data for 2011–17 from the Behavioral Risk Factor Surveillance System (BRFSS). Data for years 2011–2012 were included to control for baseline trends. Notes:The exhibit displays changes in the percentage of noninstitutionalized US adults aged 18–24 who reported that they had insurance coverage, had a primary care provider, or avoided care due to cost associated with the Affordable Care Act (ACA) rollout. Household income group is defined as an imputed percentage of Federal Poverty Level (FPL); see the Appendix for more details. All columns show regression estimates adjusted for covariates described in the text.

^*p <0.05

^**p<.01

^***p<.001

Adjusted changes from 2013–2017 in health care access for young adults under the ACA, by state Medicaid expansion status

Medicaid Expansion Status	Uninsured				No personal doctor				Avoided care due to cost
	Change 2013–2016		Change 2016–2017		Change 2013–2016		Change 2016–2017		Change 2013–2016		Change 2016–2017
	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI
No	−8.5***	(−9.7, −7.4)	3.9***	(2.7, 5.2)	−3 4***	(−4.7, −2.0)	1.3	(−1.3, 2.8)	−2.5***	(−3.6, −1.5)	0.6	(−0.6, 1.7)
Yes	−8.7***	(−9.5, −7.9)	−0.4	(−1.3, 0.5)	−4.9***	(−5.9, −3.8)	1.0	(−0.2, 2.1)	−4.1***	(−4.9, −3.3)	1.3**	(0.4, 2.1)
Differencea	0.0	(−2.1, 2.0)	4 4***	(2.0, 6.7)	0.0	(−0.04, 0.02)	0.4	(−1.4, 2.2)	−1.5*	(−2.9, 0.0)	−0.7	(−1.9, 0.6)

Source: Authors' analysis of data for 2011–17 from the Behavioral Risk Factor Surveillance System (BRFSS). Data for years 2011–2012 were included to control for baseline trends. Notes: The exhibit displays changes in the percentage of noninstitutionalized US adults aged 18–24 who reported that they had insurance coverage, had a primary care provider, or avoided care due to cost associated with the Affordable Care Act (ACA) rollout. All columns show regression estimates adjusted for covariates described in the text.

^aDifference between expansion and non-expansion states in changes over time, adjusted for covariates.

^*p <0.05

^**p<.01

^***p<.001

Adjusted changes from 2013–2017 in health care access for young adults under the ACA, by educational attainment

Educational Attainment	Uninsured		No personal doctor		Avoided care due to cost
	Change 2013–2016	Change 2016–2017	Change 2013–2016	Change 2016–2017	Change 2013–2016	Change 2016–2017
High school graduate or less	−9 6***	1.2	−4 5***	0.2	−4 1***	0.7
Some college/technical school	−8.4***	1.5**	−4 1***	2.5***	−3 2***	1.1*
College graduate (4 years or more)	−5 7***	0.6	−3.6***	−0.3	−2.3***	1.9**
Overall	−8 7***	1 4***	−4 3***	1.1*	−3.5***	1.0**

Source:Authors' analysis of data for 2011–17 from the Behavioral Risk Factor Surveillance System (BRFSS). Data for years 2011–2012 were included to control for baseline trends. Notes:The exhibit displays changes in the percentage of noninstitutionalized US adults aged 18–24 who reported that they had insurance coverage, had a primary care provider, or avoided care due to cost associated with the Affordable Care Act (ACA) rollout. All columns show regression estimates adjusted for covariates described in the text.

^*p <0.05

^**p<.01

^***p<.001