Effects of the ACA on Health Care Coverage for Adults with Substance Use Disorders

Abstract

Objective:

The authors assessed changes in health care coverage within nationally representative samples of low- and middle-income adults with and without substance use disorders following the 2014 Affordable Care Act Marketplace launch and Medicaid expansion.

Methods:

Data from the 2012-2018 (N=407,985) National Survey on Drug Use and Health identified low-and middle-income non-elderly adults with alcohol, marijuana, cocaine, or heroin use disordrs. A socio-demographically adjusted difference-in-differences analysis assessed the trends in Medicaid and individually purchased private insurance between adults with and without substance use disorders.

Results:

Between 2012-13 and 2015-16, the percentage without health insurance significantly declined for those with (27.8% to 18.7%) and without (from 22.6% to 14.6%) substance use disorders. These trends were related to gains in Medicaid and individually purchased private insurance, but not employer-based private insurance coverage. Between 2015-16 and 2017-18, however, lack of insurance among adults with (18.7% to 18.4%) and without (14.7% to 14.7%) substance use disorders was little changed.

Conclusions:

With insurance gains having stalled and the recent downturn of the US economy, there is renewed urgency to extend health care coverage to middle- and low-income adults with substance use disorders that meets their substance use and general health needs.

The benefits of health insurance have been demonstrated through several high quality studies.¹ In a randomized controlled trial, people assigned to Medicaid in relation to wait list controls experienced significant gains in financial security, improved health-related quality of life, and lower rates of depression.² There is also strong evidence that health care coverage improves the likelihood of having a usual source of care,³ accessing preventive services,⁴ receiving routine primary care,⁵ and reducing the risks of premature mortality.⁴ Because adults with substance use disorders often lack health insurance,^6,7 increasing their health coverage is an important public health goal, even apart from any potential effects on access to substance use services.^8,9

The Medicaid expansion and Health Insurance Marketplace exchange provisions of the Affordable Care Act (ACA) have the potential to increase insurance coverage to low- and middle-income adults with substance use disorders. Medicaid expansion, which extends Medicaid eligibility to residents of participating states with household incomes up to 138% of the Federal Poverty Level (FPL), contributed to early coverage gains (2012-13 to 2014-15) for low-income adults with substance use disorders.¹⁰ The Health Insurance Marketplaces offer individuals with incomes between one and four times the FPL subsidized health insurance. These tax credit subsidies could help low- and middle-income adults with substance use disorders afford coverage. Yet people with substance use disorders may have difficulties accessing and navigating Marketplace exchanges.¹¹ Although Marketplaces accounted for approximately 40% of ACA-related general population coverage gains in 2014 and 2015,¹² it is not known to what extent people with substance use disorders received insurance through Marketplaces. Despite a requirement that insurance plans sold through Marketplaces offer substance use disorders benefits at parity with medical and surgical benefits, coverage differentials may make these plans less attractive to people with substance use disorders.¹³

Following the transition from the Obama to Trump administrations in January 2017, the new administration issued an executive order seeking to reduce the “economic burden” of the ACA. Shortly after the order, advertising for federally-supported Marketplaces was reduced and outreach and consumer assistance for the 2017 Marketplace enrollment period declined.¹⁴ This was followed by reductions in federal funding for enrollment advertising for the 2018 enrollment period, a decrease in navigator support, and cancellation of $10 billion in payments to insurers for cost-sharing reductions to reimburse low-income Marketplace consumers.¹⁵ Overall, approximately 12.7 million people were enrolled in health insurance coverage through Marketplaces in 2016, 12.2 million in 2017, and 11.8 million in 2018.¹⁶

Little is known about the effects of this change in policy direction on coverage of low- and middle-income adults with and without substance use disorders. To address this knowledge gap, we compared trends in coverage among adults with and without substance use disorders who reported family income levels that would make them eligible for Marketplace premium tax credits. We examined changes in coverage from before the 2014 Marketplace launch and Medicaid expansion (2012-2013) to the last two years of the Obama administration (2015-16) and first two years of the Trump administration (2017-18).

METHODS

Data Source

The NSDUH is a cross-sectional annual US population survey sponsored by Substance Abuse and Mental Health Services Administration. NSDUH yields national and state-level representative estimates of substance use disorders for the civilian non-institutionalized population. Individuals without a household address, active-duty military personnel, and institutional residents were excluded from the sampling frame. The NSDUH data collection protocol was approved by the Institutional Review Board at RTI International. The annual mean weighted overall response rate of the 2012-2018 NSDUHs were 62.9% (range: 48.8% to 66.8%, total N=407,985).¹⁷

Background Characteristics

Based on DSM-IV criteria, NSDUH yields estimates of past year dependence on or abuse of alcohol, marijuana, cocaine, and heroin. The survey also collected information on respondent age, sex, race/ethnicity, family income, state of residence, and educational level.

Health Insurance

Health insurance status at the time of the survey interview was the outcome. We partitioned insurance status into hierarchical groups of private insurance, Medicaid, or other public insurance (Medicare, CHAMPUS, VA, Tricare or Military health) and no insurance. Respondents with private insurance were partitioned into employer-based coverage, defined as coverage “provided through work, such as through an employer, union, or professional association,” and individually-purchased private coverage defined as private coverage not provided through work. We consider individually-purchased private coverage as a proxy for Marketplace purchased plans.

Medicaid Expansion State Residence

States were partitioned by Medicaid expansion implementation status.¹⁸ By the end of 2014, 26 states and the District of Columbia had expanded Medicaid and are referred to as expansion states. The remaining 24 states are referred to as non-expansion states (Appendix Table 1).

Statistical Analysis

We examined whether health insurance coverage trends differed among adults with and without substance use disorders. Because we focused on potential effects of Marketplace tax credits on individually purchased plans, we limited the analysis to adults, aged 18-64 years, with family incomes of 100%-400% of FPL that enabled them to receive tax credits. Baseline socio-demographic characteristics were first compared among adults with and without past year substance use disorders.

To distinguish coverage trends under policies of the Obama and Trump administrations, we considered coverage from before ACA implementation (2012-2013) to the last two years of the Obama administration (2015-16) and the first two years of the Trump administration (2017-18), leaving out the year of policy implementation (2014). We used a difference-in-differences design¹⁹ to assess differences in secular trends in insurance coverage between populations with and without past year substance use disorders. Multivariable logistic regression analyses estimated changes in coverage prevalence and included the effects of categorical survey years, a substance use disorder dummy variable, an interaction term for year x substance use disorder, and covariates age, sex, race/ethnicity, region, and education. Some models compared 2012-13 to 2015-16 and others compared 2015-2016 to 2017-18 providing flexibility for the covariate distribution to be controlled separately in the two periods. Because covariate distributions varied between models, estimated insurance percentages for 2015-16 may vary between models. Adjusted difference estimates in coverage prevalence (back-transformed from marginal log-odds)²⁰ tested change in coverage over time among adults with and without substance use disorders.The interaction contrast on the predicted prevalence scale from the model provided the difference-in-differences test of whether changes over time differed between adults with and without substance use disorders. Percentages, differences, and difference-in-differences estimates were adjusted for age, sex, race/ethnicity, region, and educational level.

Because the value of Marketplace premium subsidies decline with increasing income from 100% to 400% of FPL, lower income individuals have stronger financial incentives to purchase Marketplace plans.²¹ To assess this effect, analyses were stratified by poverty level (100%-200% vs. 201%-400%). Because lower income adults in Medicaid non-expansion states compared to expansion states have fewer affordable coverage options, we compared coverage trends among expansion and non-expansion state residents with and without substance use disorders. SAS (Version 9) callable SUDAAN using PROC MULTILOG was used to account for NSDUH’s complex sample design and sample weights.

RESULTS

Background Characteristics

Among adults whose incomes were 100% to 400% of the FPL, background characteristics were compared between those with and without past year substance use disorders. As compared to people without substance use disorders, those with these disorders were more likely in each of the three study periods to be male, aged 18 to 24 or 35 to 44 years, white non-Hispanic, have lower family incomes, and not to have graduated from college (Table 1). Most of the people in the substance use disorder group met criteria for alcohol use disorder.

Table 1.

Background characteristics of income eligible adults with and without selected substance use, 2012-2013, 2015-2016, and 2017-2018.

	2012-2013					2015-2016					2017-2018
Characteristics	Past Year Substance Use Disorder		No Past Year Substance Use Disorder		P- Valuea	Past Year Substance Use Disorder		No Past Year Substance Use Disorder		P-value	Past Year Substance Use Disorder		No Past Year Substance Use Disorder		P-value
Age, Years	%	SE	%	SE	<.001	%	SE	%	SE	<.001	%	SE	%	SE	<.001
18-24	30.5	0.9	15.5	0.2		26.4	0.7	15.2	0.2		26.8	0.8	15.4	0.2
25-34	30.8	1.1	23.1	0.4		33.1	0.9	23.5	0.3		33.1	1.0	24.0	0.3
35-44	18.0	1.0	22.7	0.3		17.2	0.8	20.4	0.3		18.2	0.8	21.0	0.3
45-64	20.8	1.2	38.7	0.5		23.4	1.1	40.9	0.4		21.8	1.1	39.6	0.4
Sex					<.001					<.001					<.001
Male	67.4	1.1	47.3	0.4		65.7	0.9	47.5	0.3		66.7	0.9	47.8	0.3
Female	32.6	1.1	52.7	0.4		34.3	0.9	52.5	0.3		33.3	0.9	52.2	0.3
Race/Ethnicity					0.002					<.001					<.001
White, non-Hispanic	61.6	1.2	59.5	0.6		62.6	1.0	59.0	0.5		61.0	1.0	55.9	0.5
Black, non-Hispanic	13.3	0.8	13.1	0.4		12.1	0.7	13.5	0.3		13.1	0.7	14.1	0.3
Hispanic	19.2	1.0	19.3	0.4		19.9	0.9	20.0	0.4		19.2	0.9	21.7	0.4
Other	5.9	0.5	8.1	0.3		5.5	0.4	7.5	0.2		6.6	0.5	8.3	0.3
Education					<.001					0.017					0.042
< High school graduate	16.1	0.8	13.5	0.3		14.2	0.8	13.3	0.3		12.6	0.7	13.3	0.3
High school graduate	32.6	1.1	34.0	0.4		29.2	0.9	29.2	0.3		30.4	1.0	28.9	0.3
Some college	32.8	1.1	30.3	0.4		37.7	0.9	36.1	0.3		37.7	1.0	36.3	0.3
College graduate	18.5	0.9	22.2	0.4		18.9	0.8	21.4	0.3		19.4	0.8	21.4	0.3
Income, Mean (SE)	43,223	509	47,853	232	<.001	46,231	447	49,550	169	<.001	45,053	466	49,603	184	<.001
Substance Use Disorders, past year
Alcohol
Yes	84.6	0.9	0			83.6	0.7	0			79.8	0.8	0
Marijuana
Yes	19.1	0.9	0			20.8	0.8	0			23.8	0.9	0
Cocaine
Yes	5.0	0.5	0			5.4	0.5	0			6.2	0.5	0
Heroin
Yes	2.6	0.3	0			3.9	0.4	0			3.7	0.4	0

Data from NSDUH. SUD denotes substance use disorder. Income eligible defined as between 100% and 400% of the Federal Poverty Level (FPL). SAMHSA requires that all descriptions of all descriptions overall sample sizes based on the public-use data files and are rounded to the nearest 100 to minimize potential disclosure risk.

^ap-values are for associated chi-square tests associating the characteristic with the past year substance use disorder (yes/no)

Lack of Insurance

Trends in being uninsured were compared before (2012-13) and after (2015-16) the ACA policies were implemented and between 2015-16 and 2017-18 (Figure 1). Between 2012-13 and 2015-16, the percentage of people with substance use disorders who were uninsured declined from 27.8% to 18.7% (difference=−9.1%, 95%CI=−11.6- −6.6) while the percentage without substance use disorders who were uninsured declined from 22.6% to 14.6% (difference=−8.0%, 95%CI=−8.9- −7.1). The difference-indifferences estimate was not significant. Between 2015-16 and 2017-18, the share of each group who were uninsured was little changed.

Figure 1. Trends in no insurance coverage of low- and middle-income adults with and without selected substance use disorders, 2012-2018.

Data from NSDUH. Between 2012-13 and 2015-16, the estimated changes in the percentage without insurance were −9.1% (95%CI: −11.6, −6.6) for adults with substance use disorder and −8.0% (−8.9, −7.1) for those without substance use disorder. Between 2015-16 and 2017-18, the estimated changes were −0.3% (−2.5, 1.8) for adults with substance use disorder and 0% (−0.7, 0.7) for those without sub. Differences adjusted for age, sex, race/ethnicity, region, and education. For ease of presentation, 2015-2016 data are presented as means of modeled estimates from 2012-13/2015-16 and 2015-16/2017-18 models.

Individually Purchased and Employer Purchased Private Insurance

From 2012-13 to 2015-16, individually purchased private insurance coverage significantly increased from 5.2% to 7.6% (difference=2.4%, 95%CI=0.8-4.1) for adults with substance use disorders and from 5.4% to 8.5% (difference=3.1%, 95%CI=2.6-3.7) for those without substance use disorders (difference-in-differences=−0.7%, 95%CI=−2.4-1.0) (Figure 2). Between 2015-16 and 2017-18, there was not a significant change in individually purchased private insurance for people with or without substance use disorders. Between 2012-13 and 2015-16 and between 2015-16 and 2017-18, the proportion of people with and without substance use disorders who had employer-based private insurance was little changed (Appendix Figure 1).

Figure 2. Trends in private individually purchased insurance among low- and middle-income adults with and without selected substance use disorders, 2012-2018.

Data from NSDUH. Between 2012-13 and 2015-16, the estimated changes in private individually purchased insurance were 2.4% (95%CI: 0.8, 4.1) for adults with substance use disorder and 3.1% (2.6, 3.7) for those without sub. Between 2015-16 and 2017-18, the estimated changes were −0.5% (−2.1,1.3) for adults with substance use disorder and −0.4% (−1.0%, 0.2%) for those without substance use disorder. Differences adjusted for age, sex, race/ethnicity, region, and education. For ease of presentation, 2015-2016 data are presented as means of modeled estimates from 2012-13/2015-16 and 2015-16/2017-18 models.

Medicaid Coverage and Other Public Insurance

Medicaid coverage increased between 2012-13 and 2015-16 significantly more for people with substance use disorders (6.3% to 13.6%, difference=7.3%, 95%CI= 5.7-8.9) than for people without substance use disorders (6.2% to 10.9%, difference=4.6%, 95%CI=4.1-5.2) (difference-in-differences=2.7%, 95%CI=1.0-4.4) (Figure 3). During the following period, there was a small and statistically non-significant increase in Medicaid coverage for adults with substance use disorders (13.4% to 14.7%, difference=1.3%, 95%CI=−0.6-3.2) and an increase for those without substance use disorders (10.8% to 12.5%, difference=1.7%, 95%CI=1.1-2.3) (difference-in-differences=−0.4%, 95%CI:=−2.4-1.6).

Figure 3. Trends in Medicaid coverage among low- and middle-income adults with and without selected substance use disorders, 2012-2018.

Data from NSDUH. Between 2012-13 and 2015-16, the estimated changes in Medicaid coverage were 7.3% (95%CI: 5.7, 8.9) for adults with substance use disorder and 4.6% (4.1, 5.2) for those without substance use disorder (difference, 2.7%,1.0,4.4). Between 2015-16 and 2017-18, the estimated changes were 1.3% (−0.6,3.2) for adults with substance use disorder and 1.7% (1.1, 2.3) for those without substance use disorder. Differences adjusted for age, sex, race/ethnicity, region, and education. For ease of presentation, 2015-2016 data are presented as means of modeled estimates from 2012-13/2015-16 and 2015-16/2017-18 models.

Between 2012-13 and 2015-16, there was little change in other public insurance among adults with substance use disorders (9.8% to 9.6%) and an increase among those without substance use disorders (8.6% to 9.4%, difference=0.9%, 95%CI=0.2-1.5). From 2015-16 to 2017-18, however, there was no change in other public insurance among adults with and without substance use disorders (Appendix Figure 2).

Family Income and Individually Purchased Private Insurance

Among adults with substance use disorders, gains in individually purchased private insurance between 2012-13 and 2015-16 were roughly similar in the two income groups: 5.5% to 8.1% in the 100%-200% of poverty group (difference=2.6%, 95%CI=−0.1-5.4) and 5.0% to 7.3% in the 201%-400% of poverty group (difference=2.3%, 95%CI=0.4-4.3) (Appendix Figure 3). Among those without substance use disorders, the corresponding gain was greater among the lower income group (5.5% to 9.9%, difference=4.4%, 95%CI=3.4-5.4) than higher income group (5.3% to 7.7%, difference=2.4%, 95%CI=1.7-3.2). Between 2015-16 and 2017-18, there was little change in the percentages with individually purchased private insurance in either income group of adults with and without substance use disorders.

State Residence and Health Insurance

Among Medicaid expansion state residents, the increase in Medicaid enrollment between 2012-13 and 2015-16 was larger for people with substance use disorders than for those without substance use disorders (Appendix Figure 4). There was less change in Medicaid enrollment in non-expansion states during this period. Among adults with substance use disorders, the percentage with individually purchased private insurance significantly increased from 2012-13 to 2015-16 in non-expansion states but not in expansion states (Appendix Figure 5). Between 2015-16 and 2017-18, little change occurred in the share of people with substance use disorders with individually purchased private insurance in either expansion or non-expansion states. Similar patterns were observed among people without substance use disorders.

Discussion

Consistent with prior research,¹⁰ there was a decline in the percentage of low- and middle-income adults who were uninsured during the first two years following implementation of Medicaid expansion and Marketplace exchanges. Following change in administrations between 2015-16 and 2017-18, however, the share that was uninsured remained unchanged. This new finding was also observed among low- and middle-income adults without substance use disorders. Opposition to the ACA under the Trump administration may have contributed to ending the decline in the percentages of these two groups of low- and middle-income people without health insurance. During 2017-18, there was a shift in federal health care policy away from the ACA reflected in public statements²² and actions including a reduction in advertising Marketplace enrollment, a shortening of the enrollment period,¹⁴ and cancelation of reimbursement payments to insurers for low-income Marketplace withholding that may have slowed enrollment of lower-income people.²³

Between 2012-13 and 2015-16, trends in insurance coverage were consistent with Medicaid expansion and Health Insurance Marketplace policies. Growth occurred in Medicaid coverage and individually purchased private coverage among low- and middle-income people with and without substance use disorders. Meanwhile, coverage changed less in other public insurance programs and employee-sponsored private insurance plans that were not a focus of these two ACA provisions.

Under the Marketplace provision, people with family incomes of 100%-400% of the FPL are eligible for premium tax credits to reduce premium payments. Because the subsidies are progressive,²⁴ the increase in individually purchased private plans was expected to be greater among adults in the lower than upper group of this income range. While there was evidence of such an effect among people without substance use disorders, gains in individually purchased private coverage were similar among the two income groups with substance use disorders (Appendix Figure 4). Lower income adults with substance use disorders may have had greater difficulties than their counterparts without substance use disorders in accessing benefits available to them through the Marketplace exchanges. Detailed prospective research could help identify which outreach strategies are most helpful for which populations in achieving Marketplace enrollment.²⁵

It is not known why adults with substance use disorders were relatively less likely than those without substance use disorders to take advantage of Marketplace subsidies. Some possibilities include more challenging life circumstances, competing priorities for basic daily needs,²⁶ or more extensive reliance on navigators for enrollment assistance. During the Trump administration, federal funding for ACA navigators was reduced by 43% between 2016 and 2017 and by 72% between 2017 and 2018.²⁷

The increase in Medicaid coverage following expansion, by contrast, was proportionately greater for people with than without substance use disorders. Lower average incomes of adults with substance use disorders may have resulted in a larger proportion of them being eligible for Medicaid under ACA expansion.²⁸ Medicaid coverage plays in important role in the availability of substance use services²⁹ and general medical services³⁰ for people with substance use disorders. States expanding Medicaid eligibility have been associated with slower growth in drug overdose³¹ and substance use disorder related deaths.³² Although Medicaid coverage gains were concentrated in expansion states, a larger share of people with than without substance use disorders remained uninsured throughout the study period.

As compared to Medicaid expansion, Marketplace exchanges were associated with smaller gains in coverage for people with incomes of 100-138% of the poverty level.³³ A similar pattern was observed even among a broader income group with substance use disorders, many of whom were not income eligible for Medicaid. Higher premiums for Marketplace than Medicaid coverage or greater difficulties navigating Marketplace than Medicaid enrollment may have contributed to them having a relatively smaller increase in individually purchased private plans than Medicaid.

The current findings suggest that people with substance use disorders increased enrollment in Medicaid and individually purchased private insurance plans in the years immediately following ACA policy implementation (2014). However, because the findings capture coverage at the time of the survey, the results likely understate percentages that were uninsured at some point during each year. In a recent study, nearly one quarter of Marketplace beneficiaries dis-enrolled before the end of the year.³⁴ From the insurers’ perspective, adults with substance use disorders are among the least desirable to insure because their health care expenditures often exceed their premiums³⁵ making them a vulnerable group to disenrollment. In an analysis of Medicaid enrollees, there was greater disenrollment among adults with than without substance use problems.³⁶ Despite coverage gains, expanding coverage for lower income adults with substance use disorders, such as through incentivizing insurers to participate in Marketplaces and adjusting risk through more generous premium tax credits,³⁷ remains a key public healthcare policy challenge. In evaluating the present findings, it is important to bear in mind that most of the people with substance use disorders had alcohol use disorder. It possible that as compared to people with drug use disorders, those with alcohol use disorders faced fewer barriers to health insurance following ACA implementation.

This analysis has some limitations. First, a change in NSDUH survey design prevented examining trends in coverage of adults with prescription stimulant, sedative, or opioid use disorders. Second, NSDUH did not sample homeless individuals not living in shelters, active-duty military personnel, or people residing in institutions. Third, NSDUH did not directly assess whether insurance coverage was purchased from Marketplaces. Instead, it was assumed that private insurance not purchased through an employer, union, or professional association was individually purchased on a Marketplace. In addition, no effort was made to distinguish federally-facilitated from state-based Marketplaces. Fourth, the difference-in-difference models assume that without Medicaid expansion insurance trends in expansion and non-expansion states would have been constant over time. However, ACA provisions other than Medicaid expansion and Marketplaces, such as the employer mandate³⁸ and dependent coverage provision,³⁹ or other factors that were not controlled for in the models may have contributed to differentially changing coverage in expansion and non-expansion states. Finally, the analysis did not assess the effects of coverage on substance use treatment or other health care use. Some prior research suggests that increases in coverage alone may not be sufficient to increase substance use treatment.^10,40 Beyond health insurance, several other factors such as transportation, service availability, stigma, and perceived effectiveness of treatment may influence access to needed behavioral health services.⁴¹ A lack of local providers may help to explain why gains in coverage have not been sufficient to increase treatment.⁴²

Conclusions

In the first 2 years following implementation of the 2014 ACA Marketplace and Medicaid expansion provisions (2015-2016), there was a decline in the percentage of low- and middle-income people with common substance use disorders who were uninsured. The decline halted following the change of administrations in 2017. Although the percentage who were uninsured remained higher among low- and middle-income adults with than without substance use disorders, a lower percentage of both those with and without substance use disorders were uninsured in 2017-2018 than before implementation of the ACA provisions in 2014. As we enter a pandemic induced severe economic downturn, particularly in light of the association between financial strain and serious substance use problems,⁴³ it will be critically important to implement policies that extend health care coverage to uninsured low- and middle-income people with substance use disorders.

Highlights:

• Among US low- and middle-income adults with substance use disorders, the percentage that were uninsured declined from 27.8% (2012-13) to 18.7% (2015-16) and then remained nearly constant at 18.4% (2017-2018).

• For adults with substance use disorders, large gains in Medicaid coverage (6.3% to 13.6%) and smaller gains in individually-purchased private insurance (5.2% to 7.6%) contributed to their decrease in the percentage that were uninsured from 2012-13 to 2015-16.

• A stalling of the decrease of the percentage of uninsured low- and middle-income adults with substance use disorders underscores the importance of implementing policies to extend coverage to this vulnerable group.