Provisions in the Patient Protection and Affordable Care Act (ACA) of 2010 are likely to significantly expand access to substance use disorder treatment for low-income individuals. The ACA gave states the option of extending Medicaid coverage beginning in 2014 to all uninsured adults under age 65 with incomes up to 133% of the federal poverty level (FPL). In July 2012, the U.S. Supreme Court ruled that states will not face the penalty of losing federal funding from their traditional Medicaid programs if they choose not to expand eligibility, and some states are now likely to forgo Medicaid expansion. The Congressional Budget Office estimates that given the Supreme Court decision, the ACA’s Medicaid expansion will result in an additional seven million individuals covered by Medicaid by 2014 (
1). The expansion of Medicaid coverage is expected to increase use of substance use disorder treatment, alter the types and settings of services received, and change how treatments are financed (
2).
Although data on the population expected to be newly eligible for Medicaid as a result of the ACA are scarce, these individuals may have significant demand for substance use disorder treatment. Unlike other disabling conditions, substance use disorders are not a qualifying disability category under Supplemental Security Income (SSI) or Social Security Disability Insurance (SSDI). This has made it difficult for individuals with a substance use disorder to gain public coverage, even if the condition results in absence from the labor market. Reports from states that have previously expanded Medicaid coverage to low-income childless adults suggest high rates of substance use disorders (as well as mental illness) in this population (
3). At least two studies have examined current use of health care by the newly eligible Medicaid population, although neither examined the population with substance use disorders specifically. One study found significant problems with access to general medical care; about 60% of those expected to be newly eligible had no usual source of care and 50% had received no medical care during the past year (
4). A second study estimated that the proportion of individuals with probable depression or serious psychological distress who will be covered by Medicaid when reform is fully implemented in 2019 will increase from 12.8% to 24.5% (
5).
The goal of this study was to gather data on the need for substance use disorder treatment among persons with low incomes who are anticipated to be newly eligible for Medicaid in 2014 and to compare current rates of substance use disorder treatment in this population with rates among persons with similar incomes who are currently enrolled in Medicaid. Because it is unknown at this time which states will choose to expand their Medicaid programs, for simplicity we assumed that uninsured individuals in any state who meet the income threshold will become eligible. Understanding the characteristics of these individuals will be helpful to states designing benefit packages, making funding and resource determinations, addressing workforce issues, and developing enrollment strategies as they prepare for implementation in 2014.
Methods
Data
We used the 2008 and 2009 data from the National Survey on Drug Use and Health (NSDUH), a nationally representative survey of noninstitutionalized adolescents and adults in the United States conducted annually by the Substance Abuse and Mental Health Services Administration. The NSDUH collects information on use of alcohol and illicit drugs and consequences of use, which allowed us to estimate rates of treated and untreated substance use disorders. Rates of substance use disorders found in the NSDUH are consistent with rates found in other national surveys (
6,
7), although the National Comorbidity Survey Replication reported lower rates (
8,
9). Because Medicaid expansion will primarily affect nonelderly adults and Medicaid programs are required to cover poor children up to age 19, we restricted our analysis to individuals age 20 through 64.
Measures
The NSDUH defines alcohol and illicit drug dependence and abuse on the basis of criteria in the
DSM-IV. To be considered alcohol dependent, a respondent must report three or more of the following during the past 12 months: spent a great deal of time over a period of a month using or getting over the effects of alcohol, used alcohol more often than intended, needed to use alcohol more than before to get desired effects, was unable to cut down or stop using alcohol, continued to use alcohol even though it was causing problems with mental health or physical problems, reduced involvement in important activities due to alcohol, or experienced two or more alcohol withdrawal symptoms after alcohol use was cut back or stopped (
10,
11).
For individuals with no alcohol or drug dependence, we considered whether the individual had alcohol or illicit drug abuse. To be considered to have alcohol abuse, an individual must report one or more of the following during the past 12 months: had serious problems at home, work, or school due to alcohol use; used alcohol regularly and engaged in an activity in which being drunk may have put the individual in physical danger; got into trouble with the law repeatedly due to actions resulting from alcohol use; or experienced problems with family or friends due to alcohol use and continued to use alcohol. Similar measures were used to identify illicit drug dependence and illicit drug abuse.
To create a variable indicating whether household income would qualify the individual for Medicaid in 2014, we used the midpoint of the income ranges reported in the NSDUH, household size, and the FPL during the relevant year. We performed sensitivity analysis using the lower and upper bound in each income range and found that results did not change in a qualitatively meaningful way. Although the ACA requires Medicaid eligibility at 133% of the FPL, the law allows for a 5% income “disregard,” and thus we considered individuals with incomes up to and including 138% of the FPL to be income eligible (
12). Individuals age 20 to 22 living in a college dormitory were not included because information on family income was unavailable.
We created a measure of serious psychological distress based on the K6 scale, with a K6 score ≥13 indicating serious psychological distress (
13,
14).
We constructed six mutually exclusive insurance coverage variables in the following hierarchical order: Medicare, Medicaid, private coverage, military health care, other coverage, and uninsured. For example, if an individual had both Medicare and Medicaid, he or she was coded as insured under Medicare. Individuals covered by the State Children’s Health Insurance Program were included under Medicaid.
We categorized substance use disorder treatment on the basis of self-reported services received during the past 12 months. Outpatient treatment included outpatient treatment in a residential drug or alcohol rehabilitation facility, treatment in a mental health center or facility, or treatment in a private physician’s office. Inpatient treatment included treatment in a hospital overnight as an inpatient, treatment in a residential drug or alcohol rehabilitation facility overnight, or treatment in an emergency room. Informal treatment included services received through a self-help group, religious organization, or alcohol or drug education program.
We created an indicator variable, “substance use disorder treatment need,” to identify individuals who met the dependence or abuse criteria described above as well as individuals with no current disorder who reported being currently in treatment (either formal or informal). Consistent with prior research (
15), we chose to categorize the latter group as having treatment need because, like patients with other chronic recurring illnesses, individuals with substance use disorders experience periods of reduced or no symptoms, but symptoms can and do recur over time.
Data analysis
First, we compared individuals with incomes below 138% FPL with and without a need for substance use disorder treatment by current insurance status. Next, we compared current Medicaid enrollees with incomes below 138% FPL (N=3,809) with currently uninsured individuals with incomes below 138% FPL (N=5,049), the threshold that would qualify them for Medicaid expansion coverage as of 2014 (hereafter referred to as “future income eligible”). We included only individuals with a household income less than 138% FPL because Medicaid beneficiaries with higher income are likely to have qualified under the medically needy eligibility category. Note that the future income-eligible group included individuals who met current categorical eligibility criteria for Medicaid but had not taken up coverage. Given substantial uncertainty about implementation of the ACA, we did not model behavioral responses to the law (for example, changes in labor market participation) or account for individuals who might switch from private coverage to Medicaid or insurance take-up. Because immigration status is not available in the NSDUH, it was also not considered. We excluded current Medicaid enrollees who were dually enrolled in Medicare because Medicare is their primary payer.
For current and future income-eligible Medicaid enrollees, we calculated pooled national estimates of demographic, health, and substance use disorder characteristics and compared substance use disorder treatment rates and treatment type for individuals with a need for substance use disorder treatment. Finally, we compared the characteristics of future income-eligible individuals with and without a substance use disorder. In all cases, differences were tested by using chi square tests. All estimates were weighted to make estimates representative of the noninstitutionalized population, and variance estimates were adjusted for the complex sampling design of the survey.
Results
We examined the insurance status of all individuals meeting the income threshold of 138% FPL under the Medicaid expansion and compared individuals with and without a substance use disorder (
Table 1). We found significant differences in the distribution of insurance status (p<.01). Individuals at this income level with a substance use disorder were more likely to be uninsured than those without a substance use disorder (44.6% versus 38.5%). In addition, those with a substance use disorder were slightly less likely to be enrolled in Medicaid (21.8% versus 24.9%).
We compared characteristics of future income-eligible individuals with individuals currently covered by Medicaid (
Table 2). Future income-eligible individuals were less likely than current enrollees to be female and to have children but more likely to be employed and married. The future income-eligible group was also less likely to report being in fair or poor health (19.7% versus 27.2%, p<.01) and more likely to have a substance use disorder (14.6% versus 11.5%, p=.03). Future income-eligible individuals were significantly more likely than current enrollees to meet criteria for substance abuse (p<.01) but not more likely to meet criteria for substance dependence. In contrast, future income-eligible individuals were significantly less likely to have serious psychological distress (14.1 % versus 24.1%, p<.01).
We compared the percentage of respondents who received treatment for a substance use disorder in the past 12 months, conditional on having a need for substance use disorder treatment, among future income-eligible individuals who were uninsured and current Medicaid enrollees (
Table 3). The rates of receipt of substance use disorder treatment were low in both groups. In the future income-eligible group only 31.3% of individuals with a need for substance use disorder treatment had received any treatment. The rate was approximately 50% higher among current Medicaid enrollees, 46.8% of whom had received treatment (p<.01). When the analysis excluded informal care received outside the medical sector, treatment rates were much lower: 12.8% in the future income-eligible group and 30.7% in the currently enrolled group (p<.01).
We limited our sample to individuals who met criteria for substance dependence (
Table 3). We expected that rates of substance use disorder treatment would be lower in this group, because this subsample excluded individuals who had no current substance use disorder but who were currently receiving treatment (inclusion of individuals currently in treatment would mechanically increase treatment rates). For this subgroup, the treatment rate among the currently uninsured income-eligible group was significantly lower than the rate among current Medicaid enrollees (19.0% versus 38.7%, p<.01). When informal care received outside the medical sector was excluded, only 12.6% of the future income-eligible group and 30.2% of the current Medicaid group received any outpatient or inpatient care (p<.01). In
Table 3, we also present treatment rates for individuals who met criteria for substance abuse (other comparisons are not shown because of the small cell size); no differences in treatment rates were detected.
We compared characteristics of uninsured future income-eligible individuals and current Medicaid enrollees with and without a substance use disorder (
Table 4). We found that in both groups, those with a substance use disorder were less likely than those without a substance use disorder to be female or married and more likely to be childless and use tobacco. Among those with a substance use disorder, individuals in the future income-eligible group were less likely than those in the current Medicaid group to have concurrent serious psychological distress (28.8% versus 41.1%, p<.01). Among those with a substance use disorder, individuals in the current Medicaid group had higher overall emergency room use (56.3% versus 42.2%, p<.01) and higher inpatient use (20.8% versus 10.8%, p<.01) than those in the future income-eligible group.
Discussion
We found slightly higher rates of substance use disorders among individuals who would be income eligible for Medicaid in 2014 (household income less than 138% FPL) than among current Medicaid enrollees at the same income level. Measures of health status also suggested that Medicaid enrollees are less healthy than their currently uninsured counterparts. For example, rates of serious psychological distress were significantly higher among current Medicaid enrollees, and current enrollees were also significantly more likely to report fair or poor health status. These differences likely reflect the difficulty that individuals with a substance use disorder (in the absence of other health conditions) have obtaining Medicaid by qualifying for SSI or SSDI. Given the availability of effective and cost-effective substance use disorder treatments (
16,
17), providing insurance to these individuals may be appropriate, with the goal of increasing treatment rates.
Both the income-eligible group and the current enrollee group had significant unmet need for substance use disorder treatment, although treatment rates were about 50% higher among current Medicaid enrollees. We found that all of the difference in treatment rates was accounted for by the more severe category of substance use dependence. This finding suggests that without other changes, many individuals with a substance use disorder will remain untreated, even if they gain insurance coverage.
Beyond expansion in eligibility for Medicaid coverage, a number of provisions in the ACA are expected to improve the scope of substance use disorder treatment benefits under Medicaid. Under the new Medicaid “health home” option for people with multiple chronic conditions (including individuals with substance use disorders), Medicaid will pay for services that traditionally have not been reimbursable; the reimbursement will be at a 90% federal matching rate for the first two years after a health home is established (
18). Eligible services include care management, health promotion, postinpatient transition care, referral to social support services, and information technology to link services. The ACA also provides additional funding to improve the capacity of Federally Qualified Health Centers to provide behavioral health care, including substance use disorder treatment. These and other changes under the ACA may improve access to medication-assisted treatments, such as buprenorphine for opioid dependence and acamprosate for alcohol dependence, because these treatments are often used in an office-based setting. Yet access to insurance alone is insufficient to ensure that individuals who need substance use disorder treatment receive services. Undertreatment is also a problem in private insurance. NSDUH data suggest that the treatment rate among the privately insured—9%—was even lower than in the populations studied here (authors’ calculation, results not shown).
Although these ACA provisions suggest that treatment rates may improve for both current Medicaid enrollees and newly eligible enrollees, states will face significant challenges related to coverage of substance use disorder treatment when they implement Medicaid expansions. First, the current primary care workforce may not be adequate to treat the new influx of patients, and available primary care providers may be inexperienced with substance use disorder treatment. Second, some providers of substance use disorder treatment may find it difficult to make the transition from direct service provision financed by state funds to insurance reimbursement. Issues related to provider credentialing, consolidation of the industry, and the movement toward a person-centered integrated care model (
2) will put pressure on providers of substance use disorder treatment to transform how care is delivered. Finally, without additional state enrollment provisions, Medicaid take-up rates for the population with substance use disorders may be low (
19).
When considering the state resources currently used to treat Medicaid beneficiaries with a substance use disorder, it may be useful for states to compare demographic characteristics of the future income-eligible population with those of current Medicaid beneficiaries. As indicated in
Table 4, future income-eligible individuals with a substance use disorder were more likely than current beneficiaries with a substance use disorder to be male, employed, and married. Both groups had similar rates of self-reported fair or poor health status, but income-eligible individuals were less likely than current Medicaid enrollees to have comorbid serious psychological distress. In addition, income-eligible individuals were not significantly more likely than current enrollees to have substance dependence, a more severe disorder.
A rich literature confirms that the pathways into treatment and the success of treatments vary by gender, and different approaches are required (
20). Time constraints in an employed population, especially when jobs are unlikely to be highly flexible, may affect strategies to reduce barriers and to make treatment more accessible (
21). Spouses of newly enrolled Medicaid beneficiaries with substance use disorders will also present both opportunities and challenges, depending on whether they provide support or additional triggers for substance abuse. Tailoring available treatment options to patient characteristics may improve treatment rates and treatment effectiveness.
This study had several limitations. Data limitations include the self-reporting of substance use and mental health symptoms and the absence of information on exact income, immigrant status, or the institutionalized population. The Kaiser Family Foundation has estimated that noncitizens (lawfully present and undocumented) account for 20% of the uninsured U.S. population and that 57% of the noncitizen uninsured group have incomes <134% of the FPL (
22), which suggests that the absence of this information did not drive our results. An advantage of the NSDUH is that the data allow us to look at general prevalence rather than treated prevalence, which is important given that the expansion of insurance coverage may lead to earlier and increased identification of substance use disorders. We considered only whether individuals received any treatments and not the quantity or quality of treatment received.