Behavioral Health Coverage Under the Affordable Care Act: What Can We Learn From Marketplace Products?

The 2008 Mental Health Parity and Addiction Equity Act (MHPAEA) and the 2010 Affordable Care Act (ACA) expanded access to behavioral health services. As policy makers debate changes to the ACA, it is helpful to understand how product design evolved in this context. In the future, insurance options are likely to continue to rely on competition among insurers, making recent experience with marketplaces relevant. MHPAEA requires insurers to cover behavioral health and general medical services similarly in their commercial large-group insurance products. The ACA expanded MHPAEA to apply to coverage available through ACA-established individual health insurance marketplaces and state Medicaid programs. Because the ACA includes behavioral health services in the essential health benefits package, insurers offering products in the marketplace must cover behavioral health services at the same level as the state’s benchmark plan. Together, these policies expanded access to behavioral health treatment to an estimated 62.5 million individuals (1) and increased Medicaid and private insurance coverage in 2014 and 2015 (2,3).

Even with these regulations, marketplace plans have flexibility in designing benefit structures and implementing managed care practices. Plan design options include deductible and copayment amounts, covered services, and when to apply prior authorization and continuing review. Plans can offer narrow networks that exclude many providers or design tiered networks that charge enrollees more for using less-preferred providers.

These choices are particularly acute for individuals with behavioral health conditions and individuals and small businesses purchasing their own insurance in the marketplace. For insurers, individuals with behavioral health conditions are among the least desirable to insure (4), partly because spending on their care predictably exceeds premiums. In health economics, adverse selection occurs when “individuals who expect high health care costs, differentially prefer more generous and expensive insurance plans; those who expect low costs choose more moderate plans” (5). There is concern that insurers may offer less generous behavioral health coverage to discourage enrollment, particularly in individual and small-group products where health care users make purchasing decisions rather than human resources departments (6). Although the ACA risk adjustment provisions were designed to reduce selection incentives, behavioral health risk adjustment methods are less developed and often not effective (4,7,8). A simulation of the ACA marketplace risk adjustment method showed that it did not cover the higher costs of 80% of individuals with behavioral health conditions (9).

Thus, although MHPAEA requires coverage for behavioral health conditions at parity with general medical conditions, health plans retain incentives to select members and also have a significant amount of discretion over covered services (9,10). Coverage, network design, and cost sharing are features that health plan and marketplace Web sites make available to consumers prior to selecting a plan, unlike other features unobservable to consumers at that point (for example, prior authorization) (11).

Although the future of the ACA is unknown, the federal government may shift responsibility for marketplace oversight to states (12), and incentives to avoid individuals with behavioral health conditions are likely to continue and may strengthen. The objective of this study was to assess whether marketplace plans structure benefit design to avoid costly cases. To assess this, we compared products in the health insurance marketplaces to those available in the employer-based group insurance market (nonmarketplace) in terms of behavioral health coverage, management, and network size. The hypothesis was that for a given product, behavioral health benefits will be less generous and management will be more stringent in marketplace products than in nonmarketplace products. Results provide a baseline for comparison of proposed reforms and continued performance of marketplace products.

Methods

Data

Data were from a nationally representative telephone survey of senior health plan executives conducted between August 2014 and April 2015. Questions about the behavioral health services offered in 2014 were asked at the product level (for example, health maintenance organization [HMO] and preferred provider organization [PPO]) within each market area–specific plan. Questions were asked about the plan’s three commercial products with the highest enrollment (nonmarketplace)—primarily the employer-sponsored group market—and the top silver product (marketplace) designed for the individual market.

A panel design with replacement was used and has been described previously (13). The national sample from 2010 was augmented for 2014 with plans not previously operating in the market areas. The primary sampling units were the 60 market areas selected by the Community Tracking Study to be nationally representative (14). The second stage sampled plans within markets. If plans served multiple market areas, they were defined separately, and data were collected by market area. For some national or regional plans, respondents were interviewed regarding multiple sites. In 2014, a total of 344 eligible plans were identified and 274 responded (80%), reporting on 705 commercial products and 121 silver products. The Brandeis University Institutional Review Board approved the study.

Analytic Sample

Marketplace products are classified by actuarial value in four levels. This study focused on silver products (third level) because they were the most commonly purchased type in the marketplace, selected by 70% of participants (15) and the only level eligible for federally subsidized cost-sharing reductions. Silver products thus provide the most financial assistance for participants.

The analytic sample was created by pairing each health plan’s most commonly purchased silver product with its nonmarketplace product of the same type. For instance, if the silver product that the plan reported on was an HMO, the plan’s top nonmarketplace HMO product was selected for the analytic sample. Fifteen marketplace products from plans without a nonmarketplace product of the same type were excluded. The resulting sample consisted of 106 pairs of silver marketplace and nonmarketplace products from the same health plans, for a total of 212 products.

Measures

Product pairs were compared across health plan characteristics and management domains. Plan characteristics included product type (HMO, PPO, or point of service); contracting arrangement (external arrangements contract with a managed behavioral health organization for delivery and management of behavioral health; internal arrangements provide all behavioral health services by plan employees, by a specialty behavioral health organization that is part of the same parent organization, or by a network of providers administered by the plan), whether the plan is national (offers products in multiple states and market areas throughout the United States), profit status (non- or for profit), census region (Northeast, Midwest, South, or West), and marketplace type (administered by the state, by the federal government, or in a joint state and federal “partnership” arrangement).

Health plan management domains included covered services, prior authorization, continuing review, network design, and cost sharing. Services that plans could cover included a continuum of mental health and substance use services. Prior authorization and continuing review are policies under which providers and beneficiaries must actively seek health plan approval before initiating certain services or continuing care after a predetermined amount has been received.

Network design included measures of use of narrow and tiered networks for primary care providers and behavioral health providers. Narrow networks limit the pool of providers available to beneficiaries on the basis of characteristics such as cost, availability, and quality. Narrow networks were defined as networks composed of selected, high-value providers, restricting enrollees to use only these providers and hospitals.

Tiered networks may comprise a larger group of providers to choose from, compared with narrow networks. They function similarly to drug formularies but feature multiple cost-sharing tiers for providers; providers preferred by the health plan have lower copays or coinsurance rates, compared with nonpreferred providers. Tiered networks were defined as follows: “hospitals or physicians that meet criteria imposed by the health plan are identified and favorable financial terms are provided to patients who seek care at those providers.”

The survey asked about narrow and tiered networks for primary care providers, specialty medical providers, and specialty behavioral health providers. Finally, the survey asked how the network of behavioral health providers available for the silver marketplace product compared with the commercial network for the same product, with answer choices of smaller, larger, and about the same.

Cost-sharing measures included the type of cost sharing (that is, copays or coinsurance) for in-network behavioral health and general medical care, as well as level of patient cost sharing (average dollar amount for copays and percentage for coinsurance) for in-network behavioral health care.

Statistical Analyses

Statistical analyses used SAS, version 9.3. Results are reported at the product level and are based on nonmissing values. Observed differences in outcomes between the matched sample of marketplace and nonmarketplace plans were tested for statistical significance by using t tests for dependent samples.

Results

Among product pairs, HMO products were the most common (Table 1). Most products were offered by nonprofit companies (56%) and were not national plans (70%, N=74). Almost half of marketplace products (45%) in this sample were offered on state-based marketplaces, 37% were offered on federally facilitated marketplaces, and 18% were offered through state partnership marketplaces.

The analytic sample of plans offering both marketplace and nonmarketplace products differed in some ways from the nationally representative sample of commercial (nonmarketplace) plans (Table 1). Compared with the nationally representative sample, a significantly higher proportion of plans offering marketplace products managed behavioral health services externally (24% versus 10%). Plans with a marketplace product were also more likely than the nationally representative sample to be nonprofit and to be from the Midwest.

Each service was covered by a large majority of marketplace and nonmarketplace products (Table 2). For mental health care, all products covered inpatient, partial hospital or day treatment, and outpatient therapy. In both samples, 94% of products covered mental health crisis services, and the vast majority covered mental health residential services. For substance use treatment, all products covered inpatient or residential detoxification and inpatient and outpatient therapy. In both samples, most products also covered substance use residential rehabilitation, intensive outpatient treatment, outpatient opioid treatment programs, and 24-hour crisis services. There were not any statistically significant differences in coverage between marketplace and nonmarketplace products.

Prior authorization requirements were not significantly different between the marketplace and nonmarketplace products (Table 3). In both samples, outpatient counseling for substance use disorders was least likely to require prior authorization (13%). In marketplace and nonmarketplace products, prior authorization was most frequently applied for residential rehabilitation related to substance use disorders, with 97% of marketplace and nonmarketplace products reporting use of prior authorization.

Continuing review policies did not differ significantly between marketplace and nonmarketplace products (Table 3). Continuing review was required least frequently for outpatient counseling for substance use disorders (33%) and outpatient opioid treatment programs (51%). Nearly all products required continuing review for inpatient hospital treatment for mental health (marketplace, 96%; nonmarketplace, 95%) and residential rehabilitation for substance use disorders (97% for both).

Marketplace products were significantly more likely than nonmarketplace products to utilize narrow network designs (Table 3) for both general medical providers (marketplace, 67%; nonmarketplace, 46%) and behavioral health providers (marketplace, 16%; nonmarketplace, 11%). Tiered networks were less common than narrow networks in both provider categories but were used more frequently by marketplace than nonmarketplace products for behavioral health providers (marketplace, 14%; nonmarketplace, 8%).

Differences were found in marketplace products offered through state versus federal marketplaces (Table 4). Products offered in state marketplaces were more likely to report using narrow networks for behavioral health providers (state, 24%; federal, 3%). Although tiered networks for behavioral health providers were not used frequently, they were more common in products offered in state marketplaces than in federal marketplaces (state, 20%; federal, 5%).

Small, significant differences were found between marketplace and nonmarketplace products in the type of cost sharing for in-network general medical and behavioral health outpatient care (Table 5). Marketplace products were less likely than nonmarketplace products to require copays for both types of care (marketplace, 75%; nonmarketplace, 80%) The mean copayments for in-network specialty outpatient behavioral health care were almost identical. The mean coinsurance was 20% for marketplace products and 25% for nonmarketplace products.

Discussion and Conclusions

Despite concerns that plans might reduce benefits to avoid costly cases, the marketplace products in this sample were not more restrictive than nonmarketplace products from the same health plans in terms of their behavioral health coverage, prior authorization requirements, or patient cost sharing. This indicates that access to behavioral health services was not limited on the marketplaces by these measures. One possible explanation is that the MHPAEA regulations prevented plans from using approaches that differed greatly between behavioral health and general medical care. Another explanation is that this study was conducted during the first year of the marketplaces (2014), when financial protections for health plans, including risk adjustment, risk corridors, and reinsurance were in place. If in the future these protections are weakened, the generosity of behavioral health coverage may diminish.

Marketplace products were more likely to use narrow and tiered networks for general health services, which has important implications for individuals receiving behavioral health services in primary care settings (16). Behavioral health conditions are increasingly treated in primary care (17). Nonpsychiatrist physicians write the majority of new psychotropic prescriptions (16). Individuals with behavioral health conditions frequently have co-occurring general medical conditions. For these reasons, both primary care and behavioral health networks affect access to behavioral health care. Thus, although most products did not report restricting behavioral health provider networks, restrictions on primary care provider networks may disproportionately discourage individuals with behavioral health conditions from enrollment.

Differences in marketplace products between federal and state marketplaces are important. If ACA reform shifts responsibility to the states, the new system may look more like the state marketplaces. In this study, products offered on state marketplaces were more likely to use narrow and tiered networks for behavioral health providers, making access to behavioral health providers more challenging.

Other researchers have found that group insurance products and nonmarketplace products were more likely to be actuarially equivalent to gold and platinum marketplace products (18), suggesting that silver marketplace products are likely to be less generous than nonmarketplace products. This study did not examine premiums or deductibles, important factors in determining actuarial value; however, in terms of coverage, cost sharing, and utilization management, the silver products were similar to nonmarketplace products.

It is possible that plans choosing to enter marketplaces are systematically different from plans that do not offer marketplace products. Comparison of these findings with findings from our national sample of health plans showed lower rates of coverage for behavioral health services and some indication that plans operating in marketplaces managed access more strictly. Although coverage rates were generally high, plans that offered both commercial and marketplace products were less likely to cover residential care, crisis services for substance use and mental disorders, and opioid treatment services and were more likely to require prior authorization for outpatient services in their commercial products, compared with plans that offered only commercial products (19). Future research should examine other differences among plans by marketplace participation.

Limitations

Although selected from a nationally representative sample of private health plans, the analytic sample was not itself nationally representative. Excluded were health plans that did not operate in commercial health insurance markets except through the marketplaces, including, for example, those that specialize in the Medicaid population and have expanded to marketplace products. The large national sample of health plans was strengthened by the comparison between marketplace and nonmarketplace products from the same health plans. Most measures had complete data, and the overall survey had a response rate of 80%. Increased nonresponse on cost-sharing information may reflect the complexity of cost-sharing arrangements employed by insurers. This study examined several policies indicative of health plan coverage and generosity but did not assess premiums or deductibles. Health plan products may differ in these and other unmeasured ways, including provider payment rates, stringency of medical necessity criteria, use of quality improvement programs, and other administrative procedures (20). The findings are based on a survey of health plan executives and may not reflect the actual experience of individuals when they try to access behavioral health services (21). Executives were asked about network policies, but network size was not quantified. Therefore, we may not have identified networks that were narrow either because of a lack of provider availability or because of a lack of provider participation possibly driven by low payment rates. Clear definitions of narrow networks have not been established; results may vary depending on the definition.

Policy Implications

In 2014, insurers appeared to be relying more on network design than on coverage, utilization management, or cost sharing to address the special challenges of cost management in marketplaces. This may reflect weaker government regulation of network management than of other plan features. As the federal government shifts more responsibility for oversight to states (12), networks remain important to monitor.

Health plan products are more likely to develop narrow networks for primary care providers and in state marketplaces. The prevalence of narrow network products on state marketplaces varies, depending on methods and definition, but it is substantial (22,23). A recent study showed that narrow network marketplace products have lower premiums (24). Together with findings that products do not differ on other dimensions, these findings support the idea that networks are the key feature on which plans are now competing and may also be used to encourage selection of healthier enrollees (25).

Overall, the ACA was successful in ensuring that robust behavioral health coverage was available in marketplace plans. Today’s insurance environment is uncertain (26). The number of insurers participating in the marketplaces has declined from more than six per state in 2015 to about four per state in 2017, with many counties having one or zero insurers (27,28). Efforts to improve risk adjustment and extend reinsurance, which can help insurance companies when costs exceed premiums, may provide the stability insurers need to continuing operating in marketplaces. Without more stability, improvements in access achieved by the ACA could be eliminated even if the law is not repealed. Proposed repeal plans have envisioned eliminating mandates on individuals to purchase insurance and on plans to cover benefits defined as essential, which include behavioral health services. Without those requirements, individuals with behavioral health conditions will again be at risk. Findings from this study provide an important indicator of features to track and baseline information to which we can compare future systems.