|
The Costs Of Mental Health Parity: Still An Impediment?
Parity in mental health benefits rectifies unfairness in health insurance coverage and reduces financial risk for those with mental illness. However, increased coverage for mental illness has been seen as creating inefficiencies and increasing total spending, based largely on results from the RAND Health Insurance Experiment conducted in the 1970s. Newer evidence suggests that cost control techniques associated with managed care give health plans alternatives to discriminatory coverage for containing costs. We review both eras of research on mental health insurance and conclude that comprehensive parity implemented in the context of managed care would have little impact on total spending.
PARITY HAS BEEN THE STATED OBJECTIVE of mental health advocates since differences in mental and general health coverage first arose in the early days of private health insurance.1 The case for parity has been based primarily on the fairness argument that insurance should not discriminate against people with mental illnesses. Parity equalizes benefit design provisions, putting them on par with other medical benefits in private insurance. Advocates have had to contend with the question of why, if equalizing benefits was such an attractive idea, purchasers of private health insurance did not demand parity. "Stigma" was one answer, but economic analysis supplied another, based on the rational calculations of insurers and employers in markets subject to adverse selection. All competing health plans might "underprovide" coverage for some health conditions because of risk-selection fears, even if potential enrollees value the coverage in excess of the costs of providing it.2 Equity-in-access and adverse-selection arguments regarding parity have been articulated for many years, but until recently they have not been on the winning side of the debate. Against parity policies stood both efficiency and cost objections.3 The efficiency argument is well known in mental health services research.4 Traditional welfare economics views the demand for mental health care as representing the consumer’s willingness to pay for care. Insurance coverage reduces the price to the consumer, inducing more use; this is labeled "moral hazard." Services used only in the presence of the insurance-reduced price are not worth the cost to the consumer, assuming that the demand curve represents willingness to pay. A large literature has grown around the question of measuring this demand response. Demand response was found to be greater for mental than for physical health care, which implies that a given reduction in price resulting from insurance would create more inefficiency in mental than in physical health care use.5 This finding sets up the main efficiency argument against parity: In the interest of consumers, coverage should not be equal for physical and mental care. It is important to emphasize that these arguments about parity have to do with cost response, not the absolute level of spending. Policymakers and others have often argued that mental health care should not be covered because it is "too costly." Costly and unexpected adverse health events are what insurance should cover. Furthermore, the converse argument—"psychotherapy costs are low, so we should cover them"—is also inconsistent with the principles of insurance. Managed care matters to the economic principles bearing on coverage, as we shall see, but not because it makes coverage less costly. Managed care alters the efficiency arguments bearing on parity by changing the methods of rationing health and mental health care.6 Under managed care, benefit design provisions are only one mechanism that affects the use of mental health care. Because managed care introduces other mechanisms for controlling moral hazard and cost sharing, a health plan need not rely exclusively on such benefit-design features to control costs. Suppose, for purposes of discussion, that these mechanisms reduce a good deal of the "low value" use of health services. Then, consumer cost sharing need no longer trade moral hazard–induced inefficiencies for the loss of financial protection. If moral hazard is better dealt with by other mechanisms, benefits can be designed with a focus on risk protection. Formal analyses of payment systems confirm that benefit-design features can be assigned the task of protecting against financial risk, while policies aimed at affecting providers’ behavior, such as payment arrangements, can contend with moral hazard.7 Despite the dramatic changes that have occurred in mental health care delivery during the past fifteen years, some health insurers and employers remain wary of the cost of improving mental health benefits. Skeptics suggest that expanded mental health services will yield few benefits and generate large cost increases. In 2003, Donald Young, then president of the Health Insurance Association of America (now America’s Health Insurance Plans), described federal parity legislation as "a hidden tax" and "a misguided effort to provide additional treatment resources for a wide variety of ill-defined and difficult-to-diagnose mental disorders."8 In his message vetoing a bill to expand the Maine parity law delivered 11 April 2002, then governor Angus King said: "As we look for ways to reduce the costs of health care, we must not exacerbate the problem by adding new mandates. When you are in a hole, the first rule is not to dig any deeper."9 Fortunately, a rich body of research on response of costs to changing mental health coverage means that policymakers can base coverage decisions on evidence. Here we review this research with an eye toward consolidating an evidence base.
We distinguish two eras in our review of the empirical literature on the effects of mental health insurance. The first era encompasses research on data from the 1970s and 1980s, before managed care. The second era dates forward from the mid-1990s. Because the majority of studies are not randomized controlled trials, we opted against conducting a formal meta-analysis. Rather, we qualitatively compared the entire body of literature with attention to the methodological approaches employed and key evidence amassed in both eras. The studies conducted in these two periods differ in their primary research aims, methods, and implications for the cost of parity. Early studies focused on estimating the effect of prices on demand for ambulatory mental health care; more recent studies exploited natural experiments to evaluate parity (or near-parity) coverage expansions in the context of shifts to managed care and do not provide direct evidence on price elasticities.10 Some studies followed an "experimental" group over time and compared changes with those in an investigator-constructed "comparison" group. Some of the managed care–era studies compare people living in states that enacted parity laws with those living in other states. Because of the complexity of these real-world interventions, it is not always possible to isolate a response to price in these studies. Most importantly, research findings from these two eras differ in their implications regarding the cost of parity. Early evidence supported employers’ concern that outpatient parity would "break the bank." In contrast, the managed care studies, despite not always being able to identify price elasticity, indicate that enactment of full parity would not greatly increase total spending and would protect families against financial risks associated with mental illness. Early evidence on demand response. When the first studies of price of and demand for ambulatory mental health care were initiated in the 1970s and 1980s, it was an open question whether psychotherapy should be covered by health insurance. From the time mental health coverage was first offered as part of major medical contracts in the 1950s, insurers worried that intensive or long-term psychotherapy would drive up premiums. This concern prompted insurers to exclude or impose special limits on coverage for mental health services. These limits were tolerated in part because of the presence of a public safety-net mental health system to care for those disabled by mental illness.
Early studies sought to understand whether and to what extent demand for outpatient mental health care (with a focus on psychotherapy) differed from that for general medical care (Exhibit 1
The most definitive evidence that better mental health coverage would increase demand came from the RAND Health Insurance Experiment (HIE) (the Manning study in Exhibit 1  ), in which the elasticity for ambulatory mental health care was also estimated to be more than twice that for general medical care. The RAND HIE randomly assigned subjects to health insurance conditions, thereby minimizing the danger that insurance would be correlated with unobserved variables.11 The HIE evidence on demand response applies primarily to outpatient psychotherapy (not inpatient psychiatric care). Collectively, these five studies supported the conclusion that higher cost sharing for psychotherapy was justified. A consensus emerged by the end of the 1980s, on the basis of these studies, that validated insurers’ decisions to limit mental health coverage and confirmed policymakers’ concerns about the cost of regulating mental health benefits. The consensus was embodied in the recommendations for mental health during the 1993–94 national health reform debate. The basic benefit package proposed under the Clinton plan would have included thirty days of inpatient and residential care and thirty outpatient psychotherapy visits, with $25 cost sharing for managed care and 50 percent outpatient cost sharing for unmanaged care.12 With the research supporting the status quo of differential coverage, mental health benefits, if anything, decayed in relation to benefits for other conditions during the late 1980s into the 1990s. Data from the U.S. Bureau of Labor Statistics (BLS) indicate that private insurers augmented their limits on mental health benefits over this period. In 1981, for example, 41 percent of full-time, privately insured workers were subject to limits on inpatient mental health care, and 83 percent had limits on outpatient care.13 By 1995, 89 percent had inpatient mental health limits, and 96 percent had limits on outpatient care.14 The increased use of day and visit limits was particularly pronounced. Unlike higher deductibles and coinsurance for mental health services, use of such limits is harder to justify on efficiency grounds because when sizable cost sharing is in place, limits offer only modest savings. As a result, the decision to impose these limits has been viewed as motivated largely by incentives to risk-select rather than to address moral hazard. In fact, evidence from the Federal Employees Health Benefits (FEHB) Program suggests that adverse selection worsened over this period. In the 1980s and 1990s, the adverse selection–driven dynamics of coverage offerings in national FEHB plans approached "death spiral" proportions, with plan restrictions leading to an eventual decline in the percentage of total dollar value of claims in behavioral health services (excluding prescription drugs) from 7.8 percent in 1980 to 1.9 percent in 1997.15 Short-circuiting such market failures had been the economic rationale for state benefit mandates during the 1970s and 1980s.16 Although these mini-parity laws were popular, they only placed a floor in coverage at a low level, and only for plans subject to state insurance regulation. Furthermore, benefit mandates were often denominated in dollar terms, and medical price inflation eroded their purchasing power. Recent evidence under managed care. A second generation of research in the late 1990s studied mental health coverage in the context of managed care. These studies were more heterogeneous in their central research aims, methodological approaches, and outcomes of interest. Instead of focusing exclusively on estimating demand response to price, research addressed such questions as how total mental health spending and consumers’ out-of-pocket spending changed in response to parity-like coverage expansions. This virtue was born of necessity: The natural experiments under study were multifaceted changes, with improvements in benefit design often coupled with introduction of managed care. With the emergence of a federal debate over parity in Congress in the early and mid-1990s, this period also included debate over competing actuarial reports estimating how parity would likely affect health insurance premiums. Renewed interest in the effects of mental health benefit changes on costs was prompted in part by emerging evidence of managed care’s role in controlling overall health care spending. The nature and extent of managed care varies tremendously across health plans and over time, which makes measuring its effects difficult. (These measurement problems are exacerbated by risk segmentation if managed care plan enrollees differ from those in conventional insurance.) Evidence on the effects of managed care suggests that they have been instrumental in reducing inpatient admissions, inpatient lengths-of-stay, and total spending on inpatient care, with a concomitant increase in outpatient visit rates across the health sector.17 In the mental health context, managed care contracts often take the form of risk-sharing arrangements negotiated by employers or insurers with specialized managed behavioral health care "carve-out" firms having expertise in establishing specialty provider networks, negotiating payment rates, and managing use to affect the supply of mental health services. The carve-out industry has grown rapidly: In 2002, 164 million people were covered, compared with 70 million in 1993.18 One stated goal of second-generation mental health coverage studies was to examine whether managed care, largely done by carve-out companies, would enable more-affordable implementation of parity.
In contrast with those of the earlier era, these second-generation studies did not find large mental health spending increases attributable to parity, and all studies that addressed risk protection identified sizable decreases in consumers’ out-of-pocket mental health care spending (Exhibit 2
The consistent results of the FEHB evaluation, the Vermont parity evaluation, and the study of Massachusetts state employees provide the best evidence on how parity coverage simultaneously affects service use, total spending, and out-of-pocket spending on mental health. The FEHB parity initiative was the result of a presidential directive implemented in 2001 and constitutes the most comprehensive parity policy enacted to date. An evaluation found that MH/SA spending in the seven health plans studied was on par with or below that of other large, privately insured populations. In five of these plans, the parity policy was associated with sizable reductions in out-of-pocket spending. The Vermont parity evaluation also found that consumers paid a smaller share of the total amount spent on MH/SA services after implementation of comprehensive parity. For those with serious mental illnesses, the decrease in out-of-pocket spending following parity was particularly large: Spending declined by more than half among people spending more than $1,000 annually on MH/SA services. Within the two plans studied, the probability of using outpatient mental health services increased without prompting much spending growth. Likewise, in evaluating MH/SA benefit expansion by the state employees in Massachusetts, Ching-to Ma and McGuire estimated a minimum of 30–40 percent overall MH/SA cost reduction after the simultaneous expansion of benefits and initiation of a carve-out contract. They found decreases in consumers’ MH/SA spending, the probability of outpatient use, outpatient visits per user, and inpatient length-of-stay, with no change in inpatient admissions (but some shift to less intensive settings). Although the studies by William Goldman and colleagues and by Yuhua Bao and Roland Sturm did not examine out-of-pocket costs, their total cost and use results are similar to the broader literature in direction and significance. Goldman and colleagues examined the simultaneous introduction of an expansion in MH/SA benefits and introduction of a carve-out contract by a private employer. They identified a sizable reduction in MH/SA costs following these changes, resulting from fewer outpatient visits per user, a reduction in the probability of inpatient admissions, reduced lengths-of-stay, and lower costs per unit of service. They also detected an increase in the probability of using any mental health care over this time period. Using a difference-in-difference econometric approach to estimate the effect of parity laws enacted at the state level, Bao and Sturm found no significant difference in perceived access to care among those with differing mental health care needs in states with and without these policies.19
A key challenge in interpreting the findings from these studies involves separating the effects due to initiation of MH/SA benefit changes from those due to simultaneously occurring shifts in managed mental health care. Because supply-side rationing through managed care is inherently less transparent than control through benefit limits, we know relatively little about how various managed care tools contribute to changes in use and spending. Some studies cannot directly identify managed care changes; this is true for all multistate studies comparing people living in states with and without parity laws. Although researchers in four of the five studies were aware of shifts to carve-out contracts concurrent with MH/SA benefit changes, these research settings offer limited opportunity to peer inside the "black box" of a carve-out contract to identify how various managed care mechanisms affect outcomes. Although evidence is sparse, network incentives appear to exert subtle but powerful control over the quantity of services used. Network incentives arise from the cooperation that a plan can expect from providers who value being included "in network." Ma and McGuire attempted to decompose the quantity reductions identified among Massachusetts state employees that were directly attributable to managed care (Exhibit 2 The technology of treating mental illness also was transformed during this period. Treatment innovations and associated spending increases were most notable in psychopharmacology.22 The amount of spending on psychotropic drugs as a proportion of overall mental health spending rose from 8 percent in 1987 to 21 percent in 2001.23 Along with the increased availability of effective medications, growth in psychotropic drug use is attributable to its comparatively generous coverage. The absence of quantity increases attributable to parity across these studies is consistent with more recent actuarial estimates of the effect of parity on premiums. Actuarial estimates are calculated as the expected change in total premium as a result of parity. Studies conducted in the early and mid-1990s produced widely disparate estimates, ranging from a 1 percent to an 11 percent increase in total premiums due to parity, with the Congressional Budget Office (CBO) estimating a 4 percent increase.24 (Some of these analyses were commissioned by interest groups such as the ERISA Industry Committee, the Association of Private Pension and Welfare Plans, and mental health provider groups.) Lack of uniformity in estimated effect on premiums due to parity led to an effort to incorporate managed care effects into actuarial models using more recent cost data from the FEHB, state parity experiences, the managed behavioral health care industry, and private employers.25 After updating its estimation methods to incorporate managed care effects, the CBO scored comprehensive parity as raising group health insurance premiums by an average of 0.9 percent.26 CBO analysts also forecast a net 0.4 percent increase in total premiums after accounting for the offsetting impact of behavioral responses by health plans, employers, and workers.
A comprehensive parity bill, the Paul Wellstone Mental Health Equitable Treatment Act, is pending in Congress. In the mold of the FEHB parity policy, this legislation would prohibit higher cost sharing and deductibles or separate inpatient and outpatient service limits for in-network private mental health insurance coverage. The main argument against enacting a comprehensive federal parity law of this kind is that generous coverage would drive up mental health spending, increase premiums, and expand the number of people unable to afford coverage. Old ideas about the cost of parity die hard. In our view, the relevant research implies that parity implemented in the context of managed care would have little impact on mental health spending and would increase risk protection. Some non-cost-related objections to parity laws have been raised over the years. Parity opponents often raise philosophical objections to government mandates not limited to parity mandates. Some argue that insurance regulation is a state, not a federal, function. Doubts about the effectiveness of the treatments for mental disorders paid for under parity might also influence perceptions about the value of the legislation, although one authoritative recent review by the U.S. surgeon general addresses these concerns.27 Even if quantity of use is unchanged, an improvement in coverage would shift costs from the beneficiary to the insurer. Employers and insurers might oppose parity policies because of these shifts, even if there is no welfare loss. The bottom line is that while reasons for opposing parity might remain in the managed care era, opposition to parity on the basis of increased total spending no longer constitutes an evidence-based objection. Full parity will not cure all ills in the mental health care system.28 The very mechanisms that have weakened the traditional cost control argument against parity imply that competitive insurance markets might continue to supply inefficiently low levels of mental health care even in the presence of parity laws. Managed care tactics substitute for demand-side cost sharing. Thus, parity laws regulate one dimension of cost and access control (benefit design) and leave others (utilization review, network design, physician incentives) open for use by plans to discourage enrollment by people with mental illnesses. As David Mechanic and Donna McAlpine put it, "Parity in benefit structures means little if ADM [alcohol, drug, and mental health] care is managed more stringently than other types of health care."29 One could take comfort in the observation that parity fixes at least one problem related to equitable access to mental health treatment (benefit design). But this fallback position has problems. A plan will presumably react by tightening elsewhere if regulators force it to make demand-side cost-sharing provisions more generous. One thing is clear: Parity leads to more financial protection. More research is needed to better determine how increased financial protections due to parity might differentially affect those with more or less severe mental health conditions. However, it is important to remember that under parity, the traditional incentives to avoid enrolling people with high expected costs remain at least as strong as in the past, while the mechanisms available to health plans for affecting selection have expanded with managed care.30 A sizable majority of people with mental disorders, even among the insured, do not receive treatment in a given year.31 Remarkable scientific advances have led to the availability of various effective treatments for most mental health disorders; however, there is an urgent need to improve dissemination of evidence-based treatment into real-world practice settings. Expanding benefits under parity might help, but it does not solve the problem of unmet need or ensure use of evidence-based medicine in mental health care. Other private and governmental initiatives are better suited to advancing public policy in these areas. Passage of comprehensive parity would allow policymakers, health care managers, and clinicians to shift attention away from benefit design and toward figuring out how to get effective treatment for people who would benefit.
Colleen Barry (colleen.barry{at}yale.edu) is an assistant professor in the Department of Epidemiology and Public Health, Division of Health Policy and Administration, at the Yale University School of Medicine in New Haven, Connecticut. Richard Frank is the Margaret T. Morris Professor of Health Economics in the Department of Health Care Policy at Harvard University in Cambridge, Massachusetts. Tom McGuire is a professor of health economics in that department. Colleen Barry acknowledges support from the Robert Wood Johnson Foundation through the Changes in Health Care Financing and Organization (HCFO) Initiative. Richard Frank and Tom McGuire received support through National Institute on Drug Abuse (NIDA) Grant no. DA10233-06. Frank also acknowledges support from the John D. and Catherine T. MacArthur Foundation. The authors thank Don Metz, Howard Goldman, and two anonymous reviewers for constructive comments on an earlier draft of this paper.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Two Eras Of Economic...
Implications
