This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (11) Citing Articles via Google Scholar Google Scholar Articles by Zuvekas, S. H. Articles by Narrow, W. E. Search for Related Content PubMed PubMed Citation Articles by Zuvekas, S. H. Articles by Narrow, W. E. Related Collections Access To Care Managed Care - Mental Health Mental Health/Substance Abuse Health Spending

Mental Health

The Impacts Of Mental Health Parity And Managed Care In One Large Employer Group

Abstract

 
We examine the impacts of a state mental health parity mandate on a large employer group, which simultaneously introduced a managed behavioral health care carve-out. Overall, we find that mental health/substance abuse (MH/SA) costs dropped 39 percent from the year prior to three years after parity, with managed care offsetting increases in demand induced by parity coverage. Managed care was most effective in reducing very high inpatient use among adolescents and children. The effect of the parity mandate on access was ambiguous: While treatment prevalence rose nearly 50 percent, similar increases were observed for groups not subject to the mandate.

With mounting evidence that managed care reduces the cost of parity, many states have enacted stronger forms of parity legislation than the very limited federal Mental Health Parity Act (MHPA) of 1996. But not every state has enacted parity, and most state mandates fall well short of full parity.³ Even in states with strong mandates, many employers are exempt under the federal Employee Retirement Income Security Act (ERISA). The sunset of the original federal MHPA in 2001 shifted the parity debate back to the national level; in December 2001 Congress failed to pass new legislation regarding mental health parity, although the existing law was extended until the end of 2002. Many questions remain, not just about the cost of parity but about the ultimate impact of parity mandates on access to MH/SA treatment, especially when parity is combined with managed care. In other words, does parity make a difference for those with MH/SA needs?

We build on and extend previous studies on this topic by examining the experience of a large employer group (well over 100,000 enrollees) subject to a typical parity mandate enacted in the 1990s. In addition to providing further evidence of the minimal cost implications of parity, we use the unique features of our case study to focus on two less-settled questions: (1) What is the impact on access to MH/SA treatment? and (2) what is the impact on children and adolescents?

Previous studies provide mixed evidence of the impact of parity on access.⁴ In addition to contributing another case study, we have concurrent data available for other employer groups not subject to the parity mandate, providing us with potentially useful controls for a quasi-experimental design. Although we have concerns about the comparability of these other employer groups to our parity group, their data raise a number of important questions about whether changes in access should be attributed to parity or to managed care and general trends, such as growing demand for psychotropic drug treatment or reduced stigma.

We further exploit the large size of our case study to examine the effects of parity and managed care on different age groups. What little is known about children and adolescents suggests that they may have different use patterns and experience with managed behavioral health care.⁵

Case-study background. The state mandate covering our large employer group required full parity (the same coverage as for general medical services) for severe mental disorders only. Coverage for severe disorders thus moved from typical fee-for-service (FFS) mental health benefits to typical preferred provider organization/point-of-service (PPO/POS) coverage. The group also chose to simultaneously expand in-network coverage for treatment of nonsevere mental disorders and out-patient drug and alcohol disorders, by greatly reducing copayments and deductibles but maintaining standard mental health visit and day limits. Inpatient drug and alcohol treatment was previously required to be covered at full parity levels.

The strong economic incentives to contain the expected cost increases from the benefit expansion led the sole FFS insurer for the large employer group to carve out the MH/SA services and manage them separately when the mandate became effective.⁶ In-network providers were reimbursed on a negotiated FFS basis. Unlike previous case studies of parity and managed care, such as the Massachusetts Group Insurance Commission (GIC) program for state employees, the carve-out management was not at risk.⁷ In addition to network management, the managed care carve-out consisted of prior authorization and concurrent utilization review for all inpatient services and in-network outpatient services, with enrollees calling to obtain access to in-network services for a specified number of visits and time period.

Methods

Top Methods Data And Measures Study Results Discussion And Conclusion NOTES

 
The parity mandate combined with the resulting carve-out becomes the basis for our case study or natural experiment. We compare the experience of the continuously enrolled population from the large employer group subject to the parity mandate for one year before and three years after the implementation of parity. We focus here on changes in plan costs, utilization patterns, and access and how these changes differed among children and adolescents, employees, and spouses. The four study years are denoted Year 1 to Year 4. Parity became effective for the parity group on the first day of Year 2, so that Year 1 constitutes the one-year pre-period, and Years 2 through 4 represent the post-period. The carve-out arrangements were also introduced on the first day of Year 2 for the parity group; prior to that, MH/SA services were managed as an FFS benefit.

While enrollees in the parity group had a health maintenance organization (HMO) option, enrollment in the sole FFS plan remained stable throughout the entire period. However, to minimize the potential effects of adverse selection, we restrict analyses to the population that was continuously enrolled during the four-year study period.⁸ We find mixed evidence of adverse selection. Those we excluded because they were enrolled before but not after parity implementation had higher MH/SA costs than did those enrolled all four study years. But we also found that those we excluded because they enrolled after parity had higher costs as well. The two excluded groups tend to cancel each other out, so that we find little differences in results whether we use the population enrolled for all four years (as reported here) or all enrollees.

The first set of analyses compares changes in access, cost, and utilization from the pre-period through the post-period for the parity group as a whole, as well as for employees, spouses, and different age groups for nonspousal dependents. We also decompose the changes in plan costs attributable to (1) employees, spouses, and dependents; (2) inpatient versus outpatient treatment; and (3) utilization versus unit costs.

In addition to pre/post comparisons in the parity group, we make comparisons to other employer groups not subject to the parity mandate, collectively labeled the control group. This group comprises a number of small and medium-size employer groups that provide managed FFS coverage through the same insurance carrier that covers the parity group, during the same time period and in the same state. Most of these groups did not have HMO or other enrollment options. We initially hoped to have a sample of groups with the same managed care as the parity group and groups without, to separate the effects of the parity benefit and managed care in the parity group. However, these groups were all carved out at the same time as the parity group, eliminating any ability to separate these effects. Carve-out functions were performed by the same firm for both the parity and control groups, although not every group in the control group required authorization for outpatient services. While imperfect, the control group does provide some control for secular changes over time.

We performed standard tests of statistical significance for differences across time and between the parity and control groups, accounting for the correlation in enrollees over time. Unless otherwise noted, the numbers reported in the text are significant at the .05 level.

Data And Measures

Top Methods Data And Measures Study Results Discussion And Conclusion NOTES

 
The data are drawn from proprietary enrollment and claims files provided by the insurer covering the parity and control groups during the four-year study period. Persons age fifty-five and older during the pre-period were excluded from the analyses to minimize problems of distinguishing employees from retirees, who are likely to have very different use and cost patterns.

The parity group had approximately 75,000 members continuously enrolled during the four-year study period, with about 39,000 employees, 9,000 spouses, and 27,000 nonspousal dependents. There were approximately 13,000 continuously enrolled persons in the control group, with about 9,000 employees and spouses. We did not separately analyze different enrollee and age groups in the control group because the sample sizes were much smaller. A data anomaly in the enrollment files in the first year reduced the number of potential continuous enrollees in both the parity and control groups.

Person-level measures of MH/SA use and cost were then constructed for each year from the claims data, with MH/SA services identified by whether the carrier paid the claim under the mental health, substance abuse, or general medical benefits of the plan. International Classification of Diseases, Ninth Revision (ICD-9), diagnosis codes were the primary basis for this assignment by the carrier.

Utilization measures include inpatient admissions and days (including days spent in residential treatment facilities) and outpatient visits, which combine hospital outpatient and office-based care. "Cost" measures the amount that the carrier actually paid to providers. Payments to physicians and other professionals for treatment associated with inpatient stays were combined with the facility payments. In addition, treatment prevalence rates, a widely used proxy measure of access, were constructed from the percentage of the population with any MH/SA service use. MH/SA-related prescription drug use and costs were excluded.

Although we constructed separate measures for mental health and substance abuse, we only report the results of the combined MH/SA measures. MH/SA services exhibited remarkably similar patterns, and substance abuse represented less than 10 percent of total MH/SA costs.

Study Results

Top Methods Data And Measures Study Results Discussion And Conclusion NOTES

 
Pre/post changes. Treatment prevalence. Overall, treatment prevalence rose nearly 50 percent by the third year after the implementation of parity and the carve-out (Exhibit 1). This steady rise suggests that the increases are not an artifact of switching from an FFS plan (where some persons may not file claims because they have not met their deductible) to a PPO/POS plan with better first-dollar coverage. Treatment prevalence among children ages six to twelve rose 60 percent by the fourth year, while adolescents ages thirteen to seventeen saw their prevalence of treatment rise from 4.5 percent to 6.7 percent.

Inpatient use. Overall, the number of admissions did not change significantly during the four-year study period, starting at 5.6 admissions per 1,000 members in the pre-period and ending at 5.2 admissions in the fourth year (Exhibit 2). However, average lengths-of-stay per admission declined by two-thirds to reach nine days by Year 4. Thus, the main effect of the carve-out was to reduce lengths-of-stay rather than number of admissions. Some increase in admission rates would be expected with the parity benefit expansion.

As a consequence of much shorter lengths-of-stay and slightly lower admission rates, the total number of inpatient days for all enrollees was down by almost two-thirds by the end of the study period (data not shown). For dependents, the total number of inpatient days decreased by three-quarters, with the bulk of the decrease coming from those ages 6–17.

In addition to the declines in average number of days, there were substantial changes in the distribution of enrollees with longer lengths-of-stay (Exhibit 3). By Year 4 only 18 percent of employees and 23 percent of spouses receiving inpatient treatment spent more than fifteen days in the hospital, compared with 49 percent of employees and 55 percent of spouses in the pre-parity period.

View larger version (20K): [in this window] [in a new window]   EXHIBIT 3 Parity-Group Enrollees With Inpatient Stays: Percentage With More Than Fifteen Days

Outpatient use. Outpatient treatment use (treatment in hospital outpatient departments, emergency rooms, providers’ offices, and clinics) increased substantially from Year 1 through Year 4 (Exhibit 4). The changes in outpatient use mirror the changes in treatment prevalence (see Exhibit 1), with the percentage with any outpatient use rising 50 percent. However, the mean number of visits for those with any use remained fairly constant through the study period, resulting in a 50 percent increase in the total number of MH/SA visits.

Decomposition of cost changes. In decomposing the changes in MH/SA costs, we found that 97 percent of the cumulative decline in plan costs over the four-year period is attributable to changes in costs for dependents, with 64 percent from adolescents ages 13–17 and 27 percent from children ages 6–12. Outpatient costs rose 85 percent over the four-year period, with about three-quarters of the increase coming from more people getting treatment and the remainder from increases in per visit costs. In contrast, total inpatient costs fell 70 percent, with four-fifths of the decline attributable to reduced lengths-of-stay. Because inpatient costs accounted for 80 percent of costs in Year 1, the increases in outpatient costs in Years 2-4 were offset by declines in inpatient costs by a factor of 3.4 to 1. By Year 4 inpatient costs accounted for only 40 percent of MH/SA plan costs.

Control-group comparisons. Exhibit 5 compares the experience of the parity group to the control group. We limit the analyses to employees and spouses in both groups, because the very high initial inpatient use among dependents in the parity group may be idiosyncratic. The initial treatment prevalence, inpatient use, and out-patient use in Year 1 were at significantly lower levels in the control group than in the parity group, perhaps reflecting lower benefit levels and different levels of need and demand for MH/SA services. The control group experienced slightly lower increases (p = .06) in treatment prevalence in absolute terms compared to the parity group over the four-year study period (2.3 percent versus 1.8 percent, p = .06) but a somewhat larger increase in percentage-change terms.

Discussion And Conclusion

Top Methods Data And Measures Study Results Discussion And Conclusion NOTES

 
Consistent with previous case studies, we find strong evidence that MH/SA plan costs declined (39 percent) after the simultaneous implementation of a parity benefit and a carve-out for benefit management.¹⁰ While almost all of the cost decrease came from steep reductions in inpatient use among children and adolescents, costs for employees and spouses remained unchanged. Clearly, increased management of mental health services offset parity’s increase in benefits.

Administrative costs. It is important to note that administrative costs likely rose with the introduction of the carve-out, although we do not know how much. However, it is unlikely that additional administrative costs consumed anywhere near the treatment cost savings from managed care. Even if standard FFS administrative loading factors (generally, 11–12 percent) tripled, total plan costs for MH/SA treatment still declined by one-quarter.

Access and quality. With large decreases in the amount spent on MH/SA treatment comes the inevitable concern that access to and quality of care suffer. However, treatment prevalence rose almost 50 percent in the parity group, driven entirely by large increases in the number of people using outpatient treatment. And while more people were in treatment, the number of visits did not change, which suggests that the carve-out did not sharply limit access.

Parity itself would seem the most likely explanation for the apparent increase in access, but this is doubtful for several reasons. First, the largest increases in treatment prevalence occurred two years after the benefit expansion. According to economic theory, the benefit expansion should have led to an immediate increase in the demand for treatment but no change in subsequent years. Second, Kevin Hennessy and Howard Goldman argue that parity in benefits does not directly address the problem of stigma, which keeps many people from seeking treatment. They state that "legislating full parity will promote, but not achieve, treatment equity for those with mental and addictive disorders."¹¹ According to this argument, it is unlikely that the benefit expansion alone led to increased use. Third, despite large increases in MH/SA coverage, the rise in treatment prevalence for the parity group was only marginally greater than for the control group.

Causes of increased service use. We conclude that factors other than the benefit expansion likely drove most of the increases in treatment prevalence. Undoubtedly, nationwide trends toward greater use of MH/SA services, likely stimulated by demand for psychotropic drug treatments, as well as reduced stigma are factors but cannot fully explain the large year-to-year increases we saw.¹² Intriguingly, the managed behavioral carve-out program itself may explain much of the increased use, perhaps through better coordination and education efforts with specialists, primary care providers, and consumers.

Our finding of a significantly increased number of people using MH/SA treatment stands in contrast to previous studies, where treatment prevalence either remained largely unchanged or even decreased with parity or other benefit expansions and managed care.¹³ We note that the populations in these other studies, especially the Massachusetts state employees, tended to have much higher initial rates of outpatient treatment use than our parity group did. The carve-out firms for these higher-use populations no doubt focused on reducing outpatient use. The application of standardized treatment protocols and other carve-out technologies may have led to some convergence in treatment prevalence rates, raising relatively low-use groups upward and bringing high-use groups downward. Also, in-patient care initially accounted for 80 percent of all MH/SA treatment costs in our parity group, providing space to increase outpatient use while reducing costs.

Information gaps. Our measures of access are limited to those generally available in administrative data. While the changes in treatment prevalence are encouraging, without population-based measures of mental health need we can say little about whether the right people are being treated. We also lack information about changes in consumers’ perceptions of access. The reductions in inpatient use are consistent with other studies’ findings.¹⁴ However, concerns remain that access to inpatient care may be too restrictive under managed care, with adverse consequences for those still needing intensive treatment.¹⁵ Without more clinically based measures of quality and especially outcomes, we do not know whether people are getting the right treatment.

Caveats. Impact on children. Parity and managed care had the largest impacts on children and adolescents. All of the cost decreases came from this group, in the form of reduced inpatient use. While children and adolescents had much higher initial in-patient use than adults had, we cannot conclude whether the declines resulted from the carve-out’s curbing of inappropriate use or from restricting needed services. Measures of quality and outcomes are underdeveloped for children.

Single employer group. We note several other caveats. Like most other case studies of parity, our natural experiment involves just one employer group, which limits the ability to generalize to other populations. In particular, the high initial in-patient costs among children and adolescents in our parity group may be unique to this particular region and time period. Universal managed care techniques such as inpatient utilization review and prior authorization may have already squeezed this type of high use out of most private plans, so that we would not expect to see the same levels of cost reductions in other populations. Or it may be related to a localized drug abuse epidemic and its psychiatric consequences.¹⁶ However, our study still provides additional strong evidence of minimal cost impacts of parity, as costs for adults remained unchanged, and provides an example of the impact of parity and management through a carve-out in a lower-use population.

Control-group adequacy. While we provide control-group comparison results, we have concerns about the adequacy of our particular control group. The control group differs from the parity group in a number of ways, including much lower initial utilization, less generous benefits, and much smaller employer group sizes, which makes it a less-than-ideal comparison group. The groups’ different cost experiences under the carve-out likely reflect these differences. The limitations in our control group and the lack of comparison groups in other case studies highlight the real-world difficulties with natural experiments. At the same time, our case study underlines the importance of using comparison groups, because without the comparisons it would be easier to conclude that parity led directly to increased treatment use. Future studies, such as the evaluation of the parity benefit in the Federal Employees Health Benefits Program (FEHBP), will be greatly strengthened by use of control groups.¹⁷ Better comparison groups will be critical in disentangling the impacts of parity, managed care, and secular trends.

Despite these caveats, it is clear that large cost increases did not materialize with parity. But many questions remain from this and other case studies around the issue of whether parity truly makes a difference for those with MH/SA needs. As employers and plans continue to carve out mental health benefits, and FFS and PPO plans continue to adopt managed care techniques, it is increasingly important to understand how mental health parity affects costs, access, and ultimately outcomes within already intensely managed settings. The impacts of parity and managed care on access, quality, and outcomes remain poorly understood because of the limitations in existing measures, especially for children, and inherent limitations in claims-based studies. Finally, existing case studies tend to exclude prescription drugs, yet parity and managed care provide strong incentives for substitution and cost shifting, so that the true costs of parity remain largely unknown. We hope to shed light on this in future work with prescription drug claims data we now have available for our natural experiment, as well as to examine changes in MH/SA treatment patterns, especially for children and adolescents.

Editor's Notes

 
Samuel Zuvekas is senior economist, Center for Cost and Financing Studies, Agency for Healthcare Research and Quality (AHRQ). Darrel Regier is executive director of the American Psychiatric Institute for Research and Education (APIRE) and director of the Division of Research, American Psychiatric Association. Donald Rae is a statistician at the API. Agnes Rupp is chief of the Mental Health Economics Research Program, National Institute of Mental Health. William Narrow is director of the API Psychopathology Research Program.

The authors are grateful to the health plans and large employer group for sharing data and for the guidance and expertise of their many staff members in preparing, understanding, and interpreting the data. They also thank Steve Hill, Haiden Huskamp, Tom McGuire, Alan Monheit, and two anonymous reviewers for their many insightful comments and suggestions. An earlier version of this paper was presented at the Tenth NIMH Biennial Research Conference on the Economics of Mental Health, September 2000. The views expressed in this paper are those of the authors, and no official endorsement by the Agency for Healthcare Research and Quality, the National Institute of Mental Health, or the Department of Health and Human Services is intended or should be inferred.

NOTES

Top Methods Data And Measures Study Results Discussion And Conclusion NOTES

 

1. See, for example, W. Goldman et al., "More Evidence for the Insurability of Managed Behavioral Health Care," Health Affairs (Sep/Oct 1999): 172–181; R. Sturm, "How Expensive Is Unlimited Mental Health Coverage under Managed Care?" Journal of the American Medical Association 278, no. 18 (1997): 1533–1537[Abstract]; C.T. Ma and T.G. McGuire, "Costs and Incentives in a Behavioral Health Carve-Out," Health Affairs (Mar/Apr 1998): 53–69; W. Goldman, J. McCulloch, and R. Sturm, "Costs and Use of Mental Health Services before and after Managed Care," Health Affairs (Mar/Apr 1998): 40–52; and R. Sturm, W. Goldman, and J. McCulloch, "Mental Health and Substance Abuse Parity : A Case Study of the Ohio State Employees Program," Journal of Mental Health Policy and Economics 1, no. 2 (1998): 129–134.[Medline]
2. E.B. Keeler, W.G. Manning, and K.J. Wells, "The Demand for Episodes of Mental Health Services," Journal of Health Economics 7, no. 4 (1988): 369–392.[Medline]
3. National Advisory Mental Health Council, Insurance Parity for Mental Health: Cost, Access, and Quality: Final Report to Congress by the National Advisory Mental Health Council, NIH Pub. no. 00-4787 (Bethesda, Md.: National Institutes of Health, June 2000).
4. See, for example, Ma and McGuire, "Costs and Incentives";; Sturm et al., "Mental Health and Substance Abuse Parity"; and Goldman et al., "More Evidence for the Insurability."
5. C.R. Gresenz, X. Liu, and R. Sturm, "Managed Behavioral Health Services for Children under Carve-Out Contracts," Psychiatric Services 49, no. 8 (1998): 1054–1058.[Abstract/Free Full Text]
6. Confidentiality restrictions prevent us from describing the parity law, employer group, and insurance carrier in greater detail.
7. Ma and McGuire, "Costs and Incentives."
8. Ibid.; and H.A. Huskamp, "Episodes of Mental Health and Substance Abuse Treatment under a Managed Behavioral Health Care Carve-Out," Inquiry 36, no. 2 (1999): 147–161.[Medline]
9. Costs for spouses declined by 8 percent for the full sample (not limited to the continuously enrolled) with higher costs in both the pre- and post-periods compared with the continuously enrolled sample of spouses. Cost declines in the full sample of employees (10 percent) and dependents (73 percent) were somewhat higher than in the continuously enrolled sample.
10. Ma and McGuire, "Costs and Incentives"; and Sturm et al., "Mental Health and Substance Abuse Parity."
11. K.D. Hennessey and H.H. Goldman, "Full Parity: Steps toward Treatment Equity for Mental and Addictive Disorders," Health Affairs (July/Aug 2001): 58–67.
12. S.H. Zuvekas, "Trends in Mental Health Services Use and Spending, 1987–1996," Health Affairs (Mar/Apr 2001): 214–224.
13. See, for example, Ma and McGuire, "Costs and Incentives";; Sturm et al., "Mental Health and Substance Abuse Parity"; and Goldman et al., "More Evidence for the Insurability."
14. R. Sturm, "Cost and Quality Trends under Managed Care: Is There a Learning Curve in Behavioral Health Carve-Out Plans?" Journal of Health Economics 19, no. 5 (1999): 593–604.
15. NAMHC, Parity in Financing Mental Health Services: Managed Care Effects on Cost, Access, and Quality—An Interim Report to Congress by the National Advisory Mental Health Council, NIH Pub. no. 98-4322 (Bethesda, Md.: NIH, 1988).
16. A. Rupp, E.M. Gause, and D.A. Regier, "Research Policy Implications of Cost of Illness Studies for Mental Disorders," British Journal of Psychiatry 173, supp. 36 (1998): 19–25.
17. Hennessy and Goldman, "Full Parity."

                      What's this?

This article has been cited by other articles:

				D. A. Regier, L. F. Bufka, T. Whitaker, F. F. Duffy, W. E. Narrow, D. S. Rae, G. M. Reed, O. F. Rehman, M. Rubio-Stipec, T. Weismiller, et al. Parity And The Use Of Out-Of-Network Mental Health Benefits In The FEHB Program Health Aff., January 1, 2008; 27(1): w70 - w83. [Abstract] [Full Text] [PDF]

				S. H. Zuvekas, A. E. Rupp, and G. S. Norquist The Impacts Of Mental Health Parity And Managed Care In One Large Employer Group: A Reexamination Health Aff., November 1, 2005; 24(6): 1668 - 1671. [Abstract] [Full Text] [PDF]