Behavioral Health Insurance Parity for Federal Employees
http://www.100md.com
《新英格兰医药杂志》
ABSTRACT
Background To improve insurance coverage of mental health and substance-abuse services, the Federal Employees Health Benefits (FEHB) Program offered mental health and substance-abuse benefits on a par with general medical benefits beginning in January 2001. The plans were encouraged to manage care.
Methods We compared seven FEHB plans from 1999 through 2002 with a matched set of health plans that did not have benefits on a par with mental health and substance-abuse benefits (parity of mental health and substance-abuse benefits). Using a difference-in-differences analysis, we compared the claims patterns of matched pairs of FEHB and control plans by examining the rate of use, total spending, and out-of-pocket spending among users of mental health and substance-abuse services.
Results The difference-in-differences analysis indicated that the observed increase in the rate of use of mental health and substance-abuse services after the implementation of the parity policy was due almost entirely to a general trend in increased use that was observed in comparison health plans as well as FEHB plans. The implementation of parity was associated with a statistically significant increase in use in one plan (+0.78 percent, P<0.05) a significant decrease in use in one plan (–0.96 percent, P<0.05), and no significant difference in use in the other five plans (range, –0.38 percent to +0.23 percent; P>0.05 for each comparison). For beneficiaries who used mental health and substance-abuse services, spending attributable to the implementation of parity decreased significantly for three plans (range, –$201.99 to –$68.97; P<0.05 for each comparison) and did not change significantly for four plans (range, –$42.13 to +$27.11; P>0.05 for each comparison). The implementation of parity was associated with significant reductions in out-of-pocket spending in five of seven plans.
Conclusions When coupled with management of care, implementation of parity in insurance benefits for behavioral health care can improve insurance protection without increasing total costs.
Parity in insurance coverage for mental health services has been the Holy Grail of mental health policy for decades.1,2 Ever since President John F. Kennedy directed the Civil Service Commission to offer federal employees mental health benefits on the same basis as benefits for other medical services, parity has been a standard for excellent mental health insurance coverage. The Federal Employees Health Benefits (FEHB) Program, like other competitive insurance markets, was unable to maintain equal coverage for mental health care because of high costs.3 Strictly limiting coverage for mental health and substance-abuse care is an effective means of controlling costs, but it also limits access and distorts the insurance market. Responding to such concerns, government payers and some private payers have intervened by setting minimum standards for coverage of mental health and substance-abuse care. Parity is intended primarily to correct insurance-market failure and the unfair design of insurance benefits. Advocates have hoped that it would increase access to care. The main argument against parity has been a concern that more generous coverage of these services would result in large increases in spending.3 Opposition to legislation requiring parity of coverage has been strong, but the successes of managed care in controlling spending on mental health and substance-abuse services offer a counterweight to cost considerations.4,5,6 The use of managed care, however, raises concerns about access and quality.
In June 1999, President Bill Clinton directed the Office of Personnel Management to ensure parity of mental health and substance-abuse benefits for the FEHB Program. He also proposed an evaluation to guide federal policy — the first national study of comprehensive parity. This article reports on the effects of parity of mental health and substance-abuse benefits on access, cost, and quality in seven large FEHB plans.
Methods
Examination of Parity of Mental Health and Substance-Abuse Care in the FEHB Program
The FEHB Program has 8.5 million enrollees; approximately 25 percent are current federal employees, 25 percent are retirees, and 50 percent are spouses or dependents of current or retired employees. Enrollees select from over 350 health insurance products.7 Beginning on January 1, 2001, the Office of Personnel Management required parity of coverage for mental health and substance-abuse care, defined as coverage that is "identical with regard to traditional medical care deductibles, coinsurance, co-pays, and day and visit limitations."8 Parity applied only to in-network insurance benefits.9 Providers and beneficiaries were informed about the policy change by direct mail from the plans.
The Office of Personnel Management encouraged the plans to employ managed-care techniques to control anticipated increases associated with expanded mental health and substance-abuse coverage. Before 2001, some plans had already contracted with managed behavioral health care organizations to control costs (in a process known as a "carve-out").9
We analyzed the results of this natural experiment by using a quasi-experimental design to account for secular trends in the use of and spending on mental health and substance-abuse care not associated with implementation of parity of coverage. (Previous evaluations of parity studied a single health plan before and after the implementation of parity and were unable to account for secular trends.) We compared spending in seven large FEHB plans during the period from 1999 through 2002 with spending in a matched set of health plans without parity of coverage or changes in mental health and substance-abuse coverage from Medstat's MarketScan database. Most of the comparison plans were operated by large, self-insured employers. We matched plans according to location and type of plan.
Initially, nine FEHB plans were selected for study on the basis of location, type of plan (health maintenance organization or point-of-service plan vs. preferred-provider organization ), population size, and interest in participation. Enrollees between the ages of 18 and 64 years were included in the study. Table 1 characterizes the nine FEHB plans and the comparison plans. It shows that the parity policy improved mental health and substance-abuse benefits for seven of the FEHB plans. Two HMOs, which were close to parity in 2000, did not show a substantial change in benefits. The analysis focuses on the seven PPO plans for which an effect of the implementation of parity could be expected. (The data on the effects on HMO plans are available at www.aspe.hhs.gov.)
Table 1. Characteristics of Nine FEHB Program Plans and Their MarketScan Comparison Plans before and after Implementation of Parity of Coverage for Mental Health and Substance-Abuse Services in 2001.
We studied the responses of persons who were continuously enrolled in a plan before and after the implementation of parity of coverage. Using data from all enrollees could confound the effects of parity with those of changes in plan composition. We examined plan benefits to assess the implementation of parity and then assessed the outcomes. The key outcomes examined were the rate of use of mental health and substance-abuse services, the total spending for such services among users, out-of-pocket spending on such services, and one measure of quality of care, the duration of follow-up for treatment of depression.
Data
From the seven plans, we obtained four years of data on the design of benefits, enrollment, and medical and pharmacy claims, including two years before and two years after the implementation of parity of coverage for FEHB plans. We analyzed data from a random sample of 20,000 enrollees per plan. We also obtained data on benefits, enrollment, and claims for the matched comparison group during the same period from the MarketScan database.
Identifying Mental Health and Substance-Abuse Services
We classified inpatient and outpatient services associated with specified mental health and substance-abuse diagnoses and psychotropic medications as mental health and substance-abuse services. (A detailed description is available at www.aspe.hhs.gov.) Mental health and substance-abuse diagnoses were defined as those with diagnostic codes 291, 292, 295 through 309 (except 305.1 and 305.8), and 311 through 314 in the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). An inpatient was considered a user of mental health and substance-abuse services if the last primary diagnosis and the majority of all primary diagnoses in the inpatient record were mental health and substance-abuse diagnoses. An outpatient was considered a user of mental health and substance-abuse services if any of the following was indicated: a mental health and substance-abuse primary diagnosis, a procedure specific to mental health and substance-abuse care, or a face-to-face encounter with a provider of such care or treatment at a facility specializing in mental health and substance-abuse care. To identify use of psychotropic medications, we developed two lists: a restricted list of medications that are used only for mental health and substance-abuse disorders and an expanded list of medications that are used for both mental health and substance-abuse disorders and other conditions. Expenditures for any medications on the restricted list counted as spending on mental health and substance-abuse care. If the patient made any other use of mental health and substance-abuse services or incurred any related expenditures during the year, then expenditures for any medications on the expanded list counted as spending for mental health and substance-abuse care.
To assess the quality of care for depression, we examined data from patients with a diagnosis of major depressive disorder (codes 296.2 and 296.3). Outpatients were included only if the diagnosis appeared on at least two service dates; inpatients were included if a primary diagnosis of major depressive disorder was the reason for hospitalization.
Statistical Analysis
We estimated the economic effect of parity by the difference-in-differences method. The difference in differences is the average difference (before and after the implementation of parity) in outcomes of interest in the comparison plans subtracted from the average difference before and after implementation of parity in the FEHB plans. This approach permitted us to account for any secular trend in outcomes. Any remaining significant differences in outcome are attributed to parity.
To estimate the difference in differences, we had to address two important characteristics of data on health care spending. Most people do not receive mental health and substance-abuse care in any given year (i.e., they have zero spending), and among those who do receive such care, a disproportionate number have high levels of spending. To account for these features, we examined a number of competing approaches that have been discussed in the literature.10 After testing competing models, we settled on the two-part model because it best fitted the data. We used the generalized linear model to estimate the relation between spending on mental health and substance-abuse care and parity. After checking several link functions and distributional assumptions, we used a normal model to characterize spending. Correlation among repeated annual observations was accounted for by the use of a generalized estimating equation approach.
The first part of the two-part model used logistic regression to estimate the effect of the implementation of parity of coverage on the probability that a person would use mental health and substance-abuse services. The unit of observation was the person-year. In those regressions, we adjusted for the demographic characteristics of the person (age and sex) and the person's relationship to the policyholder (child or spouse). The age variable was used to adjust for any time trend. The key variables of interest were an indicator variable that was assigned a value of one for the postparity period and zero for the preparity period, an indicator variable that was assigned a value of one for the members of FEHB plans and zero for the members of comparison groups, and the interaction of the two indicator variables. Because the logistic model is nonlinear, the net effect of the parity policy on an outcome could not be calculated directly from the coefficient of the interaction term.11 Instead, we calculated the average effect on the probability of using mental health and substance-abuse services by employing simulation methods based on the estimated regression model. Using the bootstrap samples, we constructed 95 percent confidence intervals for our final estimates.12
The second part of the two-part model used a least-squares regression approach to analyze individual spending on mental health and substance-abuse services for those who used any such services. In this model, we used the same independent variables as in the first part, as well as indicator variables for the diagnosis for which a service user received treatment. The coefficient of the interaction term allowed us to estimate any change in spending on mental health and substance-abuse care due to the parity policy, while accounting for the secular trend in such spending among users of mental health and substance-abuse services. A generalized estimating equation was used to estimate the standard errors of the model's coefficients.
We performed a before-and-after analysis of administrative data to assess any changes in the quality of care, as measured by the duration of follow-up treatment for acute-phase depression. Receiving services (any visits for mental health and substance-abuse care or prescriptions for antidepressant medications) for four months or more is considered a guideline for the quality of treatment of acute-phase depression.13,14,15 Episodes of depression care from six of the seven FEHB plans were studied before and after the implementation of parity to assess the proportion of patients with four months or more of follow-up treatment. The national PPO (Table 1) was not studied because the data were different from those in the other plans, and the differences limited comparability. Logistic regression was used to estimate the association between the postparity period and the quality measure. We constructed a 95 percent confidence interval for the adjusted odds ratios and used a generalized estimating equation approach to account for repeated observations.
Results
Table 1 shows that the comparison plans did not have parity benefits before 2001 and that they changed very little over the course of the study. Three of the seven FEHB plans did not use a carve-out vendor as a means of managing their mental health and substance-abuse benefits before implementing parity; only PPO 2 in the Mid-Atlantic region was not carved out by 2001. Roughly half the plans that served as comparison groups were carved out before the implementation of parity; only minor changes in carving out occurred in 2001.
Table 2 reports descriptive data on rates of use of mental health and substance-abuse services and spending for service users. For all plans, the rates of use and spending increased during the study period. Table 2 also reports difference-in-differences estimates for the probability of use of mental health and substance-abuse services and for spending on such services. After accounting for secular trends in the use of mental health and substance-abuse services, we found a positive and significant effect of parity on the probability of use for one plan, Mid-Atlantic PPO 2; the increase in the rate of use of mental health and substance-abuse services in this plan was 0.78 percentage point greater than the increase in its matched comparison plan. For the remaining six plans, the estimated effect of the implementation of parity on the probability of use either was positive and not significantly different from zero or was significant but negative. Thus, the difference-in-differences analysis indicated that the observed increase in the rate of use of mental health and substance-abuse services after the implementation of parity was almost entirely due to a secular trend in the increased use of such services. The difference-in-differences estimates for spending on mental health and substance-abuse services show significant decreases in spending attributable to parity for three plans, ranging from –$68.97 to –$201.99. The estimated effects on spending for the other four plans were moderate, ranging from –$42.13 to +$27.11, and did not differ significantly from zero. Thus, this analysis offers no evidence of significant increases in spending attributable to the implementation of parity of coverage.
Table 2. Probability of Use of Mental Health and Substance-Abuse Services and Total Spending by Service Users.
Table 3 summarizes the effects of parity on out-of-pocket spending by users of mental health and substance-abuse services. In five of seven plans, the parity policy was associated with significant reductions in out-of-pocket spending, ranging from $13.82 to $87.06. There was a small but significant increase in out-of-pocket spending by service users in the national PPO.
Table 3. Difference-in-Differences Results for Amount of Out-of-Pocket Spending per User for Mental Health and Substance-Abuse Services before, as Compared with after, the Implementation of Parity.
The findings on the quality of depression treatment are presented as odds ratios derived from regression analyses. The results for all plans showed an increased likelihood of providing follow-up services to persons undergoing treatment for acute-phase depression. The improvement was significant for three plans. The odds ratio was 1.72 (95 percent confidence interval, 1.22 to 2.41) for the Western PPO, 2.33 (95 percent confidence interval, 1.31 to 4.14) for Northeastern PPO 2, and 1.60 (95 percent confidence interval, 1.19 to 2.16) for the Southern PPO.
Discussion
Our results demonstrate that the growth in the use of mental health and substance-abuse services and spending on these services in seven FEHB plans was similar to or less than that in other large, privately insured populations. The only plan that had an increase in use attributable to the implementation of parity of coverage was Mid-Atlantic PPO 2, which was the only plan that did not contract with a carve-out managed-care vendor. These results are negative, in the sense that there were few significant differences in the probability of service use or in the amount of expenditures that could be explained by the implementation of parity. A finding of negative results always raises the question of whether the effect of parity on use and spending was really limited or whether the evaluation lacked the necessary power to detect an effect. Two factors lead us to believe that the effect really was limited: the estimated differences between the results for enrollees in the FEHB Program and comparison enrollees were relatively small in magnitude. The sample sizes used in the analysis were large and were sufficient to show significant effects of similar policy measures, such as the effect of carving out mental health and substance-abuse care while holding the benefit design constant.16 For these reasons, we believe the evidence points to a finding of little or no effect of the implementation of parity of coverage for mental health and substance-abuse services on use and total spending, rather than reflecting a type II error.
Although spending increases resulting from the implementation of parity did not occur, neither did access to mental health and substance-abuse services increase. Advocates of parity might be pleased about the observed increases in the use of mental health and substance-abuse services in all plans, but these changes were consistent with the presence of secular trends and not attributable to the implementation of parity.
The parity policy reduced out-of-pocket spending by users of mental health and substance-abuse services in all but one of the seven plans. Spending decreased significantly in five plans. There was a small but significant increase in out-of-pocket spending by enrollees in the national PPO, probably as a result of the relatively high cost-sharing requirements of the prescription-drug plan for this PPO.
Some policymakers have expressed concern that any increase in the use of managed care associated with the implementation of parity might result in deteriorating quality. Studies of quality using claims data to compare adherence to guidelines for the treatment of depression and substance abuse before and after the implementation of parity have found little change (see www.aspe.hhs.gov). The measure of quality we used — the duration of follow-up after treatment for depression — did not decline in this study, and in three of the plans, it showed a small improvement. Because we were not able to perform a difference-in-differences analysis of the quality measures, it is quite possible that our results reflect secular trends that are independent of the implementation of parity. The findings are quite similar to recent findings that used Health Plan Employer Data and Information Set (HEDIS) measures of follow-up in the treatment of depression.17 (To put these results in perspective, the measure we used improved at an annual rate of 2.4 percentage points, whereas the HEDIS measures of follow-up for acute- and continuation-phase depression improved by 2.9 and 2.7 percentage points, respectively, between 2001 and 2002.)
Our study had several limitations. First, as in all quasi-experimental studies, there is a risk of nonequivalence of intervention and comparison groups, particularly with respect to time trends over the length of the period before the implementation of parity. However, the consistency of key findings for multiple matched sets of health plans is reassuring. Second, mental health and substance-abuse diagnoses may be underreported in claims data, resulting in an undercount of use of such services and spending on these services. To address this issue, we employed multiple methods of identifying the use of mental health and substance-abuse services, including a procedure code specific to such services, indication of a face-to-face encounter with a mental health and substance-abuse provider, and indication of treatment at a facility specializing in mental health and substance-abuse care. Third, the use of data only from persons continuously enrolled before and after the implementation of parity eliminated those who changed plans, and the restriction of the analysis to only seven of the FEHB plans potentially limited the generalizability of the findings. However, the study included diverse plans (with different managed-care arrangements) with more than 3 million FEHB beneficiaries from across the United States, including more than 700,000 continuously enrolled adults, who were the focus of this analysis.
The goals of parity include providing equal coverage and increased financial protection for persons with mental health and substance-abuse disorders. The primary concern has been that the existence of parity would result in large increases in the use of mental health and substance-abuse services and spending on these services. With respect to the seven FEHB plans we studied, these fears were unfounded. In addition, the goal of expanding financial protection by decreasing out-of-pocket spending was realized in all but two of the plans. These findings suggest that parity of coverage of mental health and substance-abuse services, when coupled with management of care, is feasible and can accomplish its objectives of greater fairness and improved insurance protection without adverse consequences for health care costs.
Supported by a contract with the Department of Health and Human Services (HHS-100-00-0025) to Northrop Grumman for this study; a grant from the John D. and Catherine T. MacArthur Foundation for the Network on Mental Health Policy Research (to Dr. Goldman, Dr. Frank, Dr. Burnam, and Dr. Huskamp; principal investigator, Dr. Goldman); a grant from the National Institute of Mental Health (1 K01 MH66109, to Dr. Huskamp); a training grant from the National Institute of Mental Health (T32 MH19773 08, to Dr. Barry); and a grant from the UCLA-RAND National Institute of Mental Health Center for Research on Quality in Managed Care (P30 MH068639, to Dr. Burnam and Dr. Young; principal investigator, Dr. Kenneth Wells).
No potential conflict of interest relevant to this article was reported.
We are indebted to Albert Belanger, Rita Volya, and Katya Zelevinsky of the Harvard Medical School for programming and to Joshua Noda of Westat for assistance in the preparation of the article and tables.
Source Information
From the University of Maryland School of Medicine, Baltimore (H.H.G.); Harvard Medical School (R.G.F., H.A.H., S.-L.T.N., V.A., A.B.B.) and Harvard School of Public Health (S.-L.T.N.) — both in Boston; RAND, Santa Monica, Calif. (M.A.B.,M.S.R.); Department of Veterans Affairs, Los Angeles (A.S.Y.); UCLA School of Medicine, Los Angeles (A.S.Y.); Yale University School of Medicine, New Haven, Conn. (C.L.B.); McLean Hospital, Belmont, Mass. (A.B.B.); Westat, Rockville, Md. (S.T.A., G.M.); Northrop Grumman Information Technology, Federal Enterprise Solutions, Health Solutions, Rockville, Md. (C.L.); and CSR, Arlington, Va. (M.B.).
Address reprint requests to Dr. Goldman at the University of Maryland School of Medicine, 3700 Koppers St., Suite 402, Baltimore, MD 21227, or at hh.goldman@verizon.net.
References
Goldman HH, Sharfstein SS, Frank RG. Equity and parity in psychiatric care. Psychiatr Ann 1983;13:488-491.
Frank RG, Goldman HH, McGuire TG. Will parity in coverage result in better mental health care? N Engl J Med 2001;345:1701-1704.
Frank RG, McGuire TG. Economics and mental health. In: Culyer AJ, Newhouse JP, eds. Handbook of health economics. Amsterdam: Elsevier North-Holland, 2000.
Goldman W, McCulloch J, Sturm R. Costs and use of mental health services before and after managed care. Health Aff (Millwood) 1998;17:40-52.
Ma C, McGuire T. Costs and incentives in a behavioral health carve-out. Health Aff (Millwood) 1998;17:53-69.
Rosenbach M, Lake T, Young C, et al. Effects of the Vermont Mental Health and Substance Abuse Parity Law. Rockville, Md.: Center for Mental Health Services, Substance Abuse and Mental Health Services Administration, 2003. (DHHS publication no. (SMA) 03-3822.)
FEHB handbook, 2004. Washington, D.C.: Office of Personnel Management, 2004. (Accessed March 6, 2006, at http://www.opm.gov/insure/handbook/fehb01.asp#GENERAL%20OVERVIEW.)
Call letter for contract year 2001 — policy guidance, 2000. Washington, D.C.: Office of Personnel Management, 2000. (FEHB Program carrier letter no. 2000-17.)
Ridgely MS, Burnam MA, Barry CL, Goldman HH, Hennessy KD. Health plans respond to parity: managing behavioral health care in the federal employees health benefits program. Milbank Quarterly 2006;84:201-18.
Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ 2001;20:461-494.
Ai C, Norton EC. Interaction terms in Logit and Probit models. Econ Lett 2003;80:123-129.
Efron B, Tibshirani R. An introduction to the bootstrap. Boca Raton, Fla.: CRC Press, 1998.
American Psychiatric Association. Practice guideline for the treatment of patients with major depressive disorder (revision). Am J Psychiatry 2000;157:Suppl 4:S1-S45.
Agency for Health Care Policy and Research. Depression in primary care. Vol. 2. Treatment of major depression. Rockville, Md.: Department of Health and Human Services, 1993. (Publication no. 93-0551.)
Agency for Health Care Policy and Research. Treatment of depression: newer pharmacotherapies. Rockville, Md.: Department of Health and Human Services, 1999. (Publication no. 99-E014.)
Huskamp HA. Episodes of mental health and substance abuse treatment under a managed behavioral health care carve-out. Inquiry 1999;36:147-161.
The state of health care quality 2004. Washington, D.C.: National Committee on Quality Assurance, 2004 (table 1). (Accessed March 6, 2006, at http://www.ncqa.org/communications/SOMC/SOHC2004.pdf.)(Howard H. Goldman, M.D., )
Background To improve insurance coverage of mental health and substance-abuse services, the Federal Employees Health Benefits (FEHB) Program offered mental health and substance-abuse benefits on a par with general medical benefits beginning in January 2001. The plans were encouraged to manage care.
Methods We compared seven FEHB plans from 1999 through 2002 with a matched set of health plans that did not have benefits on a par with mental health and substance-abuse benefits (parity of mental health and substance-abuse benefits). Using a difference-in-differences analysis, we compared the claims patterns of matched pairs of FEHB and control plans by examining the rate of use, total spending, and out-of-pocket spending among users of mental health and substance-abuse services.
Results The difference-in-differences analysis indicated that the observed increase in the rate of use of mental health and substance-abuse services after the implementation of the parity policy was due almost entirely to a general trend in increased use that was observed in comparison health plans as well as FEHB plans. The implementation of parity was associated with a statistically significant increase in use in one plan (+0.78 percent, P<0.05) a significant decrease in use in one plan (–0.96 percent, P<0.05), and no significant difference in use in the other five plans (range, –0.38 percent to +0.23 percent; P>0.05 for each comparison). For beneficiaries who used mental health and substance-abuse services, spending attributable to the implementation of parity decreased significantly for three plans (range, –$201.99 to –$68.97; P<0.05 for each comparison) and did not change significantly for four plans (range, –$42.13 to +$27.11; P>0.05 for each comparison). The implementation of parity was associated with significant reductions in out-of-pocket spending in five of seven plans.
Conclusions When coupled with management of care, implementation of parity in insurance benefits for behavioral health care can improve insurance protection without increasing total costs.
Parity in insurance coverage for mental health services has been the Holy Grail of mental health policy for decades.1,2 Ever since President John F. Kennedy directed the Civil Service Commission to offer federal employees mental health benefits on the same basis as benefits for other medical services, parity has been a standard for excellent mental health insurance coverage. The Federal Employees Health Benefits (FEHB) Program, like other competitive insurance markets, was unable to maintain equal coverage for mental health care because of high costs.3 Strictly limiting coverage for mental health and substance-abuse care is an effective means of controlling costs, but it also limits access and distorts the insurance market. Responding to such concerns, government payers and some private payers have intervened by setting minimum standards for coverage of mental health and substance-abuse care. Parity is intended primarily to correct insurance-market failure and the unfair design of insurance benefits. Advocates have hoped that it would increase access to care. The main argument against parity has been a concern that more generous coverage of these services would result in large increases in spending.3 Opposition to legislation requiring parity of coverage has been strong, but the successes of managed care in controlling spending on mental health and substance-abuse services offer a counterweight to cost considerations.4,5,6 The use of managed care, however, raises concerns about access and quality.
In June 1999, President Bill Clinton directed the Office of Personnel Management to ensure parity of mental health and substance-abuse benefits for the FEHB Program. He also proposed an evaluation to guide federal policy — the first national study of comprehensive parity. This article reports on the effects of parity of mental health and substance-abuse benefits on access, cost, and quality in seven large FEHB plans.
Methods
Examination of Parity of Mental Health and Substance-Abuse Care in the FEHB Program
The FEHB Program has 8.5 million enrollees; approximately 25 percent are current federal employees, 25 percent are retirees, and 50 percent are spouses or dependents of current or retired employees. Enrollees select from over 350 health insurance products.7 Beginning on January 1, 2001, the Office of Personnel Management required parity of coverage for mental health and substance-abuse care, defined as coverage that is "identical with regard to traditional medical care deductibles, coinsurance, co-pays, and day and visit limitations."8 Parity applied only to in-network insurance benefits.9 Providers and beneficiaries were informed about the policy change by direct mail from the plans.
The Office of Personnel Management encouraged the plans to employ managed-care techniques to control anticipated increases associated with expanded mental health and substance-abuse coverage. Before 2001, some plans had already contracted with managed behavioral health care organizations to control costs (in a process known as a "carve-out").9
We analyzed the results of this natural experiment by using a quasi-experimental design to account for secular trends in the use of and spending on mental health and substance-abuse care not associated with implementation of parity of coverage. (Previous evaluations of parity studied a single health plan before and after the implementation of parity and were unable to account for secular trends.) We compared spending in seven large FEHB plans during the period from 1999 through 2002 with spending in a matched set of health plans without parity of coverage or changes in mental health and substance-abuse coverage from Medstat's MarketScan database. Most of the comparison plans were operated by large, self-insured employers. We matched plans according to location and type of plan.
Initially, nine FEHB plans were selected for study on the basis of location, type of plan (health maintenance organization or point-of-service plan vs. preferred-provider organization ), population size, and interest in participation. Enrollees between the ages of 18 and 64 years were included in the study. Table 1 characterizes the nine FEHB plans and the comparison plans. It shows that the parity policy improved mental health and substance-abuse benefits for seven of the FEHB plans. Two HMOs, which were close to parity in 2000, did not show a substantial change in benefits. The analysis focuses on the seven PPO plans for which an effect of the implementation of parity could be expected. (The data on the effects on HMO plans are available at www.aspe.hhs.gov.)
Table 1. Characteristics of Nine FEHB Program Plans and Their MarketScan Comparison Plans before and after Implementation of Parity of Coverage for Mental Health and Substance-Abuse Services in 2001.
We studied the responses of persons who were continuously enrolled in a plan before and after the implementation of parity of coverage. Using data from all enrollees could confound the effects of parity with those of changes in plan composition. We examined plan benefits to assess the implementation of parity and then assessed the outcomes. The key outcomes examined were the rate of use of mental health and substance-abuse services, the total spending for such services among users, out-of-pocket spending on such services, and one measure of quality of care, the duration of follow-up for treatment of depression.
Data
From the seven plans, we obtained four years of data on the design of benefits, enrollment, and medical and pharmacy claims, including two years before and two years after the implementation of parity of coverage for FEHB plans. We analyzed data from a random sample of 20,000 enrollees per plan. We also obtained data on benefits, enrollment, and claims for the matched comparison group during the same period from the MarketScan database.
Identifying Mental Health and Substance-Abuse Services
We classified inpatient and outpatient services associated with specified mental health and substance-abuse diagnoses and psychotropic medications as mental health and substance-abuse services. (A detailed description is available at www.aspe.hhs.gov.) Mental health and substance-abuse diagnoses were defined as those with diagnostic codes 291, 292, 295 through 309 (except 305.1 and 305.8), and 311 through 314 in the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). An inpatient was considered a user of mental health and substance-abuse services if the last primary diagnosis and the majority of all primary diagnoses in the inpatient record were mental health and substance-abuse diagnoses. An outpatient was considered a user of mental health and substance-abuse services if any of the following was indicated: a mental health and substance-abuse primary diagnosis, a procedure specific to mental health and substance-abuse care, or a face-to-face encounter with a provider of such care or treatment at a facility specializing in mental health and substance-abuse care. To identify use of psychotropic medications, we developed two lists: a restricted list of medications that are used only for mental health and substance-abuse disorders and an expanded list of medications that are used for both mental health and substance-abuse disorders and other conditions. Expenditures for any medications on the restricted list counted as spending on mental health and substance-abuse care. If the patient made any other use of mental health and substance-abuse services or incurred any related expenditures during the year, then expenditures for any medications on the expanded list counted as spending for mental health and substance-abuse care.
To assess the quality of care for depression, we examined data from patients with a diagnosis of major depressive disorder (codes 296.2 and 296.3). Outpatients were included only if the diagnosis appeared on at least two service dates; inpatients were included if a primary diagnosis of major depressive disorder was the reason for hospitalization.
Statistical Analysis
We estimated the economic effect of parity by the difference-in-differences method. The difference in differences is the average difference (before and after the implementation of parity) in outcomes of interest in the comparison plans subtracted from the average difference before and after implementation of parity in the FEHB plans. This approach permitted us to account for any secular trend in outcomes. Any remaining significant differences in outcome are attributed to parity.
To estimate the difference in differences, we had to address two important characteristics of data on health care spending. Most people do not receive mental health and substance-abuse care in any given year (i.e., they have zero spending), and among those who do receive such care, a disproportionate number have high levels of spending. To account for these features, we examined a number of competing approaches that have been discussed in the literature.10 After testing competing models, we settled on the two-part model because it best fitted the data. We used the generalized linear model to estimate the relation between spending on mental health and substance-abuse care and parity. After checking several link functions and distributional assumptions, we used a normal model to characterize spending. Correlation among repeated annual observations was accounted for by the use of a generalized estimating equation approach.
The first part of the two-part model used logistic regression to estimate the effect of the implementation of parity of coverage on the probability that a person would use mental health and substance-abuse services. The unit of observation was the person-year. In those regressions, we adjusted for the demographic characteristics of the person (age and sex) and the person's relationship to the policyholder (child or spouse). The age variable was used to adjust for any time trend. The key variables of interest were an indicator variable that was assigned a value of one for the postparity period and zero for the preparity period, an indicator variable that was assigned a value of one for the members of FEHB plans and zero for the members of comparison groups, and the interaction of the two indicator variables. Because the logistic model is nonlinear, the net effect of the parity policy on an outcome could not be calculated directly from the coefficient of the interaction term.11 Instead, we calculated the average effect on the probability of using mental health and substance-abuse services by employing simulation methods based on the estimated regression model. Using the bootstrap samples, we constructed 95 percent confidence intervals for our final estimates.12
The second part of the two-part model used a least-squares regression approach to analyze individual spending on mental health and substance-abuse services for those who used any such services. In this model, we used the same independent variables as in the first part, as well as indicator variables for the diagnosis for which a service user received treatment. The coefficient of the interaction term allowed us to estimate any change in spending on mental health and substance-abuse care due to the parity policy, while accounting for the secular trend in such spending among users of mental health and substance-abuse services. A generalized estimating equation was used to estimate the standard errors of the model's coefficients.
We performed a before-and-after analysis of administrative data to assess any changes in the quality of care, as measured by the duration of follow-up treatment for acute-phase depression. Receiving services (any visits for mental health and substance-abuse care or prescriptions for antidepressant medications) for four months or more is considered a guideline for the quality of treatment of acute-phase depression.13,14,15 Episodes of depression care from six of the seven FEHB plans were studied before and after the implementation of parity to assess the proportion of patients with four months or more of follow-up treatment. The national PPO (Table 1) was not studied because the data were different from those in the other plans, and the differences limited comparability. Logistic regression was used to estimate the association between the postparity period and the quality measure. We constructed a 95 percent confidence interval for the adjusted odds ratios and used a generalized estimating equation approach to account for repeated observations.
Results
Table 1 shows that the comparison plans did not have parity benefits before 2001 and that they changed very little over the course of the study. Three of the seven FEHB plans did not use a carve-out vendor as a means of managing their mental health and substance-abuse benefits before implementing parity; only PPO 2 in the Mid-Atlantic region was not carved out by 2001. Roughly half the plans that served as comparison groups were carved out before the implementation of parity; only minor changes in carving out occurred in 2001.
Table 2 reports descriptive data on rates of use of mental health and substance-abuse services and spending for service users. For all plans, the rates of use and spending increased during the study period. Table 2 also reports difference-in-differences estimates for the probability of use of mental health and substance-abuse services and for spending on such services. After accounting for secular trends in the use of mental health and substance-abuse services, we found a positive and significant effect of parity on the probability of use for one plan, Mid-Atlantic PPO 2; the increase in the rate of use of mental health and substance-abuse services in this plan was 0.78 percentage point greater than the increase in its matched comparison plan. For the remaining six plans, the estimated effect of the implementation of parity on the probability of use either was positive and not significantly different from zero or was significant but negative. Thus, the difference-in-differences analysis indicated that the observed increase in the rate of use of mental health and substance-abuse services after the implementation of parity was almost entirely due to a secular trend in the increased use of such services. The difference-in-differences estimates for spending on mental health and substance-abuse services show significant decreases in spending attributable to parity for three plans, ranging from –$68.97 to –$201.99. The estimated effects on spending for the other four plans were moderate, ranging from –$42.13 to +$27.11, and did not differ significantly from zero. Thus, this analysis offers no evidence of significant increases in spending attributable to the implementation of parity of coverage.
Table 2. Probability of Use of Mental Health and Substance-Abuse Services and Total Spending by Service Users.
Table 3 summarizes the effects of parity on out-of-pocket spending by users of mental health and substance-abuse services. In five of seven plans, the parity policy was associated with significant reductions in out-of-pocket spending, ranging from $13.82 to $87.06. There was a small but significant increase in out-of-pocket spending by service users in the national PPO.
Table 3. Difference-in-Differences Results for Amount of Out-of-Pocket Spending per User for Mental Health and Substance-Abuse Services before, as Compared with after, the Implementation of Parity.
The findings on the quality of depression treatment are presented as odds ratios derived from regression analyses. The results for all plans showed an increased likelihood of providing follow-up services to persons undergoing treatment for acute-phase depression. The improvement was significant for three plans. The odds ratio was 1.72 (95 percent confidence interval, 1.22 to 2.41) for the Western PPO, 2.33 (95 percent confidence interval, 1.31 to 4.14) for Northeastern PPO 2, and 1.60 (95 percent confidence interval, 1.19 to 2.16) for the Southern PPO.
Discussion
Our results demonstrate that the growth in the use of mental health and substance-abuse services and spending on these services in seven FEHB plans was similar to or less than that in other large, privately insured populations. The only plan that had an increase in use attributable to the implementation of parity of coverage was Mid-Atlantic PPO 2, which was the only plan that did not contract with a carve-out managed-care vendor. These results are negative, in the sense that there were few significant differences in the probability of service use or in the amount of expenditures that could be explained by the implementation of parity. A finding of negative results always raises the question of whether the effect of parity on use and spending was really limited or whether the evaluation lacked the necessary power to detect an effect. Two factors lead us to believe that the effect really was limited: the estimated differences between the results for enrollees in the FEHB Program and comparison enrollees were relatively small in magnitude. The sample sizes used in the analysis were large and were sufficient to show significant effects of similar policy measures, such as the effect of carving out mental health and substance-abuse care while holding the benefit design constant.16 For these reasons, we believe the evidence points to a finding of little or no effect of the implementation of parity of coverage for mental health and substance-abuse services on use and total spending, rather than reflecting a type II error.
Although spending increases resulting from the implementation of parity did not occur, neither did access to mental health and substance-abuse services increase. Advocates of parity might be pleased about the observed increases in the use of mental health and substance-abuse services in all plans, but these changes were consistent with the presence of secular trends and not attributable to the implementation of parity.
The parity policy reduced out-of-pocket spending by users of mental health and substance-abuse services in all but one of the seven plans. Spending decreased significantly in five plans. There was a small but significant increase in out-of-pocket spending by enrollees in the national PPO, probably as a result of the relatively high cost-sharing requirements of the prescription-drug plan for this PPO.
Some policymakers have expressed concern that any increase in the use of managed care associated with the implementation of parity might result in deteriorating quality. Studies of quality using claims data to compare adherence to guidelines for the treatment of depression and substance abuse before and after the implementation of parity have found little change (see www.aspe.hhs.gov). The measure of quality we used — the duration of follow-up after treatment for depression — did not decline in this study, and in three of the plans, it showed a small improvement. Because we were not able to perform a difference-in-differences analysis of the quality measures, it is quite possible that our results reflect secular trends that are independent of the implementation of parity. The findings are quite similar to recent findings that used Health Plan Employer Data and Information Set (HEDIS) measures of follow-up in the treatment of depression.17 (To put these results in perspective, the measure we used improved at an annual rate of 2.4 percentage points, whereas the HEDIS measures of follow-up for acute- and continuation-phase depression improved by 2.9 and 2.7 percentage points, respectively, between 2001 and 2002.)
Our study had several limitations. First, as in all quasi-experimental studies, there is a risk of nonequivalence of intervention and comparison groups, particularly with respect to time trends over the length of the period before the implementation of parity. However, the consistency of key findings for multiple matched sets of health plans is reassuring. Second, mental health and substance-abuse diagnoses may be underreported in claims data, resulting in an undercount of use of such services and spending on these services. To address this issue, we employed multiple methods of identifying the use of mental health and substance-abuse services, including a procedure code specific to such services, indication of a face-to-face encounter with a mental health and substance-abuse provider, and indication of treatment at a facility specializing in mental health and substance-abuse care. Third, the use of data only from persons continuously enrolled before and after the implementation of parity eliminated those who changed plans, and the restriction of the analysis to only seven of the FEHB plans potentially limited the generalizability of the findings. However, the study included diverse plans (with different managed-care arrangements) with more than 3 million FEHB beneficiaries from across the United States, including more than 700,000 continuously enrolled adults, who were the focus of this analysis.
The goals of parity include providing equal coverage and increased financial protection for persons with mental health and substance-abuse disorders. The primary concern has been that the existence of parity would result in large increases in the use of mental health and substance-abuse services and spending on these services. With respect to the seven FEHB plans we studied, these fears were unfounded. In addition, the goal of expanding financial protection by decreasing out-of-pocket spending was realized in all but two of the plans. These findings suggest that parity of coverage of mental health and substance-abuse services, when coupled with management of care, is feasible and can accomplish its objectives of greater fairness and improved insurance protection without adverse consequences for health care costs.
Supported by a contract with the Department of Health and Human Services (HHS-100-00-0025) to Northrop Grumman for this study; a grant from the John D. and Catherine T. MacArthur Foundation for the Network on Mental Health Policy Research (to Dr. Goldman, Dr. Frank, Dr. Burnam, and Dr. Huskamp; principal investigator, Dr. Goldman); a grant from the National Institute of Mental Health (1 K01 MH66109, to Dr. Huskamp); a training grant from the National Institute of Mental Health (T32 MH19773 08, to Dr. Barry); and a grant from the UCLA-RAND National Institute of Mental Health Center for Research on Quality in Managed Care (P30 MH068639, to Dr. Burnam and Dr. Young; principal investigator, Dr. Kenneth Wells).
No potential conflict of interest relevant to this article was reported.
We are indebted to Albert Belanger, Rita Volya, and Katya Zelevinsky of the Harvard Medical School for programming and to Joshua Noda of Westat for assistance in the preparation of the article and tables.
Source Information
From the University of Maryland School of Medicine, Baltimore (H.H.G.); Harvard Medical School (R.G.F., H.A.H., S.-L.T.N., V.A., A.B.B.) and Harvard School of Public Health (S.-L.T.N.) — both in Boston; RAND, Santa Monica, Calif. (M.A.B.,M.S.R.); Department of Veterans Affairs, Los Angeles (A.S.Y.); UCLA School of Medicine, Los Angeles (A.S.Y.); Yale University School of Medicine, New Haven, Conn. (C.L.B.); McLean Hospital, Belmont, Mass. (A.B.B.); Westat, Rockville, Md. (S.T.A., G.M.); Northrop Grumman Information Technology, Federal Enterprise Solutions, Health Solutions, Rockville, Md. (C.L.); and CSR, Arlington, Va. (M.B.).
Address reprint requests to Dr. Goldman at the University of Maryland School of Medicine, 3700 Koppers St., Suite 402, Baltimore, MD 21227, or at hh.goldman@verizon.net.
References
Goldman HH, Sharfstein SS, Frank RG. Equity and parity in psychiatric care. Psychiatr Ann 1983;13:488-491.
Frank RG, Goldman HH, McGuire TG. Will parity in coverage result in better mental health care? N Engl J Med 2001;345:1701-1704.
Frank RG, McGuire TG. Economics and mental health. In: Culyer AJ, Newhouse JP, eds. Handbook of health economics. Amsterdam: Elsevier North-Holland, 2000.
Goldman W, McCulloch J, Sturm R. Costs and use of mental health services before and after managed care. Health Aff (Millwood) 1998;17:40-52.
Ma C, McGuire T. Costs and incentives in a behavioral health carve-out. Health Aff (Millwood) 1998;17:53-69.
Rosenbach M, Lake T, Young C, et al. Effects of the Vermont Mental Health and Substance Abuse Parity Law. Rockville, Md.: Center for Mental Health Services, Substance Abuse and Mental Health Services Administration, 2003. (DHHS publication no. (SMA) 03-3822.)
FEHB handbook, 2004. Washington, D.C.: Office of Personnel Management, 2004. (Accessed March 6, 2006, at http://www.opm.gov/insure/handbook/fehb01.asp#GENERAL%20OVERVIEW.)
Call letter for contract year 2001 — policy guidance, 2000. Washington, D.C.: Office of Personnel Management, 2000. (FEHB Program carrier letter no. 2000-17.)
Ridgely MS, Burnam MA, Barry CL, Goldman HH, Hennessy KD. Health plans respond to parity: managing behavioral health care in the federal employees health benefits program. Milbank Quarterly 2006;84:201-18.
Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ 2001;20:461-494.
Ai C, Norton EC. Interaction terms in Logit and Probit models. Econ Lett 2003;80:123-129.
Efron B, Tibshirani R. An introduction to the bootstrap. Boca Raton, Fla.: CRC Press, 1998.
American Psychiatric Association. Practice guideline for the treatment of patients with major depressive disorder (revision). Am J Psychiatry 2000;157:Suppl 4:S1-S45.
Agency for Health Care Policy and Research. Depression in primary care. Vol. 2. Treatment of major depression. Rockville, Md.: Department of Health and Human Services, 1993. (Publication no. 93-0551.)
Agency for Health Care Policy and Research. Treatment of depression: newer pharmacotherapies. Rockville, Md.: Department of Health and Human Services, 1999. (Publication no. 99-E014.)
Huskamp HA. Episodes of mental health and substance abuse treatment under a managed behavioral health care carve-out. Inquiry 1999;36:147-161.
The state of health care quality 2004. Washington, D.C.: National Committee on Quality Assurance, 2004 (table 1). (Accessed March 6, 2006, at http://www.ncqa.org/communications/SOMC/SOHC2004.pdf.)(Howard H. Goldman, M.D., )