The Costs and Effects of Parity for Mental Health
and Substance Abuse Insurance Benefits

Chapter 4 - Cost Estimates

One main goal of this study was to project the costs of providing parity for mental health and substance abuse (MH/SA) insurance benefits. We did this by using an actuarial model to predict the premium increases for three benefit options. One option gives full parity, and the other two give partial parity, for MS/SA benefits. Information about the relative costs of parity options can be used by employers and benefit managers who want to assess the tradeoffs between offering full or partial parity for MH/SA benefits.

In this chapter, we present the three parity benefit options we developed and discuss the updated actuarial model we used to determine our cost estimates of premium increases for families. We also determine increases in premiums for child health plans with parity that states may develop as a result of the Balanced Budget Act of 1997. Finally, we offer some caveats for interpreting the results.

Options for Providing Parity in MH/SA Insurance Benefits

The three benefit options we analyzed in this study were developed in consultation with this project's policy advisory panel. The panel included representatives from the business community and from mental health, substance abuse, and managed care organizations in the public and private sectors (see Appendix C).

We estimated premium increases for one full and two partial MH/SA parity benefit options. In this study, "full parity" means that insurance benefits for any group of MH/SA diagnoses must be the same as insurance benefits for medical/surgical diagnoses with respect to three areas--cost sharing (such as copayment or coinsurance amounts), service limits (such as the number of outpatient visits or inpatient hospital days), and annual or lifetime spending limits (such as annual or lifetime benefit maximums).

Partial parity means that benefits for MH/SA diagnoses must be the same as benefits for medical/surgical diagnoses in two of the three areas listed above. One of the partial parity options requires parity with respect to cost sharing and spending limits. The other requires parity with respect to service limits and spending limits. Both partial parity options comply with the Mental Health Parity Act of 1996, since they require parity with respect to spending limits.

The costs of each full and partial parity option are predicted for the following diagnosis groups and plans:

• three diagnosis groups for families (all MH/SA diagnoses, MH diagnoses only, SA diagnoses only); and all MH/SA diagnoses for children only; and

• four plan types (FFS, PPO, POS, and HMO).

The baseline (initial), partial parity, and full parity MH/SA benefit options for a FFS plan are listed in Table 4.1. Full parity plans are listed in Table 4.2. For each plan type, the full parity benefit options have unlimited MH/SA inpatient days and outpatient visits. The cost- sharing requirements are the same as the cost-sharing requirements for medical/surgical benefits for the typical plan. For FFS plans, the cost-sharing requirement for covered services under the full parity option includes a 20 percent coinsurance payment for inpatient and outpatient services.

The typical plan was determined based on a review of the literature (O'Grady, 1996; KPMG Peat Marwick, 1997). In addition, our expert of actuaries and economists reviewed our assumptions about the benefit packages.

TABLE 4.1
PARTIAL AND FULL PARITY BENEFIT OPTIONS FOR A FEE-FOR-SERVICE PLAN

MH/SA Service	Baseline Benefits (Typical FFS Benefits)	Partial Parity		Full Parity
		Parity in Service Limits	Parity in Cost Sharing
Inpatient Hospital	30 days 20% coinsurance	Unlimited days Days 1-30: 20% coinsurance More than 30 days: 50% coinsurance	30 days 20% coinsurance	Unlimited days 20% coinsurance
Outpatient	20 visits 50% coinsurance	Unlimited visits 50% coinsurance	20 visits 20% coinsurance	Unlimited visits 20% coinsurance

TABLE 4.2
FULL PARITY BENEFIT OPTIONS

MH/SA Service	Fee-For-Service	PPO		POS		HMO
		In-Network	Out-of-Network	In-Network	Out-of-Network
Inpatient Hospital	Unlimited days 20% coinsurance	Unlimited days 10% coinsurance	Unlimited days 30% coinsurance	Unlimited days fully covered	Unlimited days 20% coinsurance	Unlimited days covered in full
Outpatient Services	Unlimited visits 20% coinsurance	Unlimited visits 10% coinsurance	Unlimited visits 30% coinsurance	Unlimited visits $10 copayment	Unlimited visits 20% coinsurance	Unlimited visits $10 copayment

Method for Estimating Costs

We used actuarial cost models developed by the HayGroup to estimate the costs of the full and partial parity benefit options. The model in our study improves on models in previous actuarial studies in several ways. First, it was recently revised by the HayGroup. This revised version includes expense data from managed behavioral health care companies, separate expense data for substance abuse services, and separate data on MH/SA expenses for children. Earlier versions of the HayGroup model did not include these features.

Second, the assumptions we used to estimate the premium increases, as defined in chapter 3, were reviewed by this project's expert panel of actuaries and economists. Many of our initial assumptions were revised according to their comments. Third, our assumptions incorporate new data and case study findings that were not available in 1996. Finally, we produced separate estimates for a wider range of benefit options, diagnosis groups, and health plan types than previous studies. The updated actuarial model and assumptions are described more fully in Sing and Hill (1998a).

The Estimation Process

The estimation process is as follows. The models predict premiums for health plans by using data on the benefit packages of the full and partial parity benefit options. These data include information on covered services, service limits, and cost-sharing arrangements. The model then builds in assumptions about administrative costs, the level of utilization management, and patients' responses to changes in their out-of-pocket costs.

The cost of each parity benefit option is the difference between the predicted premium for that option and the estimated premium for a "baseline plan." In this study, a baseline plan is a typical health plan covering medical/surgical and MH/SA services. For each plan type, it is the benefit package that has the most enrollees (the statistical mode). For example, a typical FFS plan covers 30 days of inpatient care and 20 outpatient visits for MH/SA services (O'Grady, 1996). In comparison, a full parity FFS plan would cover unlimited inpatient days and outpatient visits for MH/SA services.

Appendix B illustrates how to compute the premium increase due to a benefit option with increased MH/SA insurance benefits. The new premium is the sum of the baseline MH/SA and medical/surgical expenditures plus the increase in MH/SA expenditures. The percentage premium increase is the difference between the new premium and the baseline premium, divided by the baseline premium. A large percentage change in MH/SA expenditures causes a small change in premiums, because MH/SA expenditures account for only a small part of premiums (4 percent to 6 percent of the expense data in the HayGroup model, depending on the type of health care plan).

The HayGroup Actuarial Model

The HayGroup actuarial model has been used extensively to study the effects of proposed policies for the federal government. Earlier versions of this model were developed under contract with the Congressional Research Service (CRS). These versions were used to predict the costs of the Mental Health Parity Act of 1996 and the Domenici-Wellstone amendment to the Health Insurance Reform Act of 1996. The model was recently updated in consultation with the National Institute of Mental Health. The updated model includes data from managed behavioral health plans, separate expense data for children, and substance abuse data (Hay/Huggins Company, Inc., 1997b).

Sturm (1997) recently criticized an older version of the HayGroup model. He believes that the model overstates the costs of parity because it does not adequately account for managed care in the delivery of MH/SA services. This criticism is no longer relevant since the revised model includes data and assumptions for managed behavioral health plans.

Assumptions about PPO and POS plan network use and provider discounts. The HayGroup actuarial model is adjusted to incorporate three features of PPO and POS plans. These features are (1) network provider discounts, (2) coverage for in-network and out-of-network services, and (3) the effects of utilization management by POS gatekeepers.

Enrollees in PPO and POS plans pay lower out-of-pocket costs when they use network providers. These lower costs encourage enrollees to use these providers. Enrollees in PPOs can self-refer to any provider they wish to see. However, many enrollees in POS plans are assigned to a primary care network provider called a "gatekeeper." The gatekeeper must authorize all in-network service use (Jensen, et al., 1997). Providers in PPO and POS networks agree to charge a discounted price for the services they provide to PPO and POS plan enrollees.

For PPO and POS plans, the HayGroup model assumes that 70 percent of care is given by network providers. The model also assumes that the plan receives a 15 percent discount from network providers. For POS plans only, the model assumes that the use of in-network services is further reduced by 12 percent due to services denied by gatekeepers. The model assumes that POS out-of-network service use increases by 15 percent. This is based on the assumption that some POS plan enrollees will seek treatment out of network (and pay a higher coinsurance rate) when the gatekeeper denies in-network care.

Types of MH/SA treatment. The model does not separately compute expenditures for psychotherapeutic drugs, intensive nonresidential care, and SMI. This is because usable expense data for these services are not available. However, expenses for intensive nonresidential care services and partial hospitalization are included in the model's inpatient and outpatient expense data.

Estimated Premium Increases for Families

The model predicts that full parity for all MH/SA diagnoses will raise family premiums for a composite of plans by 3.6 percent (Table 4.3). By "composite" we mean a weighted average of fee-for-service (FFS), preferred provider organization (PPO), point-of-service (POS), and health maintenance organization (HMO) plans. Mental health care accounts for most of this increase (3.4 percent).

Although MH/SA expenditures would increase by 75 percent, the premium increase is 3.6 percent because MH/SA expenditures are only 4 percent to 6 percent of health expenditures at baseline, depending on the type of plan. Premium increases are the largest for FFS plans and PPOs (5.0 and 5.1 percent).⁵ Premium increases are lower for tightly managed HMOs (0.6 percent) (Table 4.4).

⁵ Total premiums rise slightly more in PPO plans than in FFS plans because MH/SA expenditures are a larger proportion of the PPO premium (4.3 percent) than of the FFS premium (3.9 percent).

Our tables do not report separate estimates for the parity options for serious mental illnesses (SMI), since the model cannot compute premium increases for changes in these benefits. However, a very "rough" estimate for SMI parity options can be obtained by pro-rating the predicted cost increases for the mental health parity options.

One way to do this is to use findings from two studies conducted by Milliman and Robertson, Inc. One study (Melek and Pyenson, 1996b) estimated that parity for SMI, as defined in S.298 (which did not pass), would increase premiums by 2.5 percent. The other study (Melek and Pyenson, 1996a) estimated that parity in benefits for all mental health diagnoses would increase premiums by 2.8 percent.

These studies suggest that expenses for SMI represent 89 percent of the increase in expenditures for all mental health diagnoses due to parity, since 2.5 percent is 89 percent of 2.8 percent. Therefore, to get a rough estimate of the increase in premiums due to parity for SMI, we can assume that the premium increase for SMI parity is 89 percent of the premium increase for mental health parity.

TABLE 4.3
AVERAGE PREMIUM INCREASES AGGREGATED ACROSS PLAN TYPES

MH/SA Diagnoses	Average Premium Increase
	Parity in Cost Sharing	Parity in Service Limits	Full Parity
MH/SA	0.4	1.2	3.6
MH only	0.3	1.1	3.4
SA only	0.1	0.03	0.2

NOTES:

1. The premium increases for FFS, PPO, POS, and HMO plans were aggregated by assuming the following distribution of enrollees among plan types:

   FFS	20%
   PPO	30%
   POS	20%
   HMO	30%

2. The table indicates increases in family premiums. Family premiums are computed from the individual adult and child premiums using the following formula:

   Family premium = (1 + .84 * 1.08)*adult premium + 1.22* child premium

   This formula assumes that 84 percent of employees are married, the cost of coverage for the spouse is 1.08 times more than the cost for the employee, and that there are 1.22 children per family on average. The demographic assumptions are based on data from the Current Population Survey. The cost of coverage for a spouse relative to the employee is based on data for a typical plan (Hay/Huggins Company, Inc., 1997a).

3. Premium increases do not necessarily add up within columns due to rounding.

TABLE 4.4
PERCENTAGE INCREASES IN TOTAL PREMIUMS BY DIAGNOSIS AND PLAN TYPE

Diagnosis	Percentage Increase in MH/SA Expenses			Percentage Increase in Total Family Premium
	Parity in Cost Sharing	Parity in Service Limits	Full Parity	Parity in Cost Sharing	Parity in Service Limits	Full Parity
FFS
MH/SA MH only SA only	15.4 13.5 1.9	41.7 40.5 1.2	126.8 119.8 7.0	0.5 0.4 0.1	1.4 1.3 0.05	5.0 4.8 0.3
PPO
MH/SA MH only SA only	17.0 14.8 2.2	40.0 38.7 1.3	117.7 111.4 6.3	0.6 0.5 0.1	1.5 1.4 0.1	5.1 4.8 0.3
POS
MH/SA MH only SA only	0.2 0.2 0.00	33.9 33.4 0.4	64.6 63.1 1.6	0.00 0.00 0.00	1.7 1.7 0.02	3.5 3.4 0.1
HMO
MH/SA MH only SA only	6.1 5.5 0.6	3.7 3.7 0.00	11.6 10.9 0.7	0.3 0.3 0.03	0.2 0.2 0.00	0.6 0.6 0.04

NOTE: Premium increases do not necessarily add up within columns due to rounding.

Parity for substance abuse benefits. Full parity for substance abuse would increase expenditures on substance abuse care by about 26 percent (not shown), but the total premium for the composite plan would increase by 0.2 percent. This premium increase is low relative to the premium increase for all mental health diagnoses for two reasons.

First, and most importantly, few people would use expanded benefits. Consumers of long-term substance abuse treatment are rare in the privately insured population, because few are employed. For employed people who do receive substance abuse treatment, few require long-term care. Wesson (1995) reports that most detoxification patients do not need inpatient or residential treatment. Furthermore, inpatient treatments are short-term, so that 30 days per year of inpatient substance abuse benefits are reasonable.

Second, our projections use a baseline plan with more generous substance abuse benefits than the typical plan. We assume that the lifetime benefit maximum for both mental health and substance abuse treatment is $50,000 for the typical plan before parity. This implies that the lifetime benefit maximum for substance abuse services only for the typical plan is less than $50,000 (by the amount of the mental health benefits used).

Under the Mental Health Parity Act of 1996, the lifetime benefit maximum for mental health treatment is $1 million, but no benefit maximum is specified for substance abuse services.

For purposes of this study, we assume that there is a $50,000 lifetime benefit maximum for substance abuse services. Therefore, the benefit maximum for substance abuse treatment under the act is greater than the benefit maximum for substance abuse treatment before the act.

Partial parity options. The premium increases predicted by the model for the partial parity options as defined in this study are much lower. For all MH/SA diagnoses, the composite premium increases are 0.4 percent or less if there is parity in cost sharing. If there is parity in service limits, the composite premium increases are 1.2 percent or less.

The partial parity premium increases are higher when there is parity for service limits compared with parity for cost sharing because there is a greater increase in benefits when there is parity for service limits. Specifically, for most plan types, parity for service limits increases the number of covered inpatient hospital days from 30 days to 365 days and increases the number of covered outpatient visits from 20 visits to an unlimited number of visits.

In contrast, when there is parity for cost sharing, there is generally no change (or a relatively small change) in the out-of-pocket expenses for inpatient care because the typical health plan already offers parity with respect to inpatient cost sharing. For outpatient visits, there is a decrease in the FFS coinsurance rate of 50 percent to 20 percent, and a decrease in the HMO copayment amount from $20 to $10.

Estimated Premium Increases for Child Health Plans

Under the Balanced Budget Act of 1997, states will receive block grants to fund health insurance for uninsured, low-income children (Mann and Guyer, 1997). States may either expand Medicaid eligibility and provide full Medicaid benefits for uninsured children, or they may establish or expand a separate state program for children.

Estimating the costs of parity in this program is difficult for two reasons. First, we do not know what the baseline benefit packages are. States that do not expand Medicaid may use any of three standard benefits packages or a package that is actuarially equivalent to one of these. Second, we do not know the prevalence of MH/SA disorders among the covered children. States have great latitude in choosing which children are covered. Within the scope of this project, we could not fully address the costs of parity for these programs for uninsured children, but we make predictions that are likely to be conservative estimates of the true costs of parity for separate state programs for children. Our estimates are not relevant for Medicaid expansions.

We produced actuarial estimates of the costs of full and partial parity for all MH/SA diagnoses for currently insured children of employees of medium and large employers (Table 4.3). The premiums are estimates of the average costs of coverage for one child. Unlike employer-sponsored insurance, the premiums do not include any costs for covering adults.

We estimate that full parity for child health plans would increase MH/SA expenditures for currently insured children with FFS coverage by 158.9 percent and the total premiums by 7.0 percent. In contrast, full parity for children's MH/SA benefits in a tightly managed HMO would increase premiums by 0.8 percent. We do not present a composite premium, because we do not know what types of delivery systems states might use for their block grant programs.

The estimated increases in MH/SA expenses and premiums for children are greater than the estimates for families shown in Table 4.1 because MH/SA expenses account for a greater portion of children's premiums. Data on children's MH/SA expenses indicate that children use such services at a much lower rate than adults, but children who use MH/SA services have, on average, higher expenditures than adults (Sturm, 1997; Grazier and G'Sell Associates, 1997).

The estimates in Table 4.5, which are based on data for children who are currently privately insured, are likely to be lower than the actual costs of parity for uninsured children. If states provide insurance coverage to currently uninsured children, these children are likely to use more MH/SA services than those who are currently insured. For example, Frank, et al. (1994) estimate that, if the uninsured population (adults and children) were given insurance coverage, the number using MH/SA treatment services would be 0 percent to 5 percent higher than the currently insured population.

However, the costs of covering uninsured children are uncertain, especially because states may choose to cover only some uninsured children. In addition, the state may establish a program with more or less generous medical/surgical benefits than those typical of medium and large employers, depending on how the state applies the benefit standards law.

TABLE 4.5
INCREASES IN CHILDREN'S MH/SA EXPENDITURES AND PREMIUMS FOR FULL PARITY BY PLAN TYPE

Plan Type	Percentage Increase in MH/SA Expenses			Percentage Increase in Total Child’s Premium
	Parity in Cost Sharing	Parity in Service Limits	Full Parity	Parity in Cost Sharing	Parity in Service Limits	Full Parity
FFS	14.4	56.3	158.9	0.5	2.2	7.0
PPO	15.6	53.2	143.7	0.7	2.3	7.0
POS	0.3	38.0	81.3	0.00	2.2	4.9
HMO	6.0	3.5	11.2	0.4	0.2	0.8

NOTE: Estimates based on privately insured children, using typical benefit packages of medium and large employers. The premiums are estimates of the average costs of coverage for one child. Estimates do not adjust for likely greater service use among uninsured children or differences between private insurance benefits and those of health block grant programs for children.

Interpreting the Estimates

Readers should keep several features of the model and our assumptions in mind when interpreting the predicted premium increases. First, this model (and other actuarial models) does not account for employer responses to parity mandates. Employers could respond to an anticipated premium increase due to a parity mandate by increasing employee contributions, dropping health insurance coverage, dropping or reducing coverage for MH/SA services, reducing other benefits, or increasing management of MH/SA services. These responses would lead to a lower premium increase than that estimated by this model. Therefore, our estimates indicate only the initial premium increase due to parity.

Second, these estimates are made with a baseline benefit package that is more generous than those used in previous actuarial estimates. Our baseline benefit package for each plan has a $1 million lifetime spending limit for mental health services, which reflects the Mental Health Parity Act of 1996. However, the baseline packages for previous actuarial estimates have a much lower lifetime spending limit for MH/SA services (such as $50,000). If our baseline packages had this lower limit, our projected premium increase for a composite of plans would be 4.2 percent instead of 3.6 percent.

Third, these estimates are based on the characteristics of a typical health plan for each plan type. But among each plan type there is great diversity in benefit and management levels.

Fourth, in this study we estimated premium increases for family coverage. Many previous studies estimated premium increases for single adults. This model predicts lower premium increases for single adults than for families. According to the model, full parity in MH/SA benefits for single adults would raise premiums for the composite plan by 3.1 percent (as opposed to 3.6 percent for families). For FFS plans, the model estimates a 4.3 percent premium increase for single adults, compared to 5.0 percent for families.

The adult-only premium increase for expanded MH/SA benefits is lower than the estimate for family coverage because the relative cost of MH/SA coverage for children in this model is higher than the relative cost of non-MH/SA benefit coverage. For non-MH/SA benefits, the revised HayGroup model assumes that the relative cost of children to adults is 58 percent (i.e., for every $100 in adult costs, children cost $58). For MH/SA benefits, the relative costs of children to adults is about 68 percent.

Table of Contents | Previous | Next