Tap 15 — TAPs <<<Documents<<<Home

Chapter 6 of TAP 15: Forecasting the Cost of Chemical Dependency Treatment Under Managed Care: The Washington State Study

Chapter 6—Estimating the Effects of Managing Care

Managed care achieves savings by reducing utilization, duration of treatment, and costs. All three variables in the actuarial calculation could change when managed care is introduced into a plan. Different organizations and philosophies of managing care will achieve different levels of savings, so changing managed-care organizations or concepts will affect the actuarial cost. A managed-care organization will also achieve different levels of savings for different populations, so experience with one covered population does not necessarily transfer directly to another.

Essential to an accurate estimate of the effect of managing care on services and insureds not previously managed is to find data from a program that is as similar as possible in managed-care style, covered population, and benefit package to the plan under study. For policy reasons, Washington State wished to use the American Society of Addiction Medicine criteria as the basis for managing care. Unfortunately, there are few large data bases that reflect the society's criteria, and all of them are outside Washington.

Washington currently uses less inpatient hospital care (the most expensive modality) without managed care than managed-care firms have achieved nationally, primarily because nonhospital treatment is more widely available in the State than it is nationally. The State and its actuary decided against using data from national managed-care organizations to estimate utilization distribution. Although data bases restricted to instate insured lives are much smaller and therefore less reliable, the State decided that they would be superior to the national data in their ability to reflect the distribution of modalities likely under the State's Health Care Reform Act.

Washington and its actuary concluded that overall admission rates for chemical dependency treatment would not change when managed care is introduced, but the distribution of these admissions between modalities or services would shift somewhat to favor less expensive care. This conclusion allowed the actuary to modify the utilization calculation, using a two-step model. First, a utilization rate was determined for all chemical dependency services taken together. Since managed care was not expected to alter the overall utilization rate, the actuary could use data from both managed-care and non-managed-care plans for this step, reducing reliance on the small data bases. Second, a distribution of admissions among the various services was determined for the subgroups for which the State had adequate data. The data for this step were from two managed-care organizations whose styles were close to the policy ideal. Multiplying the first factor by the second created utilization rates for each service, using overall utilization predictions from large data bases and deriving the effects of managed-care on modality utilization distribution from appropriate managed-care plans. The actuary used similar techniques to estimate duration of service.

For the uninsured and medicaid populations, there were no local or national data on the effects of managed care. The severity of dependence among poorer populations might be greater due to previous lack of treatment, which would result in more frequent utilization of residential modalities for the uninsured and medicaid populations, and longer durations. On the other hand, younger populations have less time to develop severe disease stages, so medicaid and uninsured groups could include fewer severely dependent persons. Given that the use of some form of managed care is widespread for insured patients and virtually absent for uninsured and medicaid patients, it is impossible to verify either conjecture. There is more use of residential treatment and there are longer stays among medicaid and uninsured patients than among insured patients (Table 6–A), but these differences could be due to managed care or to greater severity. Lacking better data about severity among uninsured and medicaid subgroups, the State assumed the same distribution of utilization among modalities for all groups; that is, it assumed that managed care would affect all groups equally. The only differences in utilization were due to differences in prevalence.

Cost Shifting

Effects of managed care on cost-of-care data are complicated by cost shifting. Managed-care firms achieve part of their cost savings by forcing service providers to accept lower payments, sometimes even below the average cost of care. Providers may accept these arrangements because they can fill otherwise empty beds or slots, enabling them to spread fixed overhead over a larger base and thus reduce their average cost. Even if the low payments are insufficient to cover the variable, marginal costs, providers may still accept the arrangement. They can compensate for the below-cost payments by raising charges to plans or individuals who are able and willing to pay more than their share. This amounts to an in-direct subsidy of managed-care patients by non-managed-care patients. The public sector also pays less than provider cost, taking advantage of the fact that the provider can raise fees to non-publicly supported patients.

As more and more plans switch to managed care and seek to have costs shifted elsewhere, there are fewer and fewer nonmanaged plans and individuals to whom costs can be shifted. Unless providers can find previously undiscovered efficiencies, they eventually must either refuse to accept patients in the plans or go bankrupt. If the plans cover enough individuals, there are virtually no patients outside managed care who are paying the shifted costs. At this point, cost shifting ends and the actuarial cost rises.

The Washington State study was part of a health care reform effort that was aimed at universal coverage. Under the State's plan, all patients statewide would be under managed care. Once the plan was fully implemented, no cost shifting would be possible. Washington therefore needed to calculate net-cost-per-person-per-month (PMPM) estimates that had no cost shifting while using data from environments where cost shifting is rampant.

Phase-In

In Washington's case, the estimate was further complicated by the fact that coverage was to be phased in over 4 years, so the ratio of various groups would change from year to year. This meant that some cost shifting would still occur during phase-in and that the amount of cost shifting would vary, depending on which subpopulations were added each year. Cost shifting would reach zero only when all subgroups were included in the plan.

The State's actuary came up with a methodology for estimating the changes in cost per unit that would result from the additions of various populations to the plan. The actuary first assumed that the chemical dependency treatment system is currently efficient (that is, that any cuts in payments would have proportionate effects on quality or quantity of treatment) and that total current provider profits are reasonable. These assumptions meant that the average current payment should not change as the plan is implemented, although payments for individuals might increase or decrease as they are added to the plan and cost-shifting factors change. Thus, the absorption into the plan of a group that had previously borne the burden of cost sharing would result in a decrease in the group's payment and an increase in the payments for everyone else, but the net revenue to the providers would be the same.

The actuarial cost of the plan thus becomes a weighted average of the actuarial costs for all the subgroups in the covered population. The weighting has to take into account the size of each subgroup and its utilization and duration of stay. Washington's actuary achieved this by estimating a PMPM for each subgroup separately, at the subgroup's current average cost; this step weighted properly for utilization and duration. The actuary then averaged PMPM's, weighting them by group size (this weighted average PMPM is called a community rate). Since the groups were to be phased in over 4 years, the actuary used different population sizes for each phase-in year. The result was a PMPM estimate (before inflation) that increased by 1 percent from the first to the second year, decreased 1 percent for the third year, stayed flat for the fourth year, and then decreased 2 percent for the fifth year. Table 4–B displays the community rate for each year of implementation, after the effects of 5-percent annual inflation are included.

Copayments and Deductibles

Actuarial costs are affected by patient participation requirements, such as copayments and deductibles. Copayments (or simply "copays") are fees paid by patients for each service they receive under a plan. Deductibles are minimum payments that patients must make, above which the plan makes all payments. Usually, the deductible is renewed annually; the patient starts each year at zero and pays for services until he or she reaches the deductible limit, at which point the plan kicks in.

Copays and deductibles reduce the amount that a plan pays for services that it covers. The effect is computed in a straightforward fashion: copays are applied to the average cost per unit, and deductibles are applied to the total annual cost. To return to our actuarial equation, copays are incorporated as follows:

To apply a deductible to a single service in the plan, the equation is modified as follows:

Most plans apply deductibles to all services simultaneously, so payments made toward one service apply to the deductible for the whole. The actuarial effect of deductibles in such cases is computed at the end of the process, when the weighted community rate for all services is computed.

Washington wanted copays and deductibles as a means of sharing the cost of services with the patient, provided that the copays were not greater than those charged for general medical care. It was not essential to determine in advance of the study whether a copay or a deductible would be employed and at what level; this was one factor whose effect on PMPM the actuary could easily estimate.

For Washington, the more difficult issue was trying to determine the income level below which copays would be reduced or waived. No policy decision had been made regarding copay waiver income levels for general medical care, and none seemed likely in the near future. Sensitivity analysis indicated that this would not be a trivial assumption. To complete the study, the State assumed that medicaid and low-income patients would have no copay, knowing that some of them would pay at least a partial copay, and that uninsured persons would have full copay, although some would be entitled to free care.

Elasticity of Demand

Copays and deductibles can also affect utilization and duration of services. If patients have to pay part of the cost of treatment, they tend to use it less, and the more they have to pay, the less inclined they are to use it. The degree to which utilization and duration of a treatment service respond to the amount of copay or deductible is called the elasticity of demand for the service. Services that are very sensitive to the amount of patient participation in payment are called elastic, and those that respond only slightly to changes in patient participation are termed inelastic.

Washington did not change its estimates for utilization and duration of treatment services for its calculation of the effects of different copays. The Washington study relied on a review of socioeconomic studies by the Rutgers University Center of Alcohol Studies for information on elasticity of demand for chemical dependency treatment services.¹ This review concluded that for dependencies other than alcohol, demand for treatment is highly inelastic: no matter what the patient has to pay, demand for treatment remains roughly the same. Lacking any similar studies on alcohol utilization, Washington assumed that demand for treatment of alcohol dependencies would be similarly inelastic.

This is an important assumption, for many legislators and policymakers believe that demand for chemical dependency treatment services is in fact very elastic. They think that many patients of chemical dependency treatment centers are really not very sick and are happy at an insurer's (or the government's) expense. The data contradict this view. The fact that demand for services is highly inelastic indicates that those individuals who have decided to seek treatment are in fact so desperate that high costs do not deter them.

Sensitivity Analysis

It is not always necessary to pursue additional data or more sophisticated synthetic estimates in order to eliminate or improve assumptions. Some assumptions are not worht the time and expense to improve because they affect the PMPM estimate very little. For such assumptions, a good ballpark guess is sufficient.

Once the basic estimating model is built, the actuary can estimate the sensitivity of any assumption simply by varying the assumption over the probable range of values and observing the change in the PMPM. When Washington State and its actuary were debating an estimate of the duration of hospital-based inpatient treatment for the medicaid population, the actuary calculated PMPM estimates for three values for duration of treatment: a "shortest likely" average stay, a "most probable" average stay, and a "longest likely" average stay. The actuary found that the differences in PMPM were a matter of only a few cents and that it matters very little which estimate for duration by medicaid populations the State prefers to use. The impact of variations in duration of hospital-based inpatient care for medicaid patients of PMPM is small because of the small population eligible for medicaid (about 10 percent) and the low use of this modality by the plan (about 10 percent). Because of this low sensitivity, doubling the length of stay for hospital-based care for medicaid recipients increased the community PMPM by only 1 percent.

Note

¹ See James W. Langenbucher, Barbara S. McCrady, John Brick, and Richard Esterly, 1994, Socioeconomic Evaluation of Adictions Treatment, pp. 3-10. The authors cite Hallen (1981), but do not include a complete reference.

Last Updated 11-7-02