Chapter 4—Major Elements of an Actuarial Study

As noted in Chapter 2, an actuarial study is an estimate of the cost of a specific plan. An actuarial estimate is usually expressed in terms of cost of a specified package of benefits per covered person per month, or "net cost per member per month" (PMPM). This form of the estimate is traditional for insurance companies, since it translates directly into a monthly premium, which is how insurance companies typically charge their enrollees. This estimate may also be referred to as a "premium" or "capitation," although these terms are more properly applied to the purchase price for actual plans, not the actuarial estimated cost to the plan maker. An insurer or a health maintenance organization might set a premium or capitation higher or lower than the actuarial estimated PMPM, based on marketing judgments.

In creating the PMPM estimate for a plan, an actuary first estimates separately the cost of each service in the benefit package. For example, if the chemical dependency treatment benefit is to include inpatient treatment and outpatient treatment, the actuary first estimates a PMPM for inpatient treatment and a separate estimate for outpatient treatment and then combines the two estimates with appropriate weightings to get a final single PMPM. (Washington State's actuarial study had six separate covered treatment services: hospital-based inpatient, freestanding short-term residential, long-term residential, regular outpatient, intensive outpatient, and methadone treatment.) Separate estimates may also be needed for various segments of the population, depending on what is known about the population segments' use of services. As with the various benefits, a final analysis weights all the separate estimates and combines them into a single PMPM estimate for the entire covered population.

Actuarial Data

Actuarial estimates are based on empirical data wherever possible. The data in an actuarial study typically include payment and enrollment figures that the actuary has obtained from insurers and medicaid and medicare data obtained from State government or the Federal Government.

The Washington State study used both national and statewide data. Less confidence was placed in the national data, because national data may reflect factors not relevant to individual States. For example, State poverty rates may be significantly lower than the national norm. National data also may not reflect influences that are of great local importance, such as the collapse of a particular industry or local laws prohibiting alcohol possession or sales. In some situations, State data were unavailable or were based on too few covered lives to be reliable, making national data, despite the drawbacks, the best available.

An actuarial estimate requires data about three factors: the annual utilization of each service in the benefit package by the covered population, the duration of each service (the average number of units used per admission), and the average cost per unit for each service. The basic actuarial formula is as follows (its terms are defined below):

Utilization

"Utilization" means the frequency of admission to treatment service by the covered population (or a particular covered population segment). This usage of the term is different than the typical sense found in chemical dependency treatment, where it means the percentage of capacity of a program that is actually delivering services or occupied by patients. The actuarial sense of "utilization" is the proportion of a covered population that will use a particular service during a specified period (usually 1year). Utilization is computed by dividing the total number of admissions expected from the covered population by the size of the covered population. Utilization is usually expressed in terms of the number of admissions per 1,000 insured lives (covered persons) per year.

Utilization is different for each service in a benefit package. Some covered individuals are admitted only for outpatient care, some are admitted only for inpatient care, and some are admitted only for detoxification; many are admitted to multiple modalities. This service mix varies from State to State, depending on therapeutic traditions and facilities available. For example, most detoxification and short-term "in-patient" treatment in Washington is done in freestanding residential facilities rather than in hospitals. Washington's utilization patterns are different from those of the rest of the country, and they probably will remain so after the implementation of health care reform.

Washington needed its actuary to use local practices and data to estimate a separate utilization rate for each modality of service in the benefit package. Washington's actuary calculated separate utilization estimates for six modalities (Table 4–A): medically managed inpatient (meaning hospital-based detoxification and treatment), nonmedically managed 30-day residential treatment (in non-hospital-based facilities), non-medically managed long-term residential treatment (over 30 days), intensive outpatient treatment (including "day treatment" and "partial hospitalization"), regular outpatient treatment, and methadone treatment (including both detoxification and maintenance; methadone treatment is referred to as "opiate dependency treatment" in Appendix B, to incorporate other approved substitute chemotherapies that may be used in the future).

Utilization data are typically derived from historical admission patterns from a known population. Insurance payment records, medicaid records, and medicare records are typical sources. Of these, insurance records are typically the most numerous and most complete. Since insurance records are maintained by organizations that historically have relied heavily on actuarial studies and know that they will need them in the future, they are reliable, complete, and numerous enough to generate estimates that are likely to prove accurate.

Washington has a large and fairly detailed set of data from its computerized client information system. The State hoped to use these data in determining actuarial costs, but in practice, this system provided little of value. In general, State and local substance abuse management information systems are weak data sources for deriving utilization estimates. These systems may record the number of admissions (plus detailed data regarding persons admitted), but they cannot record data on the eligible members of the covered population who have not been admitted. The size and composition of the population from which these admissions are drawn can only be roughly estimated. This lack of "denominator" data means there is too much uncertainty regarding the rate of admissions. If State substance abuse management information systems are to be used to estimate utilization, an additional data source must be available from which accurate denominators can be drawn.

Duration of Service

In actuarial studies, "duration of service" is a consistent measure of service intensity and answers the question, How many units of service does the average patient receive per admission? For residential and inpatient modalities, duration of service is expressed as days of treatment. For outpatient services, duration is expressed in terms of number of events, meaning individual or group counseling events. (The analysis assumes that the ratio of individual counseling to group counseling does not change. Such an assumption may be challenged, particularly where the plan anticipates strong cost containment measures.) Methadone treatment may be measured in events or days. Washington State chose days for metha-done, because the data on "days" were considered more accurate than the data on "events," and choosing days avoided the need to assume minimal variation in the ratio between individual counseling and group counseling.

Table 4–A.—Washington State Utilization by Insured Population

____________________________________________________________________

_______________________________________________________________________

Insurance payment data bases typically include duration-of-service data, as do medicaid data bases at the State level. Washington was able to use encounter data from its substance abuse management information system (the problem of inadequate data on the covered population does not affect estimates of duration, because the duration calculation is the number of events divided by the number of admissions).

Cost of Care

Actuarial "cost of care" is the average amount paid per unit of service by the patient or third-party payer. It is not the same as an average treatment program's cost of providing care; rather, the actuarial cost is the average amount that the insurance company or other payer has to pay to obtain the care from the treatment provider. Because insurers like to negotiate discounts, the actuarial cost is typically less than the treatment program's usual and customary charge for a unit of care. The actuarial cost in many cases may be lower than a treatment facility's average cost of providing care. Fa-cilities must make up for such below-average payments by overcharging other patients.

Actuaries prefer using insurance payments as the basis for estimating cost of care because they are trying to forecast payments for specified periods. From the point of view of an insurer, a treatment provider's average costs or usual and costomary charges are only of concern insofar as they affect the amounts that the insurer has to pay.

Provider average costs and usual and customary charges are clearly critical to consider: driving providers into bankruptcy is in no one's best interest. Still, there were compelling arguments that payment data were the best to use. Insurance payment data are easier to obtain and more reliable than data on provider average costs or usual and customary charges, and there are drawbacks (described below) to both provider charges and provider costs as measures of the cost of care.

Usual and customary charges of providers are a function more of marketing judgments than of clinical or economic judgments. For example, a treatment facility may establish a usual and customary charge much in excess of costs, competitive rates, or actual collections simply to leave open the possibility of collecting a large fee if anyone is willing to pay it. Actual provider costs would be free of such marketing influence, but these data are difficult to obtain. Questions about reasonableness, necessity, allowability and allocation of costs, return on investment, marginal versus average costs, and other cost accounting issues can make it very hard to estimate actual cost.

Finally, there is reason to believe that provider costs are determined more by available revenue than by anything else. To a point, providers can accommodate reduced revenues by being more efficient—trimming overhead and reducing profit margins. Once these efficiencies are reached, providers can only respond to lower revenues by reducing the amount of patient services or the quality of care, or both. Higher revenues produce the opposite effect. If providers are efficient, they are probably providing the best quality of care possible for the amount of revenue they have. If the current level of quality is acceptable, revenue should be a good proxy measure of the cost of providing treatment.

The distinction between actuarial costs and provider average costs proved nettlesome in discussions between the State and service providers concerning the Washington study. Many service providers were angered and disheartened by an estimate of actuarial costs for insured and medicaid populations that was clearly below their own facility operating costs. To some extent, these concerns were reiterations of complaints about inadequate reimbursement under current reimbursement mechanisms. Providers were suspicious that the actuarial study was just another way for State government to cut provider payments.

Health care reform introduced some circumstances where these concerns have a new urgency. Chemical dependency treatment experiences a common health care phenomenon known as cost shifting, where some patients (or their insurers) pay more for a particular service and others pay less, depending on the market "muscle" of the payer. The cost-of-care data that Washington obtained from insurance payment data bases, medicaid data bases, contract and budget documents, rate studies, and other sources yielded sharply different average costs for different portions of the covered population, reflecting this cost shifting. Washington's Health Care Reform Act, however, called for universal coverage, with everyone paying a single "community rate." The problem was, Which rate should the State use? If the lower rate were chosen, providers would go broke; if a higher rate were chosen, plans would be paying for shifted costs that would no longer occur.

The State's actuary developed a model to combine all these different sources and estimates into a PMPM estimate that compensated for the end of cost shifting. First, the actuary adjusted the best sources into the same time period, using data on medical inflation rates. Next, the actuary weighted each rate or cost estimate, based on the number of admissions expected at each rate during each year of implementation, in order to arrive at an average cost estimate. Since Washington's health care reform phases in various groups over a 4-year period, the weightings are different for each year.

Washington's actuarial cost-of-care estimate included a few items in addition to the payment to the treatment provider. Allowances for the insurer's administrative costs, the cost of the managed-care organization, and profits were factored in. For simplicity, Washington did this with an across-the-board marginal overhead factor. The Washington study assumed a rather low, 15-percent overhead factor, anticipating that other aspects of State health care reform would contain overhead costs.

Inflationary increases must be loaded into the cost-of-care formula to bring historical data (usually at least 1 year old) into the forecasting period (usually at least 1 year into the future). For Washington, there was some concern over which inflation rate to use. Actuaries normally use only the medical component of the Consumer Price Index when estimating inflation for health care costs. In the past, Washington's legislature had used the whole Consumer Price Index, which reflects changes in the general cost of living, for budgeting increases in rates for chemical dependency treatment facilities. Increases in the cost of medical care historically have exceeded increases in the general cost of living by about a factor of 2. Except for medically managed programs such as hospital-based inpatient treatment or detoxification, chemical dependency treatment costs are not increasing faster than the general rate of inflation. However, Washington chose to estimate conservatively, using the medical component of the Consumer Price Index, which averaged a 5-percent increase in costs each year (Table 4–B).

PMPM ^aIncludes a 5-percent increase from the preceding year for inflation.

Back to Top