Community Support

the Evaluation Center@HSRI Toolkit

Estimating Per Unit Treatment Costs for Mental Health and
Substance Abuse Programs

III. DEFINING AND COUNTING SERVICE USE (STEP 1)

1. Identifying Services

In most cost evaluations, the services used by every person in the study are collected. A list is then compiled of each type of service provided to be used as a starting place for determining just which per unit costs must be determined. If no one in the study spent anytime in a ClubHouse program, then ClubHouse costs do not have to be calculated. If, on the other hand, study participants were admitted to three different hospitals, then the per diem costs of each of these hospitals would have to be included.

To have an idea of the range of possible services provided in a given area it is useful to review service taxonomies maintained by public agencies for administrative and planning purposes. Service taxonomies may be found in:

State plans,
State mental health authority contracting manuals or other documentation,
Billing system documentation including CPT codes ³,
List of covered mental health benefits maintained by insurers or managed care companies.

Table 1 is an example of a taxonomy of services from a SAMHSA funded multi-site managed care study.

2. Defining Services

Being clear about the nature of each service to be costed is essential. Conceptual definitions must be grounded in concrete information. A mental health service such as a "crisis bed" may be different in different sites. Thus, it is helpful to define a treatment program along two dimensions: what does the treatment consist of and how is a unit of treatment typically defined? If you begin by determining the exact nature of the program for which component costs are being estimated, you will avoid the apples and pears problem in making comparisons later. For example, in two vocational rehabilitation programs, one per unit cost is $15/hour and another is $60/day. Are they the same program (the same clinical goals), but one costs more than the other or are they different programs (different clinical goals) which happen to cost the same? Without more knowledge of the programmatic differences, you may have accurately captured the costs, but will be unable to interpret the findings that result from the analyses of your data.

Conceptual service definitions should consist of information about aspects of services like the ones listed below.

Staff to client ratios (fewer clients per staff member = more intensive services)
Types of clients served (level of impairment, age, etc.)
Expected mean length of service (days, weeks, months)
Goals of the service, and (recovery, employability, clinical stability)
Types of activities within services (job training, skills of daily living, etc.)

For community-based services not defined elsewhere, we recommend that investigators interview the service director and the financial administrator as first steps in defining the service. The point of being clear about the nature of the service is to avoid misinterpreting cost and outcome data associated with service effects. It is important to be comparing apples with apples. A program with the same definition (and in the same setting, and in the same region of the country) is likely to cost about the same.

Some years ago, we needed to determine the cost of a particular day hospital service. One member on our team decided to investigate the per unit cost of day hospitals nationally. She interviewed service directors all over the country. The questions she asked were:

1.  What is the treatment goal?
2.  Are there clinical indicators for inclusion in or exclusion of individuals from the service?
3.  Are there clinical indicators of successful client response to treatment?
4.  What is the direct clinical staff to patient ratio?
5.  What level of professional training do the staff have?
6.  What is the average length of stay in the service?
7.  Is there a limit to the amount or duration of this treatment that clients can get?
8.  Must providers in this service co-ordinate or plan treatment with other providers or is the service self-contained?

We found wide variation in the answers. The same questions should be asked of any treatment service, including inpatient treatment, so that per unit cost variation can be appropriately interpreted. The responses to these questions (see Table 2) allowed us to define a particular "day hospital" service. It also allowed us to make a decision about whether or not the cost of the "day hospital" in Boston could be substituted for the cost of a day hospital service in Walla Walla or Little Rock. The staff to client ratios, the level of staff professionalism and the numbers of support and administrative staff proved to be the primary indicators of the cost per unit of service.

Managed care is the catalyst for innovation in service approaches, but these very innovations can create confusion in arriving at service definitions. In many states, service innovation is moving well ahead of the bureaucratic procedures for renaming and categorizing services. The result is that out-of-date service names may be used to label new services. The evaluator must be both nimble and skilled enough to detect and closely examine variations in service definitions and also keep abreast of new services, including how and where they are accounted for in administrative documents.

3. Units of Treatment

Any approach that estimates service costs based on units of service provided must define a unit of service for every service studied. Units of service differ with the type of service. Examples for different services are listed in Table 3. Unit of service definitions can usually be found along with service definitions.

4. Sources of Data

The cost of a service, program or system for some number of persons over some period of time is estimated by multiplying unit cost times the number of units of service delivered in the period of time specified. Counts of units of service delivered can be based on different sources. Each of these sources has advantages and disadvantages. The sources and their advantages and disadvantages are listed in Table 4. No data source should be accepted as adequate without investigation. Wolff et al recommend that secondary data (such as paid claims) be cross-validated by comparing them with primary data from client charts.

Table of Contents | Previous | Next