This Web site is a component of the SAMHSA Health Information Network
|
This Web site is a component of the SAMHSA Health Information Network |
 |
| | | | | | ||||||||||
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This Web site is a component of the SAMHSA Health Information Network. |
|
Community Supportthe Evaluation Center@HSRI Toolkit
Estimating Per Unit Treatment Costs for Mental Health and
Substance Abuse Programs IV. ESTIMATING PER UNIT COSTS (STEP 2) In this section we will offer several methods which can be used to arrive at a per unit cost estimate. These methods vary in their level of precision and each has limitations which are described in Table 5. On Table 5, the methods of estimating are arranged by relative level of effort and precision. The greatest effort and the most precision is in constructing per unit costs that are provider specific. The least effort and most imprecise approach is using pre-existing estimates. We include the latter here because there may be a case when this approach is both feasible and appropriate, but in general, we discourage the use of pre-existing estimates. The evaluator must find a judicious compromise between the specific conditions of the evaluation and the recommendations in this toolkit. If you choose one method of estimating per unit costs and stick to it, you might find that some types of treatment or services will not be included. This might happen if you use paid claims and want to also include services not covered by the health care plan. Alternatively, you might choose to use several different methods, but then data interpretation may be difficult. It is likely that no single data source may be the right one for the study you are doing. In some ways, estimating per unit costs is like solving a mystery and you are the detective. We can provide some general rules, but each case is different and a satisfactory solution is a balance between case-specific circumstances and adherence to the recommendations in this toolkit. Below, we present some methods for determining per unit costs. Regardless of the method chosen, you should specify the inputs, to the extent that you are able, that are the basis of your estimates. Even if you use pre-existing or payment based estimates, it will be important to understand what inputs were included or how payments were determined. Costs based on these data sources can only be understood and compared if their components are known. Inputs can include:
The per diem cost of hospital care is a direct function of the amount of resources used. When more resources are used, such as more nursing staff per patient, then the per diem cost will probably be higher. When comparing costs of two hospital-based programs, it is important to know whether the differences are due to programmatic differences or simply due to differences in the cost of resource use for virtually the same program. An example of cost differences as a result of programmatic differences might be two inpatient programs which serve two different types of patients which, in turn, may affect the level of resource use: one program may be in a locked unit and another in an unlocked unit. Locked units admit patients who tend to require close observation because they are a danger to themselves or others. This monitoring may require additional nursing staff (i.e., more resources and therefore raise the cost). An example of the latter might be a program that treats the same patients in the same type of program but pays higher salaries for the same staff (e.g., competition for nursing staff in a large city may result in higher salaries than salaries paid in a small community for the same training, experience and duties). a. Hospital Per Diems: Cost Reports Information about hospital expenditures can be obtained from the hospital "Medicare cost report" submitted to the state rate setting agency. Virtually all states have such an agency and cost reports are public information. These reports are very detailed and represent each hospital's record of expenditures, revenues and occupancy rates over a given year. Based on this report, the agency approves a per diem rate for each type of "service" (i.e., what the hospital can charge for a one day stay). For example, per diem approved charges are different for surgery, for intensive care units and for obstetrics. We caution the reader to avoid using the "approved" charge in lieu of costs. N. Wolff and T. Helminiak (1993, pp. 171-172) describe charges as "merely list prices of which third-party payers will pay some fractional amount (which is defined as payments). . . Medicare cost reports for Madison community hospitals reveal cost-to-charge ratios for service cost centers that vary from less than 0.5 to more than 2.0. Charges were found not only to be a poor absolute indicator of costs, but also a poor relative factor as well." Cost reports do permit some calculations of the costs per day that avoid some (but not all) of the problems described above. If there is a separate and distinct psychiatric unit, cost information will be reported separately. The cost report is organized in much the same way income tax schedules are organized. The first few pages summarize all income and expenditures and each successive schedule summarizes more detailed information used to complete the calculations needed to support the preceding page. The most detailed data are at the end of the report. The cost report is not an estimate of future costs, but a report of the year's actual revenues, expenditures and services delivered. Using the summary pages of the hospital cost report (see Tables 7 and 8), divide the total expenditures for the psychiatric unit by the total number of bed days actually used (not the total possible number of bed days). Using information from these summary pages for Forest Green Hospital, we calculate the per diem shown in Table 6. Total expenditures for psychiatric unit = $2,846,106 (Schedule II, col. 7, expense by service) Hospital cost reports provide only room and board information, and do not give costs for ancillary services (e.g., lab tests, medications). In general, the room and board costs represent about 90-95% of the total costs of psychiatric inpatient episodes. If you have access to the average payment for these other services (through the billing office of the hospital), you can estimate the additional costs incurred per patient per episode. Another cost associated with inpatient care, is visits by the attending physician or any medical consultants. These services are billed directly by the physicians themselves and are difficult to estimate without some record of the number of visits (often daily during confinement) and the number of consultant visits. If you do not have information on physician visits, adjust room and board costs to approximate a comprehensive rate. We suggest adding 10% to the daily cost of room and board for ancillary charges and physician visits: $407 per day plus 10% = $448 per day This is a good example of taking a short cut to estimating the costs on top of room and board. Because we know they are a small part of the overall per diem, this proxy will result in only a small margin of error. b. Hospital Per Diems: Paid Claims Using paid claims and dividing the number of days in the hospital stay into the dollar amount of the paid claim provides a figure that is likely to underestimate the actual cost of care if the payer is the government (Medicare or Medicaid) and to overestimate the cost if the payer is a commercial insurer. When using paid claims, it is important to remember that the claim is for the "room and board" costs, not any of the person-specific treatments that may have been part of the stay. These are billed through outpatient billing codes and must be added to the cost of room and board (see above). Surgical procedures (rare in psychiatric clients) are billed separately by physicians. We have calculated state hospital per diems (separately for each facility) using the state records of annual operating and capital expenditures for these facilities as the numerator and dividing it by the number of total inpatient days for the year. The facility records we used were aggregated at the inpatient level, so it was not possible to determine inpatient-level-of-care-specific daily rates. This meant that long-term care and acute treatment had the same per unit cost. In many state hospitals almost everything (not just the room and board) is included in the daily rate, and thus costs do not vary by patient as they do in private hospitals, even though the resources used by each patient may differ. Sometimes state hospital costs may be available through the rate setting agency but if not, they may have to be calculated from records of expenditures or provided by the state mental health agency. In Massachusetts, the fringe on all salaries of state employees is paid by another separate state agency, so we must take care to add 29% to personnel for the cost of the fringe benefits, even though it does not show up on the books of the Department of Mental Health. Capital costs are often not included and should also be added. The additional costs in the Boston area (Cannon, McGuire, Dickey, 1985) are about 5% but range as high as 10%. The mean percent increase attributable to capital costs across eight VA sites (Rosenheck, Frisman and Neale, 1994) was 5.8%. 2. Day Hospital Treatment If the day hospital is in a hospital and serves acutely ill patients diverted from inpatient care, then a very rough estimate of the cost per day is possible by taking the inpatient cost per day (as calculated for acute inpatient treatment, above) and assuming that one at least one half of the rate (and possibly as much as two thirds) represents costs that can be attributable to the daytime (8 hour) period during which the day hospital is functioning. Consulting with the service director and the financial administrator will improve the precision of the estimate. This estimate provided below is based on the definition we provided earlier. If we just took one third of the rate (8/24 = 1/3), we would have to assume that inpatient staffing didn't vary by shift (it does) and that all the administrative and other support functions associated with the inpatient unit would be reduced by two thirds (it isn't). It is far more likely that these overhead costs attributable to the day hospital would remain almost as high, regardless of the number of days per week or the number of hours per day the day hospital operated. Using the per diem from Forest Green hospital (see Table 6) the calculations would then be: $2,846,106/6992 = 407 (expenses divided by patient days) + 10% = $448 This rough calculation assumes patients attend for 8 hours a day. This may not be the case. Furthermore, the cost of treating an episode of illness in a day hospital needs to take into account the days the day hospital is open for business. Unlike inpatient units, many day hospitals are closed on weekends and holidays. Some day or partial hospital programs (the names seem fairly interchangeable) are actually not in hospital settings, but are considered a community-based program. When this occurs it is wise to have a good definition of the program, as it is likely to be geared to less acutely ill individuals. However, in this era of privatization, agencies sometimes have contracts to run programs that are sited on hospital grounds or otherwise nearby. If the contract is available to you along with the annual actual expenditures associated with the contract (the "actuals"), then those expenditures can be divided by the total number of actual number of patient days (or service units) over that year to arrive at a fairly good estimate of the per unit cost. These services include, but are not limited to, psychotherapy and medication visits with a clinician. For purposes of determining the cost, the treatment philosophy is not important, but the experience and training of the provider delivering the service would influence the per unit cost of that service as well as the amount of time spent with the client. One hour of psychotherapy with a psychiatrist costs more than one hour of psychotherapy with a social worker. Less obvious, and much more complex is the question of value: what is the cost (comparing two professional types) to treat an episode of illness? The answer begins with knowing what the per unit cost of psychotherapy is for each category professional type. We will consider four types of outpatient treatment in this section, assuming all of them are delivered by the same agency: individual therapy, group therapy, family therapy and medication visits. (Of course, all of these "acute" treatments may be modes of support and maintenance treatment for long-term clients, but if the professional delivering the care and the time spent is the same, then the cost is the same.) The per unit cost of these services will be partly a function of the level of professional training of the staff and the amount of time provided to a particular client. To estimate the costs without actually using the general ledgers of the agency to do the cost-finding, we suggest the following approaches. a. Using Information Supplied by an Agency or Clinic In consultation with the agency, gather data about the staffing (Full Time Equivalents by professional level), the proportion of total budget that is devoted to non-personnel costs (e.g., an estimate of indirect and overhead costs), the number of units of service provided, by service type, the average number of people served in a group therapy session and the average number of people served in family therapy. Use this information to estimate the agency expenditures for these services and to arrive at a per unit cost by dividing the program expenditures by the number of units of service provided (See example on the following page.). Before you can make the necessary calculations, ask if the rent or capital costs are included in the overhead (non-personnel expenditures)? Are there volunteers? (Can the value of their time be estimated?) Does the clinic budget include expenditures for services other than the one for which you are estimating costs? (Can these resources be subtracted?) To proceed, we assume that rent is paid and accounted for in the expenditures provided by the agency they gave you, but there is one volunteer who works two days a week who acts as an advocate for consumers who want information and help getting state or local benefits. To estimate the salaries (and fringe) of staff, use the Occupational Wage Survey available from The Department of Labor, which gives state and regional specific average salaries. The volunteer time is valued at the rate (including fringe benefits) this volunteer would be paid to do the same job. If there is no information about the job level, use the entry-level wage of a person doing similar work in another setting. Annual Expenditures: $750,000 personnel5 Agencies may provide service use in one of two ways. One way is to enumerate the time spent by clinicians by type of treatment. The list below is units of service by clinician. Typically treatment is billed in 15-minute segments allowing flexibility in treatment duration. Thus, "9,500 group therapy" means 9,500 15-minute units of group meetings.
To calculate the cost per unit of service, divide the agency budget by the total number of service units actually delivered: $1,000,000 / 35000 = $28.57/ 15 minute unit for individual therapy and med visits (when only one client at a time is served). When more than one client is seen at one time by a therapist, then the cost per unit goes down. For group therapy, assume that on average 6 persons are in a group session, then $28.57/6 = $ 4.76 per 15 minute session per person (consult with program personnel to determine the actual average number of group members). For family therapy, assume 4 persons are seen together, then $28.57/4 = $7.14 per 15 minute session per person. (For actual mean group sizes, interview the program director at the agency.) A second way that agencies might provide service use is by client visit. In this case, Using this approach, a 15-minute session costs $9.71. This approach results in markedly lower rates for individual therapy and medication visits, but higher rates for group therapy and family therapy. This probably underestimates the cost of delivering individual therapy and medication and assumes all therapies have equal costs per 15 minutes. This is usually not the case. Professionals who deliver individual therapy or medication may have higher salaries. Regardless of the approach, you can see from these examples that we cannot differentiate professional levels or experience. All the per unit costs assume the average personnel cost. In order to be more precise, it would be necessary to have access to the agency personnel records and job descriptions to accurately assign specific personnel costs to specific service provision activities. (Even a general ledger will not tell you how each person spends their time, just which cost center carries their expense. Furthermore, individual staff members typically provide a variety of services). These per unit costs could be adjusted using differences in salary levels by professional, if two additional pieces of information are available: first, the proportion of services provided (as units) by each type of professional and second, the service use data collection documents the profession of each service provider. There is no point in going to the extra effort to calculate profession-specific cost data if you don't know which specific services were actually delivered by which professionals. b. Secondary Data: Billing Information A different approach is to use billing data from the agency (this is not exactly the same thing as using paid claims data from the payer). This approach is feasible if the agency provides a limited number of services and is able to account for the funding of "free" care (e.g., care not billed to any payer and delivered "free" to the client). Using this approach, one could call the agency and ask for reimbursement rates for specific services delivered; ask the billing office what different payers pay for different services: Medicaid is usually lowest, Medicare higher and commercial carriers or "the Blues" the highest. To arrive at a more precise estimate of the cost per unit, use the average payment weighted by the proportion of clients in each payer category. Also, ask if the clinic is entirely supported by reimbursement or are there contracts or deficit spending to provide "free" care? For purposes of our calculations (see example below), assume that the state reimburses the agency for free care each year. In the year of interest, that amount was assumed to be the equivalent of Medicare rates and amounted to 10% of the revenues of the agency equivalent to reimbursement at that rate. For the procedure for determining the best estimate, we will use individual therapy as our example, but all types of services delivered by the agency can be estimated the same way.
Then, to weight the average payment, by proportion of payer revenue, [22 x .35] + [26 x .30] + [29 x .20] + [32 x .15] = 26 Thus, we estimate the average cost to provide 15 minutes of acute outpatient individual therapy is $26.00. The same method could be used to estimate group, family and medication treatment. 4. Long-Term (non-acute) 24-hour care When collecting service use data, it is important to determine the level of care provided, distinguishing long-term 24-hour care from acute inpatient care. Long-term care includes institutional programs, such as nursing homes or domiciliary care. An example of the latter might be clients in state hospitals who require 24-hour nursing care in a hospital setting, but not at the level of intensity of acute inpatient units. We make special note of this here because the cost of this non-acute round the clock care is likely to be far less costly than acute inpatient treatment, but records of admissions and discharges to long-term care units in state hospitals may be difficult to distinguish from acute admissions. This is especially true in state hospitals with multiple treatment units that serve patients with different needs. Daily rates for long-term care in a state hospital might be available from the state department of mental health. If only one rate is available, interview the hospital superintendent to help develop a definition that would permit an estimate based on what rates are available. It is likely that overhead will be less (but not much) for this type of unit, but staffing and medical backup may be significantly lower. Knowing how the staffing patterns differ from the acute inpatient care should help guide the estimation process. 5. Long-Term Community Support Services Ambulatory community support services vary widely across the country, but most public services include some type of daytime program. These programs are among the most difficult to define for purposes of cost comparison, in part because several different types of activities (and levels of care) are sometimes provided within a single program and in part because the use of labels, such as day hospital or clubhouse, has not been standardized. Other programs include vocational rehabilitation, supported employment and case-management. Often programs providing community support are funded by the state or local mental health agency, rather than being reimbursed by health insurance. These funds might be provided to a local agency on contract to the public entity or that entity might directly provide the service. When a private agency is delivering care on a contract, the agency or the contractor should have annual expenditures available along with the number of units of service provided through the contract. In Massachusetts, the Department of Mental Health categorizes all services by program code, which provides a measure of standardization across programs within code numbers. Remember that you need to inquire about whether the rent or capital costs are included in the overhead (non-personnel expenditures). Are there volunteers? Does the contract include expenditures for other services? To calculate the per unit cost of a day program (as an example of community support programs): Actual annual expenditures: $60,168 Some programs have records of the number of hours attended by each consumer. If this is the case, then the per day unit cost can be broken down by per hour costs and more precisely attributed to each attendee. Many systems of care for the seriously mentally ill have found that rapid response to psychiatric crises can mitigate the potential disruption that occurs and also avoid admission to an inpatient unit. In Massachusetts, crisis teams are called upon when a consumer is at risk for hospitalization. They provide rapid assessment of the situation and based on a clinical evaluation of the client, recommend and refer to the least restrictive level of care. If we were to evaluate the cost effectiveness of these teams, then it would be important to understand if and how team activities varied. A crisis "contact" might last several hours, might involve travel, or might involve only a phone call between a provider and the team. Rural teams often spend considerable time on the road while Boston area teams are more likely to be Center-based, but travel occasionally to local hospital emergency rooms. Yet, the Department of Mental Health lumps all the contracts for these teams into one program type, "crisis evaluation" when they report their annual expenditures. The per unit cost, if calculated for each contract separately, varies by region and some of that variation is due to differences in how the crisis evaluation team activities are structured. To calculate the per unit cost of a team: Annual expenditures (actuals or contract) for a set of teams in a given region of the state = $250,000 Determining the costs of the use of hospital emergency rooms is particularly difficult for several reasons. First, costs vary according to the time spent and the procedures provided. These costs vary widely, by client, and cannot be meaningfully averaged. Keeping in mind the caveats listed above, hospital cost reports can be used to determine mean per unit costs in much the same way as the inpatient per diem was calculated. Using the same type of approach, we calculate the per unit cost as: Expenditures for emergency room = $8,184,340 (Schedule II, col. 7) In the Boston study of housing for homeless mentally ill adults, case management was a major feature of the program studied, so the level of precision in the cost estimates was important. The agency that employed the case managers had a subcontract from the larger state-financed housing program. This contract included a budget that specified the salaries of the case managers, their supervisor and the agency overhead. Interviews with the state agency (the contractor) and the local agency finance officer supported the plan to use the contract budget figures as the first step in the cost-estimation plan. In addition to personnel costs, the subcontract also included other operating costs of the program including case-manager travel costs; the case management program's share of the agency operating costs, and a share of the facility capital costs. To these numbers, we added a portion of the salary that represented the time of the supervisor's supervisor (not in the subcontract, but paid directly by the state). In this example, "salary" includes fringe benefits. Note, how several of these items are joint costs and are apportioned among many services in addition to case management. In this case, the apportionment was calculated as a percent of the program personnel salaries of the agency salaries. Sometimes apportionment is done using the square feet of space the program uses or some other meaningful unit depending on the work of the unit. The sum of these annual costs was adjusted by subtracting the time spent (10%) on research related activities. Next, the adjusted total was divided by the total number of client direct contact hours over the year. The information of face to face visits was taken from the case-manager logs. The result is an estimate of the cost of providing an hour of face to face case management. The actual calculations were: Personnel plus fringe benefits $323,901 Remember that using contact hours as the denominator means that we have accounted for all the time the case manager spends on behalf of the "average" client. A more precise estimate of the cost of a case manager's time for a particular client would require an accounting of all the time a case manager spent on behalf of that client (both directly and indirectly, including administrative meetings and supervision) and then that amount of time would be multiplied times the cost per hour to deliver case management services. This hourly rate might be a third of the rate derived above. Rent for apartments are normally set so as to compensate the landlord for vacancies as well as other expenses. When apartments are single-occupancy, costs can be simply estimated from the monthly rent . Supported apartment expenditures consist mainly of the occupancy costs. Personnel costs are, by design, much lower than living sites with staff. In the case of the housing study, apartments were distributed among a dozen buildings in Boston, virtually all under the control of the Boston Housing Authority (BHA). Due to the number of buildings involved and the variation in apartment size and within-building location, real estate agents' estimates of fair market rents for each unit were not obtained. Instead, the U.S. Department of Housing and Urban Development (HUD) fair market rents for Section 8 housing were used. These are set at the 45th percentile of rents (including utilities) in the city. About 80% of the clients were assigned to efficiencies, according to availability, and the remainder to one-bedroom apartments. These proportions were used to weight the HUD ceilings for studio apartments (single room with small kitchen and bath) and one-bedrooms for 1994, yielding an average monthly rent of $580. We choose to use this average amount rather than apartment-specific amounts because apartment was a function of random assignment, not individual choice. In order to calculate the "support" part of supported housing, we included staff time related to housing (in other words, not treatment support). A part time housing liaison for the BHA apartments was assigned to maintain contact with the BHA on-site manager in each location (a cost already included in the average BHA rent), who met with residents on a monthly basis for education, crisis resolution, and other individualized help; and was available 24 hours a day to property management staff. In addition, a housing specialist furnished and set up the individual apartments and cleaned them when vacated. The salaries of these two individuals, including fringe benefits and payroll taxes were added to the cost. In addition, a staff person in one building was hired for several months to reduce problems of drug dealing during the night. The staff person's total compensation, for this responsibility, was estimated at about $5,000/year (10% of his salary). This amount was distributed among clients in proportion to the number of client-months each year. The only start-up expenditures associated with apartments in the housing study were for new furniture, lamps and a television. These costs were allocated to the first month of rental tenancy. Although it is more common to treat furniture as a capital expenditure, which are amortized over the course of their lifetime (typically 5 years), we found, in this study, that furniture often had to be replaced more often which led to our decision to allocate these costs to the first month of rental tenancy. In the Boston housing study, the number of nights in a shelter for all the housing study subjects accounted for less than 2% of the total nights for all clients in the study, the cost per night was calculated as the annual budget for the shelter divided by the number of clients who used the shelter over the course of the year. This method is not very precise, but if shelter costs were a large part of a cost evaluation, then more care would need to be applied to obtaining the cost per night. For example, shelters are far more likely to be crowded in the winter, leading to more staff and possibly higher costs per diem in the winter because of fluctuations in staffing and other related resources used (e.g., higher utility costs). Often the cost of living on the streets or in the home of a friend or relative is priced at zero, although the true cost of such stays is higher. Those living on the streets may in fact incur some local government expenditure, such as police contact, and living with friends results in some additional use of resources at that house. In another study, we used SSI monthly payments as a proxy for subsistence costs and divided the monthly payment by 30.4 days to arrive at a daily cost of living on the street or in someone else's house. Market rent should reflect resources used by the landlord to make the apartment available on the market, including less than optimal occupancy, real estate taxes, maintenance and repair costs. When some, but not all, of the utilities are included in the rent, those utilities that are not included must be added on. When the client shares the apartment with a roommate, the rent and utilities are divided by 2, assuming that each incurs half of the resource use (and thus is libel for half the expenses). The following description of how to cost a police contact is taken directly from Nancy Wolff and Thomas Helminiak's work (1993) and provides guidance on how to estimate other non-mental health costs. It is provided to assure readers that the same guidelines for estimating those costs are the same for non-mental health services. They begin the discussion by reminding the reader that a contact needs to be clearly defined as it could mean a telephone contact with a dispatcher, an arrest, an incident that leads to a written report or any other incident. They chose to define contact as " a case-logged incident, where the officer responds to a complaint or an observed infraction and the contact generates an officer report." They based their per unit cost estimate on "the annual budget expenditures plus the estimated imputed value of building space that was allocated to the provision of primary police and follow-up investigative services. The population-based average cost estimate for a police contact was then adjusted to account for the higher use of police department resources by the severely mentally ill. Unit costs ...were estimated to be 17% higher than cost of police involvement with other individuals." Provided by local police headquarters $148 per contact |
|
| Home | Contact Us | About Us | Awards | Accessibility | Privacy and Disclaimer Statement | Site Map |
