Micro-Cost Methods: Pseudo-Bill

VA utilization data may be combined with unit costs from non-VA sources to estimate the cost of patient care. This is referred to as the pseudo-bill method, because the itemized list of costs is analogous to a fee-for-service bill for healthcare services. To estimate costs in this way, the analyst needs data on the services utilized by the patient, and the cost of each service.

Itemized information on outpatient services can be obtained from the VA outpatient National Patient Care Database at the Austin Automation center, as described below. This database includes the Current Procedures and Terminology (CPT) codes assigned to each outpatient encounter. The cost of providing this care can be estimated from Medicare reimbursement rates developed by the healthcare Financing Administration (HCFA). Although Medicare does not reimburse for all services that have been assigned a CPT code, the analyst may still wish to assign a cost to care that Medicare does not pay for.

We first describe sources of data on VA outpatient care and methods for estimating the cost of outpatient procedures. These include employment of Medicare relative values, the use of non-Medicare values, and the types of assumptions that may be needed to find cost when neither source of relative values is available. We then consider "conversion factors," the rates which are used to convert a relative value to a reimbursement amount.

This discussion then turns to methods for finding the cost of acute hospital stays, and why a clinical cost function may be a more feasible method than a pseudo bill for finding this cost. Examples of studies that used the pseudo-bill method are given as references.

1. Sources of data on VA outpatient care

The most available source of information on VA outpatient care is the National Patient Care Database (NPCD). The procedure files from this database includes a unique patient identifier, the date of service, the location of care (the DSS identifier, formerly called the clinic stop), and the CPT codes assigned to the encounter. This file is accessible to users of the VA Austin Automation Center. Details on the contents of this file, and how to access it, may be found at the VA Information Resource Center (VIReC) website.

Outpatient visit data are also kept in the VISTA data system at each VA medical center. This information is the source of the data in the NPCD. Since it is difficult to extract VISTA, we recommend that the NPCD be used instead.

Outpatient laboratory

The NPCD outpatient procedure file includes CPT codes for laboratory procedures, but it may not contain all information recorded in VISTA. Research is needed to learn exactly what information is transmitted to NPCD, and what information is only available in VISTA.

Methods of Extracting Data from VISTA

Utilization data may be extracted from VISTA using one of the following methods:

1. Write a report with the Fileman report writing tool included in the VISTA system.
2. Access individual patient records in VISTA, capture screens of data with the log features of a terminal emulation program, and then extract needed data from the file of captured screens using the Monarch programming language.
3. Request electronic patient summaries of data for specific patients, and extract these with the Monarch programming language.
4. Install a custom MUMPS program on each facility's VISTA system.
5. Use the data extracted from VISTA that appears in the DSS production system.

Outpatient pharmacy

The VA Pharmacy Benefits Management database includes information on pharmaceuticals dispensed to each VA outpatient, and the cost of the prescribed drugs. The cost estimate does not include the cost of dispensing. VA researchers cannot directly access the PBM file; they must make a request of the Pharmacy Benefits Management program. (Details on the contents of this file, and how to access it, may be found at the VA Information Resource Center (VIReC) website.

The PBM data does not include the cost of dispensing pharmaceuticals. Medicaid pays pharmacies a fee of $2 to $5 per prescription to cover this cost. Several years ago the VA estimated that its average cost to fill a prescription was $2.50.

2. Determining the Relative Value of Outpatient Services

To find the cost of outpatient services, we recommend that Medicare reimbursement rates developed by the healthcare Financing Administration (HCFA) be used. Medicare regulations report a relative value for CPT codes for physician, anesthesiology, and laboratory procedures. A conversion factor translates these relative values into a reimbursement amount. This relative value is converted to a reimbursement rate by multiplying the relative value by a conversion factor. For the year 2000, Medicare pays physician and laboratory providers about $36.00 for each unit of relative value; anesthesiologists are reimbursed about $17.00 per unit.

Medicare Relative Values

Medicare has adopted the Resource Based Relative Value System to reimburse providers for services provided to Medicare patients. Care is characterized by a CPT code. Each CPT code has an associate set of relative values (RVUs), consisting of Facility, Physician Work, and Malpractice RVUs.

• Global Fees

  The RVUs for some procedures reflect a global fee that reimburses the providers for the procedure, for visits that occur immediately before the procedure, and for visits after the procedure was performed. For each procedure, HCFA indicates the length of the follow-up period that the global fee covers. Global fees give providers an incentive to be efficient, and use the minimal number of visits. The cost analyst has a different concern, wishing to adopt a method in which the cost estimate reflects the number of visits that occur before and after the procedure. For this reason, we recommend that the analyst not apply global fees, but whenever possible, apply separate RVUs for the procedure and the pre-procedure and post-procedure visits

• Facility vs. non-facility RVUs

  HCFA has created two sets of RVUs for the facility component of care. These are "facility," created to reimburse hospitals, nursing homes, and ambulatory surgery centers, and "non-facility", which were created to reimburse office-based practices. The difference is these two sets of RVUs is generally small, but it can be substantial for certain procedures. The RVU for office-based practice is higher, evidently because HCFA feels that extra expense is incurred when complex procedures are done in a setting where they are not normally performed. The relative values for each CPT code may be downloaded from the Center for Medicare and Medicaid Services (CMS) web site. The analyst should apply the facility weights for care at VA medical centers. It is uncertain whether the office setting should be applied for care provided by VA satellite clinics, or whether VA databases provide sufficient information to determine where care is actually delivered.

Services Not Reimbursed by Medicare

Many of the CPT codes used by VA do not appear on the list of services that are reimbursable by Medicare. There are several reasons for this, including the following:

1. Medicare does not pay for the service,
2. the VA code is not specific,
3. the VA code is out of date,
4. the VA code is not a valid CPT code.

In general, we believe that a cost should be assigned to all services, even if they are not reimbursable by Medicare. It is the challenge of the analyst to make as few assumptions as possible in assigning the RVU (and thus estimate the cost). We consider each of these problems in turn.

• Gap Codes

  There are some services which Medicare does not pay for. These services have a CPT code, but HCFA has not assigned them a relative value unit. These CPT codes are known as "gap codes." Relative values for these procedures were developed by the Cambridge Health Economics Group, a private firm that has been acquired by Ingenix. A book and an electronic file are available from this company.

• Unspecified Procedures

  CPT codes for unspecified procedures, typically ending with the digits "99", do not have a Relative Value Unit assigned to them. Among the top 30 CPT codes used by VA are the following unspecified codes for laboratory tests:

  85999 - Unlisted hematology and coagulation procedure
  84999 - Unlisted chemistry procedure
  83999 - Unlisted miscellaneous pathology test

The analyst can assign an RVU by assuming that the unlisted code represents a typical procedure of its type. Thus the RVU would be the weighted mean RVU of the procedures of that type, with the weight the relative frequency the procedure is used by VA. Using this method we would assign an RVU to 85999 "Unlisted hematology and coagulation procedures", based on the VA weighted mean RVU for hematology and coagulation procedures, that is, all procedures reported in the range of CPT codes 85002-85810.

Other CPT Codes without RVUs

For some procedures, no RVU is available. For example, 99078, Group Health Education, is among the top 30 most frequently used CPT codes by VA, but it does not have a Medicare or a gap RVU. The analyst could use the RVU of a similar CPT code, or assign the mean RVU of all CPT codes used in the that clinic stop.

Some VA services are characterized by an invalid CPT code. Although a healthcare payer would reject the bill for such a service, the analyst may want to include the cost of this care in a pseudo bill. For frequently used codes, the analyst should check the code to see why it is invalid. For example, code 90724 was one of the top 30 codes used by VA in 1998, has been deleted, and has been supplanted by a series of new codes for influenza vaccine, 90657-90660. All of these new codes have been assigned the same gap code relative value unit.

If the VA assigned CPT code appears to be entirely invalid, the analyst can drop the code from consideration, or assign the mean RVU of all CPT codes used in the that clinic stop.

3. Determining the Conversion Factors

There are a number of issues to be considered in applying conversion factors to estimating the reimbursement for a relative value unit. These include:

• Geographic Adjustment

  Medicare provides separate geographic adjustment factors for the physician services, malpractice, and the facility component. These adjustments can be used to estimate costs in a given region. Alternatively, a national average can be used. The national average geographic adjustment factor is approximately one.

• Professional and technical component

  For some laboratory and radiology services, the RVUs are split into professional (physician) and technical components. This allows fees to be split between the provider and interpreter of the test. Care must be taken in apply these components to avoid double counting the RVU and associated reimbursement.

• Non-Physician providers

  HCFA has established slightly different payment rates for non-physician providers; an accurate pseudo bill may wish to reflect these differences.

In addition, the analyst may wish to adjust the conversion factor to reflect differences between VA costs and Medicare reimbursement. This additional adjustment is analogous to the cost-to-charge ratio, with the total Medicare reimbursement for all services provided by that VA department representing its "charges", and it costs, from a source like the VA Cost Distribution Report, representing the costs.

For more information, see FAQ D1: What is the cost of physician services for inpatients?

4. Inpatient Pseudo-Bills

It is very difficult to create a pseudo-bill for VA hospital stays. VA does not thoroughly document medical procedures provided to inpatients, nor does it assign CPT codes to physician services provided to inpatients. A list of many of the resources used in an inpatient stay can be extracted from the VISTA system, but there are gaps in what is covered. Information on inpatient laboratory and pharmacy utilization in VISTA may be incomplete. Additional information about services provided in VA hospital stays can be found in the Surgical and Procedures databases in the VA Patient Treatment File. Data include most surgical procedures, and some medical procedures, recorded as ICD-9 procedure codes. These gaps might be filled via direct observation of the resources used in the hospital stay.

If a list of resources is obtained for a VA hospital stay, there is still the problem of finding the appropriate charges for each item on the list. Charges from a non-VA hospital could be used only if the items listed are directly comparable. It will be difficult to create a resource list for a VA hospital with a level of detail that corresponds to the charge schedule of a non-VA hospital. There are no relative values associated with ICD-9 procedure codes. Strong assumptions must be made to convert ICD-9 procedures to a CPT codes; the CPT system has a much finer level of detail, with many CPT codes associated with a single ICD-9 procedure code.

Two studies from the United Kingdom have determined that hospital costs can be accurately measured with a reduced list of utilization measures. A study of community and home-based psychiatric care found that 5 out of 21 measures accounted for more than 90% of the costs (Knapp & Beecham, 1993) . A study of colorectal cancer found that 4 out of 14 types of utilization accounted for 91.6% of total costs (Whynes & Walker, 1995) . These studies suggest that a full pseudo bill may not be necessary, and that a reduced list of utilization may be used to determine hospital costs, if the markers of resource use are chosen appropriately.

An alternative to either the pseudo-bill or the reduced list cost estimate is to estimate a clinical cost-function. If a suitable source of non-VA data can be found, a function can be used to estimate the relationship between cost-adjusted charges and measures of utilization. Its parameters can be used to translate VA utilization data into a cost estimate. Its chief advantage is that it requires less VA data than is needed to prepare a pseudo-bill.

The cost function approach has been compared to an itemized pseudo-bill (Kukull et al., 1986) . It was found that a function that used days of stay, intensive care unit days, number of lab tests, and number of surgeries predicted 97.7% of the variance in these imputed charges.

For more information on this method, see our discussion of clinical cost functions.

5. Examples Applying Pseudo-Bills to Estimate VA Cost

The following studies are examples of where pseudo-bill methods are used to find the cost of VA healthcare. Non-VA data sources of relative value include the rate of charge rates of an affiliated university medical center (Schneiderman, Kronick, Kaplan, Anderson, & Langer, 1992; Wasson et al., 1992), the payment rates from a typical healthcare payer (Kessler, Kessler, & Myerburg, 1995), and the charge rates allowed by Medicare (Volicer et al., 1994; Wade et al., 1996; Wasson et al., 1992).

References

Kessler, D. K., Kessler, K. M., & Myerburg, R. J. (1995). Ambulatory electrocardiography: a cost per management decision analysis. Arch Intern Med, 155(2), 165-9.

Knapp, M., & Beecham, J. (1993). Reduced list costings: examination of an informed short cut in mental health research. Health Econ, 2(4), 313-22.

Kukull, W. A., Koepsell, T. D., Conrad, D. A., Immanuel, V., Prodzinski, J., & Franz, C. (1986). Rapid estimation of hospitalization charges from a brief medical record review: evaluation of a multivariate prediction model. Med Care, 24(10), 961-6.

Schneiderman, L. J., Kronick, R., Kaplan, R. M., Anderson, J. P., & Langer, R. D. (1992). Effects of offering advance directives on medical treatments and costs. Ann Intern Med, 117(7), 599-606.

Volicer, L., Collard, A., Hurley, A., Bishop, C., Kern, D., & Karon, S. (1994). Impact of special care unit for patients with advanced Alzheimer's disease on patients' discomfort and costs. J Am Geriatr Soc, 42(6), 597-603.

Wade, T. P., Virgo, K. S., Li, M. J., Callander, P. W., Longo, W. E., & Johnson, F. E. (1996). Outcomes after detection of metastatic carcinoma of the colon and rectum in a national hospital system. J Am Coll Surg, 182(4), 353-61.

Wasson, J., Gaudette, C., Whaley, F., Sauvigne, A., Baribeau, P., & Welch, H. G. (1992). Telephone care as a substitute for routine clinic follow-up. Jama, 267(13), 1788-93.

Whynes, D. K., & Walker, A. R. (1995). On approximations in treatment costing. Health Econ, 4(1), 31-9.