Cost accounting supports clinical evaluations.

COST CONTAINMENT

This article describes how a hospital cost accounting system was used to evaluate the financial impact of the use of a new and expensive drug at a 700-bed academic medical center.

Hospital cost accounting systems have changed considerably in recent years. Prior

to the advent of Medicare's prospective payment system, the Medicare Cost Report was the only approximation of cost accounting in most hospitals. As hospital payment methods have changed, so have the concepts of cost accounting for hospitals.(a) Today, most hospitals that have invested in new cost accounting systems have done so to improve productivity and cost management, to evaluate the profitability of various services or products, or to evaluate managed care contracts.(b)

Concurrent with the advances in cost accounting have been the changes in clinical practice as new procedures, technologies and products emerge. Since many of these new products are extremely expensive, hospitals have had an increasingly vested interest in evaluating their financial and clinical impact before or shortly after their introduction.

Traditionally, analysts have used costs-to-charges ratios developed from data taken from the Medicare Cost Report to approximate costs at the patient level. These cost estimates have suffered from a number of biases that have been noted elsewhere.(c) Today's new cost accounting systems can provide much better evaluations, as illustrated in the following example of Thomas Jefferson University Hospital (TJUH), a 700-bed academic medical center in Philadelphia, Pennsylvania, and how it used its cost accounting system to evaluate the financial impact of a new and very expensive drug, Zofran.

The cost accounting system

The clinical financial management system (CFMS) at TJUH integrates clinical and financial data from the medical records, billing and cost accounting systems. The cost accounting component of the system uses a bottom-up approach that begins with cost determinations at the procedural level. The CFMS system utilizes relative value units determined by management engineering studies to estimate each cost component of a procedure. Unit costs are divided into fixed and variable as well as direct and indirect components. For example, the cost for a complete blood count would include direct variable costs for labor and supplies, direct fixed costs for equipment and space, and a fair share of indirect overhead costs such as medical records, utilities, or housekeeping. Unlike the traditional step-down allocation method used in the Medicare Cost Report, the CFMS system offers a customized allocation of indirect costs, first to departments and then to procedures within a department.

The CFMS system uses medical record information to identify sub-groups of patients with various clinical characteristics. The billing system provides utilization data, indicating the number and type of procedures or services these patients received. The cost accounting system provides the unit costs for each procedure. These costs are aggregated to determine an average cost per patient.

The CFMS system is normally used by financial administration staff to monitor the hospital's performance by payer, product line, and service and can be used for planning, marketing, and quality assurance. Data from the system also can be used to evaluate third-party contract proposals or the impact of regulatory changes. In addition, the system's capabilities increasingly are in demand for clinical evaluation studies.

The new product

Zofran is a new pharmaceutical product that can prevent the severe episodes of nausea and vomiting associated with some cancer chemotherapy treatments and is believed to be a more effective treatment than traditional anti-nausea and vomiting medications.(d) Because of its efficacy and the absence of significant side effects, Zofran was quickly adopted by physicians for use as the primary medication of its type, and within months of its commercial debut it had been added to the formulary of many hospitals across the United States. One major teaching hospital reported that expenditures for Zofran alone consumed 10 percent of the hospital drug budget within the first year of use.(e) By 1992, Zofran had become the number one drug in the pharmacy budget at TJUH; Zofran expenditures for FY92 were almost $400,000.

The rapid growth of the use of Zofran might not have been of concern were it not for its substantial cost. Zofran is much more expensive than the traditional therapy it replaced.

The cost analysis

TJUH first used its CFMS system to determine the cost-per-case for traditional anti-emetic therapy. A list of the 12 drugs most commonly used to treat nausea and vomiting was developed by physicians who treat cancer patients at TJUH. This was an important component of the analysis because prior to the availability of Zofran, TJUH physicians often had to prescribe four or five different combinations of medications.

Once a list of drugs was developed, the total cost of these pharmacy products for all patients in DRG 410 (maintenance chemotherapy) was obtained through the CFMS system, and the average cost-per-case was calculated. This cost was found to be about $30 per admission. Using the same process, it was determined that the average pharmacy cost-per-case for patients who received Zofran was $300. This ten-fold difference in cost-per-case prompted a more detailed analysis.

Although Zofran costs significantly more than traditional therapy, physicians at TJUH felt that patients tolerated their chemotherapy better when they received Zofran. They argued that the benefits of Zofran might include lower total costs if length of stay was reduced. The CFMS system was used to evaluate this hypothesis.

The CFMS system identified three groups of patients for the cost analysis. Patients in all three groups were those who had been admitted for maintenance chemotherapy over the same time span. (A one-year time frame from six months before to six months after Zofran first became commercially available was selected.)

The first group (ALWAYS) was patients who received Zofran during all of their admissions during the time period. The second group (NEVER) was patients who never received Zofran during an admission in that time period. The third group (BOTH) was patients who received Zofran on at least one admission but not all admissions.

Exhibit 2 compares costs and utilization statistics for the ALWAYS vs. NEVER groups. No statistically significant difference in the average cost per case for the ALWAYS group vs. the NEVER group was found, despite the fact that anti-emetic therapy for the ALWAYS group was 10 times more costly than for the NEVER group.

The high cost of Zofran for the ALWAYS group was offset by a reduction in average length of stay (2.3 days vs. 3.4 days). This difference in length of stay was significant at the 5 percent level. As the exhibit shows, the ALWAYS group also had lower utilization of various ancillary services.

Exhibit 3 shows the distributions of payer mix and cancer type for the ALWAYS vs. NEVER groups. Statistical tests confirmed that the two groups were not similar in terms of cancer site or payer. By comparing Medisgroup severity scores, it was determined that the two groups were comparable in terms of severity of illness. However, the differences in cancer site and payer mix might explain the differences in resource use observed.

TABULAR DATA OMITTED

Exhibit 3: Payer mix and cancer site distribution comparison of DRG 410
admissions (10/90-9/91)

                           Never received   Always received
Cancer site distribution       Zofran            Zofran

Bone and tissue                7.48%             4.41%
Colorectal                     3.74%             2.94%
Female genital                 4.67%             7.35%
Gastrointestinal              16.82%            13.24%
Leukemia/lymphoma             16.36%             4.41%
CNS                            0.50%             4.41%
Respiratory                   28.04%            39.71%
Skin                           2.80%             0.00%
Urinary                        3.27%             0.00%
Head and neck                  0.00%             2.94%
Other                          9.81%            20.59%
Unknown                        6.54%             0.00%

Payer mix distribution

Blue Cross                    27.10%            41.18%
Commercial                    16.82%            20.59%
Managed care                   4.67%             4.41%
Medicaid                       2.34%             2.94%
Medicare                      43.00%            29.41%
Healthpass                     1.40%             0.00%
Self                           0.00%             1.47%
CHAMPUS                        1.40%             0.00%
Unknown                        3.27%             0.00%

An analysis of a subgroup of patients with the same type of cancer would reduce any disease-related variation; the largest group was made up of patients with respiratory cancer. The analysis was repeated for this subgroup and the results showed that the ALWAYS group still had shorter lengths of stay, resulting in no significant difference in total cost per case between the ALWAYS and NEVER groups.

It was noticed during the analysis that a large number of patients in the NEVER group was admitted by one physician and that this physician's patients had longer lengths of stay than those for other physicians. The analysis was repeated, excluding this physician's patients.

The length of stay was still significantly lower for the ALWAYS group, although the absolute difference between ALWAYS and NEVER was less than it was in the original analysis. The length of stay within the respiratory cancer TABULAR DATA OMITTED TABULAR DATA OMITTED group, however, was not significantly different. This part of the analysis suggests that physician practice patterns are an important factor in explaining the shorter lengths of stay for Zofran patients. Unfortunately, the sample size was too small to repeat the analysis by physician.

The next part of the analysis focused on the group of patients in DRG 410 who had at least one admission with Zofran and others without Zofran (BOTH). Their admissions with Zofran (WITH) were compared to their non-Zofran admissions (WITHOUT). This comparison was done to minimize any effects that might be due to differences in cancer site, severity, or physician practices. The results are summarized in Exhibit 5.

As shown in the exhibit, there was a slightly shorter length of stay for those admissions where Zofran was used, but the difference was not statistically significant. As in the other analyses, the average cost-per-case was not significantly different for the WITH vs. WITHOUT admissions despite the higher cost of Zofran.

It is concluded that the introduction of Zofran at TJUH did not result in a significant increase in total hospital costs because many patients who received Zofran had shorter lengths of stay. This is an important finding for institutions where reimbursement for chemotherapy patients is on a fixed rate, per case basis.

Future applications

The results of this study were shared with the TJUH pharmacy and therapeutics committee and illustrate how the CFMS system could be used for a variety of clinical-economic analyses. The number of requests for CFMS system data, particularly from physicians, nurses, pharmacists, and other clinicians, has grown steadily as more and more departments realize the potential to utilize this information to evaluate their clinical practices.

As the number of new products grows, and cost containment pressures increase, the need for clinical cost evaluations also will increase.(f) Good cost accounting systems will help hospitals access better, more reliable cost data for the type of analysis described here.

a. Razaee, Z., "Examining the effects of PPS on cost accounting systems," Health-care Financial Management, March 1993, pp. 58-62.

b. Nemes, J., "Tight margins lead hospitals to cost accounting systems," Modern Health-care, December 17, 1990, pp. 22-27.

c. Krueger, D.J., and T.A. Davidson, "Alternative Approaches to Cost Accounting," Topics in Health Care Financing, Summer 1987, pp. 1-9.

d. Milne, R.J., and R.C. Heel, "Ondansertron: Therapeutic use as an antiemetic," Drugs April 1991, pp. 574-595.

e. "F-D-C Reports: The Pink Sheet," September 23, 1991.

f. Meyer, B.R., "Biotechnology and therapeutics: Expensive treatments and limited resources. A view from the hospital," Clinical Pharmacology & Therapeutics, April 1992, pp. 359-365.

Caryl E. Carpenter, PhD, is assistant professor, Widener University, Chester, Pennsylvania, and a member of HFMA's Metropolitan Philadelphia Chapter.

Linda C. Weitzel, MBA, is director of financial planning, Thomas Jefferson University Hospital, Philadelphia, Pennsylvania, and a member of HFMA's Metropolitan Philadelphia Chapter.

Nelda E. Johnson, Pharm.D, is project director, health policy & clinical outcomes, Thomas Jefferson University Hospital, Philadelphia, Pennsylvania.

David B. Nash, MD, MBA, is director, health policy and clinical outcomes, Thomas Jefferson University Hospital, Philadelphia, Pennsylvania.