Hormontablette (Östrogen und Gestagen = kombinierte Pille). Eine Packung enthält in der Regel 21 Tabletten, von denen täglich eine ungefähr zur gleichen Zeit einge-nommen wird. Danach folgen in der Regel sieben Tage Pillenpause. In dieser Zeit tritt normalerweise eine Blutung ein. Der Empfängnisschutz bleibt in den pil enfreien Tagen bestehen. Die 1. Pil e der ersten Packung wird am ersten
6a-r_knoer.pmdReports Benchmarking database
Evaluating a benchmarking database and
identifying cost reduction opportunities
by diagnosis-related group
SCOTT J. KNOER, RICHARD J. COULDRY, AND TANYA FOLKER Abstract: Pharmacy cost
Index terms: Acyclovir; Ad-
Am J Health-Syst Pharm.
ospital administrators are under constant pres- Several widely used benchmarking databases have been sure to find new ways to more efficiently use approved by JCAHO as mechanisms for monitoring Hresources. Critical pathways, formulary manage- outcome-related data. HBS International (HBSI)7 has
ment, and productivity consultants have all been used been accepted by JCAHO, and MECON-PEERnext8 is to reduce pharmacy costs, with varying degrees of being reviewed for acceptance (Anderson-Miles E, MECON, personal communication, 1999 Apr 27).
Benchmarking, which involves identifying best There is substantial interest on the part of pharmacy practices and the components that lead to them, has administrators nationwide in having accurate data that gained popularity as a method for assessing an organi- can be used for organizational comparisons. Questions zation’s productivity. This tool allows hospitals to com- as to the validity of the data invariably arise as manag- pare their practices with a benchmark performer and to ers are held accountable for the numbers presented on examine the internal processes that result in a given behalf of their departments in benchmarking products.
outcome.4 One appropriate benchmarking approach is These questions can lead managers to explore the to undertake a comparative analysis of organizations recognized as leaders in the area identified for improve-ment.5 Available pharmacy benchmarking
The Joint Commission on Accreditation of Health- databases
care Organizations (JCAHO) has recently incorporated Several large databases are currently available or in benchmarking into its hospital survey. JCAHO’s Oryx the process of being developed. In addition to MECON- program, which was introduced in February 1997 and PEERnext and HBSI Action, Hospital Pharmacy Data began in June 1998, requires that performance mea- Quarterly (HPDQ) (Aspen Publishers, Gaithersburg, sures be monitored as part of the accreditation process.6 MD) and the University HealthSystem Consortium SCOTT J. KNOER, PHARM.D., M.S., is Pharmacy Operations Manager, Texas Medical Branch at Galveston, Suite 1.100, Clinical Sciences University of Texas Medical Branch at Galveston. RICHARD J.
Building, 301 University Boulevard, Galveston, TX 77555-0701, COULDRY, M.S., is Assistant Director of Pharmacy and Clinical Assistant Professor and TANYA FOLKER, PHARM.D., is Clinical Phar-macist, University of Kansas Medical Center, Kansas City.
Copyright 1999, American Society of Health-System Phar- Address reprint requests to Dr. Knoer at the University of macists, Inc. All rights reserved. 1079-2082/99/0601-1102$06.00.
1102 Am J Health-Syst Pharm Vol 56 Jun 1 1999
Benchmarking database Reports
(UHC) Clinical Database are among the benchmarking Use of the UHC Clinical Database at KUMC
products available to hospital administrators.
Resource-use information from the UHC Clinical MECON is a consulting firm that specializes in Database was recently circulated to managers and prac- health care productivity improvement and cost reduc- titioners in the various cost centers at KUMC. Hospital tion. MECON-PEERnext is a Web-based system that administration initially focused on DRG 302 (kidney evolved from MECON-PEERx. It contains operational transplant) to identify potential cost savings for the and financial data for more than 650 hospitals in the medical center. The pharmacy department performed United States.7 This database attempts to standardize an in-depth validation of the pharmacy costs reported pharmacy workload by assigning weights to various by UHC for this DRG and identified cost-cutting oppor- pharmacy tasks. A multiplier assigns a relative weight to tunities by contacting the hospitals listed as top-per- various pharmacy workload indicators and results in forming hospitals in the UHC database.
the pharmaceutical care unit (PCU). Hospital data are The UHC Clinical Database listed the average KUMC then categorized into a standard format, and compari- pharmacy cost per discharge for DRG 302 as $8546. This son data are shared with reporting organizations. Data cost is substantially above the $2764.60 average of the from similar institutions, the upper 25th percentile, five top-performing institutions as reported in the UHC and the lower 75th percentile are listed for comparison.
Clinical Database for FY96 (Table 1).
HBSI Action is a health-system benchmarking data- Identifying a cost reduction goal required verifica- base similar to MECON-PEERnext. It contains financial tion that the average accountable costs in the database and operational information on more than 600 hospi- were valid. It is imperative that administrators use accurate data when making decisions that can affect There are several challenges associated with translat- staffing, new programs, and, ultimately, departmental ing benchmarking data into useful information. Al- performance. A project was developed to compare actu- though attempts are made to standardize data entry, al pharmacy costs for DRG 302 with the costs shown in categories may be ambiguous and subject to different the UHC Clinical Database. The project was expanded interpretations. Procedural workload variations be- to include assessment of appropriate drug use for DRG tween institutions add ambiguity to the data collection 480 (liver transplant) to find opportunities to reduce techniques of MECON. The variety and quantity of data pharmacy costs and to ensure appropriate medication prevent presentation in easily comprehended graphs or tables, and some data indicators (e.g., doses billed perpatient day) have limited practical utility. In the context of reimbursement for specific diagnosis-related groups The project involved analyzing the data collection (DRGs), the number of doses billed does not necessarily techniques used by UHC, analyzing KUMC pharmacy correlate to revenue collected by the hospital.
cost and charge data for DRG 302, analyzing kidney Aspen Publishers recently mailed 2000 surveys re- transplant patients for appropriate drug use, seeking questing facility and workload information to pharmacy existing critical pathways or guidelines for this patient directors for use in HPDQ. Aspen believes that pharmacy population, contacting the top-performing institutions administrators are interested in obtaining pharmacy- to compare drug use, working with the hospital team specific benchmarking data and that a database pro- (including physicians, nurses, care coordinators, and duced by a neutral third party rather than a consulting others involved with guideline development) to opti- firm will be more appealing to pharmacy managers.10 mally use pharmaceuticals, and presenting the results The first HPDQ report was generated in July 1998 and to hospital administration and relevant physicians.
contained data derived from 100 hospitals. Future sur- An understanding of the data collection techniques veys can be mailed in or completed on the Internet.
UHC, a buying group composed of 70 university hospitals, has expanded its efforts into the benchmark- ing arena. The UHC Clinical Database for fiscal year Pharmacy Cost Information for University of Kansas
1996 (FY96) contains data on 60 member hospitals.
Medical Center (KUMC) and Other Hospitals Taken
from University HealthSystem Consortium (UHC)
Cost information for these 60 hospitals is broken down Clinical Database
by cost center and reported for the top 20 DRGs. Ex- penses for these cost centers are reported for each member hospital, and comparisons with the five top- performing hospitals (in terms of cost per DRG) for each During the research for this project, the University of Kansas Medical Center (KUMC) used MECON-PEERx and the UHC Clinical Database in an effort to reduce aDRG 302 = diagnosis-related group 302 (kidney transplant).
Vol 56 Jun 1 1999 Am J Health-Syst Pharm 1103
Reports Benchmarking database
and formulas used by UHC to arrive at KUMC cost data housekeeping, and cafeteria are the major categories of was necessary to assess the validity of the costs shown overhead that the hospital allocates to each cost center.
in the UHC Clinical Database. UHC headquarters was Total pharmaceutical costs for FY94, the year for consulted by telephone to determine UHC’s methods which UHC determined pharmacy CCRs for its FY96 of data collection. UHC provided cost-to-charge ratios UHC Clinical Database, were $22,428,614. Adding (CCRs) and wage-adjustment formulas upon request.
overhead increased costs to $27,764,594, and reported The KUMC billing office and the pharmacy informa- charges were $39,771,338 for this period. This gave tion system specialist were consulted to ascertain phar- KUMC a CCR of 0.6981 as filed with HCFA. The audited macy billing procedures. The information specialist HCFA CCR used in the UHC Clinical Database was also provided a list of pharmacy charge formulas, which 0.6744, while the actual CCR for KUMC before over- were used to compare KUMC’s pharmacy CCRs with head was added was 0.5639. UHC then applied a wage- those reported by UHC. Lists of patients in DRGs 302 adjustment factor to the cost data associated with each and 480 were obtained through queries of the hospital hospital (Table 1). KUMC had the second highest wage- information system, and charges were queried directly adjustment factor (0.9538) among the hospitals with by pharmacy personnel through the pharmacy com- which it was compared. Determining a CCR from HCFA puter system. The financial office at KUMC provided data is a complex process involving multiple steps of fiscal data for FY94–96, and critical pathways for kidney data manipulation. Each time this number is manipu- and liver transplants were obtained from the respective lated, potential error is introduced into the data.
departmental manuals. The five top-performing hospi- The final CCRs used by UHC in analyzing pharmacy tals in terms of cost per DRG were selected by UHC and data among peer institutions are shown in Table 1. A are shown in the UHC Clinical Database for bench- high CCR inflates the reported costs of the institution.
marking comparisons. Transplant pharmacists and The CCR for KUMC is considerably higher than that of pharmacy administrators were contacted by telephone any of the hospitals with which it was compared.
To verify the validity of the CCR used by UHC, the actual CCR was calculated for all patients in DRG 302. A list of all kidney transplant patients from FY96 was It was determined that the cost information reported generated from the hospital information system, and in the UHC Clinical Database came from charge data two separate reports were compiled from the pharmacy generated by KUMC at the revenue code level. To computer system. The reports listed all pharmacy costs understand how charges are derived, it is necessary to and charges generated for each patient. The actual and review how a drug is charged at KUMC. The pharmacy UHC-reported CCRs for DRG 302 for FY96 at KUMC are computer system automatically receives updated acqui- 0.461 and 0.647, respectively. This translates into a sition costs from the KUMC wholesaler whenever con- 28.74% difference between the CCR estimated by UHC tracts are updated. Unique revenue codes are attached and the CCR that was calculated from actual patient to each product listed in the computer system; these codes are then grouped into classes based on drug The next step in the process was to identify medica- category. When orders are entered, a charge formula is tions that could significantly affect the costs of provid- applied to the product on the basis of its class and a ing pharmaceutical services to patients in DRG 302.
charge is generated (Table 2). This charge amount is Twenty-four kidney transplants were performed at then downloaded to the hospital information system KUMC in FY97. The medication profiles for these 24 for billing purposes; this is the level of data that UHC patients were obtained for review. FY97 patients were receives. This information is then converted back to chosen so that the most recent therapy trends could be cost data by applying a CCR (0.647 for FY96 for KUMC identified. Several factors affecting reported pharmacy costs were discovered, including high use of lympho- UHC derives its CCRs from information provided to cyte immune globulin (Atgam, Pharmacia & Upjohn), the Health Care Financing Administration (HCFA) onthe HCFA Cost Report Worksheet C, Part I, columns 1–9.
Potential for error in UHC data was identified because Table 2.
Components of University of Kansas Medical Center
UHC used CCRs from FY94 to analyze data for FY96.
Pharmacy Charge Formulaa for Various Products
Investigation of CCRs revealed that pharmacy costs were not strictly based on variables controlled by the phar- Overhead is allocated to the various hospital cost centers (e.g., pharmacy) and is factored into the CCRs reported to HCFA for KUMC (20% of actual costs for FY96). Capital, benefits, number of telephones, pur- chasing, plant operation, administration and general, aPharmacy charge formula = (Cost × Multiplier) + Fee.
1104 Am J Health-Syst Pharm Vol 56 Jun 1 1999
Benchmarking database Reports
problems with billing discharge medications, and dif- were therefore rewritten to exclude IVIG dosing in the ferences in the classification of anesthesia agents by An investigation of lymphocyte immune globulin Pharmacy is only one of 35 cost categories reported protocols revealed a similar disparity between the usage in the UHC Clinical Database. No anesthesia costs were patterns of KUMC and the top-performing hospitals.
reported to UHC by KUMC, although all other hospitals Two of the top-performing hospitals for DRG 302 (hos- reported costs averaging $564 for this category. A re- pital A and hospital B) followed noninduction protocol view of charges for kidney transplant patients revealed guidelines for their kidney transplant patients. Lym- that anesthesia drugs accounted for $178 worth of phocyte immune globulin was used only if three days of charges at KUMC. Multiplying this number by UHC’s high-dose steroid therapy had failed. The KUMC kid- CCR of 0.647 yields a cost of $115. For an accurate ney transplant protocol calls for lymphocyte immune comparison with other institutions, anesthesia costs globulin to be given if a patient is anuric or if no drop in should have appeared in a separate section of the UHC serum creatinine (SCr) concentration is seen within 48 report rather than under pharmacy costs.
hours of the transplant, and cyclosporine is held until a A review of patient profiles showed that KUMC drop in SCr concentration is seen. Once cyclosporine is reported costs for discharge medications averaging restarted, daily trough concentrations are drawn for the $1839 on the inpatient bill. Conversations with repre- entire patient stay. Lymphocyte immune globulin costs sentatives of the five top-performing hospitals revealed and cyclosporine laboratory costs were much higher at that KUMC was the only hospital that charged dis- KUMC than at hospital A or hospital B. Intravenous charge medications to the inpatient bill. This account- medication costs for kidney transplant patients at ed for a 21.5% inflation of inpatient pharmacy costs as KUMC with or without lymphocyte immune globulin were $73,028 and $19,662, respectively. Communica- KUMC also charged patients $135 for self-medica- tion with best-practice hospitals resulted in KUMC tion teaching. Conversations with representatives of identifying a potential $53,000 cost saving related to the top-performing hospitals revealed that other hospi- tals did not charge for this service. Applying UHC’s CCR It was also discovered during the guideline review of 0.647 to this charge yields a calculated cost of $87.
that patients in DRG 480 were receiving cyclosporine This cost did not appear on the charge summaries of immediately postoperatively by the nasogastric route; comparison hospitals in the database.
however, these patients were also receiving i.v. acyclo- Factoring out anesthesia, discharge medications, vir and azathioprine. When this was brought to the self-medication teaching, and an inappropriate CCR attention of the liver transplant team, guidelines were gives a more accurate picture of pharmacy costs at rewritten so that azathioprine would be given nasogas- KUMC. Table 3 shows that the actual cost of medica- trically immediately postoperatively to patients who tions for DRG 302 was 46% lower than the cost reported can tolerate oral or nasogastric medications, and acy- clovir was taken out of the guidelines altogether.
Because of the practice overlap of transplant phar- macists among kidney, liver, and bone marrow trans- Discussion
plant patients, we made several observations about Calculated pharmacy costs for DRG 302 at KUMC pharmaceutical use in populations other than DRG ($4635) were 46% lower than those quoted by UHC 302. Conversations with pharmacists at the top-per-forming institutions confirmed that there was a dispar-ity between the amount of both intravenous immune globulin (IVIG) and lymphocyte immune globulin used Costs of Medications for Diagnosis-Related Group
by KUMC and the top-performing hospitals. Protocols 302 (Kidney Transplant)a
involving IVIG for liver transplant patients were re- viewed, and guidelines for use of lymphocyte immuneglobulin in kidney transplant patients at KUMC were analyzed on the basis of clinical practice information This analysis revealed that the KUMC liver trans- plant protocol included treating all patients with cy- Difference (28.74%) between calculated and tomegalovirus infection with IVIG 500 mg/kg every other day for two weeks (245 g for a 70-kg patient). At a aUHC = University HealthSystem Consortium, CCR = cost-to-charge cost of $15 per gram, a total of $3675 is incurred per ratio, KUMC = University of Kansas Medical Center.
patient. Conversations with practitioners at the top- bAdjusted total costs = total costs less anesthesia costs, self-medica- performing hospitals revealed that IVIG use for these tion teaching costs, and discharge medication costs.
cActual costs = adjusted total costs less difference between calculated patients was not a standard practice. KUMC guidelines Vol 56 Jun 1 1999 Am J Health-Syst Pharm 1105
Reports Benchmarking database
($8546). Although it is not possible to determine the actual costs for the comparison hospitals in the data- There are also ambiguities in the way various products base, it is fair to assume that the numbers reported also are classified. KUMC was the only hospital that did not contain inaccuracies. Telephone conversations with report anesthesia agents separately. These products need the kidney transplant pharmacist at hospital B revealed to be uniformly reported in order to limit error in the that patients received cyclosporine free of charge in FY96 as part of a large investigational study. Free cy- Although the UHC Clinical Database did not accu- closporine would significantly reduce pharmacy costs rately reflect pharmacy costs at KUMC, it is not without for DRG 302. The considerable variation between actual merit. Undertaking an in-depth review of costs associat- and reported drug costs makes it difficult to draw mean- ed with various procedures can provide useful data. By ingful comparisons between organizations.
reviewing KUMC guidelines, contacting peer institu- A major source of error potential in the UHC data- tions, and questioning current drug therapy practice, base is the inability of the system to directly report cost substantial cost reduction opportunities were identified.
data. At KUMC, the pharmacy department is the only Discontinuing IVIG use in liver transplant patients could source of actual cost data for the products it provides to potentially save the hospital $121,275 a year. Changing patients. UHC does not have access to this information; lymphocyte immune globulin guidelines for kidney therefore the data it uses must undergo a complex transplants is associated with a potential cost saving of process of being transformed from cost data to charge data and then back to cost data. Error is introduced at This project provided objective data to the hospital that lymphocyte immune globulin use at KUMC is far Another source of potential error with UHC CCRs is greater than that at peer institutions. It also identified that the most current data are not being used; CCRs are opportunities for pharmacists to affect drug costs by based on HCFA data that are two years old. The integri- suggesting that azathioprine be given orally or nasogas- ty of cost data calculated from charge data is question- trically rather than intravenously on postoperative day 1 able at best. As managed care continues to expand, hospitals are increasingly reimbursed a fixed amount Although potential cost savings of $53,000 were iden- per DRG or receive a flat rate per member per month.
tified through better protocols for lymphocyte immune Under this system, the amount a hospital charges its globulin use, it is unclear what effect current KUMC patients has little bearing on actual revenues; therefore, lymphocyte immune globulin guidelines have on the total cost of care to the patient for DRG 302. It is possible Another source of concern with the CCRs reported to that KUMC has a lower readmission rate for acute rejec- HCFA is that they incorporate hospital overhead. Al- tion episodes, which could potentially lead to an overall though the actual cost of providing pharmaceutical ser- lower cost for these patients. Readmission data were not vices in a hospital includes hospital overhead, this infor- tracked; complex formulas for determining overall pa- mation is not presented in the reports circulated to the tient outcomes were beyond the scope of this study.
various departments. If hospital administrators do not Although the pharmacy department often makes sug- know how the costs in benchmarking databases such as gestions about drug therapy, the thorough analysis of MECON-PEERnext, HBSI Action, and the UHC Clinical specific populations by the entire hospital, as well as the Database are derived, they could make misinformed support of hospital administration, led to many pharma- decisions or set unreasonable goals for departments.
cy recommendations being incorporated into protocols.
The fact that each hospital has its own unique charge Benchmarking databases can provide an opportunity formulas adds even more confusion to the data. Many to re-evaluate procedures at an institution and can help noncontrollable variables are introduced during at- identify opportunities for cost reduction and improve- tempts to standardize data for different institutions. For ment in patient care. They can also facilitate networking example, not all pharmacies bill for i.v. tubing. KUMC by providing a list of peers at other institutions who can does not pay for its i.v. pumps; the hospital instead pays serve as invaluable sources of information.
a premium on i.v. sets when it purchases them and Identifying top-performing organizations and at- recoups these costs by charging a higher fee for all i.v.
tempting to learn from them in an effort to improve medications it dispenses. Upon surveying other hospi- organizational performance is a mandatory part of con- tals, we discovered that some institutions buy their tinued success in health care. As data collection tech- pumps and that this cost is not passed on directly niques improve and outcome models are established, through an increase of pharmacy charges.
benchmarking has the potential to substantially improve Other potential variables that reflect differences in the quality of care provided by health care institutions.
pharmacy department costs include responsibility forradiopharmaceuticals and contrast media. These are very Conclusion
expensive products that can add considerable cost to a Data in the UHC Clinical Database were not repre- department if they are included in pharmacy inventory sentative of pharmacy costs at the University of Kansas 1106 Am J Health-Syst Pharm Vol 56 Jun 1 1999
Benchmarking database Reports
Medical Center for DRG 302 (kidney transplant), over- stating pharmacy costs by 46%. However, benchmark- 5. Szeinbach SL. Benchmarking. Consult Pharm. 1993; 8:260-1.
6. Joint Commission on Accreditation of Healthcare Organiza- ing was found to be a useful tool for identifying oppor- tions Home Page [resource on World Wide Web]. URL: http:// www.jcaho.org/perfmeas/oryx/oryx_frm.htm. Available fromInternet. Accessed 1999 Mar 28.
7. HBSI Home Page [resource on World Wide Web]. URL: http:// References
hbsi.com/html/jcaho.htm. Available from Internet. Accessed 1. McAllister JC. Collaborating with re-engineering consultants: maintaining resources in the future. Am J Health-Syst Pharm.
8. MECON Home Page [resource on World Wide Web]. URL: http://www.mecon.com/mecon/abmecon/newsletter.htm.
2. Clare M, Sargent D, Rhodes M et al. Reducing health care Available from Internet. Accessed 1999 Mar 28.
delivery costs using clinical paths: a case study on improving 9. HBSI Home Page [resource on World Wide Web]. URL: http:// hospital profitability. J Health Care Finance. 1995; 21(3):48-58.
hbsi.com/html/action/htm. Available from Internet. Access- 3. Horn SD. Unintended consequences of drug formularies. Am J Health-Syst Pharm. 1996; 53:2204-6.
10. Hospital Pharmacy Data Quarterly survey. Aspen Publishers.
4. Gannon K. Benchmarking new route to excellence. Hosp Vol 56 Jun 1 1999 Am J Health-Syst Pharm 1107
(combined response) (lamivudine) (adefovir dipivoxil) (entecavir) 1 Strader DB, Wright T, Thomas DL, and Seeff LB. Diagnosis, management, and treatment of hepatitis C. 2 Seeger C, Mason W. Hepatitis B virus biology. Microbiol Mol Biol Rev, 2000, 64:51-68. 3 Scaglioni PP, Melegari M, Wands JR. Biologic properities of hepatitis B viral genomes with mutations in the precore prom