Engaging physicians in cost savings initiatives: eliminating medication cost bias in the selection of inpatient medications.

Higher cost medications or therapeutic regimens pose a serious question for physician executives and other hospital managers who must determine if the cost/benefit ratio of a newer technology drug helps the organization achieve its value proposition of providing optimal care.

Even when a hospital has robust data, engaging physicians to help with these decisions is often an elusive goal. The stakes are increasing as we move from routine antibiotic choices to evaluation of new, and often expensive, biotechnology-based medications or devices, some with costs that can run into thousands of dollars.

Hospitals typically approach the problem of increasing drug costs with a silo-structured analysis called a drug utilization review (DUR). (1) DUR is inefficient and ineffective because it fails to address several important issues including complications, length of stay, resource consumption, clinical outcome and satisfaction.

In response to the limitations of the DUR, we developed a methodology using each participating hospital's commonly available administrative databases to provide a detailed comparison of the total costs of managing similar patients receiving either low molecular weight heparin (LMWH) or unfractionated heparin (UFH).

The administrative databases we used were maintained primarily to support key administrative functions, including registration and billing, staff management, materials management, accounting and financial decision support. Clinical relevance was added by combining this data with clinical data from the UB-92 (2) and the hospital's pharmacy information system. The data elements are displayed in Table 1 (pg. 16).

The hope of many senior hospital managers--that these administrative databases could also be used to win physician collaboration in the hospital's efforts to control expenses through clinical guideline development--remains largely unfulfilled.

Explanations for this failure are various, and are most often laid at the feet of the physicians, citing their overarching demands for autonomy, authority and income preservation. Our experience suggests that other factors are more important and that the explanation is more complicated.

Methodology and findings

From 2001 to 2003 we studied the use of the two important anticoagulant medications for the prevention and treatment of venous thromboembolic disorders (VTE) in 15 hospitals across the U.S. (3)

The study used commonly available, hospital-specific administrative databases and a detailed application of the cost/charge ratio at the individual cost center level to compare the total cost of treating inpatients with relatively inexpensive UFH (about $1.00 per day to purchase) to the total cost of treating similar patients with LMWH (about $26.00 per day).

This comparison is highly relevant because VTE is a ubiquitous problem (4) with substantial morbidity, mortality and high costs and because LMWH has grown to become one of the top five line item expenses in acute care hospital pharmacies. (5) We also interviewed physicians and other key stakeholders to determine their attitudes about the use of the target drugs and how they approach the adoption of new drugs and practice patterns.

For each participating hospital, we looked at all inpatient admissions for a period of two years or more and determined in which diagnosis-related groups (DRGs) it was more cost-effective to use LMWH and in which it was more cost-effective to use UFH.

The data were formally reported to each hospital in a format designed to facilitate use in the hospital's quality management (QM) processes. All participating hospitals demonstrated substantial savings using LMWH that were not anticipated based on the difference in acquisition costs.

All participants demonstrated the potential for additional future savings in many, but not all DRGs and patient categories. (6) The findings are summarized in Table 2 (7) (pg. 17).

Clinical interviews

We conducted interviews at 14 of the 15 hospitals prior to data analysis. The interview process assumed that properly selected key stakeholders could identify the factors that facilitated or limited the institution's ability to achieve clinical effectiveness (8) in the use of these medications.

Some questions were identical for all interviewees but others were tailored for the various disciplines interviewed (clinicians, quality management personnel, senior managers and information services employees). Medical and quality management leaders at all 15 hospitals indicated that their hospital had not achieved an optimal level of clinical effectiveness in the use of UFH and LMWH, sometimes despite considerable effort and expense.

The process of modifying prescribing patterns for inpatient anticoagulation often begins with pharmacy managers or administrators asking why the less expensive alternative, UFH, cannot be used more frequently and LMWH less frequently.

Physicians respond that medication cost can only be evaluated in the broader context of risks and complications, patient satisfaction, staff satisfaction, cost of administration and, most importantly, outcomes like length of stay, morbidity and mortality. As a result, the total cost of care associated with a drug, not its acquisition cost, is the important metric.

Even when cost issues can be resolved, clinical resource management programs remain ineffective at engaging physicians in collaborative efforts to reduce or control hospital expenses by changing practice patterns.

Our interviews of physicians, quality management personnel, nurses, administrators and managers identified eight barriers associated with physician resistance to modifying their practice patterns (see Table 3.) Here's a look at five of the most significant:

[ILLUSTRATION OMITTED]

1. Time management

The impact on physician time management was the most frequently cited barrier to changing practice patterns, and the most adamantly expressed. The progressive increase in mandatory non-revenue-producing activities and the inexorable reduction in compensation for each unit of service over the past 20 years combined to produce intense physician scrutiny of any activity that threatens to reduce compensated productivity further.

Consequently, physicians will not easily give up use of an older, more familiar drug, like UFH, to learn the intricacies of a newer one, like LMWH, and to explain them to nurses, patients and family members. Observation and logic suggest that this barrier will continue to worsen until physicians adopt tools and workflow strategies that permit them to work more efficiently.

[ILLUSTRATION OMITTED]

2. Autonomy

Issues of autonomy and authority, cited by participants almost as frequently as time constraints as barriers to change, are important but are not the "hot button" they were 10 or 20 years ago.

Physicians who began practicing when absolute physician authority was infrequently questioned are turning the reins of medical staff leadership over to a younger generation. These new opinion leaders distrust institutions and hospital administrators even more than those who preceded them, but they do not bristle as much at the idea of justifying their decisions or at participating as members of multi-disciplinary care teams, if time allows. They are beginning to share decision-making authority with other members of the multi-disciplinary care team.

[ILLUSTRATION OMITTED]

3. Habits and knowledge

Habit and familiarity were also cited as barriers to the adoption of newer, more clinically effective medications and care processes. Many interviewees differentiated these barriers from time constraints, indicating that the comfort of doing the familiar helped to offset the relentless increase in apprehension felt by many physicians in response to lawsuits, regulations and managed care scrutiny.

4. Knowledge deficit

Habit and familiarity are also linked to the fourth barrier, one we termed "knowledge deficit." UFH has been part of the therapeutic armamentarium for more than 60 years, but only in the past few years has the American medical community united around the concept of a weight-based protocol for the administration of therapeutic intravenous UFH. Furthermore, there is still controversy over the appropriate dosing interval for subcutaneous prophylactic UFH.

LMWH has been available for about 10 years and there are several different preparations available. They differ by FDA-approved indications and dosing; monitoring is not readily available and there is not yet agreement over use in chronic renal insufficiency.

Consequently, many physicians have chosen not to adopt LMWH fully, even when the clinical evidence is clearly in its favor. This decision is both a result of, and a cause of insufficient knowledge about LMWH among some physician groups.

[ILLUSTRATION OMITTED]

5. Data perceptions

The fifth barrier--the perception of inadequate or unreliable data about clinical practice patterns--was difficult to understand initially, since many of the participating hospitals routinely provided physicians with detailed information about their clinical resource utilization.

This apparent paradox was clarified when we presented their data to several of the participants. With the exception of physician executives and a few clinicians involved in their hospitals' quality management programs, the physicians we worked with were not familiar or comfortable with the use of administrative databases.

Today, evidence-based medicine emphasizes randomized, clinical trials and consequently, clinicians look suspiciously at data from administrative databases. They are confused by the data's lack of specificity, by the uncontrolled variables and by the need to work from the generalized data towards the specifics of care for a group of patients.

The "controls" are often thousands of other patients, cared for at distant institutions, and, the issue of comparability inevitably arises. As long as this problem of physician perception of administrative data persists, the data itself will remain a barrier to the acceptance of clinical resource management efforts.

Each of these five--time constraints, autonomy and authority, habit and familiarity, knowledge deficit and lack of familiarity with administrative databases among physicians--is a barrier that limits the effective use of clinical resource management data to support the development of guidelines and protocols and to change physician inpatient prescribing patterns.

Recommendations

Our data demonstrated that the newer, more costly medication, LMWH, was frequently more cost-effective than the less expensive, older alternative, UFH. Savings varied by hospital, DRG and clinical category, but the trends were strong and statistically significant.

Quality management personnel and administrators understand that only clinicians can decide how to treat patients, but the quality management team can offer clinical resource management data as an adjunct to the process of making those decisions.

If physicians work with others on modifying their own practice patterns, time management must always be in the forefront--both operationally and as a goal. Most physicians reject complicated or multi-page guidelines and they will not support or use a hospital Intranet that requires them to wait to access a terminal, endure a lengthy sign-on process and then toggle through numerous pages to get useful information.

Specific order sets and evidence-based treatment protocols are time savers and more appealing tools than multi-page clinical pathways, Critical pathways that focus on a crucial, but short segment of the admission are also more physician friendly and probably achieve more medical management goals than detailed, day-by-day guidelines.

The hospital, responsible for care delivery processes, also plays an essential role in how physicians embrace resource management. Physicians hampered by inefficient manual systems are not good partners in cost-containment initiatives (and the same is true for nurses, pharmacists, technicians and other hospital employees).

Fewer secretaries, limited hours and staffing, delayed orders and phone calls from nurses who don't know patients well enough to transfer information effectively are among the problems that result from financial cutbacks. While these management strategies reduce the hospital's expenses they also shift responsibilities to physicians, who find themselves doing secretarial work or making decisions based on incomplete information.

In response, physicians wait for all the tests to be completed, for the results to be posted to the chart and for pre-discharge education to be completed. While they wait, lengths of stay and resource utilization increase.

Administrators need to increase resources for development of automated systems that improve ordering, scheduling, reporting and information exchange, and staff nursing units adequately to engage physician cooperation in achieving optimal cost effectiveness.

Physician autonomy and authority remain under attack by insurance companies, managed care companies, regulators and legislators. Because many physicians perceive no other venue in which to protest these changes, they often focus their frustrations on the hospital. As a result, our interviewees reported that abuses of physician authority and resistance to accountability were common while achieving real collaboration was uncommon.

Senior administrators who understand the medical staff bylaws and their mandate to ensure high quality care, and who hold medical staff leaders to them, enhance physician self-esteem. This encourages more effective collaboration, through which more cost-effective, quality care is ultimately achieved.

Habit, familiarity the "knowledge deficit," and their impact on medical management practices, are best considered together. Education is the obvious remedy; how to educate is the challenge. Time-honored methods like grand rounds, department conferences and half-day seminars have diminishing impact in these times of high-speed practice.

Newer tools are beginning to supplant them. They share the common elements of simple, random accessibility ("just in time" learning); variations in detail (choose a table, summary or more comprehensive information), privacy (some attending physicians are uncomfortable learning how to use a medication in the nursing station) and timeliness (they are kept up to date, sometimes with data from the hospital itself).

Hospitals have invested substantially in administrative management systems but most have seen little return on those investments in terms of more effective medical management programs. One common early strategy--training physician "superusers" on the system--brought little success, judging by the hospitals we studied.

Hospitals may achieve greater success in clinical resource management by providing basic knowledge about administrative databases and teaching an understanding of how such data is used legitimately. Hospitals could then establish incentives that bring a substantial majority of physicians face-to-face with these data regularly.

For example, if radiologists want a new angiography suite, an effective approach would be to show them not only the breakeven analysis, but also how to use administrative data to develop care strategies that improve the investment's financial projections.

Physicians and administrators all have a stake in ensuring the most clinically effective and cost-effective management of their patients. The complicated task of choosing between a newer, more expensive medication and an older, less expensive medication is one of the most common situations in which this problem plays out.

Analysis using the clinical resource management process we developed gives both clinicians and administrators the tools they need to make truly informed decisions-based on clinical evidence and cost benefit analysis.

Table 1 -- Data Elements

A. From Uniform Billing System (Core Patient Data)

Hospital identifier
Patient identifier (a masked or encrypted number unique to each patient
(i.e., medical record number) to allow identification of readmissions
and, if needed, to allow a hospital to retrieve records)
Admission identifier (a masked or encrypted number unique to each
hospital stay; i.e., billing or patient control number)
Admission and discharge dates
Patient date of birth
Patient sex
Birth weight, in grams (for neonates only)
Admission source and type (UB-92 codes)
Discharge status (UB-92 codes)
Principal and 14 secondary diagnoses (including E-codes)
Diagnosis present at admission
Flag (where available) for each secondary diagnosis
Principal and 14 secondary procedures (ICD-9-CM procedure codes and
dates)
Primary payor
Attending physician ID and specialty
Operating physician ID and specialty
Total charges
Additional data items a hospital would want included in the database
returned to them

B. From Pharmacy and/or Billing System (detailed billing data for all
departments)

Admission identifier for linking to the Core Patient Data for the same
admission
Charge code
Quantity
Charges
Date of service

C. Medicare Cost Report

An electronic copy ("PI" file) of the hospital Medicare cost report
(HCFA 2552) for the years corresponding to the patient data, or the most
recent two years available, for estimating hospital costs for these
patients.

Table 2

Aggregate Data Summary
Time Period: 4Q 1999-IQ 2003
Cost of Inpatient Care (2002 dollars): > $5.75 Billion
Number of Participating Hospitals/Health Systems: 15
Number of Inpatient Admissions in Database: > 720,000
Number of Patients Receiving UFH: 90,846
Number of Patients Receiving LMWH: 933,656

Category              # Of  # Of Pts       Less expensive   Potential
                      DRGs  (UFH/LMWH)     (UFH or LMWH10)  for
                                                            Additional
                                                            Savings (Y
                                                            of N)

All Patients          174   90,846/33,656  LMWH             Y
Hip Replacement         1      410/2,485   LMWH             Y
Knee Replacement        1      325/2,371   UFH              N
Unstable Angina        10    8,278/2,204   LMWH             Y
DVT-PE                  3      870/406     LMWH             Y
Abdominal Surgery      16    6,290/1,382   UFH              Y
Medical At-Risk (10)   44   17,852/10,063  LMWH             Y
Medical
Cardiovascular         19   14,300/8,920   LMWH             Y
Surgical
Cardiovascular         14   30,340/964     LMWH             Y
Other Orthopedic       12    1,503/3,123   LMWH             Y
Rheumatology            6      748/652     UFH              Y
Medical
Abdominal (GI)          9    2,667/1,213   LMWH             Y
Respiratory            14    4,866/3,489   LMWH             Y
Neurology               9    3,858/1,305   LMWH             Y
Dermatology             5    1,169/758     LMWH             Y

Table 3

Barriers to Changing Practice                Frequency

Time Management                                12/14
Autonomy and Authority Issues                  11/14
Habit and Familiarity                          11/14
Knowledge Deficit (regarding LMWH)              9/14
Insufficient/Unreliable Data About Clinical
Practice Patterns in the Hospital               6/14
Higher Acquisition Cost of LMWH                 4/14
Tradition                                       3/14
Cross-Cultural Stress"                          2/14

References

1. John GW, Spieler JL Jr. Drug utilization review: A practical approach. Hosp Pharm, 1981, 16(11), 587-90, 595-8.

2. Accepted for publication elsewhere.

3. This study, the Clinical Effectiveness Initiative (CEI), was developed and carried out by Aon Consulting's Life Sciences Consulting Practice under an unrestricted grant from Aventis Pharmaceuticals. The process is not presently commercially available.

4. Anderson FA Jr, Wheeler HB, Goldberg RJ, Hosmer DW, Patwardhan NA, Jovanovich B, et. al. A population-based perspective of the hospital incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. The Worcester DVT Study. Arch Intern. Med. 1991; 151:933-8.

5. Top 200 Prescription Drugs for Hospitals in 2002. Drug Topics, 2003, 7, 29.

6. Detailed results are in preparation for publication.

7. Enoxaparin, only. Insufficent data for other LMWHs.

8. Doing the right thing for the patient, at the right time, in the right setting, for the right reasons, using the right resources to get the right results.

9. Enoxaparin, only. Insufficient data for other LMWHs.

10. Longer-stay medical DRGs.

11. For example, following the merger of two medical staffs, one allopathic and the other osteopathic: the leadership group was unwilling to promote the use of guidelines for fear that the development of guidelines would be interpreted by one group as the other group dictating how to practice medicine.

RELATED ARTICLE: Details of the Study

For each participating hospital, CEI looked at all inpatient admissions for a period of two years or more. Costs were determined by summing each patient's charges at the revenue code level and applying a cost/charge ratio at the cost center level after ensuring that all charges (expenses) and costs were properly aligned.

The composite cost of each case included all direct and indirect expenses. Cases were collected by DRG and were aggregated into cohorts that make sense to practicing physicians--i.e., into groups corresponding to FDA indications for LMWH and other clinical groups in which anticoagulants are commonly administered. A cost comparison was made in each DRG in which at least ten 10 patients received UFH and 10 received LMWH, and the difference was tested for statistical significance using standard tests.

Differences (i.e., "savings" or "losses" using LMWH in lieu of UFH) were reduced by 50 percent (an estimate of variable cost) and summed for the number of cases receiving LMWH. Annualized savings and future projections were also calculated.

Quality management and clinical personnel looking at the data could identify savings or losses already achieved and could analyze those savings/losses by key cost center. Comparison to the aggregate national database facilitated the identification of individual management benchmarks as well as opportunities to improve costs through operational efficiency and changes in practice patterns.

By Richard M. Weinberg, MD, CPE, Nimisha Vyas Bhatt, MPH and F. Randy Vogenberg, RPh, PhD

Richard M. Weinberg, MD, CPE, is medical director, quality improvement and patient safety officer, UMDNJ-University Hospital in Newark, N.J. He can be reached at 973-972-6780 or [e-mail].

[ILLUSTRATION OMITTED]

Nimisha Vyas Bhatt, MPH, is assistant vice president of Aon Consulting, Life Sciences Practice, a world leader in employee benefits, compensation, communication and management, human resources outsourcing, and health care and biotech consulting. Based in Somerset, N.J., she can be reached at 732-271-2670 or [e-mail]

F. Randy Vogenberg, RPh, PhD, is senior vice president of Aon Consulting, Life Sciences Practice, a world leader in employee benefits, compensation, communication and management, human resources outsourcing, and health care and biotech consulting. Based in Providence, R.I., he can be reached at 401-553-6609 or [e-mail].

[ILLUSTRATION OMITTED]