Reducing hospital lengths of stay.

There are six general ways in which physicians have been influenced to change their practice patterns: education, feedback, participation, administrative rules, financial incentives, and financial penalties. *

Education has been effective in changing physician behavior, but it is not as simple as providing physicians with information. Educational interventions are most successful if the physician perceives a need to change and if the education is provided in a face-to-face encounter with a respected professional colleague.

Feedback is suggested by the application of cybernetic theory; by providing feedback, the system will correct itself. Strategies using feedback are most successful if the data are individualized and delivered personally by a respected clinical leader and if there is consensus regarding appropriate standards.

If physicians participate in planning and implementing programs to change their behavior, the change is much more likely to be effective. Participation has worked in changing physician behavior in organizations in shich physicians are themselves heavily involved and invested in the success of the organization, such as in a group model HMO.

Examples of administrative rules designed to changed physicians' behaviors are limited formularies, mandated second opinions, preadmission screening, and required forms for referrals or for requests for service. Such administrative programs run the risk of creating a backlash from physicians. It is also possible that physicians will circumvent the new regulations by prescribing services that can substitute for the restricted ones.

Financial incentives have been used to appeal directly to the economic desires of physicians. Gatekeeper systems in which primary care physicians are financially rewarded for efficient care, programs that reward doctors in closed panel HMOs with bonuses for achieving surpluses or holding down costs, and direct financial incentives to reduce lengths of stay have all been tried with variable success. The arguments against using such incentives are based upon the fear that needed services will be withheld. Furthermore, Medicare has regulations that preclude hospitals from providing any financial incentives for their medical staffs to reduce lengths of stay.

Financial penalties have been tried in Canada, New Mexico, New York City, and Massachusetts and seem to be effective either as a threat, when combined with education, feedback or peer review pressure, or on their own account. Medicare appears to believe that penalties can be effective.

In two organizations, greatly separated geographically and quite different in many ways, I have been a party to significant attempts to alter physician behavior and reduce lengths of stay. The two organizations are compared in figure 1, below. Pacific Medical Center (PacMed), Seattle, Wash., the first of my examples, was, until 1981, a U.S. Public Health Service Hospital. In 1981, the Medical Center was transferred to local community governance under a newly chartered, not-for-profit, public authority of the city. Simultaneously, the medical staff became the first multispecialty group practice in the state, Pacific Health Associates of Seattle (PHASE).

About 80 percent of the patients at PacMed were military personnel and their dependents, predominantly retired. These federal beneficiaries had 100 percent6 health care coverage with no copayments, deductibles, or fees. We provided them care under a contract with the Department of Defense. Initially, this contract reimbursed the hospital at per-diem and per-visit rates approved by the state's rate-setting commission and payed the physicians on a fee-for-service basis. All physicians in PHASE were on a fixed salary, with no financial incentives for productivity or efficiency. We spent several months developing a productivity incentive program to take advantage of our fee-for-service environment but never implemented it because we anticipated our reimbursement arrangement would change to managed care.

Our goal by 1984 was to become an efficient group-model managed care system. This led us to spin off several satellite clinics and start an HMO. In

[TABULAR DATA OMITTED]

addition, by reducing hospital expenses, we could dedicate more of our resources to building our clinic system and expanding our patient care base.

The second organization is Saint Vincent Hospital, a large community teaching hospital affiliated with the University of Massachusetts Medical School in Worcester. The majority of the active medical staff of Saint Vincent Hospital, more than 150, are physicians employed by a large multisite, multispecialty clinic, Fallon Clinic. More than 70 percent of Fallon Clinic's patients and 40 percent of Saint Vincent patients are members of the Fallon Community Health Plan, a group-model HMO providing care to more than 120,000 enrollees in central Massachusetts. Fallon offers both a regular plan and a senior plan. The senior plan is a Medicare risk contract initiated as one of the first federal demonstration projects.

In addition to Fallon Clinic physicians, Saint Vincent's open medical staff includes solo practitioners; hospital-based, full-time faculty; and a 30-person group practice, PrimaryCare Physicians (PCP), distributed in small offices around Worcester. During 1990, PCP and Fallon Clinic merged, with no major changes in either their patient mix or their sites of practice. The Fallon and the Saint Vincent organizations actively pursued merger discussions, and, later in 1990, formally affiliated.

Twenty percent of the admissions and nearly 40 percent of the patient days at Saint Vincent Hospital are for Medicare patients under the DRG system. These patients are cared for by physicians from all segments of our medical staff: Fallon, PCP, and solo and faculty physicians. Two years ago, our Medicare length of stay was more than 13 days. Under prospective payment, this length of stay, long even by northeastern standards, was costing the hospital up to five million dollars per year. Concurrently, the Fallon senior plan, with an identical case mix severity, had a length of stay of under 9 days, demonstrating that it was possible to provide more efficient inpatient care to this patient population. Although we recognized that reducing Medicare lengths of stay would have a negative impact upon individual physician reimbursement, we set out to reduce them through the implementation of concurrent utilization review.

The contingency theory of management says that the best way to change an individual's behavior is contingent on the nature of the organization, the type of people involved, and the nature of the task. In general, because physicians deal with ambiguity and operate independently, a decentralized and participatory style is predicted to be the most successful approach.

At the Pacific Medical Center, a participatory approach was in fact adopted. All strategic decisions were made jointly by the hospital and the group practice. Physicians were equally represented at the Medical Center Sttering Committee. The directors on the PHASE board were predominantly practicing physicians. Joint board retreats were held yearly with the hospital board. Decisions to launch satellite clinics and start an HMO were jointly reached by the hospital and PHASE. The HMO was a joint venture between the hospital and the group practice. Physicians were involved in all resource allocation decisions that had an impact upon their practice. We used feedback about how the organization was doing overall. However, we never accumulated data about doctor-specific lengths of stay. We used education only to tell the whole organization how it was doing. We widely distributed quarterly utilization statistics and monthly financial data. We took collective pride in our accomplishments through frequent celebrations.

Figure 2, above, shows lengths of stay for PacMed, plotted yearly from 1982 through 1986. Significant reductions were achieved in both 1985 and 1986. This was after lengths of stay appeared to be creeping up or at least plateauing from 1983 to 1984, Remember that PacMed was under cost-plus and fee-for-service reimbursement through 1984.

Figure 3, right shows hospital growth during the same period. Between 1983 and the end of 1986 we saw a steady rise of users of the system, from fewer than 25,000 to more than 43,000 users--an increase of 80 percent. In 1985 and 1986, the hospital and group were reimbursed by the Department of Defense on the basis of the total number of users. We paid a lot of attention to the ratio of days per 1,000 users, shown in figure 4, right. This provided us with the best barometer of overall efficiency in the use of hospital beds. It was also considered a useful prognosticator of our future success with the HMO.

A participatory approach would be desirable at Saint Vincent Hospital but very difficult because the medical staff is so diverse. Feedback had been used previously, and most all of the physicians with prolonged lengths of stay knew who they were. However, their behaviors were not moving toward the norm. Attempts to educate the medical staff about the problem had been made, but to no avail. We had excellent data-reporting systems, both Medisgroups and CFIS. But the data were all being reported retrospectively, after the opportunity to influence physician behavior had passed. The utilization management nurses were attending "overstay" meetings with the social workers and the financial officers, and everybody was wringing their hands about "these damn doctors" who were keeping their patients too long.

When I took over the Quality Assurance/Utilization Management Department of the hospital in 1989, I hired a new director, a nurse with experience in clinic, HMO, and PPO settings but not in hospitals. I asked her, the head of medical records, and the head of social services to jointly design a concurrent utilization management system. Several months prior to this, we had held the first retreat with all the full-time chiefs and senior administrators of the hospital. One of the major goals that emerged from this retreat was to reduce lengths of stay. The chiefs had committed themselves to doing whatever was necessary to achieve this goal. The concurrent utilization program was implemented in April 1989. Beforehand, we obtained the concurrence of the executive committee of the medical staff and informed the hospital board and the medical staff about the program through presentations and memos.

It was really quite simple. Medical records personnel would place a bright red sheet in the chart stating the expected length of stay from the admitting diagnosis. The utilization review nurses would review all cases daily. If it appeared that a patient no longer needed to be in the hospital or if the diagnostic or therapeutic plans were slow in their implementation, the nurse reviewer would call the attending physician. Anything but a cooperative response was followed by a phone call from the chief of service. This was usually effective, but, in a few instances, my associate or I called the physician directly, asking what was causing the delay or why the patient was still in the hospital. (My associate was six feet two and, in addition to being chief of obstetrics and gynecology, was still active with the Airborne U.S. Army Rangers; we sometimes called him "Rambo.")

This intervention is a combination of an administrative and feedback approach. Unlike with previous attempts to use feedback, the physician was presented with a situation in real time and asked to respond promptly. The message was delivered by the utilization review nurse, the chief, the associate vice president of medical affairs, or me. Initially, the chiefs, my associate and I were called upon often to deliver the message. But soon the nurses gained more confidence and became better known, and the physicians became more responsive to their suggestions. The key to success, I believe, was the willingness of influential physicians in leadership positions to become involved when needed.

Figure 5, above, depicts the average lengths of stay for Medicare patients at Saint Vincent Hospital by quarters for the past two years. Administratively necessary days are excluded from the analysis because the nursing home bed shortage has been getting worse in Massachusetts over the past two years and these days are reimbursed under Medicaid. Figure 6, above, compares the data from the year prior to the intervention to the year during the intervention, quarter by quarter. This was the first year there was no seasonal increase in lengths of stay during the winter months.

The decrease in lengths of stay between these two years is 1.2 days, for a total savings of 4,600 days for the year. This resulted in the availability of more than 12 additional beds per day. In the two winter quarters, October through March, the reduction of average lengths of stay from 13.4 days to 11.2 days represents an additional 24 beds per day. This had a positive effect on decreasing ambulance diversion from our emergency department, which can be seen in figure 7, right, which plots hours of ambulance diversion per month from October 1988 to March 1990. In addition, Saint Vincent Hospital has experienced a 5 percent increase in admissions and has realized an increased profitability of $2 million per year that can be directly attributed to more efficient use of beds.

* Eisenberg, J. Doctors' Decisions and the Cost of Medical Care. Ann Arbor, Mich.: Health Administration Press, 1986, Chapters 5-7.

At the time this article was written, Paul H. Rockey, MD, MPH, FACPE, was Senior Vice President for Medical Affairs, St. Vincent Hospital, Worcester, Mass. He is now Associate Dean for Clinical Affairs, Southern Illinois University School of Medicine, Springfield, Ill. Dr. Rockey is a member of the College's Society on Managed Care Organizations.