The emerging role of the physician in administration.

The recent upsurge in the involvement of physicians in the administration of health care programs and organizations has raised important issues about the role of physicians in management and about appropriate preparation for that role.[1,2] Physicians have, of course, held management positions since hospitals became viable health care organizations. In most cases, however, their roles were much more circumscribed than those emerging today. Medical care was far less complex, hospitals were the only organizations in the field, and hospitals were considered "doctors' work shops," with little need for strategic management. There were few really important organizational decisions made by these physician managers, and it was, in fact, quite difficult to make a serious management mistake. Even public health administration, long the domain of physicians, was often more political than managerial. It is interesting to note that the major reports on physician training written as late as the mid-1960s made no mention of management roles.[3]

The role of the physician in management today is vastly different from that of the past. The major difference stems from the fact that the field is undergoing a shift from a primary focus on the one-to-one relationship between the physician and patient to an emphasis on populations and cost effective ways of organizing services for those populations.[4] Increased technology, cost issues, restrictive payment systems, and a more knowledgeable and demanding public have helped precipitate this shift, but, in many ways, it is a natural extension of other changes that have or are taking place in our society. Rather than a "sky is falling" calamity, it is simply the health services field evolving from its cottage industry into organized forms of care with shared decision-making structures.

Viewed from this perspective, it is easy to see why health care management roles have gained stature and why physicians are becoming more attracted to those roles. The field is becoming highly structured and power is shifting from the individual clinician to hospitals, group practices, HMOs, insurance plans, and other organizations. Some physicians feel that this change is long overdue and perceive it as an opportunity for them to influence major improvements in health care by shifting to at least part-time management roles. Others take these roles in an attempt to maintain control over health care delivery or at least their own destinies. Still others are being recruited to management roles by nonphysician CEOs who want to strengthen their management capabilities to deal with clinical issues.[5-8]

Many physicians are finding the transition to management roles more difficult than expected. What appeared to be a job that simply required "using good judgment" turns out to be a very complex and demanding undertaking, requiring skills that are foreign to clinical training. Consequently, the partners (the employing organizations and the physicians) are often disappointed, and turnover is high.

While, in part, this reflects the early stage of development of the physician manager concept and the lack of agreement regarding expected performance, there is a legitimate concern that physicians are not being prepared well for these roles and therefore are creating a great deal of discordance. The problem of appropriate preparation, in good part, relates to the wide range of management roles being assumed by physicians. These roles vary from department heads in multispecialty group practices, where there may be 10-12 physicians and nurses, to CEOs or medical directors of large hospitals or health plans. One of the major problems is that physicians, because of their clinical reputations, are often being hired into senior generalist management positions in large scale organizations that are far beyond their management capabilities. As the unique contributions clinicians bring to the management arena become more clearly defined, the roles will more realistically match their talents, and entry into the field quite likely will be in a narrower corridor of role clusters.

These role clusters will largely focus on the patient care process, physician practice styles, and the cost effectiveness of health services. It is in this arena that clinicians can make unique contributions by integrating clinical insights into management roles. Moreover, this is where most health care organizations are seriously lacking in management capacity and desperately need help. Schumacher recently identified the lack of managerial talent in hospitals that can deal quantitatively and analytically with clinical quality of care and cost effectiveness issues as one of the "major shortfalls in the institutional and human resources needed for the 1990s."[9] We agree fully, but would broaden the concern to include several other health care organizations. Medical group practices, HMOs, community clinics, and nursing homes face the same set of management challenges. This is where physicians can greatly enhance the management capacity of a health care organization, and this is where the majority of physicians seeking managerial positions probably should enter the field.

Viewed from this perspective, the management competencies needed by physicians to augment their clinical skills fall into three areas. First, they need to shift from a one-to-one patient and physician perspective to a population-based analytic focus. Individual patient needs must remain a central concern, but they can only be dealt with cost effectively by viewing them in the context of population-based disease patterns, disease staging, expected outcomes, and the life cycles of diseases and illnesses. While several disciplines can contribute to the training of physician managers in this area, epidemiology should play a central role. This is not to say that they should abandon their clinical roles. At this level of medical management, it is very important for physicians to stay actively involved in clinical practice.

The second important skill area deals with acquisition and analysis of information. At a minimum, this includes knowledge of probabilities, random variations in data, and the risks associated with making decisions based on small or truncated data sets. It also includes measurement issues and the special measurement problems presented by the health care field. Physicians must be equipped to "own" their data rather than rely on others for analysis. This means that they must understand quantitative methods and be able to evaluate alternative analytic methods, even if someone else conducts the statistical analysis. It also means that they must have sufficient training in quantitative methods to be able to creatively manipulate the data to uncover hidden patterns and new trends. A fundamental requirement in this regard is the ability to read the relevant journals and evaluate both what is and is not being said by the authors. A good physician manager at this level of the health care organization cannot simply delegate these responsibilities to subordinates and expect to maintain a valued role in the patient care decision-making process.

Finally, incumbents in these roles must be good decision makers. In fact, their ability to make sound decisions will largely determine whether or not they get promoted into generalist management positions. Here, the decision sciences are key. Having good quantitative skills is fundamental but is not by itself sufficient. These physicians must gain a broad perspective on decision making such as that provided by cost-benefit analysis, must develop a sense of timing, and must gain an appreciation of the consequences of decision vacuums. The human element must also play a prominent role in ensuring a balanced decision-making process. Consensus-building skills are imperative, especially as they relate to professional medical practice arenas. While good basic skills in these areas should be evident before a physician is recruited into management, these skills can be improved through problem-solving exercises using the case method. It must again be noted that while consensus-building skills and the ability to influence physician behavior are important attributes, they are of little value, and may be dangerous, unless the course of action being pursued is based on sound analytic judgments grounded in quantitative methods.

To assume that physicians somehow have inherent management capabilities that enable them to enter the field as generalist managers in top executive positions is a disservice to the physician and to the employing organization. Rather, they should be entering the field in second- or third-level management roles that use their clinical knowledge. Those who demonstrate over time that they have acquired broader management capabilities will progress to generalist roles. The demonstrated ability to view issues from a strategic and integrative perspective and to make mature judgments and sound decisions is key to this process.

The necessary preparation for physicians to move to generalist management roles is much different from that for the entry-level positions described above. Here the emphasis should be on designing new organizational forms, strategic management, human resource development, communication skills, conflict resolution, and resource management. Defining organizational goals and developing a culture and a reward system that supports them is part of this role. Ability to work effectively in the policy arena is also important, both in terms of positioning the organization strategically and in shaping the health policy environment.

Before shifting to full-time generalist management, physicians should engage in a graduate-level educational program leading to a master's degree in management. There are several ways to obtain this training. Part-time MBA programs available from local universities and colleges present very good opportunities, although they need to be augmented by courses in health policy and health economics. Graduate programs in health administration offered through some form of off-campus study also offer potential sources of management training for physicians who cannot devote full time to the program. However, as physicians become serious contenders for generalist management roles and become important players in the health care management field, they will find it increasingly important to devote at least a year to full-time study at the master's degree level. This will free them from geographic constraints and will enable them to seek out the best programs available and interact with nationally recognized researchers and teachers. It will also greatly enhance their socialization into the field of health care management. Moreover, when employing organizations gain a better appreciation of the value-added effects of graduate training in management for physicians, they quite likely will provide the funding necessary to make full-time participation in these programs possible.

In the remainder of this article, we have outlined six management problems that we feel an entry-level physician manager should be able to deal with successfully without asking for help. If physicians find that they are unable to deal with these issues, they should seek additional training before considering a management role.

Issue Number 1

According to Bailar and Smith, the nation's age-adjusted mortality rate from cancer rose from about 170 per 100,000 population in 1962 to 185 per 100,000 in 1982, despite intense and growing efforts to improve the treatment of cancer.[10] The National Cancer Institute reported in 1984 that, with improved detection and treatment, the United States was on the verge of attaining a five-year survival rate of 50 percent. What questions do these seemingly contradictory statements raise about incidence rates, survival rates, and mortality rates? What could account for the statements? Answer: Incidence rates can be affected by changes in diagnostic criteria that come about through advances in screening and detection as well as through true increases in the disease. No test is perfect. If the test is applied widely, there will be false positives. It is also likely that, in earlier stages of the disease, this problem might be accentuated.

Survival rates can be affected by the percentage of cases that are true positives, changing the natural history of the disease, the timing of diagnosis, and the treatment. If, over time. more of the cases are false positives, survival rates will improve. This is also true if cases are recognized at an earlier stage, if the disorder naturally becomes less virulent, or if treatment is improved. Simple survival rates are not able to separate out these different effects.

Population-based, age-adjusted mortality rates can be affected by changes in the natural history of the disease and by improved treatment. They are not affected by the stage at time of diagnosis or by increased incidence due to false positives.

The improved five-year survival rate is probably best explained by advances in screening and detection that have included some cases that, in fact, do not have the disease and by identification of cases at an earlier stage in their disease. The increasing mortality rate, along with the fact that, other than for lung cancer, there appears to be little evidence that true cancer cases are increasing, suggests that the improved five-year survival rate is not due to improved treatment.

Issue Number 2

You want to use indirect assessment techniques to estimate the number of people in your HMO with untreated elevated blood pressure. Assume that the choice of your plan by individual enrollees has no selection bias. List four pieces of information that you would need to be able to come up with this estimate. Answer: All of the following data elements would be necessary to derive the estimate:

* Racial make-up of your HMO.

* Ratio of females to males in your HMO.

* Number of HMO enrollees now being treated.

* Number of enrollees in each 10-year cohort.

* Prevalence of hypertension in the general population by race, sex, and age.

Issue Number 3

Treatments A and B have the following costs and outcomes:

Treatment            A          B
Costs             $30,000    $15,000

Life Expectancy  5.0 years 4.0 years

Quality of Life 0.80 0.90

1. What is the incremental cost-effectiveness ratio (CER)?

2. What is the incremental cost-utility ratio (CUR)? Answer:

1. Incremental CER = Cost of treatment A - Cost of treatment B Effectiveness of A - Effectiveness of B = $30,000 - $15,000 = $15,000/5-4

2. Incremental CUR = Cost of treatment A - Cost of treatment B QALYs of treatment A - QALYs of treatment B = $30,000 - $15,000 = $37,500/5(0.80) - 4(0.90)

Issue Number 4

A hospital establishes clinical profiles of its medical staff for management of acute myocardial infraction. The total charges by physician for an average case range from $7,230 to $14,892. The medical director targets physicians in the top decile for an informal discussion about the concern of growing hospital costs for acute myocardial infarction patients and is gratified to note that these physicians are not in the top 10 percent when the following quarter's audit is available. Was the intervention successful? How can the results be explained?

Answer: One could not conclude that the intervention was successful. This example illustrates the common problem of directing an intervention at the extreme end of a distribution and mistakenly concluding that the intervention was successful when the end of the distribution moves toward the mean. This phenomenon is termed "regression toward the mean" and is likely to occur without intervention. Extreme values are less likely to be as extreme when observed at a later time. This would require a control group with which to compare the mean change and the range of charges for management of acute myocardial infarction.

Issue Number 5

The CEO of your hospital has asked you to talk with Dr. A about the length of stay of his Medicare patients hospitalized for total hip replacement. Four physicians are hospitalizing patients under this DRG. Their average lengths of stay for the past year are as follows:

Dr. A -- 14.3 days

Dr. B -- 12.5 days

Dr. C -- 11.1 days

Dr. D -- 12.5 days

What additional data will you need in order to make a judgment about these differences? What would be your advice to your CEO?

Answer: Both the distribution of lengths of stay (or variance) and the number of cases for each physician are necessary to determine whether or not the average lengths of stay are statistically significant. Even if there are statistically significant differences using the raw data, you will need to correct for severity factors before concluding that true differences exist.

Issue Number 6

As the medical director of a prepaid group practice (PGP) HMO, you are interested in comparing the expected expenditures for its enrollees to the expected expenditures for enrollees in independent practice association (IPA) and fee-for-service (FFS) health plans.

In the health services research literature, you find a study that reports the following findings from a regression of health expenditures on a variety of enrollee characteristics, including the type of health plan to which the enrollee belongs (IPA, PGP, or FFS):

Variable   Coefficient   t-statistic
Constant    10.0             4.7
Age          5.2             3.5
Sex        103.5             1.8
(1=female, 0=male)
PGP        -51.0            -2.5
(1=PGP enrollee, 0=otherwise)
IPA        -12.7            -2.2
(1=IPA enrollee, 0=otherwise)

1. What are the predicted health expenditures for a 40-year-old male in each of the three types of health plans?

2. Does the table contain enough information to determine whether the difference in predicted expenditures for PGP and FFS enrollees, controlling for age and sex, is statistically significant? How about the difference between PGP and IPA enrollees?

3. Would your ability to charge premiums (or receive payments) that varied by age and sex alter your use of the results? If so, how? Answer:

1. FFS prediction = 10.0 + 5.2(40) = $218 PGP prediction = 10.0 + 5.2(40) - 51.0 = $167 IPA prediction = 10.0 + 5.2(40) - 12.7 = $205.30

2. For the PGP-FFS difference, the answer is yes. The t-statistic on PGP measures the significance of the $51 difference between PGP and FFS enrollees. For the PGP-IPA difference, the answer is no. You would need the variance-covariance matrix of the estimated coefficients for PGP and IPA to know whether the $38.30 difference between IPA and PGP enrollees was statistically significant.

3. Suppose the purpose of examining the table is to estimate the expected profit for your PGP plan versus competing IPA and FFS plans. To do that, you have to estimate your (and their) expenditures and compare those predicted expenses to your (and their) premiums. If you and your competitors are able to charge higher premiums for females and older enrollees, the only difference in expenditures you care about is that embodied in the PGP and IPA coefficients. The remaining difference in premiums can be made up through premium adjustments. If you must charge the same premium for everyone, you probably will want to calculate expected expenditures for the three plans using your best estimate of the age and sex of individuals the plans are likely to enroll.

References

[1] Ruelas, E., and Leatt, P. "The Roles of Physician-Executives in Hospitals: A Framework for Management Education." Journal of Health Administration Education 3(2):151-69, Spring 1985. [2] Wallace, C. "Physicians Leaving Their Practices of Hospital Jobs." Modern Healthcare 17(10):40-1, 44, 48, 55-7, May 8, 1987. [3] The Graduate Education of Physicians, Report of the Citizens Commission on Graduate Medical Education. Chicago, Ill.: American Medical Association, Aug. 1966. [4] Ottensmeyer, D., and Smith, H. "Patterns of Medical Practice in An Era of Change." Frontiers of Health Services Management 3(1):3-29, Aug. 1986. [5] Barr, J., and Steinberg, M. "Professional Participation in Organizational Decision Making: Physicians in HMOs." Journal of Community Health 8(3):160-73, Spring 1983. [6] Hillman, A., and others. "Managing the Medical Industrial Complex." New England Journal of Medicine 315(7): 511-3, Aug. 21, 1986. [7] Nash, D. "Hospitals and Their Medical Staffs: High Anxiety." Frontiers of Health Services Management. 4(3):24-6, Spring 1988. [8] Schenke, R. "The Physician Manager and the Evolving Health Systems." In The Physician in Management, Schenke, R., Ed. Tampa, Fla.: American College of Physician Executives, 1980. [9] Schumacher, D. "Organizing for Quality Competition: The Coming Paradigm Shift." Frontiers of Health Services Management 5(4):4-30, Summer 1989. [10] Bailar, J., and Smith, E. "Progress Against Cancer?" New England Journal of Medicine 314(19):1226-32, May 8, 1986.

John Kralewski, PhD, is Williams Wallace Professor of Health Services Research and Director, Institute for Health Services Research, School of Public Health, University of Minnesota, Minneapolis. Terence Wingert, MD, is a doctoral student and research associate at the Institute for Health Services Research and a member of the attending staff at Ramsey Medical Center, St. Paul, Minn.