Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
CHANGING PHYSICIAN BEHAVIOR: In Search of the Little Blue Button* John R. Kimberly Summary This paper addresses the topic of non-economic inf luences on physician behavior as part of the Institute of Medicine's study on strategies for reform of physician payment. The paper is not an exhaustive review of the literature; rather it is intended to synthesize what is known in a manner that highlights potential policy options. The underlying question of interest is how might what is known about non-economic inf luences on physician behavior inf luence policies designed to change that behavior. Where is the little blue button? Five categories of non-economic influences are particularly noteworthy: education, feedback, managerial and organizational constraints, peer pressure, and patient demands. The potential value of education is based on the assumption that there is a connection between cognitive change and behavior change. If physi clans know, for example, that the benefits of a particular procedure do not come close to the costs of that procedure, the hope is that they will be disinclined to use it. In the absence of such knowledge, the argument goes, no behavior change should be expected. Is having the knowledge, however, suff icient for producing the desired change? Feedback is a potentially powerful shaper of behavior. The notion is straightforward. In the normal course of practice, physicians get feedback in a variety of informal, generally unsystematic ways. Were formal, systematic feedback on selected aspects of their practice regularly available to physicians, they would be in a position to judge whether they wanted to change any of these aspects. For the motivational potential of feedback to be realized, the physician must be able to compare not only his present practice patterns with his own past patterns but also, and perhaps more importantly, he must be able to compare his pattern with the patterns of other physicians in comparable sett ings. *I would like to thank many of my colleagues at the Leonard Davis Institute for Health Economics at the University of Pennsylvania for contributing to the pool of ideas f ram which this paper emerged. They are blameless for the product but entirely culpable for the stimulation and provocation. 30
As professionals, physicians are susceptible to inf luence by other professionals whom they particularly respect. Although cliff icult to quantify its effects, peer influence appears to shape important dimensions of clinical practice such as test ordering, preferences for drug perscription, and technology utilization. Peer influence also shapes behavior in more subtle ways, and undoubtedly influences, at least at the margins, important non-work values and attitudes as well. The contexts in which physicians practice medicine are changing rapidly. One of the most striking changes is the increasingly organizational character of those settings. As a result, physicians are more subject to managerial and organizational constraints than at any time in the past. Their work will be more carefully scrutinized, and organizational priorities will have to be integrated with professional priorities. The traditional autonomy that physicians have enjoyed will inevitably diminish somewhat as their work is subject to managers' needs for coordination and control. To the extent that physicians become employees of organizations (which they themse Ives may or may not own I, they become subject to new constraints and controls which will influence their behavior, economic and non-economic. Education, f eedback and peer inf luence do not appear to be signif icant, enduring influences or behavior in and of themselves. Rather, their effects are most powerful in concert. Initiatives designed to change physician behavior, therefore, ought not to focus exclusively on one of them. To the extent that any given intervention can incorporate all three sensibly, it stands a better change of having a more profound, more enduring impact. For this reason, the increasingly organized character of medical practice is significant. Organizational settings where physicians are using their professional skills in close proximity to other physicians and where other links may be established among them provide unusually good opportunities to incorporate all three effectively. Patient demands are perhaps less amenable to policy initiatives, but are nonetheless an influence on physician behavior which deserves mention. Assuming that the vast majority of physicians want to practice good medicine, and assuming that an important part of their sense of professional competence comes from the affective quality of their relations with patients, they may be more open to influence from their patients than we might believe. As patients become more sophisticated' ask more questions, and begin to assume greater responsibility for personal health and medical decisions, we can expect the nature of the inf luence process in the patient/physician encounter to change. These five categories of non-economic influences must be considered in the context of the changing environment in which the physician practices medicine. The changing demographics of the population, the increased emphasis on efficiency in the settings in which they practice, increased competition among physicians for patients, and the spread of -31-
corporate medicine all signal rather dramatic changes in the matrix of incentives to become a physician and in the structure of rewards for being a physician. Policy initiatives intended to influence physician behavior must be built on an awareness of differences among physicians currently in the profession and on their effects on the process of recruitment of new blood to the profession. Introduction This paper is designed to stimulate reflection on some of the key non-economic influences on physician behavior. It is not a comprehensive review of research on that and related topics. Such a review might be useful, but would require excursions into the disciplines of psychology, social psychology, sociology, and anthropology as well as the medical literature. The more limited objectives of this paper required a more selective review strategy. Two questions were posed initially by the topic, and the answers helped shape the approach taken. First, what are ~non-economic. influences? The answer to this question proved elusive. An effort to develop a definitive answer would have perhaps been theoretically interesting, but as a practical matter would have led us far afield. Because the boundaries between economic and non-economic influences are hazy at best, we decided to focus on factors whose primary influence appeared to be non-economic. Even this distinction is imperfect, but it permitted us to move off dead center. Theoretical purists may yearn for more rigor, and undoubtedly the issue deserves more attention than we were able to give it. A second question posed at the outset was what kinds of physician behaviors should receive attention? It is tempting to distinguish between clinical and non-clinical behavior, and, indeed, much of what has been written uses this distinction either explicitly or implicitly. Studies of physician use of new technology, test ordering, or drug prescribing, for example, focus on clinical behavior. The distinction is useful as a way of talking about what has been written (the vast majority of which looks at clinical as opposed to non-clinical behavior) but may be less useful in thinking about policy options. We will return to this issue later in the paper, but for now let us raise the possibility that the distinction unnecessarily constrains the way we think about how behavior might be influenced. With these two questions at least confronted--if not resolved--the literature was reviewed selectively, principal themes were identified, and implications were explored. The three main sections of the paper which follow contain the essence of this work. Section I I, Primary Influences, briefly reviews some of the basic research and then synthesizes the most policy-relevant work into the five headings noted above e Fortunately or unfortunately--it is a matter of perspective--the world does not hold still. In Section III some of the more significant 32
changes in the environment of the medical profession and its practice are noted. The purpose is to highlight the fact that new policy initiatives need to be based on informal predictions about the direction and impact of these changes rather than on the existing order. Moving targets are more difficult to hit, to be sure, than stationary ones--particularly when there is considerable uncertainty about their direction and velocity--but treating a moving target as though it were stationary is to guarantee a miss. The final section of the paper suggests an alternative way of thinking about influences on physician behavior and proposes a set of research questions that need to be asked if one goal is to move beyond present levels of understanding. Primary Influences Career Choice. A great deal of ef fort has been invested in understanding the various kinds of career choices that physicians make. The choices themselves are varied: choice of medicine as a profession; choice of medical school; choice of internship and residency; choice of specialty and sub-specialty; choice of academic medicine versus private practice; choice of location of practice; choice of setting for practice; and choice of kinds of patients to serve. Studies that have been done are primarily correlational or associational, that is, the outcomes of these choices for individual physicians are correlated across large numbers of physicians in an ef fort to detect patterns. Exceptions are studies of medical education and its influence on how future physicians are socialized into the profession.] These studies tend to be done in the field and examine the decision process up close. The primary findings, of course, are that medical school is a powerful socializing agent, that future physicians are often strongly influenced by faculty role models, and that one consequence is often a devaluing of general practice or primary care and an over-emphasis on specialty practice. Non-clinical professional behavior. In this category are studies which in practice explore a relatively limited range of behaviors but which to be more useful should broaden that range. Examples of this kind of behavior are variability in physician travel to professional meetings and variabilitly in reading of scientific literature related to professional practice. Such issues are rarely the primary focus of study. More often these variables are included as part of a study of something else, and the results are generally correlational and thus of limited value in other than a descriptive sense. Studies with these variables included are relatively few in number. Fewer still are studies of non-clinical professional behaviors of considerable current interest to policy makers, such as political activity in professional associations or involvement In administrative or managerial roles. Even descriptive information here would be helpful. 33
The basic conclusion is that, for whatever reasons, non-clinical professional behavior has not attracted much research attention. Whether more attention would be productive is a question well worth addressing. Clinical decision-making. A great deal of research has been done in this arena, primarily because of the potential attributed to changing certain aspects of clinical behavior for controlling the costs of medical care.4 If physicians would order fewer laboratory tests, prescribe fewer drugs, and be less inclined to use expensive technology of questionable efficacy, the aggregate impact on costs would be significant, the argument goes. There is considerable evidence that the problem of overutilization is real, particularly with respect to test ordering and use of technology. The explanations vary. Younger physicians tend to order more tests than older ones. Internists in academic health centers tend to order more than in other settings. Is the explanation age-centered, setting-centered, or both? With respect to technology, the problem appears to be the demand of physicians for the latest developments, independent of demonstrated clinical efficacy.6 Once the technology is available in an institution, there are pressures to use it. What explains the rapid diffusion of essentially unproven technologies? The dominant explanation appears to be, in the hospital setting, physician pressure. Yet one has to wonder whether this explanation ignores some of the complexity of the physician/institution relationship. At the very least, one has to wonder how all this may change as we move from cost-based to prospective reimbursement. Given the widespread belief that change in clinical behavior is something that is worthwhile pursuing, a number of studies have reported the results of experiments designed to produce such change. Some experiments have involved educating physicians about the costs of the tests they order.' The idea is that, absent such information, there is no reason to expect physicians to change. Knowing costs, however, might affect their decisions. Other experiments have used feedback to physicians about certain aspects of their clinical behavior compared with others in their institution and others with similar patients in similar settings.8 The idea here is that if physicians observe differences they will be motivated to change. The assumption, of course, is that physicians who see themselves, for example, as ordering comparatively more tests for their patient with similar problems will be encouraged to order f ewe r tests . Less attent ion has been f ocused on the case of the physician who uses relet ively fewer tests. Will he be encouraged to order more? The results of these studies provide hope if not certainty. Educational efforts have proven successful, if only for a short time. Feedback has proven to be a motivator of change, although not equally powerful under all circumstances and for all physicians.9 Variation in -34-
utilization of resources is real, dif ferences in the clinical behavior of physicians accounts for a large part of this variation, and this behavior is susceptible to change. Precisely how and under what conditions remains somewhat cloudy. Other studies cite the importance of opinion leadership in influencing the clinical behavior of physicians. Perhaps the best-known of these studies focused on patterns of diffusion of a new drug among physicians and discovered that a very few well-respected physicians had an enormous influence on their colleagues.l° These physicians were, in effect, gatekeepers. When they began prescribing the drug, the green light quickly flashed to their colleagues. The implications of this study certainly have not been lost on the pharmaceutical industry. Physicians in most respects are like anyone else. They have role models. They mirror the values of and try to emulate those whom they respect. They have self-doubts and anxieties. They are, in other words, susceptible to a whole range of influences on their behavior, many of which are economic but many of which are not. In a very real sense, the nature of a physician's work and more important, the attributions we make toward it make it difficult for us to see the uncertain, trial-and-error, non-scientific context of the world he inhabitse We are thus less able to appreciate the power of the mot ivational levers that are available and are perhaps less able to see the levers themselves. As many observers have noted, the contexts in which most physicians work most of the time are loaded with incentives for greater rather than reduced use of resources.ll The challenge is to understand the incentives, the contexts, and human behavior well enough to know which levers are there and which to pull. The social context and normative understructure of physician behavior. Some researchers have tried to understand physician behavior by observing it in situ. The use of anthropologically-inspired ethnographic methods has its drawbacks, as the recent controversy over Margaret Mead's early writings suggests.12 Nevertheless, there are some themes that emerge from the research that has been done that deserve careful attention. Most significant is the power of the informal normative structure that evolves wherever physicians work together. The informal norms define the ranges of acceptable and unacceptable behavior on a whole series of dimensions. They define what clinical mistakes are and how they will be handled. They define a code of ethics for that setting.13 And undoubtedly they define the parameters within which economic incentives will operate. What is particularly noteworthy is that these norms operate outside of formal rules and regulations. They are the equivalent of local interpretations. They may operate in concert with the formal rules or they may act as a buffer between those rules and what phys icians actually do. And a [though there are undoubtedly certain norms that operate at the level of communities, the most powerf ul influences on physician behavior on a day-to-day basis are those that operate in their immediate environment. As the sites where physicians 35
work assume an increasingly organizational character, this will mean that influences will be felt primarily at the organizational and departmental levels. Implications. We have skimmed over the surface of what is really a substantial body of literature on physician behavior. Much of this literature is descriptive; relatively little of it is straightforward in its implications for policy intended to change that behavior. Nevertheless' we know enough at this point both about physician behavior and human behavior in general to be able to extract f ram what has been written a limited set of variables on which to place our bets. Assuming that the goal is to change physician behavior, we should bet on a combination of education, feedback, peer pressure, organizational and managerial constraints, and patient demands. It is of ten suggested that because of the early and powerful influence of medical education on subsequent career choices, the most direct and effective way to change physician behavior is to change the structure, context and process of medical education. I do not disagree that such change would be helpful. However, I believe that the lasting influence of the experience may be over-estimated and the difficulty of producing major change in medical education may be under-estimated by those advocating this strategy. As a practical matter, therefore, I choose to emphasize the f ive variables noted above. Efforts to rethink the structure, content and process of medical education should continue. In the near term, there are other initiatives that may be taken as well. By education, I mean increasing the knowledge that practicing physicians have about selected aspects of their practice. It has been demonstrated that overuse of resources can be reduced by a program deliberately designed to educate users about such things as cost.14 Lack of awareness of cost may contribute to overuse. Education may help solve the problem. The question which has arisen is how long we can expect education effects to persist. Given that the effects do not appear to be permanent, the conclusion is that education may be helpful, it certainly ought to be incorporated more widely, and that it is not sufficient to achieve the des i red resu 1 t . By feedback, I mean providing physicians with information about selected aspects of their own past and current patterns of practice and the practice of others in similar settings with similar patients so that they can see how their patterns have changed and where they stand relative to their peers.15 The presumption is that once they are aware of these two dimensions they may be motivated to change. The absence of feedback affords them little opportunity to raise the question about whether change should be contemplated or to make the judgment about where change should take place and why. As with education, one can question the permanence of the effects of feedback. More to the point, however one can also ask how feedback should be designed in order to enhance its -36-
effectiveness. Here the worlds of medicine and management intersect in what might prove to be an interest ing and product ive way. The influence of peers is powerful. The theme runs throughout the literature on physician behavior in more and less explicit ways. And it certainly pervades more general theories of human behavior. To oversimplify the message only somewhat, physicians, like other people, are ~nf luenced in a variety of ways by their peers. Two aspects of peer inf luence are particularly relevant here. First is the importance of what have been called Opinion leaders.. These are the surprisingly small number of individuals to whom others look for signals about what is legitimate and valued. Identification of opinion leaders and changing their behavior is a potentially powerful and efficient way to engender broader-based behavior change. 16 Second is the development of what we referred to earlier as the normat ive understructure of physician behavior. These norms are held, transmitted, and enforced by peers. Change in these norms is unlikely to occur without widespread peer approval and support. The implication here, though perhaps obvious, or perhaps because it is obvious, is often overlooked by those who would create change. The result repeatedly is change in veneer without change in substance. Managerial and organizational constraints may be the focal point for education, feedback, and peer influence. None of those three in isolation is likely to produce the enduring changes in physician behavior that many are searching for. To the extent that current trends for physicians to exercise their profession in increasingly organized settings continue, opportunities to influence their behavior will flower. Possibilities for linking educational efforts with regular, systematic feedback and capitalizing on the power of peer influence are dramatically enhanced when physicians are working in close proximity and are sub ject to a common admi nistrat ive system, no matter how loosely or tightly def ined that system may be. Enough is known about designing education and performance feedback systems and about capitalizing on the power of peer pressure to permit policy initiatives to take advantage of the increasingly organized character of the practice of medicine. Patient demands are less amenable to policy initiatives (although second opinion experiments may have encouraged patients to be less reticent to question their own physician) than the foregoing, but their inf luence on physician behavior deserves recognition. In fact, as one ref lects on the social context of medical practice and how physicians develop feelings of personal competence and self-worth, the influence of patients on physi clan behavior may be greater than we think. Two trends are noteworthy. First is the increasing sophistication of the patient. Patients appear to be more likely to ask questions about diagnostic and therapeutic regimens now than previously. They appear, in other words, to be less likely to be passive ~ n their relationships with physicians than has been the case historically. To the extent that this trend -37-
continues, physicians will be increasingly constrained to think through and justify their clinical decisions to their patients. One might expect that this phenomenon alone would inf luence physician behavior. The real question, of course, has to do with the direction of this inf luence. Assuming that it is in the physician ' s interest to be respons ive to patient demands ~ in part as a consequence of increasing competition among physicians for patients and the need to protect markets, the second trend alluded to above), one can imagine a price-sensitive, procedure-skeptical patient inf luencing the physician and, in the aggregate, the system, in an ultimately cost-containing way. On the other hand, one can imagine the physician interested in being responsive to patient demands giving a price-sensitive, procedure-oriented patient what he or she wants (within limits) thereby contributing to the cost spiral. Both scenarios are over-simple, of course, but the point is that the patient may have considerable leverage on physician behavior. If this possibility is acknowledged, then one policy orientation might be in the direction of producing price-sensitive, procedure-skeptical patients. What the most efficient vehicles for doing this are is open to discussion, and many different proposals eve been advanced. The basic point is that the inf luence of the patient on physician behavior is substantial and its potential ought to be incorporated into thinking about policy options. To summarize, the most signif icant non-economic influences on physician behavior are education, feedback, peer pressure, managerial and organizational constraints and patient demands. Working these one at a time, or not appreciating their joint effects will diminish substantially their potential to produce significant, enduring change in the way physicians do their business. Furthermore, exploration of their effects has been limited primarily to clinical behavior. Less is known about influences on career choice, non-clinical professional behavior, and behavior apart f rom the profession which may affect professional behavior. Each of these behavior domains could be more throughly researched with great profit. Clinical uncertainty. In a view that leads in a parallel direction for policy, one relatively common explanation for overutilization of resources is the way in which physicians seek and use information in clinical practice. Faced with uncertainty and not wanting to co~Tunit er rors of omission the argument goes, the physician tends to gather as much information as possible in the course of dealing with the patient, regardless of cost. The threat of malpractice further intensifies this tendency. Absent incentives to the contrary, physicians will continue to seek more information than they need in many cases to do their work, thus continuing the strong upward pressure on costs. What might help develop counter pressures? One answer is for the profession to begin to define practice guidelines more carefully and more extensively. Development of such guidelines by acknowledged leaders could interact in a potentially useful way with education and feedback. Appropriateness of deviation 38
f ram the guidelines would be judged by peers. This device would not remove clinical uncertainty, but would help the individual physician judge what const itutes sound medical practice in particular cases. Development of practice guidelines is fraught with problems, to be sure. But a variety of developmental ef forts are currently underway, and they are likely to increase. Creation of the DRGs required a similar ef fort, and institutional management pressures will encourage similar activities. Thus, there appears to be a convergence of forces moving in the direction of the establishment of practice guidelines. These guidelines could influence physician behavior positively if developed and implemented properly. That is the policy challenge. Environmental Changes There is no need here to rehearse the myriad changes that are taking place in the environment of medicine.l7 Of particular interest is the increase in the supply of physicians J the increasing application of the logic of managerial efficiency to the work that health professionals do, the proliferation of organizations as the shone baser for physicians, the infusion of private capital into the health care industry, and the growth of multi-institutional arrangements of many sorts linking previously independent, autonomous health care providing institutions one to another. Opinions differ on the desirability of many of these changes, a fact which surprises no one. At least one implication of these changes, however, is clear. The context of the practice of medicine in 1986 will be strikingly dif ferent f ram the context only 10 years earlier. That context will see much closer scrutiny of physicians' work at a very micro level, greatly increased pressures to keep costs down, much less solo practice, and much more Corporate medicine.~18 It would take considerable time to do full justice to the implications of the changing environment for patterns of physician behavior. For present purposes, I would like to make only two points. First, most physicians are well aware of the fact that change is in the wind and that their lives will be affected by the shape of these changes. This awareness will produce a mixture of anxiety, anger, withdrawal, hostility and enthusiastic participation from the profession, whose members hardly speak with one voice. Second, whatever policy changes are contemplated should be based on emerging realities rather than outdated models. At some point, the solo practitioner will be exceptional. Like it or not, organized settings will be statistically the most dominant employer of physician labor. This fact will provide abundant opportunities to influence physician behavior. The challenge will not be to find the little blue button which, when pushed, will suddenly line up economic and non-economic influences in a sensible, enduring fashion. The button simply does not exist and never will. The 39
challenge is to use what is known about economic and non-economic inf luences on physician behavior creatively and responsibly. New Research Initiatives In this final section of the paper, I would like to sketch the broad outlines of an alternative way of looking at and doing research on physician behavior which, if more f ully developed and seriously pursued, might result in greater understanding of physician behavior and the parameters that changes in policy might be expected to inf luence. One str' king feature of much of the research on the topic is that it is cross-sectional. Snapshots are taken, attributes of those snapshots are correlated, and dynamic tendencies are inferred. The pitfalls of such approaches are often acknowledged, yet alternative strategies have seldom, to my knowledge, been pursued. Cross-sectional correlational studies provide useful descriptive portraits. They do not, however, enable one to determine the etiology of observed patterns. How and why did the patterns turn out as they did, and what implications do the answers to these questions have for how the patterns are likely to look tomorrow? Or next year? The answers to questions such as these are both important and elusive with present research orientations. What might a more productive orientation look like? A combinat ion of introspection and even the most casual of conversations with others strongly suggests that people have dif ferent priorities at different times in their lives, that they are motivated by different concerns, and that their behavior is most difficult to understand, let alone influence, without some insight into what mix of priorities has and has had significance for them. Should we expect things to be any different for physicians? It is not unreasonable to expect that physicians fresh out of training will have dif ferent priorities and will respond to different incentives than physicians who have been in practice for 20 years and whose children are through college. At any given time the population of physicians includes people who are at very dif ferent points in the development of thei r professional careers and personal ambitions. It would be unrealistic to assume, therefore, that this population would respond with anything 1 Ike uniform) ty to changed incentive structures. The research challenge is to develop a greater appreciation for the dynamics of physician personal and career development. Are there ~typical. developmental patterns? If there are, what are their characteristics and how can we account for differences between clusters? What are the implications of these differences for changing behavior? Might not different strategies be more effective for different clusters? It is here that I return to the distinction between clinical and -40-
non-clinical behavior about which ~ expressed some concerns earlier in the paper. The distinction is useful in the context of describing research on physician behavior, the vast majority of which focuses on clinical behavior. Both the distinction and the focus, however, tend to compartmentalize artifically the life of the physician. As all of us, the physician may play many roles: wage-earner; spouse; parent; homeowner; politician; friend; social critic; taxpayer; investor. The list is long. The influences on behavior are multiple and often conflicting. To focus exclusively on clinical behavior is to focus on one arena and perhaps to ignore or underestimate the power of other arenas and how they do or do not fit together as shapers of motivation and behavior. The increasing supply of physicians, the increased emphasis on productive efficiency, and the increasingly organized character of much of medical practice will change the way in which the physician views the ex ist ing opportunity structure. Research and policy need to appreciate these changes and their personal and career impl ications . ~ developmental view of the meaning and consequences of being a physician will enhance this appreciation. 41
References 1. See, for example, H. Becker, Boys in White. Chicago: University of Chicago Press, 1962; R.K. Merton, G. Reader and P. Kendall, The Student Physician. Cambridge, MA: Harvard University Press, 1957. ~ Luft, H.S., Health Maintenance Organizations: Dimensions of . Performance. New York: John Wiley & Sons, 1981. 3. Hick, S.S., Sussman, S., Anderson-Selling, L. et al, physician turnover in eight New England prepaid group practices: an analysis. ~ Medical Care, 21: 323-338, 1983. 4 . Estes , E. H. Jr., The behavior of heal th prof ess ionals : Impact on CQSt and quality of care.. -Perspectives, 3:7-13, 1983; Lyle, C.B., Biarchi, R.F., Harris, J.H. and Wood, Z.L. Teaching cost containment to house officers at Charlotte Memorial Hospital.. Journal of Medical Education, 54:856-862, 1979; Eisenberg, JaMe and Williams, S.V., Cost containment and changing physicians' practice behavior. Journal of the American Medical Association, 246 2195 - 22O1J 1981e 5. Daniels, M. and Schroeder, S.A., Variation among physicians in use of laboratory tests: II. Relations to clinical productivity and outcomes of care.. Medical Care, 15:482-487, 1977; Eisenberg, J.M., The use of ancillary services: A role for utilization reviewer Medical Care, 20:849-861, 1982. 6. See, for exmple, Russell, L.B. Advances and Their Diffusion. Washington, DC: Brookings, 1979. 7. Eisenberg, J.M., Educational program to modify laboratory use by house staff.. Journal of Medical Education, 52:578-581, 1977; Griner, P.F. and Lipzin, B., ruse of the laboratory in a teaching hospital. Annals of Internal Medicine, 75:157-163, 1971. 8. See, for example, Wennberg, J.E., Blowers, L., Parker, R. et al, Changes in tonsillectomy rates associated with feedback and review. ~ Pediatrics' 59: 821-826, 1977. 9. Dyck, F.J., Murphy, F.A., Murphy, J.K. et al, Effect of surveillance on the number of hysterectomies in the province of Saskatchewan. New England Journal of Medicine, 296:1326-1328, 1977. 10. Coleman, J.S., Katz, E. and Menzel, H. Medical Innovation: A Diffusion Study. Indianapolis: Bobbs-Merill, 1966. -42-
11. See, for exmple, Eisenberg, J.M. Modifying physician patterns of laboratory use. ~ in Connelly, D.P., Benson, E.S., Burke, M.~., and Penderson, D. teds. ~ Clinical Decisions and Laboratory Use. Minneapolis: University of Minnesota Press, 1982: 145-158. 12. Freedman, Derek, Margaret Mead and Samoa : The Making of a Myth. Harvard University Press, 1983. 13. Bask, C. Forgive and Remember. Chicago: University of Chicago Press, 1979; Eisenberg, J.M. Sociologic influences on decision-making by clinicians.. Annals of Internal Medicine, 90:957-964, 1979. 14. See, for example, Greenland, P., Mushlin, A., and Griner, P.F. Discrepancies between knowledge and use of diagnostic studies in asymptomatic patients.. Journal of Medical Education, 54:863-869, 1979. 15. Schroeder, S.A., Kenders, K., Cooper, J.K. et al, ruse of laboratory tests and pharmaceuticals: variation among physicians and the effect of cost audit on subsequent use.. Journal of the American Medical Association, 225:969-973, 1973. 16. For an application of the principle to medical practice, see Stoss, J.K. and Bole, G.G. ~ ~ ~ ~ ~ for the primary care physician.. 22:787-791, 1979. Continuing education in rheumatoid arthritis Arthritis and Rheumatism, . 17. A comprehensive analysis of many of these changes can be found in Starr, P., The Social Transformation of American Medicine, New York: Basic Books, 1982. 18. See Kervasdoue, J., Kimerly, J. and Rodwin, V. teds. ~ Illusion. Berkeley: University of California Press, 1984. -43- The End of an
