Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 173
Effectiveness Research and Changing Physician Practice Patterns Harold C. Sox Our goal has been to learn the circumstances in which technologies are effective. Once that goal has been accomplished, however, we must meet a second goal: that of altering physicians' behavior so that they implement research findings appropriately and consistently. My purpose is to discuss two assertions. First, it is difficult to be sure that changes in doctors' practice habits are due to published recommendations. Second, some of the resources of the Effectiveness Initiative should be earmarked for studying the factors that influence physicians to adopt new ways of practicing medicine and for testing interventions designed to promote change. EFFECTS OF RESEARCH ON MEDICAL PRACTICE The relationship between a specified research result and changes in medical practice is very complex. Diffusion, which is a term for the adoption of new medical technology, also applies to altered ways of using technology, such as might result from an effectiveness study. The determinants of diffusion include the following (1, pp. 178-181~: · Prevailing medical theory: A change in medical practice is more likely to be adopted if it builds on existing theory and medical logic. · Ease of learning a new practice style: How much effort is involved in changing habits that have been ingrained and polished through years of practice? · The importance of the clinical problem: Is the problem one that is likely to lead to death or disability for one's patients? If so, the physician is more likely to make the effort. · Advocacy by a professional leader: There is evidence that opinion 173
OCR for page 174
174 EFFECTIVENESS AND OUTCOMES IN [IEALTH CARE leaders in the medical community can influence their colleagues to adopt new practices. · Characteristics of the adopting physician: Have physicians' training prepared them to grasp new concepts quickly and see the implications for their patients? Do they have the ability to change from one style to another? · The practice setting: Does the physician belong to a group practice in which there is a lot of peer pressure to change? Are there financial pressures to change, either to do more procedures in fee-for-service practice or to do fewer in a prepaid practice? Are the new technologies available in the practice setting? . The physician's control over decision making: Do physicians have direct control over the decisions to acquire new technology or to make it easier or more difficult to obtain access to the technology? · The results of formally evaluating the technology: This component is the one with which the Effectiveness Initiative is most concerned. The evidence that formal evaluation affects medical practice will be discussed later. · The effectiveness of the channels of communication of evaluation findings: If physicians are not aware of the results of a formal evaluation of a tech- nology, its influence will be much diminished or delayed in taking effect. Both the professional and the popular media are important in disseminating information about technology, and the influence of the popular media on patients' expectations of their physicians is a topic that is particularly neglected. CLINICAL TRIALS The recognized standard of evidence for clinical effectiveness is the clinical trial with randomized controls. Do clinical trials influence medical practice? Fineberg examined several studies that attempted to trace the influence of a clinical trial (1, pp. 185-195~. To evaluate these studies, Fineberg first established standards of evidence that change in practice style was attribut- able to research results. First, what is the baseline pattern of using the technology? Is there a trend among practicing physicians that is due to factors unrelated to the research results? Second, is there evidence, perhaps obtained through surveys, that physicians are aware of the research results? Third, do the research results imply that a change in practice style should occur? Fourth, is there a temporal relationship between the assessment appearing in the medical literature and the subsequent changes in medical practice? Fineberg applied these research standards to 28 studies, of which only ten were suitable for analysis. The others failed because the study results did not have clear implications for practice, because there were no data on practice style both before and after the assessment was published, or because
OCR for page 175
APPLICATION TO CLINICAL PRACTICE 175 there were no quantitative data on the frequency of using the technology that was being studied. Fineberg9s study showed that only two of these ten studies contained strong evidence that the published technology assessment affected practice style. These two studies were among four in which there was evidence of a marked change in practice style. In two of these studies, the randomized trial preceded the change in practice, which is fairly strong evidence that the randomized trial had something to do with the change in practice; in the other two studies, the randomized trial did not precede the change in practice. In five studies, there were small changes that were consistent with trends in practice style, and in one there was no shift in practice style at all. Fineberg's study shows that one of the more powerful forms of medical knowledge, the results of a randomized clinical trial, had little measurable effect on practice style. CONSENSUS DEVELOPMENT CONFERENCES The second example shows that a program aimed at effecting change, the National Institutes of Health (NIH) Consensus Development Conferences, had little measurable effect on practice (2~. The goal of these conferences is professional and consumer consensus about the best way to use a technology. The RAND Corporation studied the effect of four of these conferences on clinical practice in hospitals. Table 1 shows the four conditions studied and a selection of the recommendations of the NIH Consensus Development Conference on these topics. The RAND investigators used the recommendations of these conferences as the standard of care against which to compare what they observed in patient records in a randomly selected sample of hospitals. They measured TABLE 1 RAND Study of NIH Consensus Conferences Condition Selected Recommendations Breast cancer Standard is total mastectomy with axillary dissection in Stage 1 or 2 Breast cancer An estrogen receptor assay should be performed on each primary tumor Cesarean section A trial of labor in low-risk women with a previous C-section Unstable angina Unstable angina patients should have a coronary angiogram on their first hospital admission for the condition SOURCE: Adapted from Kosecoff et al. (2~.
OCR for page 176
176 EFFECTIVENESS ED OUTCOMES IN HEALTH CARE compliance with these recommendations in three time periods: for one year starting two years before the conference; for one year starting one year before the conference; and for one year staring nine months after the conference. Table 2 shows the percent of cases in which there was compliance with the recommendations of the consensus conference. There was no trend toward increased compliance with the recommendation to perform total mastectomy with axillary dissection in Stage I and Stage II breast cancer. The RAND investigators studied compliance with a recommendation to test for estrogen receptors in breast cancer. There was a strong trend among practicing physicians toward increased compliance during all three periods of observation. In addition, there was a significant increase in compliance following the consensus conference, as compared with the entire period prior to the conference. The consensus conference appeared to have made a difference. A trend could be observed throughout the three periods toward compliance with a recommendation that low-risk pregnant women with a previous Ce- sarean section be allowed a trial of labor. However, the consensus confer- ence had no measurable effect on this decision. There was no trend toward increased use of angiography in patients with unstable angina and no evidence that the consensus conference had any effect. The RAND investigators made three additional observations. First, com- pliance was less than 50 percent during the year following the conference for 6 of the 11 criteria. Therefore, compliance with these criteria was low. Second, there was an overall trend toward increased compliance throughout the three time periods. Thus, physicians were generally aware of the chang- ing standards of practice, regardless of how much attention they paid to the consensus conference recommendations. Finally, when the RAND investi TABLE 2 RAND Study of Compliance with Recommendations of NIH Consensus Conferences Compliance (Percent) Case Period 1 Period 2 Period 3 Mastectomy and axillary dissection 74 79 84 Estrogen receptors 54 78 86a b Trial of labor 6 11 29a Angiography 14 29 24 ap < 0.05 over entire period only. bp < 0.05 for before-after. SOURCE: Kosecoff et al. (2~.
OCR for page 177
APPLICATION TO CLINICAL PRACTICE 177 gators examined compliance for each indication and each condition (not just the sample illustrated in Table 1), the rate of change in compliance actually slowed when time period 2, before the conference, was compared to time period 3, after the conference. All in all, the NIH consensus conferences had a limited immediate effect on practice style. Doctors do change. They no longer do gastric freezing or Halstead radical mastectomies. They discharge patients with uncomplicated myocardial infarction in one week rather than three weeks, which was the practice 20 years ago. Each of these changes is consistent with the findings in a series of empirical studies. Medical practice is moving ahead, albeit slowly, in a direction that is consistent with research results. How do we reconcile this change with our difficulty in establishing a cause-and-effect relationship between specific studies or consensus recommendations and change in practice style? The resolution of this paradox will require better understanding of the factors that influence adoption of new practice styles. WHAT SHOULD BE DONE The present circumstances provide an opportunity that may not come our way again soon. There has never been greater motivation to understand how to change medical practice, and there is adequate support to begin the task. We need much more research on the determinants of change in physicians' practice style, and now is the time to begin. Second, we should look to the professional societies to identify effective clinical policies. Their recommendations should be based on research results, with clear delineation of the logic leading from the research findings to the recommendations. If the professional societies are to play this central role, they should spend some time working together on a common approach to developing guidelines. At present, no two organizations use the same methods. When two organizations come to different conclusions about the same technology, fruitful discussion about how to interpret the data may be impeded by disagreement about the methods that were used in coming to a conclusion. A common methodology should promote respect for each other's efforts and lead to useful dialogue. We need to intensify efforts to motivate change. Organizations that pay for health care will be doing their part to motivate physicians, but we need more vigorous programs to teach physicians how to deal responsibly with fiscal pressures. The most efficient way to accomplish this task may be to aim these programs al pnys~c~ans wno are recognized by their peers as the opinion leaders in their professional community. We need to keep our patients informed of research results that will affect them. Well-informed patients can exert considerable influence at the time of a close-call decision, either encouraging or frustrating attempts to practice
OCR for page 178
178 EFFECTIVENESS AND OUTCOMES IN HEALTH CARE a lean style of medicine. There have been well-documented successes in using the popular media to influence people to reduce their cardiovascular risk. The same approach might help people to acquire a more realistic understanding of the benefits and risks of patient care technologies. Leaders in government, industry, and medicine must look upon the Effec- tiveness Initiative as a long-term investment. Finding the truth about what works in the practice of medicine will be an ongoing task, in which constantly improving research methods are aimed at evolving technologies. The chal- lenges are as daunting as those involved in basic biomedical research. It will take a decade or more for the Effectiveness Initiative to achieve research results that physicians can use to change the way that they practice medicine. These research findings will not have their intended impact unless there is an intensive effort to understand the factors that influence physicians to change. REFERENCES 1. Fineberg, H.V. Effects of Clinical Evaluation on the Diffusion of Medical Technology. Pp. 176-210 in Assessing Medical Technology, Mosteller, F. ed. Wash- ington D.C.: National Academy Press, 1985. 2. Kosecoff, J., Kanouse, D.E., Rogers, W.H., et al. Effects of the National Institutes of Health Consensus Development Conferences on Physician Practice. Journal of the American Medical Association 258:2708-13, 1987.
Representative terms from entire chapter: