National Academies Press: OpenBook

Assessing Medical Technologies (1985)

Chapter: 4. Effects of Clinical Evaluation on the Diffusion of Medical Technology

« Previous: 3. Methods of Technology Assessment
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 176
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 177
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 178
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 179
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 180
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 181
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 182
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 183
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 184
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 185
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 186
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 187
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 188
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 189
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 190
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 191
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 192
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 193
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 194
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 195
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 196
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 197
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 198
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 199
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 200
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 201
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 202
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 203
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 204
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 205
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 206
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 207
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 208
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 209
Suggested Citation:"4. Effects of Clinical Evaluation on the Diffusion of Medical Technology." Institute of Medicine. 1985. Assessing Medical Technologies. Washington, DC: The National Academies Press. doi: 10.17226/607.
×
Page 210

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.

~' Effects of Clinical Evaluation on the 7 Diffusion of Medical Technology Patterns of medical practice often di- verge from recommendations based on controlled clinical evaluations. This chap- ter views such discrepancies in light of the many forces in addition to clinical evalua- tion that influence the adoption and aban- donment of medical technology. The central question for the chapter can be stated simply: What effect does the evaluation of medical technologies have on their diffusion? The next two sections of the chapter introduce some of the complex- ities of this question and present an ap- proach to assessing the literature that tries to answer it. Following a review of litera- ture about the effect of evaluation on the diffusion of medical technology, the chap- ter summarizes its principal conclusions and offers a few recommendations. The relation between evaluation and diffusion is part of a larger issue of the con- tribution of technology assessment to im- proved health. This review emphasizes the connection between evaluations and physi- Harvey V. Fineberg prepared this chapter. clan behavior, although recognizing that health benefits in many medical situations ultimately depend on the behavior of pa- tients as much as or more than that of phy- . . slclans. EVALUATION In this review, the impact of two general types of evaluation are considered: (l) pri- mary assessments of the consequences of a medical technology and (2) synthetic as- sessments of the implications for clinical practice of the available primary evidence. Both primary and synthetic assessments take a variety of forms. Primary assess- ments range from judgments based on per- sonal experience to multicenter random- ized clinical trials. Synthetic activities range from review articles to meetings of experts for the purpose of reaching consen- sus on a controversial issue. This section deals with the effect of eval- uation on medical care decisions, usually the decisions of physicians. Relatively few studies quantitatively assess the influence of community-based epidemiological stud- ies, data banks, or case studies on changes 176

EFFECTS OF CLINICAL EVALUATION in physician practice. Many primary and synthetic evaluations are used by regula- tory bodies like the Food and Drug Admin- istration or by third-party payers in reach- ing their policy decisions; these evaluations indirectly and powerfully impinge on clin- ical practice and should be considered by someone. In this chapter no attempt was made to cover the effect of evaluations on decisions about the organization, adminis- tration, and support systems of health in- stitutions, although these too have an indi- rect bearing on medical practices. Nor has an attempt been made to catalog the im- pact of epidemiologic assessments on social policy (as in the areas of toxicology and en- vironmental health) or on public behavior (as in the decline in cigarette smoking) re- lated to health. The emphasis is on physi- cian practices and on the influence of vari- ous forms of clinical evaluation in changing those practices. Primary clinical evaluations could be ar- ranged in a hierarchy according to their freedom from bias, for instance, with the randomized clinical trial (RCT) at the top and then, moving downward, controlled (nonrandomized) studies, series of patients without controls, and personal recollection unaided by systematic record keeping. If even the weakest forms of evaluation count in our lexicon of evaluation, then most clinical practice is based on an evaluation. To refine the question posed at the outset of the chapter, we would like to know whether (and, possibly, to what extent) the more rigorous and powerful forms of pri- mary clinical assessment are more influen- tial than less rigorous forms in shaping medical practice and the policy decisions that affect the use of medical technology. In some instances, of course, a particular technology cannot be studied using the stronger methods, for example, when an RCT is not feasible because of sample size. 177 Diffusion refers to the spread of an inno- vation over time in a social system (Rogers and Shoemaker, 1971~. The concept in- cludes new practices being adopted and old practices being abandoned. Built into the notion of diffusion is the expectation that social change is not instantaneous and that some difference in practice among physicians at a moment in time is therefore reasonable and likely. Many studies of dif- fusion in the social sciences examine situa- tions in which measurable expansion or contraction in a practice occurs over the duration of the study rather than where patterns of practice (possibly including marked variation across individuals, insti- tutions, or geography) are relatively sta- ble. In many studies of diffusion, the correct- ness of knowledge available to the poten- tial adopter is taken for granted. An inno- vation or practice is regarded as objectively and knowably good or bad. The majority of these studies do not relate the nature and quality of evaluative evidence to the spread of a practice over time. Instead, dif- fusion studies tend to focus on characteris- tics of the innovation, characteristics of the potential adopters, communication chan- nels (bringing information to the adopter), the decision-making process, institutional features, and environmental forces that bear on the spread of a practice. Notions of evaluation, if introduced at all, tend to en- ter as attributes of the innovation (for ex- ample, the ease with which it can be tried on an interim basis) rather than as an inde- pendent determinant of the rate or extent of diffusion. Investigators concerned with the impact of scientific evidence on physician beliefs and practices frequently examine the state of practice at a single point in time rather than as a diffusion process over time. The prime interest in many of these studies has been to assess the knowledge and judgment

178 of physicians rather than to judge critically the effectiveness of a clinical trial in reach- ing or convincing physicians. This discus- sion is especially concerned with studies that relate evaluation and specific evalua- tion methods to changes in practice over time. Determinants of Diffusion Many factors bear on the adoption and abandonment of medical technology. The following discussion identifies 10 sources of influence. The first 4 (prevailing theory, attributes of the innovation, features of the clinical situation, and the presence of an advocate) are relatively insensitive to change by policymakers. The next 3 (prac- tice setting, decision-making process, and characteristics of the potential adopters) may be subject over time to some policy in- fluence. The remaining 3 (environmental constraints and incentives, conduct and methods of evaluation, and channels of communication) are relatively susceptible to influence by policymakers. Each factor is discussed briefly, with the greatest atten- tion given to the last group. Prevailing Theory Prevailing theory and accepted explanations for empirical phenomena appear to have a strong influ- ence on the acceptance of new ideas. Pre- vailing theories may delay the acceptance of ultimately proved innovations. Stern (1927) cites a number of classic examples, such as the resistance to smallpox vaccina- tion by those who held that improved sani- tation was the main cause of a decline in the smallpox rate, disbelief in the manifold consequences of syphilis by those who held to the theory of duality of tuberculosis; and refusal to recognize puerperal fever as a contagious disease by those who subscribed to atmospheric, cosmic, and telluric influ- ences on health. Twentieth century exam- ples include long-delayed acceptance of sa- licylates in the treatment of rheumatoid ASSESSING MEDICAL TECHNOLOGY arthritis (Goodwin and Goodwin, 1982) and a number of advances in cardiopulmo- nary medicine and surgery (Comroe, 1976~. In other cases, appeals to prevailing theory as a rational basis for belief about etiology and treatment of disease appeared to have hastened the acceptance of unsub- stantiated practices that were ultimately discarded. This occurred, for example, in the case of gastric freezing for the treat- ment of duodenal ulcers (Fineberg, 1979) and the conduct of subsequently discarded operations such as surgery for the endo- crine glands and surgery for constipation (Barnes, 1977~. Marks (Ideas as Social Re- forms: The Legacies of Randomized Clini- cal Trials, unpublished report, 1983) ar- gues that relatively new methods of evaluation (like RCTs) are themselves an innovation whose acceptance is influenced by prevailing theory about the nature of clinical evaluation and its role in medical decision making. The Innovation Innovations vary in the benefits and costs they offer the physi- cian and in their compatibility with the physician's experience and style of prac- tice. Diffusion of new practices is presum- ably enhanced by the extent to which they are easy to use, require little effort to learn, impose little change in practice style, are highly remunerative and satisfying, and have no clinically worthy competitors. The Clinical Situation An innovation that solves an important clinical problem and is seen as highly pertinent to practice is likely to be adopted more readily than an otherwise equally attractive innovation that addresses a less pressing or pertinent situation. Advocacy Successful diffusion of new practices often has been attributed to an authoritative advocate who promotes the innovation (Globe et al., 1967; Barnes, 1977; Fineberg, 1979~. Forceful advocates

EFFECTS OF CLINICAL EVALUATION have wrongly encouraged practices that were subsequently abandoned as ineffec- tive as well as practices that were eventu- ally proved effective. An authority figure who is correct in strongly promoting or op- posing one innovation may turn out to be wrong about a later innovation (Stern, 1927; Comroe, 1976; Fineberg, 1979~. The Potential Adopter Many studies of diffusion and of variation in medical prac- tices seek to explain patterns of practice in terms of physician attributes such as their technical skills, demographic characteris- tics, professional characteristics, socio- metric status, and attitudes toward inno- vation. In principle, changes in medical school admissions policies and in access to various types of specialty training could in time alter attributes of the physician popu- lation. The Practice Setting Several features of the setting and environment in which physicians practice can influence their use of medical technology. Physicians in group practices appear to adopt innovations more rapidly than physicians in solo prac- tice (Williamson, 197S). The size and teaching status of hospitals appear to influ- ence hospital acquisition of equipment, making possible new physician practices (Russell, 1979~. The pattern of practice among colleagues influences the way phy- sicians use available medical technology (Freeborn et al., 1972~. The Decision-Making Process Some medical practice decisions are wholly within the domain of the individual practi- tioner. Others are group decisions, and yet others require a concomitant or prior insti- tutional decision. A decision-making pro- cess that involves more people is likely to require a longer time to reach a conclusion. In a study of three anesthetic practices, the one that required a collective and institu- tional decision (scavenging for waste anes- 179 thetic gases) entailed several years longer delay between awareness and change in practice than was the case for the other two practices in which the physician could take action as an individual (Fineberg et al., 1978, 1980~. Environmental Constraints and Incen- tives Regulatory agencies and medical care insurers exercise direct and indirect control over the diffusion of many medical practices. Examples may be cited at the federal, state, and local levels. At the fed- eral level, the Food and Drug Administra- tion sets standards for the approval of new drugs and medical devices; the Health Care Financing Administration makes in- surance coverage decisions for Medicare patients that may prompt similar action by other third-party payers; and the Centers for Disease Control set the antigenic con- tent and recommended usage of vaccines, among other recommendations for prac- tice. At the state level, certificate-of-need programs, at least in principle, directly in- fluence hospital equipment and services, many of which impinge on clinicians' prac- tices. Locally, institutional review bodies and quality assurance efforts affect deci- sions about medical practices. The climate of malpractice litigation also may alter a clinicians' reliance on certain medical pro- cedures. All such environmental forces shape the opportunities and incentives for change in medical practice. Evaluation and Methods of Evalua- tion The factor that is of central concern in this chapter is the role of formal evalua- tion in shaping the behavior of physicians. Evaluation may act directly on the percep- tions of physicians; it may influence ex- perts who in turn influence physicians (through a channel of communication); or it may influence the policy decisions of reg- ulatory bodies (such as the Food and Drug Administration), or of third-party payers (such as the Health Care Financing Ad-

180 ministration), and hence alter the environ- ment in which medical practice decisions are made. In this chapter, all three possible chains of influence on physician practices are examined. Many clinicians are not well prepared to deal with quantitative methods in formal evaluations (Berwick et al., 1981~. Debates about the merits of particular evaluations may be the expression of fundamental dis- agreement about the nature and role of controlled clinical trials in medicine (Fein- stein, 1983; Bonchek, 1979; Marks, unpub- lished report, 1983~. Channels of Communication A sub- stantial number of diffusion studies in medicine have examined the ways in which physicians learn about new prac- tices. Investigators are interested in which sources of information and which channels of communication are most influential. Be- cause different channels of communication can to a degree be selected by clinical in- vestigators and potentially enhanced by policymakers, research in this area war- rants elaboration. A large body of early work on channels of communication concerned the dissemi- nation of drugs among medical practition- ers, especially the influence of face-to-face sales representatives and of social networks among physicians (Sherrington, 1965~. These early studies found that direct per- son-to-person contacts by drug company representatives were more influential than other forms of advertising (Caplow, 1954~. This finding has been reaffirmed in recent studies showing that personal representa- tion by pharmacists or, even more effec- tively, by other physicians can influence doctors to be more prudent drug prescrib- ers (Avorn and Soumerai, 1983; Schaffner, 1983~. Studies of how doctors learn about new medical practices, based on physician sur- veys, have found medical journals, discus- sion with colleagues, and continuing edu- ASSESSING MEDICAL TECHNOLOGY cation each to be regarded as important sources, with journals most consistently cited as high (Fineberg et al., 1978; Man- ning and Denson, 1979, 1980; Stross and Harlan, 1981; Market Facts, 1982; Jordan et al., 1983~. One study of physician awareness of pertinent findings published in a journal of a specialty different from their own found that most of those who were aware of the findings learned about them from consultants or colleagues (Stross, 1979~. In a study of three practices in anesthesiology, the channels of commu- nication (papers published in journals, col- leagues, and continuing education) dif- fered more in how many physicians they reached than in their persuasiveness to change practice (Fineberg et al., 1978~. Persuasiveness depended more on the na- ture of the clinical finding being communi- cated than on the channel of communica- tion. Whether formal medical training appears relatively important in conveying new knowledge (Jordan et al., 1983) or rel- atively unimportant (Manning and Den- son, 1979) may depend mainly on the re- cency of the innovation and on the age of the potential adopter. Other studies are be- ginning to assess the influence of National Institutes of Health (NIH) consensus con- ferences on specific physician practices (The Rand Corporation, 1983; Jacoby, Biomedical Technology Information Dis- semination and the NIH Consensus Devel- opment Process, unpublished report, 19834. In thinking about the implications of these studies, several additional points should be borne in mind. First, channels of communication that are perceived to be most effective may be specific to particular types of innovation (e. g., different for drugs than for surgical procedures) and the particular physician audiences (specialists or younger practitioners may be attuned to different channels than generalists or older physicians). Second, what physicians say or believe

EFFECTS OF CLINICAL EVALUATION influences them may differ from what ac- tually influences them (Avorn et al., 1982~. Third, investigators are unlikely to draw favorable conclusions about the impor- tance of channels of communication that they omit from their survey instruments. If, for example, a survey questionnaire lists "journals, books, conferences, and contin- uing education" as possible responses, then "talking with colleagues" is unlikely to be found important. If a study restricts itself to the social network of physicians in the transmission of information, then the role of journals is not likely to appear very prominent. Fourth, if a channel of communication has been less effective, it may be because few physicians have been exposed to that channel or because that channel is intrinsi- cally unconvincing compared with other sources of information. Continuing medi- cal education, for example, appears to have lesser perceived impact than medical journals. The policy implications are quite different if one believes that the weak showing is because of low exposure to con- tinuing education than if one believes that physicians are not convinced by what they see and hear at such educational programs. Fifth, as time passes after the initial re- lease of new information, different chan- nels of communication may become rela- tively more important as conveyors of information to physicians. The initial re- lease of new findings usually come in the form of a publication or presentation at a professional meeting. Soon after the re- lease of new findings, then, journals (or public news media) may be especially prominent sources (Stross and Harlan, 1981; Market Facts, 1982~. Later, col- leagues and medical teaching conferences may become more prominent than they were earlier (Fineberg et al., 1978; Stross and Harlan, 1979; Jordan et al., 1983~. Thus, in a study of the importance of cl~- ferent channels of communication, find- ings may depend in part on the timing of 181 the study relative to the time the innova- tion was introduced. All physicians are challenged to discern what they need to know from the sea of new medical information that surges around them. Physicians do appear open to the idea of receiving direct mail summaries of new medical findings (Market Facts, 1982~. The success of the brief monthly, The Medical Letter, has spawned a flock of imitators, and at least one medical text- book (Scientific American Medicine) pro- vides monthly updates with short summar- ies of key findings relevant to practice. Channels of communication that convey pertinent, concise information would seem to have a comparative advantage. The ex- panding availability of personal computers in physician homes and offices offers a new medium that can potentially convey evalu- ative information on new and current medical practices. Of course, the physi- cians most resistant to changing their med- ical practices may also be the last ones to install a home computer. Measures of Diffusion and Sources of Data Studies of diffusion and evaluation in- volve a variety of dependent variables, partly determined by the specific objec- tives of the study, partly by the nature of the medical practice being studied, and partly by the data available to the investi- gator. Interest here is in evidence about the di- rect and indirect effects of evaluation on (1) physician beliefs, (2) expert opinion, and (3) clinical practices. Measures of the first typically rely on surveys or interviews with practitioners and express results as a proportion who adhere to certain beliefs at a particular time or at different points in time. Studies of expert opinion may also use survey methods or rely on literature re- views or on recommendations in textbooks, review articles, or other guidelines written

182 by experts. Studies of medical practices may draw upon many data sources: (1) surveys of physicians or of institutions; (2) information gathered from patients or from the general public, as in the Health Interview Survey of the National Center for Health Statistics (NCEIS); (3) patient medical records; (4) pharmaceutical rec- ords; (5) information from manufacturers of equipment or of medical devices; (6) na- tional practice registries, such as the Na- tional Disease and Therapeutic Index; (7) insurance payment records; and (8) infor- mation from state and national data files on hospitalization and medical proce- dures, such as the NCHS Hospital Dis- charge Survey, reports of the Commission on Professional Hospital Activities, and several statewide data consortia. While admiring the ingenuity of many investigators in finding pertinent data, we should bear in mind limitations and biases lurking in these various measures. Surveys are subject to selection and recall biases. Manufacturer sales data may trace the dis- semination of new equipment, although that is not necessarily coincident with the extent of its use. Records from selected in- surers may be incomplete, and national registries may fail to reflect changing pat- terns of disease classification and new, im- portant though less than major, changes in patient management. The dependent variable in a diffusion study may be expressed as a count or as a proportion. For example, a certain num- ber of pills, number of prescriptions, or number of devices is supplied each year; or a certain fraction of physicians used a drug or a fraction of hospitals have purchased a new piece of equipment. Embedded in each of these dependent variables is a pop- ulation of potential adopters of the prac- tice being studied. The potential target population is explicitly the denominator in a fraction and is left unspecified in the case of counts. In either case, the interpretation of a study depends on an appropriately de- ASSESSING MEDICAL TECHNOLOGY fined target population that is either stable over time or is correctly adjusted over time, as, for example, the target pool of pa- tients varies or the number of trained clini- cians changes. Diffusion as Affected by Evaluation Diffusion may be considered a process of growth and decay over time. In an ideal- ized case, unambiguous new findings of an unequivocally superior innovation, or clear-cut determination of a definitively inferior current practice, would be instan- taneously communicated to all pertinent adopters who would promptly make the appropriate change in practice without any constraints or disincentives. The diffu- sion pattern would be an extremely sharp rise in the case of adoption and an ex- tremely sharp fall in the case of abandon- ment (Figure 4-1A and 4-1B). Evaluations often are not clear-cut, and any of the 10 factors that influence diffu- sion can introduce friction into the system. Empirically, a number of innovations have been found to follow an S-shaped pattern of diffusion (Figure 4-2A). A traditional so- ciometric model accounting for such a pat- tern postulates early adopters and opinion leaders who influence increasing numbers of other physicians to adopt a new prac- tice, leaving some resisters in the end who fail to adopt the innovation (Rogers and Shoemaker, 1971~. Other models invoking different determinants of diffusion (fea- tures of the innovation itself, of the setting for use, of the decision making process, etc.) could also be constructed to account for an observed S-shaped pattern of diffu- sion. Thus, an observed pattern of diffu- sion does not typically indicate the relative contributions of the various possible deter- minants of diffusion. Several studies of the spread! of medical equipment and institutional innovations have shown approximately S-shaped pat- terns of spread with slower initial rise fol-

EFFECTS OF CLINICAL EVALUATION A. Vertical Adoption — IIJ z LLJ ~ ~ O O ~ B. Vertical Abandonment Unequivocally Superior Innovation TIME . Definitively Inferior Current Practice TIME FIGURE 4-1A and 4-1B Idealized diffusion pattern. lowed by an accelerated phase (Fineberg et al., 1977; Fineberg, 1979; Russell, 1979; Office of Technology Assessment IOTA], 1981~. Some studies of the diffusion of new drugs have found a pattern of spread that is initially very rapid (Figure 4-2B) (Warner, 1975; Fineberg and Pearlman, 1981~. Fewer empirical studies have examined the phase of abandonment. Some cases show gradual abandonment (Figure 4-2C) (Fineberg et al., 1980~; others, involving drugs found to be unsafe, show rapid de- cline in usage shortly after release of the findings (Figure 4-2D) (Finkelstein and Gilbert, 1983~. An assessment of the impact of evalua- tion on practice requires a standard for judging impact. Two standards seem plau- sible, one that might be called rational be- havior, and a second that might be called expected behavior. With rational behavior as the standard, emphasis is on the extent to which practice conforms to the findings of evaluation, meaning all adopt for posi- tive evaluations and all abandon for nega- tive evaluations. With expected behavior 183 as the standard, emphasis is on changes in the pattern of diffusion that can be related to evaluation. A clinical evaluation might affect both the rate of adoption or abandonment of a practice and the extent of its ultimate use. Since these changes may be in a direction consistent with the findings of evaluation, though falling short of full conformance, the expected behavior standard is less de- manding than the rational behavior stan- dard. Investigators who base their conclu- sions on a rational standard often intend to judge the behavior of physicians, not the credibility of an evaluation method. An ex- pected behavior standard implicitly recog- nizes that evaluation is only one of many determinants of diffusion. Seeking to establish a relation between an evaluation and a change in diffusion can take the form of answering five kinds of questions: 1. What is the baseline pattern of diffu- sion? In other words, what pattern of adoption or abandonment of pertinent

184 ASSESSING MEDICAL TECHNOLOGY A. S-Shaped Adoption 6 ~ Lo Z LL ~ ~ O O ~ oh 6 ~ 111 Z ~ 111 Q ~ O O ~ B. Convex Upward Adoption / - TIME ~ C. S-Shaped Abandonment TIME - / TIME - D. Concave Upward Abandonment TIME - FIGURE 4-2A through 4-2D Empirical diffusion patterns. clinical practices would be expected to oc- cur over time if the evaluation in question had not been carried out? 2. What do results of the evaluation, when correctly interpreted, imply about what constitutes appropriate medical care? Do the implications require a change in the use of a medical technology? In ef- forts to discriminate among the effects of different types of evaluation, are the impli- cations from one type of evaluation differ- ent from the implications of others? 3. Are there changes in physician awareness and in the pattern of practice that are consistent with the findings of evaluation (or of one type of evaluation)? 4. What is the temporal relation be- tween evaluation and changes, if any, in the pattern of practice? 5. Is there additional evidence (such as interviews, bibliographic citations, opin- ion surveys, etc.) supporting a connection between an evaluation and change in prac- tice? Answers. to these questions can provide circumstantial evidence about the relation between evaluation and practice, making

EFFECTS OF CLINICAL EVALUATION a connection seem plausible or implausi- ble, though not established in a rigorous way. A controlled trial of the effectiveness of controlled trials has not been done and is hard to imagine. EVIDENCE ABOUT THE EFFECTS OF EVALUATION ON DIFFUSION The principle body of work used in this analysis consists of 48 papers, reports, books, and other documents that assess the relation between evaluation and medical practice. This highly diverse literature cuts across many medical specialty areas. (A summary of the literature is appended.) Though doubtless incomplete, this collec- tion is sufficiently rich to provide a basis for discussion. Several recent reviews con- cerned with the effects of clinical trials on medical practice aided the bibliographic search (Controlled Clinical Trials, Sep- tember 1982; OTA, 1983; Hawkins, Eval- uating the Benefit of Clinical Trials to Pa- tients, unpublished report, 1983~. The terminology in referring to this lit- erature can be confusing because it consists of studies of the effects of other studies. In the remainder of this chapter, the term study will be reserved for an analysis of the effects of one or more clinical trials, con- sensus exercises, or other forms of clinical evaluation on physician behavior in a par- ticular clinical situation. We use the words practice or clinical practice to mean physi- cian behavior. When referring to a clinical evaluation that is the subject of a study, we will refer to it as a clinical trial or, when appropriate, as a randomized clinical trial (RCT). The discussion is organized in two major categories depending on the directly mea- sured effect of evaluation: (1) effects cli- rectly on physician behavior and (2) effects directly on regulators or third-party pay- ers. 185 Evaluation and Physician Behavior All but two of the reviewed studies deal at least in part with the relation between clinical evaluation and the knowledge, be- liefs, and decisions of physicians. The kinds of evaluation whose effects are examined in these studies fall into two broad groups: primary evaluations, such as randomized clinical trials, which acquire and present new clinical findings; and synthetic evalu- ations, such as consensus conferences, which integrate and interpret available primary evidence. Thirty-eight studies deal at least in part with primary evalua- tions. The Impact of Primary Evaluations on Physicians In attempting to organize evi- dence about the relation between clinical trials and physician decision making, it is useful to distinguish two analytic strategies that may be adopted in a study (Gamier et al., 1982; OTA, 1983; Hawkins, unpub- lished report, 1983~. The first strategy be- gins with a clinical trial or trials and at- tempts to trace its effects on physician awareness or behavior. The second begins with a set of practices or innovations and traces back to the various kinds of evalua- tion that contributed to its development, dissemination, or abandonment. The next two subsections discuss studies that follow the first strategy and examine the effects of clinical trials on physician awareness and clinical decisions. The third subsection re- views studies that follow the second strat- egy and attempt to trace the origin of med- ical opinion or practices. Effects of Clinical Trials on Physician Awareness Two studies are devoted mainly to assessing physician awareness of findings from RCTs (Stross and Harlan, 1979, 1981~. In the first, family physicians and internists attending a continuing med- ical education (CME) course (on an un- specified subject) were asked whether they

186 were aware of findings about the treat- ment of diabetic retinopathy that had been published 18 months earlier in the Ameri- can journal of Ophthalmology. Two- thirds were not (72 percent of family prac- titioners and 54 percent of internists). Of the minority who did know about the study, two-thirds said they learned about it from an ophthalmologist or colleague. A1- though 70 citations to the controlled trial of treatment for diabetic retinopathy ap- peared in the medical literature between 1976 and 1979, none before 1978 was pub- lished in a general American medical jour- nal unrestricted in geographical or subject scope (Dunn, 1981~. In this case, numerous citations did not ensure the effective com- munication of a scientific finding. More- over, physician awareness would turn out to be no guarantee of improved medical practice. A later evaluation found that 60 percent of internists could not properly di- agnose proliferative diabetic retinopathy and so would be unable to recognize pa- tients to whom the initial RCT applies (Sussman et al., 1982~. In a second study of physician aware- ness, Stross and Harlan (1981) surveyed physicians attending CME courses about their knowledge of results from the Hyper- tension Detection and Follow-up Program (HDFP). This time, 40 percent of family physicians and 60 percent of internists, re- spectively 2 months and 6 months follow- ing publication of the RCT in the Journal of the American Medical Association in 1979, said they were aware of the findings, and most of those had learned about it from the literature. These figures represent awareness among physicians at single points in time for each physician group. Hence, the figures are not revealing about the diffusion over time of knowledge about this RCT, much less about its impact on clinical practices. As in the case of diabetic retinopathy, a later study would raise questions about the translation of this RCT into effective clinical practice, in part be- ASSESSIN(7 MEDICAL TECHNOLOGY cause of inappropriate medication pre- scribed by physicians and in part because of patient nonadherence to prescribed medication regimens (Wagner, 1981~. The management of hypertension illus- trates some of the pitfalls in attempting to draw conclusions about the effects of clini- cal evaluations on physician awareness and medical practice. In 1977, the Joint Na- tional Committee on Detection, Evalua- tion, and Treatment of High Blood Pres- sure recommended an individualized approach to treating patients with mild hypertension. A 1978 survey of physicians in New York City found that 92 percent were routinely starting antIhypertensive medication for patients with mild hyper- tension, a far more aggressive treatment strategy than recommended by the Na- tional Committee (Thomson, 1981~. Re- sults from the Hypertension Detection and Follow-up Program subsequently lent sup- port to the more aggressive treatment strat- egy previously followed by more than nine out of ten physicians in New York City. In this case, a widespread practice that ap- peared unjustified at one time was borne out by a subsequent randomized trial; thus, demonstration of efficacy by an RCT may follow as well as precede prevailing patterns of practice. The optimal strategy for treatment of mild hypertension con- tinues to be controversial, and value judg- ments about the conformance of physician practices to a particular management strategy would seem more hazardous today than perhaps they once appeared. A set of surveys recently commissioned by the National Heart, Lung, and Blood Institute (NHLBI) investigated physician awareness of findings from two specific RCTs and their effects on practice (Market Facts, 1982~. This study design is notewor- thy in that it represents an exceptional ef- fort to obtain data on physician knowledge and attitudes both before and after release of findings in 1980 from an RCT (the Aspi- rin Mvocardial Infarction Study FAMISH

EFFECTS OF CLINICAL EVALUATION trial) that examined whether aspirin pre- vents recurrence of myocardial infarction. This RCT found no benefits to aspirin, in contrast to some earlier clinical trials of antiplatelet therapy (Friedewald and Schoenberger, 1982~. The majority of phy- sicians in the survey remained unaffected by findings from the AMIS trial and con- tinued to recommend aspirin for patients following a myocardial infarction. The second subject of the survey was an RCT, the Coronary Drug Project (CDP), that had been published more than 4 years be- fore the initial survey. This RCT had raised doubts about the efficacy and safety of lipid-lowering drugs in preventing re- currence of myocardial infarction. The majority of physicians in the survey either never prescribed lipid-lowering drugs (14 percent) or used them only as secondary therapy (72 percent), and physicians who were aware of the CDP results were less likely than other physicians to use these drugs. A recent paper about the impact of the CDP draws in part on pertinent findings from the surveys sponsored by the NHLBI (Friedman et al., 1983~. Of the more than 1,700 physicians interviewed, 45 percent said they had no familiarity with the CDP, 39 percent said they were familiar with the CDP in general, and 17 percent said they were familiar with its specific findings. When sorted by specialty group, 16 per- cent of cardiologists, 46 percent of inter- nists, and 70 percent of general practition- ers had no familiarity with the CDP. Physicians claiming some or full familiar- ity with CDP were approximately twice as likely as physicians with no familiarity with CDP to believe that specific lipid low- ering drugs are not effective or safe. Another survey of physician awareness of new clinical knowledge focused on three well-established findings in anesthesiology (Fineberg et al., 1978~. In this study there appeared to be a marked acceleration in the spread of knowledge among anesthesi- 187 ologists, following initial publication of one finding, though not for the other two. Change in practice lagged to varying de- grees after awareness and was longest for the one practice (scavenging waste gases) requiring institutional action. Effects of RCTs on Physician Prac- tices Twenty papers attempt to assess the effect of specific randomized clinical trials on medical practice (see Appendix A). This group of papers is of particular interest be- cause RCTs are technically the strongest form of evaluation. Several papers cover more than one clinical practice, and a number of practices are covered in more than one paper. On balance, the 20 papers deal with the effects of RCTs on 19 clinical practices. If we count as one study each instance in which a paper assesses the effects of one or more RCTs on a particular clinical prac- tice, the 20 papers represent 28 studies (Ta- ble 4-1~. One paper (Chalmers, 1974) as- sesses the effects of randomized clinical trials on four medical practices. Three other papers (Friedewald and Schoen- berger, 1982 Fisher and Kennedy, 1982; Market Facts, 1982), respectively, cover four, two, and two practices in the area of cardiovascular disease, accounting for a to- tal of 8 studies. The remaining 16 papers each deal with a single clinical practice. Five clinical practices (internal mammary artery ligation, gastric freezing, treatment of breast cancer, antiplatelet agents for the prevention of recurrent myocardial infarc- tion, and management of mild hyperten- sion) are each examined in 2 papers. The effect of the Coronary Drug Project on the use of lipid-lowering drugs and of the Uni- versity Group Diabetes Project (UGDP) on the use of oral hypoglycemics are each cov- ered in 3 studies. Examination of these 28 studies proceeds along a line of questioning like that out- lined at the end of the section on diffusion

188 ASSESSING MEDICAL TECHNOLOGY TABLE 4-1 Studies of the Effects of Randomized Controlled Trials on Medical Practice Practices Study No. References Date Action Medical Condition 1 Barsamian 1977 Internal mammary artery Angina pectoris ligation 2 Chalmers 1974 Stilbestrol Pregnancy 3 Chalmers 1974 Bed rest Viral hepatitis 4 Chalmers 1974 Bland diet Duodenal ulcer 5 Chalmers 1974 Oral hypoglycemics Diabetes mellitus 6 Chassin 1983 Length of hospitalstay Myocardial infarction 7 Combs et al. 1983 Referral for treatment Senile macular degeneration 8 Fineberg 1979 Artificial lung Respiratory failure 9 Fineberg and Hiatt 1979 Gastric freezing Ulcer disease 10 Finkelstein and Gilbert 1983 Tolbutamide Diabetes mellitus 11 Fisher and Kennedy 1982 Coronary artery bypass Cardiovascular graft surgery disease 12 Fisher and Kennedy 1982 Internal mammary artery Angina pectoris ligation 13 Friedewald and Schoenberger 1982 Drug prescription Mild hypertension 14 Friedewald and Schoenberger 1982 Antiplatelets Cardiovascular disease 15 Friedewald and Schoenberger 1982 Beta-blockers Cardiovascular disease 16 Friedewald and Schoenberger 1982 Lipid-lowering drugs Cardiovascular disease 17 Friedman et al. 1983 Lipid-lowering drugs Cardiovascular disease 18 Haines 1983 Neurosurgery Neurosurgical conditions 19 Market Facts 1982 Lipid-lowering drugs Cardiovascular disease 20 Market Facts 1982 Antiplatelets Cardiovascular disease 21 McPherson and Fox 1977 Treatments Breast cancer 22 Miao 1977 Gastric freezing Ulcer disease 23 Moskowitz et al. 1981 Treatment Alcohol withdrawal 24 OTA 1978 Hyperbaric oxygen Cognitive deficits 25 OTA 1983 Treatments Breast cancer 26 Thomson et al. 1981 Drug prescription Mild hypertension 27 Warner et al. 1978 Oral hypoglycemics Diabetes mellitus 28 Wilson et al. 1982 Antimicrobial prophylaxis Gastrointestinal surgery as affected by evaluation. Based on the view of the authors of 25 studies, 17 prac- tices have had RCTs with clear implica- tions for medical practice. Of these 25 studies, 5 deal with two RCTs (UGDP study of treatments of diabetics and the HDFP study of hypertension) where the clinical implications of the RCT are con- troversial, though not in the view of the authors of the papers cited. The RCTs in two areas (use of antiplatelet agents to pre- vent recurrent myocardial infarction and a collection of randomized trials on neuro- surgical procedures) do not have clear im- plications for practice in the views of the authors of the 3 studies that deal with RCTs in these two areas (see Table 4-2~. Among the 25 studies in which RCTs are judged to have had clear implications for practice, 17 (dealing with 13 practices) provide some quantitative information about the frequency of use of the medical practice in question. Of these 17 studies, 7 describe the use of a practice at one point in time. The remaining 10 studies report quantitative information about patterns of

EFFECTS OF CLINICAL EVALUATION TABLE 4-2 Effects of Specific RCTs on Medical Practice F. A. Does RCT have clear implications for practice: No—(3) t2] Yes—(27) t17] B. Is practice pattern quantitatively reported? No—(8) [7] Yes—il7) t13] Is practice measured over time or only at one point in time? Over Time One Point (10) t7] (7) t7] D. Does practice level or pattern conform to RCT findings? il)tl] No (4~4] (5)i4 i Somewhat (1)~11 (9~7] (4)[4] Yes (2~2] (3~3] E. Did RCT precede measured change in pattern of practice? (1~1] No (1~1] Probably not (1~1] Probably yes (7)~5] (6~[4] Yes (2~2] (2~2] Do RCT results differ from results of other (less strong) forms of evaluation or from traditional practice? Uncertain (4~3] Somewhat (7)~5] (3~4] Yes (1~1] (2~2] NOTE: A total of 28 studies (in parentheses) and 9 practices (in brackets) are considered here. practice over time and thus provide evi- dence about the relation between RCTs and diffusion of medical technology. This latter group of 10 studies covers seven medical practices: oral hypoglyce- mics for diabetics (Chalmers, 1974; Warner et al., 1978; Finkelstein and Gil- bert, 1983~; referrals for treatment of se- nile macular degeneration (Combs et al., 1983~; gastric freezing (Fineberg, 1979~; lipid-lowering drugs (Market Facts, 1982; Friedman et al., 1983~; length of hospital stay for myocardial infarction (Chassin, 1983~; coronary artery bypass graft surgery (Fisher and Kennedy, 1982~; and treat- ment of breast cancer (OTA, 1983~. One study found no change in practice consis- tent with the findings of an RCT; this 189 study examined application of the UGDP study in 1970 (Chalmers, 1974~. Five stud- ies found a moderate shift in practice con- sistent with the findings of previous RCTs. These dealt with the treatment of breast cancer between 1972 and 1981 (OTA, 1983~; with coronary artery bypass graft surgery between 1974 and 1979 (Fisher and Kennedy, 1982~; with the use of lipid- lowering agents between 4 and 6 years af- ter publication in 1975 or the CDP results (Market Facts, 1982; Friedman et al., 1983~; and with prescriptions for oral hy- poglycemic agents between 3 and 7 years after publication of the UGDP study (Warner et al., 1978~. In the first two of these four cases (treatment of breast cancer and coronary artery bypass graft surgery), some evidence from non-RCTs favored the same directions of practice as did the RCTs. Four studies (Fineberg, 1979; Finkelstein and Gilbert, 1983; Combs et al., 1983; Chassin, 1983) found a marked shift in practice consistent with the find- ings of RCTs, though in the two cases of gastric freezing (Fineberg, 1979) and length of stay for myocardial infarction (Chassin, 1983) evaluation by RCT did not precede the change in the pattern of prac- tice. After applying all of these analytic fil- ters; there remain 2 of the original 28 stud- ies of the effects of specific RCTs on medi- cal practices in which the RCT has clear implications for practice, the pattern of practice reported quantitatively over time conforms fully to the RCT findings, the RCT preceded the change in the pattern of practice, and findings from the RCT differ from the results of other forms of evalua- tion. One of these is an unpublished study of referrals of patients with senile macular degeneration at one opthalmological treat- ment center (Combs et al., 1983~. Twice as many patients with treatable disease were referred in the 6 months following release of the results of the senile macular degener- ation trial as had been seen in the 6 months

190 before the trial. The second study found that the use of tolbutamide began to de- cline promptly and sharply after release of the UGDP results in 1970 (Finkelstein and Gilbert, 1983~. Studies of oral hypoglycemics, of treat- ment for breast cancer, and of lipid-lower- ing drugs highlight the importance of the passage of time in drawing conclusions about the effects of RCTs on clinical prac- tices. In the first few years after release of the UGDP study, doctors apparently sub- stituted other hypoglycemic agents for tolbutamide, because the use of all hypo- glycemics remained steady through 1973 (Chalmers, 1974) despite the fall in use of tolbutamide beginning in 1971 (Finkel- stein and Gilbert, 1983~. By 1977, use of all hypoglycemics had declined by half (Warner et al., 1978; Chalmers, 1982~. Similarly, conclusions about the effect of RCTs on the treatment of breast cancer that are based on the pattern of treatment in 1970 (McPherson and Fox, 1977) differ from conclusions based on trends during the subsequent decade (OTA, 1983~. In the period between 4 and 6 years after publica- tion of results of the CDP in 1975, the pro- portion of physicians who said they never prescribe lipid-lowering drugs rose from 10 to 18 percent (Market Facts, 1982~. This occurred despite the fact that the propor- tion of physicians demonstrating full knowledge of COP findings remained low (6 percent of all cardiologists and 3 percent of all physicians). Such results support the importance of secondary spread of new findings from one or multiple primary publications through such channels as medical conferences and interaction with colleagues. The relation between the level or pat- tern of a clinical practice and the results of a clinical trial can reasonably be described only as of a given time after the release of findings from the trial. The full effects of an RCT on practice may take some years to manifest. The occurrence of other events ASSESSING MEDICAL TECHNOLOGY (new technical developments, evaluations, environmental changes, etc.) in the in- terim may make it more difficult to con- nect changes in practice with an RCT that has in fact been influential. Some difficulties in discerning temporal and causal connections between clinical evaluations and changes in medical prac- tice are illustrated by the case of length of hospital stay for myocardial infarction. In this instance of medical decision making, changing patterns of practice, randomized trials, and evidence from nonrandomized studies all overlap in time in a way that de- fies inferences of causal relations. The av- erage length of hospital stay for myocar- dial infarction has declined steadily in the United States, falling from almost 19 days in 1968 to less than 13 days in 1980 (Figure 4-3~. During this time, marked variation in length of stay from one region of the coun- try to another persisted. Superimposed on the trend line for hos- pital length of stay in Figure 4-4 are the du- rations of hospitalization examined in ran- domized and nonrandomized controlled trials of hospital stay for myocardial in- farction (Chassin, 1983; Pryor et al., 1983~. Virtually all studies have found no statistically significant differences in out- come between shorter and longer hospital stays (Chassin, 1983; Pryor et al., 1983~. Because the evidence from all trials points in the same direction, it is impossible to distinguish the impact of one or another trial on clinical practice. Moreover, be- cause of the limited size of available trials, it is possible that earlier hospital discharge may pose some health hazard despite the array of evidence showing no statistically significant differences between shorter and longer hospital stays (Chassin, 1983~. On the other hand, it is also possible that ear- lier discharge confers as yet unestablished benefits. The trend in practice may be the result of multiple factors, including chang- ing theories about cardiovascular healing and external pressures to shorten hospital

EFFECTS OF CLINICAL EVALUATION 20 En - 15 Oh ~ 10 - In o I \ United Stan - ~ Northeast ~ West \ _ - - - O : ~/1 ~ 1 1 1 1 1 1 1 1 1 1 1 1 1 19561966 1968 1970 1972 1974 1976 1978 1980 YEAR FIGURE 4-3 Average length of hospital stay for myocardial infarction. stays, quite apart from evidence produced by controlled clinical trials. It is evident from inspection of Figure 4-4 that many physicians have been discharging patients with myocardial infarction earlier than had yet been demonstrated as safe by ran- domized trials. Simply because a study fails to demon- strate beyond a reasonable doubt that a clinical trial has influenced the diffusion of a practice does not mean that the clinical trial was not actually influential. For ex- ample, it is possible that early randomized trials helped to prevent the unwarranted diffusion of the artificial lung (Fineberg and Hiatt, 1979) and of hyperbaric oxygen to treat cognitive deficits in the elderly 191 (OTA, 1978~. When an RCT reinforces the value of an existing practice, this may be beneficial in preventing unwarranted de- partures from the standard practice and in influencing nonconfirming physicians to adopt the demonstrably superior practice. In some cases the patterns of practice over time conform partially to the findings of randomized trials. In some of these cases, such as coronary artery surgery, treatment of breast cancer, and use of lipid-lowering drugs, it seems highly likely that RCTs have influenced clinical prac- tices. When multiple RCTs and possibly other studies as well all suggest changes in practice in a similar direction, it may be difficult to discern the particular effect of a

192 ASSESSING MEDICAL TECHNOLOGY 25r 20 15 CD - 10 CL cn I _ _ O 5 _ ~ Uncontrolled trials n=8 O Non-randomized, controlled trials n = 4 O Randomized trials n = 6 O O \ United State:\ - \ Northeast - \O West \ 0 A\ to o \ - ~ O O 1 //1 1 1 1 1 1 1 1 1 1 1 1 ~ 1 1 19561966 1968 1970 1972 1974 1976 1978 1980 YEAR FIGURE 4-4 Average length of hospital stay for myocardial infarction. single study; yet, circumstantial evidence supporting the eventual influence of the collection of studies can be strong. The Coronary Artery Surgery Study en- rolled nearly 25,000 patients undergoing angiography at 15 institutions (Fisher and Kennedy, 1982~. The proportion of pa- tients with one-vessel disease who under- went surgery declined from 38 percent in 1974-1975 to 30 percent in 1978-1979, a result consistent with the findings of ran- domized trials that had appeared in the in- terim. Perhaps more striking than the over- all effects of these randomized trials is the marked variation in reliance on surgery at participating hospitals. The proportion of angiographically examined patients under- going surgery ranged from 31 to 70 percent at the different hospitals, an observation consistent with highly variable responsive- ness to randomized trials in different set- tings. In the case of breast cancer, radical mas- tectomy declined from 70 percent of opera-

EFFECTS OF CLINICAL EVALUATION tive procedures in 1970 (McPherson and Fox, 1977) to about 3 percent in 1981 (OTA, 1983~. The dominant practice in 1981, accounting for 70 percent of opera- tive procedures, was the modified radical mastectomy, though randomized trials suggested that simple mastectomy with lo- cal irradiation would be equally effective (OTA, 1983~. The number of prescriptions for lipid- lowering agents rose between 1970 and 1975, then declined by more than 50 per- cent between 1975 and 1980 (Friedman et al., 1983~. A number of RCTs reported be- tween 1970 and 1980 support the overall conclusions that lipid-lowering drugs do not demonstrably improve the course of cardiovascular disease (Buchwald et al., 1982~. Prominent among these trials are the CDP discussed earlier and the World Health Organization (WHO) (European) trials published in 1978 and 1980 showing that clofibrate, the most widely used lipid- lowering agent, is ineffective in the pri- mary prevention of myocardial infarction and possibly adds to the risk of mortality (Buchwald et al., 1982; Friedman et al., 1983~. Among these RCTs, only the CDP has been the subject of a specific inquiry about awareness among U.S. physicians, and it is uncertain which trials have had the greatest impact on physician practices. While physician familiarity with CDP findings is limited, and many continue to prescribe lipid-lowering agents, there has been a gradual shift in practice along the lines suggested by these RCTs. Types of Evaluation That Precede Ac- cepted Medical Practices A number of studies attempt to assess the roles played by different types of evaluation prior to the general acceptance of a medical practice. Of special interest are studies that compare the influence of randomized and of non- randomized clinical trials. According to an opinion expressed by the Office of Tech- nology Assessment, randomized clinical 193 trials have been applied to 10 or 20 percent of medical practices (OTA, 1983~. In some areas of medical practice, such as the de- velopment of cancer chemotherapy, thou- sands of randomized trials have been con- ducted (Armitage et al., 1978~. The frequency of randomized trials in other areas of medicine has been increasing (Chalmers et al., 1979~. In the opinion of many oncologists and researchers involved in evaluations of can- cer treatment, randomized trials have gen- erally been more useful than nonran- domized trials in the development of cancer therapies (Armitage et al., 1978, Rockette et al., 1982~. However, several studies that have looked quantitatively at the origins of current therapeutic practices in several types of cancer found nonran- domized studies to have played a dominant role in the development of therapy. For ex- ample, Gehan (1982) found that nonran- domized trials, more frequently than RCTs, were the source of currently ac- cepted treatments for acute leukemia. In their oncology center, Gamier et al. (1982) found the greatest consensus around treat- ments for head and neck cancer that had not been evaluated by randomized trials. Part of the reason may be that randomized studies are especially likely to be applied to areas of controversy. Other studies also suggest a nondomi- nant role for randomized studies in shap- ing medical practices. Ingelfinger et al. (1974) reviewed 23 controversies in inter- nal medicine that had been debated in print 8 years earlier. Though interest in some had waned, few if any had been re- solved by a clear-cut clinical trial. Gilbert et al. (1977) traced the introduction of 107 innovations in surgery and anesthesia and found that one-third were randomized trials. In reviewing the literature about four discarded surgical practices, Barnes (1977) noted a lack of randomized trials at both the beginning and end of the life cy- cles of the discarded procedures. Christen-

194 sen et al. (1977) compared the treatments for duodenal ulcer recommended in text- books with evidence from randomized tri- als. They found a number of treatments, including the use of antacids, continued to be recommended without having been demonstrated effective in randomized studies. Antacids in sufficient amounts subsequently were shown to be more effec- tive than placebo in promoting ulcer heal- ing (Peterson et al., 1977~. Like the earlier example of treatment for mild hyperten- sion, this sequence of studies shows that ac- ceptance of a medical practice sometimes precedes a demonstration of efficacy in randomized trials. The Impact of Synthetic Evaluations Synthetic evaluations encompass a variety of efforts to interpret and integrate infor- mation obtained from primary evalua- tions. Synthetic evaluations include review articles, decision analyses, cost-effective- ness assessments, informal and formal group processes to reach consensus, and many more. It is, of course, difficult to dis- cern the effects on practice of review pa- pers apart from the primary evaluations on which they are based. Although decision analysis and cost-effectiveness analysis are being applied to medical problems with in- creasing frequency, much more has been written about the methods and limitations of such analyses than has been written about the adoption of their results by phy- sicians. Because this chapter concerns the effects of evaluation on clinical practices, evidence will not be reviewed about the ef- fects of programs that entail regulatory sanctions, like hospital utilization review. Our primary focus in this section is on stud- ies of the impact on physician knowledge and practice of expert consensus and guide- lines. A major study of the effectiveness of the National Institutes of Health Consensus Development Program is currently under way (The Rand Corporation, 1983~. This ASSESSING MEDICAL TECHNOLOGY study is intended to trace the dissemination of conclusions from NIH consensus confer- ences into the medical community and to measure changes over time in reliance on practices that have been the subject of a consensus conference. Its results are ex- pected to be available in 1985. At least one survey has measured physi- cian awareness of the occurrence and con- clusions from two NIH consensus confer- ences by means of telephone surveys 2 weeks before each conference and 6 weeks after publication of conclusions in the Journal of the American Medical Associa- tion (lacoby, unpublished report, 1983~. Approximately 9 percent of physicians in the survey were aware of the consensus de- velopment program at NIH. Two weeks before the consensus conference on com- puted tomography, 16 percent of physi- cians in pertinent specialties said they knew about the upcoming conference. At a similar time before the consensus confer- ence on hip joint replacement, 7 percent of pertinent specialists reported being aware of the upcoming conference. Six weeks af- ter publication of results, 4 percent of sur- veyed physicians reported being aware of conclusions from the conference on com- puted tomography, and 1 percent said they were aware of the conclusions from the conference on hip replacement. Most of those who knew about conference conclu- sions said they had read about them in a professional journal. An earlier section described a study of physician attitudes and practices following a consensus conference on hypertension (Thomson et al., 1981~. Physicians were more aggressive in treating mild hyperten- sives than was recommended by the con- sensus conference, though such practice was consistent with the findings of the sub- sequently published HDFP trial. Since the mid-1970s, the Blue Cross and Blue Shield Association has sponsored a program designed to produce medical con- sensus on appropriate practices and to im-

EFFECTS OF CLINICAL EVALUATION prove physician use of medical resources. In 1977, this Medical Necessity Program announced consensus on 42 outmoded di- agnostic tests and surgical procedures that should no longer be performed. Member Blue Cross and Blue Shield Plans were ad- vised to discontinue routine payment for these procedures and to require physicians requesting reimbursement to provide a medical justification. A review of insur- ance claims for the federal employee health benefits program in 1975 and 1978 revealed a decline in claims for listed surgi- cal procedures of 26 percent and a decline in diagnostic test claims of 85 percent (Blue Cross and Blue Shield Association, 1982~. These results are suggestive of an effect, though without any earlier points of refer- ence, it is difficult to know how much of the decline in tests and surgery might be at- tributable to the pronouncement by the Medical Necessity Program in 1977. In early 1979, the Medical Necessity Program discouraged routine use of admission test batteries, and most Blue Cross and Blue Shield Plans are reportedly working with hospitals to implement this policy (Blue Cross and Blue Shield Association, 1982~. In October 1982, the Medical Necessity Program announced and widely dissemi- nated new guidelines for respiratory care that were developed in association with the American College of Physicians, the Amer- ican College of Surgeons, and the Ameri- can Academy of Pediatrics. The aim of this new phase of the program is to reduce un- ~ . . . necessary reliance on respiratory care ser- vices, estimated to account for two to four billion dollars of annual hospital costs (Blue Cross and Blue Shield Association, 1982~. More recently (1984), Diagnostic Imaging Guidelines were released in asso- ciation with a number of medical specialty societies, primarily the American College of Radiology. . . ~ A number or professional associations have undertaken programs to improve the use of medical resources, relying heavily on 195 guidelines generated by expert panels. One of these, the Clinical Efficacy Assessment Project of the American College of Physi- cians (ACP), uses expert committees, con- sultants, review by relevant medical pro- fessional societies, and approval by the Board of Governors and Regents of ACP to synthesize the literature and prepare posi- tion papers on the safety and efficacy of a variety of clinical practices. These papers have begun to appear periodically in the Annals of Internal Medicine. The Public Health Service publishes a variety of clinical guidelines for preven- tion, diagnosis, and treatment of diseases of public health importance. Such recom- mendations appear, for example, in Mor- bidity and Mortality Weekly Reports. Widely distributed to subscribers, these recommendations are surely influential, though their overall effects on the diffusion of new practices has not been systemati- cally assessed. One survey of public health physicians responsible for tuberculosis con- trol found greater conformity to public health service guidelines on chemopro- phylaxis than on diagnosis and treatment (Leff et al., 1979, 1981~. EVALUATION AND POLICY ON REGULATION AND REIMBURSEMENT Requirements of the Food and Drug Ad- ministration (FDA) to establish the safety and efficacy of new drugs are doubtless among the strongest promoters of con- trolled clinical trials in the United States. Until the FDA approves a new pharmaceu- tical, it is not generally available for use by physicians. FDA's insistence and reliance on randomized controlled trials in the li- censing of new drugs probably constitute the most potent source of influence of RCTs on the diffusion of new medical technology in the United States. Although the FDA approves particular uses of a new drug, physicians are gener- ally free to prescribe a licensed drug as they

196 ASSESSING MEDICAL TECHNOLOGY see fit. Studies of specific new drugs, such uted to decisions by Medicare and other in- as cimetidine, have found them to be used surers not to reimburse for the procedure frequently in ways that are not explicitly (OTA, 1978~. approved by the FDA (Cocco and Cocco, 1981; Schade and Donaldson, 1981; Fine- berg and Pearlman, 1981~. Friedman et al. (1983) reviewed changes in FDA labeling for the lipid-lowering drug, clofibrate, between 1969 and 1982. During this period, indications for the drug were progressively restricted and warnings of adverse reactions expanded. The authors ascribe major changes in 1979 both to the Coronary Drug Project pub- lished in 1975 and to the WHO (European) clofibrate trial published in 1978. Further labeling changes in 1982 were due primar- ily to new findings in the WHO trial pub- lished in 1980. When the Health Care Financing Ad- ministration receives a request for reim- bursement of a new medical procedure about which it is undecided, it may seek the advice of the Office of Health Technol- ogy Assessment of the Public Health Ser- vice about the medical acceptability of the practice. During 1979 and 1980, the Public Health Service made recommendations of nonreimbursement for 21 of 50 procedures it reviewed. The Health Care Financing Administration accepted all these recom- mendations. One study commissioned in 1980 by the now defunct National Center for Health Care Technology estimated that decisions not to reimburse four procedures (dialysis for schizophrenia, hyperthermia for cancer, radial keratotomy for myopia, and endothelial cell photography) pro- duced savings to the Medicare program of $312 million over a 10-year period (Center for the Analysis of Health Practices, 1981~. We are aware of no systematic review of the contribution of various types of clinical evaluation to the recommendations pro- duced by the Public Health Service. Ac- cording to the OTA, an RCT showing no benefit from hyperbaric oxygen treatment for cognitive deficits in the elderly contrib- CONCLUSIONS AND RECOMMENDATIONS Clinical evaluation is one among many factors bearing on the diffusion of medical technology. Evaluations often seem to be overwhelmed by the other nine determi- nants of physician behavior discussed in this chapter. Improving the care of pa- tients requires both improved methods of evaluation and more effective translation of the results of evaluation into practice. Evaluations are likely to exert a greater im- pact on diffusion if they are buttressed by attention to other controllable factors, such as channels of communication and en- vironmental constraints and incentives, that affect the adoption and abandonment of medical technology. The discussion in this chapter leads to the following recommendations (in italics). · Strengthen the weaker methods of evaluating medical practice and increase use of stronger methods. Those who are working to improve methods for evaluat- ing medical practices should attend to strengthening the weaker methods, such as case studies and nonrandomized trials, in order to enhance their reliability as guides to clinical action. At present, stronger forms of evaluation, such as controlled trials, are not notably more successful than weaker forms in shaping medical prac- tices. Recognizing the widespread use and apparent influence of these weaker meth- ods of evaluation, efforts to improve both the methods and their execution might have a substantial impact on clinical care. The advantage of scientifically stronger forms of evaluation is not that they ensure greater impact on clinical practice; the ad- vantage is that whatever impact the evalu-

EFFECTS OF CLINICAL EVALUATION ation carries will be more likely to benefit the patient. · Study the diffusion of medical prac- tice concepts and procedures so as to un- derstand how to speed adoption of good practices and discourage use of those that are less effective or harmful. Government agencies, private foundations, and other interested organizations should support re- search to enhance understanding of the process of diffusion of medical practices. Instances of successful and of unsuccessful dissemination and instances in which eval- uation does and does not prominently ap- pear to influence diffusion should offer guidance. The research program should place special emphasis on those factors af- fecting diffusion, such as the channels of communication and the environmental constraints, that are at least partially con- trollable by public policy and organiza- tional decisions. Researchers and funding agencies should support research to develop and test more acceptable and effective ways to has- ten dissemination of demonstrably benefi- cial practices and to promote abandon- ment of inappropriate practices. This research is aimed also at finding how cur- rent policies and programs influence medi- cal practices. The range of possible inter- ventions that might be assessed includes (1) education of physicians and medical students about evaluation methods, princi- ples of decision making, and the effects of specific clinical practices; (2) innovative methods of communicating the results of clinical evaluations to physicians and other decision makers, including the possible use of electronic media such as home com- puters and medically oriented cable televi- sion stations; (3) feedback to physicians about their performance compared with established standards of practice or peer performance; (4) changes in administra- tive procedures, affecting such practices as standing orders, test requests, and use of prophylactic antimicrobials, that can lead 197 to more thoughtful clinical decisions; (5) changes in reimbursement policy, in- cluding the possibility of material rewards to medical institutions and to physicians as incentives to improve the use of medical re- sources, (6) requirements or quotas limit- ing access to medical practices that are used excessively, that are unproved, or that require special facilities or training to use optimally; and (7) changes in the regula- tory authority or procedures of such agen- cies as the federal Food and Drug Adminis- tration and state Determination of Need programs. · Establish lines of responsibility for making better medical practice a conse- quence of evaluation of medical technol- ogy. Responsibility for promoting a closer connection between the results of evalua- tion and medical practice is widely shared among clinicians, medical educators, re- searchers, professional bodies, research organizations, journal editors and pub- lishers, medical care institutions, equip- ment and pharmaceutical manufacturers, third-party payers, and regulatory agen- cies. These individuals and organizations have mixed interests and differing perspec- tives, though all are in principle dedicated to improved health care for the public. The diffuseness of responsibility for translating the results of evaluation into improved health care is one motivation behind pro- posals for a public-private entity sponsored by the Institute of Medicine and for addi- tional forms of organization discussed in this report. Each responsible party can and should take action that would strengthen the link between the results of evaluation and prac- tice. For example, medical educators can place greater emphasis on training medical students and physicians to be discerning consumers of clinical evaluations and to be knowledgeable about medical decision making. Editors of professional journals can enhance the technical review of sub- mitted papers by involving more statisti- .

198 clans, epidemiologists, and other method- ologists in their review processes. Medical specialty societies can develop and dissemi- nate authoritative guidelines for the use of medical technology within the purview of their clinical expertise. Third-party payers can support demonstration projects to re- duce the inappropriate use of clinical prac- tices. APPENDIX 4-A: SUMMARIES OF STUDIES ASSESSING MEDICAL TECHNOLOGY Many of these recommended actions are already under way in some degree, though they need to be extended, reinforced, and coordinated. A public or public-private body charged with responsibility can iden- tify priorities, enhance communication among the interested organizations, and promote systematic efforts to strengthen the connection between sound evaluation and the diffusion of medical technology. Reference Topic (Year) Studies of Patterns of Clinical Evaluation and of the Effect of Clinical Evaluations on the Opinions of Experts Drugs to lower Market Facts blood lipo- (1982) proteins Type of Evaluation Source of Data Studied on Effects Findings Aspirin to re- duce recur- rence of myocardial infarction Market Facts i 1982) Three practices Fineberg et al. in anesthesi- (1978) ology Treatment of diabetic retinopathy Specific RCT: Cor- onary Drug Proj- ect (CDP) of NHLBI Survey of 1,700 physi- Physicians aware of CDP clans before and af- results were less likely to ter release of AMIS prescribe lipid-lowering results; all inter- drugs (consistent with views followed study results). publication of CDP results. Specific RCT: Aspi- Survey of 1,700 physi- Knowledge of AMIS not rin Myocardial clans before and af- clearly related to atti- Infarction Study ter release of AMIS tudes toward aspirin use (AMIS) of results; all inter- (AMIS results showed no NHLBI views followed benefit to aspirin; other publication of studies had shown posi- CDP. tive trends favoring aspi- rin). Epidemiologic sur- veys; physiologic findings nonran- domized con- trolled trial Stross and Harlan (1979) Specific RCT: Dia- betic Retinopathy Study Mailed questionnaires Marked acceleration in to anesthesiologists, at least 5 years af- ter publication of evaluations. Survey of physicians attending CME course 18 months after publication of study. spread of knowledge fol- lowing initial publica- tion of one finding, not for the other two find- ings. Journals domi- nated as source of infor- mation for this one finding. Colleagues were equally important for the other two findings. 28 % of family practition- ers and 46 % of internists were aware of study finding. Two-thirds of those aware learned from an opthalmologist or colleague.

EFFECTS OF CLINICAL EVALUATION Hypertension Stross and Harlan Specific RUT: Hy- Survey of physicians (1981) pertension Detec- attending CME tion and Follow- course. up Program 199 40% of family physicians and 60 % of internists were aware of study findings, respectively, 2 months and 6 months following release of study findings. Drugs to lower Friedman et al. Coronary Drug Personal interviews 55 % of all physicians bloodlipo- (1983) Project (main with 1,785physi- (84% of cardiologists; proteins results published clans (587 cardiolo- 54% of internists; 30% in 1975) gists, 584 internists, of GPs) were aware of 614 general practi- CDP in general or fa- tioners tGPs]) (See miliar with its specific Market Facts, findings. Physicians with 1982~. some or full knowledge of CDP are approxi- mately twice as likely as those with no familiarity of CDP to believe lipid- lowering drugs are non- effective or safe. Discharge of Berg and Salis- Nonrandomized Telephone survey of 73 % have inappropriate low-weight bury (1971) trial with com- 102 board-certified criteria for hospitaliza- infants parison group pediatricians. lion, 2 years after initial (supplemented by study published. absence of con- tradictory evi- dence in litera- ture) Stilbestrol in Chalmers (1974) Controlled versus 7 textbooks of obstet- 6 of 7 textbooks have rec- pregnancy uncontrolled rics, 10 years after ommendations in agree- to prevent studies controlled trials ment with controlled abortion studies showing no bene- fit of stilbestrol. Treatment of Christensen et al. Randomized con- Recommendations of Treatments not demon- duodenal ul- (1974) trolled trials con- textbooks in 1964 strafed to be effective in cer ducted between compared to 1974 clinical trials (antacids 1964 and 1974 and anticholinergics continue to be recom- mended, and other drugs (demonstrated to be effective in RCTs) are ignored. 23 controver- Ingelfinger et al. Variety, 10 years af- Literature review Few if any controversies sies in inter- (1974) ter initially pub- were resolved by a clear- nal medicine fished cut clinical trial. Oral hypogly- Marks, unpub- UGDP Published articles, re- Views persistent controver- cemic agents fished report views, editorials, sies as disputes over cri- in diabetics (1983) letters. teria for judgingevi- dence, weights accorded different types of evi- dence (RCTs and other), and convictions about RCTs as the single proper standard for judging merit of prac- tices.

200 APPENDIX 4-A Continued ASSESSING MEDICAL TECHNOLOGY Reference (Year) Type of Evaluation Source of Data Studied on Effects Findings Studies of Patterns of Clinical Evaluation and of the Effect of Clinical Evaluations on the Opinions of Experts Secondarypre- Marks, unpub- 7 RCTs Published articles, re- Views persistent controver- vention of fished report views, editorials, sies as disputes over cri- myocardial (1983) letters. teria for judging evi- infarction dence, weights accorded different types of evi- dence (RCTs and other), and convictions about RCTs as the single, proper standard for judging merit of prac- tices. Antiplatelet Friedwald and 2 RCTs (aspirin Views of authors. Aspirin not demonstrably agents to Schoenberger myocardial in- effective; methodologic prevent re- (1982) farction study controversies dominate current and anturane anturane trial. myocardial reinfarction trial) infarction Cancer ther- UICC reports; Ar- RCTs and other Views of authors. RCTs have been more use- apy mitage et al. controlled trials ful than non-RCTs in (1978) developing cancer treat- ments. Treatment of Rockette et al. RCTs Views of authors. RCTs have strong effect on breast can- (1982) shaping breast cancer cer treatment. Neurosurgical Haines (1983) 51 RCTs in the lit- Views of authors. Many trials have serious procedures erasure published methodologicshortcom- since 1944 ings and few if any have resolved important clini- cal questions. Therapy for Gehan (1982) Published trials Review of literature. Nonrandomized (rather acute leuke- than randomized) stud- mia ies have been the pri- mary means of establish- ing effectiveness of new therapies between 1948 and 1971. Therapy for Gamier et al. Available literature Opinion of hospital In most cases where there head and (1982) experts plus review is a consensus about neck cancer of literature. treatment at the investi- gators' hospital, it is the result of nonrandomized studies. Innovations in Gilbert et al. Published evalua- Literature review. Of 107 papers assessing in- surgery and (1977) tions of innova- novation, approximately anesthesia tions in surgery one-third are RCTs. and anesthesia Less well-controlled studies are more positive about innovation.

EFFECTS OF CLINICAL EVALUATION 201 Discharge of Berg and Salis- Comparison group (Blue Cross) Insur- 90% of low-birth-weight low birth bury (1971) (concomitant) ance records. infants are kept in hospi- weight nonrandomized tal longer than neces- trial (plus absence sary; 2 years after publi- of contradictory cation of initial study. evidence in litera- ture) Oral hypogly- Chalmers (1974' RCT (University National Disease and No decline in use of all comic agents Group Diabetes Therapeutic Index oral hypoglycemics 3 in diabetics Project) (NDTI). years after UGDP results published in 1970. Warner et al. Use of all hypoglycemics (1978~; Finkel- drops by 50 % between stein and Gil- third and seventh years bert (1983) after UGDP results pub- lished. Use of tolbu- tamide shows prompt and sharp decline soon after publication of UGDP results. Stilbestrol in Chalmers (1974) 6 controlled trials, Reported marketing 50,000 women per year re- pregnancy one randomized studies. ceived stilbestrol in late to prevent between 1946 1960s. abortion and 1955, show no effect; uncon- trolled studies re- port positive results Bedrest in viral Chalmers (1974) 2 controlled trials Medical records of 10-15 years after the first hepatitis (randomization hospitalized pa- definitive study 49% of unspecified) show tients. university hospital pa- no benefit to tients and 67 Coo of com- bed rest munity hospital patients still being kept at bed rest. Bland diet for Chalmers (1974) 8 studies (type un- Medical records of 35 of 38 physicians admit- duodenal ul- specified) show hospitalized pa- tingpatientswith diag- cer no benefit for ul- tients. nosis of ulcer order cer healing from bland diets (a practice bland diet not substantiated by studies). Tetracyclines Ray et al. (1977) in children Reports of drug tox- Insurance (Medicaid) 5 % of all prescriptions icity in children records in Tennes- (7,000 prescriptions) for see. children under 8 years of age were for tetracy- clines.

202 APPENDIX 4-A Continued ASSESSING MEDICAL TECHNOLOGY Reference Topic (Year) .Studi~..s of Patt~rn.s of Clinical Evaluation and of the Effect of Clinical Evaluations on the Opinions of Experts Following publication of VA RCT results in 1975, there was a slight de- cline in the proportion of patients with 1-vessel and 2-vessel disease re- ferred for surgery (an ef- fect consistent with find- ings of the study): 1974- 1978- 1975 Type of Evaluation Source of Data Studied on Effects Findings Coronary ar- Fisher and Ken- 7 RCTs tery bypass nedy (1982) graft surgery Prophylactic Wilson et al. RCTs antimicrobials (1982) in gastroin- testinal sur- gery Three practices Fineberg et al. on anesthesi- (1978) ology Referrals for treatment of senile macu- lar degener- ation Internal mam- mary artery ligation Combs (1982) Barsamian (1977~; RCT Fisher and Ken- nedy (1982) Gastric freez- Miao (1977) RCTs and other ing studies Fineberg (1979) Referrals to surgery for CASS Registry within 4 years of Veteran's Adminis- tration (VA) RCT. Survey of surgeons in Scotland. Epidemiologic sur- veys; physiologic findings; nonran- domized con- trolled trials Mailed questionnaire to anesthesiologists, at least 5 years af- ter publication of evaluations. RCT showing bene- Records of Witmer fit of treatment Ophthalmological Institute. Views of authors. Manufacturer records. 1979 1-vessel 38 % 2-vessel 53 % 3-vessel 63.9 % 29.6% 46.7 To 64.3% Most use prophylactic anti- microbiotics in accor- dance with RCTs; 25 % believe definitive proof lacking. Proportion who had adopted new practices at time of survey ranged from 65 to 85 % . Delay between awareness and change was longest for the one practice requir- ing institutional action (scavenging waste gases) . In 6 months following an- nouncement of study results compared to 6 months previous, the number of patient refer- rals tripled and the number of treatable cases doubled. RCT definitively showed procedure to be ineffec- tive. (No quantitative data on utilization of the procedure before defini- tive studies.) RCT definitively showed procedure to be ineffec- tive. Sale of devices stopped sev- eral years prior to ap- pearance of definitive study.

EFFECTS OF CLINICaL EVALUATION CT scanning Creditor and Car- Availableliterature Hospital records. rett (1979) 203 Acceptance of computed tomographic (CT) scan- ning preceded controlled evaluation. Extracorporeal Fineberg and RCT Views of authors. Early randomized trial support for Hiatt (1979) said to prevent spread of respiratory technology. · rr. . 1nsuttlclency Amniocentesis Omenn (1978) Multi-center con- Views of authors. Early multicenter trials trolled trial said to promote wider dissemination. Hyperbaric O2 OTA (1978, 1983) RCT Views of authors. RCT finds procedure inef- for cognitive fective and dampens deficits in physician use. the elderly Treatment of McPherson and RCTs National rates of sur- Physicians persist in using breast can- Fox (1977) gery cited in other radical mastectomy de- cer works. spite evidence from RCTs that simple mas- tectomy plus irradiation is at least as successful. Treatment of OTA (1983) RCTs Survey of surgical Practice has changed in breast can- patterns by Ameri- the direction, though cer can College of Sur- not the degree, indi- geons. cased by RCTs, Percent of breast cancer patients 1972 1981 Radical mastectomy 50 % Modified radical 30 % 70 % Lumpectomy 3 % 8 % (Remainder) presumably represents simple mas- tectomy and possibly other treatments. Beta-blockers OTA (1983); 41 RCTs Views of authors. Small RCTs all show trend after Friedwald and favoring use of beta- myocardial Schoenberger blockers. Widespread infarction (1982) use probably preceded evidence from RCTs. Length of stay Chassin (1983) RCTs other con- NCHS Hospital Dis- LOS for myocardial in- (LOS) for trolled studies charge Survey. farction (MI) in the myocardial and non-RCTs United States declined infarction by one-third between 1968 and 1980. Many studies find shorter stays as safe as longer, though results are not conclu- s~ve.

204 APPENDIX 4-A Continued ASSESSING MEDICAL TECHNOLOGY Reference Topic (Year) Studies of Patterns of Clinical Evaluation and of the Effect of Clinical Evaluations on the Opinions of Experts Type of Evaluation Source of Data Studied on Effects Findings Treatment of alcohol withdrawal Moskowitz et al. RCTs (1981) Drugs to lower Friedewald and blood lipids Schoenberger (CDP) (1982) Hypertension Friedewald and detection Schoenberger and follow- (1982) up Drug to lower blood lipids (Coronary Drug Proj- ect) Aspirin to re- duce recur- rence of myocardial infarction (AMIS) RCT (CDP) Views of authors. RCT (HDFP) Views of authors. Market Facts (1982) RCT Market Facts RCT (1982) Modern medi- Lambert (1978) Varied cat mistakes Discarded sur- Barnes (1977) gical proce- dures Drugs to lower blood lipids (Coronary Drug Proj- ect) Survey of physicians. Physicians practice was consistent with findings in RCTs prior to ap- pearance of review arti- cles making same recom- mendations. Gradual reduction in use of clofibrate consistent with findings in study. Study should have broad Survey of physicians 4-5 years after trial showing risks and lack of benefit from lipid-lowering agents. Survey of physicians before and after study showing no benefit from aspi- rin following myocardial infarc- tion. Review of literature. Evaluations in liter- Literature between ature 1880 and 1942. Friedman et al. RCT (COP) (1983) National Disease and Therapeutic Index. impact, leading to more aggressive treatment of hypertension. Majority of physicians ei- ther never prescribe lipid-lowering drugs (14 Coo ~ or use them only as secondary therapy (72 % ). 47 % said they were using lipid-lower- ing drugs less often than in the past; 67 % use these drugs more often. Majority of physicians re- mained unaffected by AMIS findings, continu- ing to prescribe aspirin for patients following myocardial infarction. Many medical mistakes oc- cur because of the ab- sence of proper early evaluation. Eventually discarded oper- ations were character- ized by lack of control . , . experience ana In sev- eral cases were sustained in the literature over decades. Prescriptions for all lipid- lowering drugs in the United States rose from 1.5 million in 1970 to 2.3 million in 1975 then fell to 1 million in 1980.

EFFECTS OF CLINICAL EVALUATION 205 Personal interviews with 1,785 physi- cians from 1979 through 1981 (see Market Facts, 1982). Percentage of physicians prescribing lipid-lower- ing drugs for post-MI patients: 1979 1980 1981 Never prescribe 10.2 17.2 18.0 TRY only as 2° therapy 73.8 74.0 69.2 Sometimes Rx as 1° therapy 15.9 8.8 12.7 Number of physicians 859 296 621 Studies of the Effect of Synthetic Assessment on Knowledge of Physicians NIH consen- Jacoby, unpub- Consensus confer- Telephone surveys of 8-9 % of physicians were sus confer- lished report ence 700 physicians in aware of the Consensus ences on CT (1983) pertinent specialty Development Program scanning areas 2 weeks prior at NIH. Two weeks be- and total hip to conference and 6 fore each conference, replacement weeks after publi- 16 °70 of physicians knew cation of confer- ence results in the Journal of the American Medical Association. about the upcoming conference on CT; 7 % knew about the confer- ence on hip joint re- placement. Six weeks after publication of results, 14 % were aware of the conference on CT and 4% aware of the conclusions; 7 % were aware of the con- ference on hip joint re- placement and 1% aware of its conclusions. Most of those aware of study findings had read about them in profes- sional journals. Studies of Effect of Synthetic Assessments on Clinical Practices Guidelines Leff et al. (1979); Recommendation Mail survey of 28 mu- Greater conformity to for tubercu- Leff and Bre- from Public nicipal tuberculosis guidelines on chemopro- losis control win (1981) Health Service, control officers. L}IIV1~1L ~ .1.~1.1 Ad L11~1~ ~- American Tho- racic Society, American Lung Association phylaxis than on diagno- sis and treatment. Medical neces- Blue Cross and Consensus on 42 Insurance claims in Number of claims paid for sity, Blue Blue Shield As- outmoded prac- 1975 and 1978 for tested surgical proce- Cross and sociation (1982) tices, announced Federal Employee cures declined 26%; Blue Shield and disseminated Health Benefits claims paid for listed di- in 1977 Program offered by agnostic procedures de- Blue Cross and . clined 85 % . Blue Shield.

206 APPENDIX 4-A Continued ASSESSING MEDICAL TECHNOLOGY Reference Topic (Year) Type of Evaluation Source of Data Studied on Effects Findings Studies of Effect of Synthetic Assessments on Clinical Practices Approved uses for cimeti- dine Fineberg and Pearlman (1981) Schade and Donaldson <1981) Cocco and Cocco (1981) NIH consensus Thomson et al. conference (1981) on high blood pres- sure, held in 1977 Treatment of alcohol withdrawal FDA-approved uses NDTI data on drug Many uses of cimetidine in use. practice are not ap- proved by FDA. Medical records. Many uses of cimetidine in practice are not ap- proved by FDA. Consensus confer- ence Moskowitz et al. (1981) Survey of physicians in ambulatory set- tings in New York City. Review articles Survey of physicians. Many uses of cimetidine in practice are not ap- proved by FDA. 90 % of respondents were routinely treating pa- tients with mild hyper- tension in contrast to in- dividualized approach advocated by consensus statement. Physician use practices that were found effec- tive in RCTs and use them before they are recommended in review articles. Studies of the Effects of Clinical Evaluations and Synthetic Assessments on Regulation and Reimbursement PHS adviso- Center for Analy- Review by PHS Estimates of authors. The PHS through National ries to sis of Health Center for Health Care HCFA Practices (1981) Technology made rec- ommendations of non- reimbursement for 21 of 50 procedures reviewed during 1979-1980. All recommendations were accepted by HCFA, though procedures at HCFA did not necessar- ily assure uniform appli- cation throughout the country. Decisions not to reimburse four proce- dures (dialysis for schiz- ophrenia, hyperthermia for cancer, radial kera- totomy for myopia, and endothelial cell photog- raphy) are estimated to produce savings of $312 million to the Medicare program over a 10-year period.

EFFECTS OF CLINICAL EVALUATION Hyperbaric ox- OTA (1978, 1983) RCT ygen for cognitive deficits in the elderly Drugs to lower Friedman et al. blood lipids (1983) 207 Views of authors. RCTs (CDP and WHO clofibrate trial) RCT shows procedure in- effective; facilitates deci- sion by Medicare and other insurers not to re- imburse. Review of FDA label- Changes in labeling, iden- ing changes for clo- fibrate (Atrom id- S) between 1969 and 1982. tifying more side effects and progressively re- stricting indications for use, are believed due in part to CDP (which as- sessed use in patients with previous MI; results with clofibrate published 1975) and to WHO clofibrate trial (which assessed primary prevention of MI; results published 1978 and 1980~. Major labeling changes in 1979 are at- tributed to both CDP and WHO trial; further restrictions in 1982 re- flect most recent find- ings from WHO trial, with CDP results serving as background. REFERENCES Armitage, P., D. Bardelli, and D. A. G. Galton, et al. 1978. Methods and Impact of Controlled Thera- peutic Trials in Cancer, Part I. UICC Technical Re- port Series, 36, Geneva. Avorn, J., M. Chen, and R. Hartley. 1982. Scien- tific versus commercial sources of influence on the prescribing behavior of physicians. Am. J. Med. 73:4-8. Avorn, J., and S. B. Soumerai. 1983. Improving drug-therapy decisions through educational outreach: A randomized controlled trial of academically based "detailing." N. Engl. J. Med. 308:1457-1463. Barnes, B. 1977. Discarded operations: surgical in- novation by trial and error. Pp. 109-123 in Costs, Risks and Benefits of Surgery, J. P. Bunker, B. A. Barnes, and F. Mosteller, eds. New York: Oxford Uni- versity Press. Barsamian, E. M. 1977. The rise and fall of inter- nal mammary artery ligation in the treatment of an- gina pectoris and the lessons learned. Pp. 212-220 in Costs, Risks and Benefits of Surgery, J. P. Bunker, B. A. Barnes, and F. Mosteller, eds. New York: Oxford University Press. Berg, R. B., and A. J. Salisbury. 1971. Discharging infants of low birth weight: Reconsideration of cur- rent practice. Am. J. Dis. Child. 122:414-417. Berwick, D. M., H. V. Fineberg, and M. C. Wein- stein. 1981. When doctors meet n\umbers. Am. J. Med. 71:991-998. Blue Cross and Blue Shield Association. 1982. An- nouncement of new phase of medical necessity pro- gram. October 12. Memorandum. Chicago, Ill. Bonchek, L. 1979. Are randomized trials appropri- ate for evaluating new operations? N. Engl. J. Med. 301:44 45. Buchwald, H., L. Fitch, and R. B. Moore. 1982. Overview of randomized clinical trials of lipid inter- vention for atherosclerotic cardiovascular disease. Controlled Clinical Trials 3:271-83. Caplow, T. 1954. Market attitudes: A research re- port from the medical field. Harvard Bus. Rev. 30: 105-12. Center for the Analysis of Health Practices, Har- vard School of Public Health. 1981. Impact on health costs of NCHCT recommendations for nonreimburse- ment for medical procedures. Monograph Series. Washington, D.C.: National Center for Health Care Technology.

208 Chalmers, T. C. 1974. The impact of controlled trials on the practice of medicine. Mt. Sinai J. Med. (NY) 41:753-759. Chalmers, T. C., B. Silverman, E. P. Shareck, A. Ambroz, B. Schroeder, and H. Smith, Jr. 1979. Ran- domized controlled trials in gastroenterology with particular attention to duodenal ulcer. Pp. 223-255 in Report to the Congress of the United States of the Na- tional Commission on Digestive Diseases, Vol. 4, part 2B, DHEW Pub. No. (NIH) 79-2885. Washington, D.C.: Department of Health, Education, and Wel- fare. Chalmers, T. C. 1982. A potpourri of RCT topics. Controlled Clinical Trials 3:285-98. Chassin, M. P. 1983. Health Technology Case Study, 24: Variations in Hospital Length of Stay: Their Relationship to Health Outcomes. OTA-HCS 23. Washington, D.C.: U.S. Congress, Office of Technology Assessment. Christensen, E., E. Juhl, and N. Tygstrup. 1977. Treatment of duodenal ulcer: Randomized clinical trials of a decade (1964-1974). Gastroenterology 73: 1170-1178. Cocco, A. E., and D. V. Cocco. 1981. A survey of cimetidine prescribing. N. Engl. J. Med. 304:1281. Combs, J. L., S. M. Cohen, S. L. Fine, R. P. Mur- phy, and A. Patz. 1983. The spectrum of senile macu- lar degeneration (SMD) at the retinal vascular center. Presented at the Forty-second Scientific Meeting of the Residents Association of the Wilmer Ophthalmo- logical Institute, Baltimore, April 20-22, 1983. Cited by Hawkins (1983~. Comroe, J. H., Jr. 1976. Lags between initial dis- covery and clinical application to cardiovascular pul- monary medicine and surgery. Report of the Presi- dent s Biomedical Research Panel, Appendix B. Part I. DHEW Pub. No. (OS) 76-502. Washington, D.C.: U.S. Government Printing Office. Creditor, M. C., and J. B. Garrett. 1977. The in- formation base for diffusion of technology: Computed tomography scanning. N. Engl. J. Med. 297:49-52. Dunn D. R. F. l98l. Di.s.semination of the nub- __7 _ fished results of an important clinical trial: An analy- sis of the citing literature. Bull. Med. Libr. Assoc. 69:301-306. Feinstein, A. R. 1983. An additional basic science for clinical medicine: II. The limitations of random- ized trials. Ann. Intern. Med. 99:544-550. Fineberg, H. V., G. S. Parker, and L. A. Pearlman. 1977. CT scanners: Distribution and plan- ning status in the United States. N. Engl. J. Med. 297:216-218. Fineberg, H. V., R. Gabel, and M. Sosman. 1978. The acquisition and application of new medical knowledge by anesthesiologists: Three recent exam- ples. Anesthesiology 48:430-436. Fineberg, H. V. 1979. Gastric freezing: A study of ASSESSING MEDICAL TECHNOLOGY diffusion of a medical innovation. Pp. 173-200 in Medical Technology and the Health Care System: A Study of the Diffusion of Equipment-embodied Tech- nology. Washington, D.C.: National Academy of Sci- ences. Fineberg, H. V., and H. H. Hiatt. 1979. Evalua- tion of medical practices: The case for technology as- sessment. N. Engl. J. Med. 301:1086-1091. Fineberg, H. V., L. A. Pearlman, and R. A. Ga- bel. 1980. The case for abandonment of explosive an- esthetic agents. N. Engl. J. Med. 303:613-617. Fineberg, H. V., and L. A. Pearlman. 1981. Sur- gery for peptic ulcer disease in the United States: Trends before and after the introduction of cimeti- dine. Lancet 1: 1305-1307. Finkelstein, S. N., and D. L. Gilbert. 1983. Scien- tific evidence and the abandonment of medical tech- nology: Study of eight drugs. Working Paper #1419- 83. Alfred P. Sloan School of Management, Massachusetts Institute of Technology, Cambridge, Massachusetts. Fisher, L. D., and J. W. Kennedy. 1982. Random- ized surgical clinical trials for treatment of coronary artery disease. Controlled Clinical Trials 3:235-58. Freeborn, D. K., D. Baer, M. R. Greenlick, and J. W. Bailey. 1972. Determinants of medical care utili- zation: Physicians' use of laboratory services. Am. J. Public Health 62:846--53. Friedewald, W. T., and J. A. Schoenberger. 1982. Overview of recent clinical and methodological ad- vances from clinical trials of cardiovascular disease. Controlled Clinical Trials 3:259-270. Friedman, L., N. K. Wenger, and G. L. Knat- terud. 1983. Impact of the coronary drug project find- ings on clinical practice. Controlled Clinical Trials 4:513-22. Gamier, H. S., R. Flamant, and C. Fohanno. 1982. Assessment of the role of randomized clinical trials in establishing treatment policies. Controlled Clinical Trials 3:227-34. Gehan, E. A. 1982. Progress in therapy in acute leukemia 1948-1981: Randomized versus nonran- domized clinical trials. Controlled Clinical Trials 3: 199-208. Gilbert, J., B. McPeek, and F. Mosteller. 1977. Progress in surgery and anesthesia: Benefits and risks of innovative therapy. Pp. 124-169 in Costs. Risks. , , and Benefits of Surgery, J. P. Bunker, B. A. Barnes, and F. Mosteller, eds. New York: Oxford University Press. Globe, S., G. W. Levy, and C. M. Schwartz. 1967. Science, Technology, and Innovation. Contract NSF 8667 (Battelle-Columbus Laboratories). Washington, D. C.: National Science Foundation. Goodwin, J., and ]. M. Goodwin. 1982. Failure to recognize efficacious treatments: A history of salicy- late therapy in rheumatoid arthritis. Perspect Biol. Med. 25:78-92.

EFFECTS OF CLINICAL EVALUATION Haines, S. J. 1983. Randomized clinical trials in neurosurgery. J. Neurosur 12:259-264. Ingelfinger, F. J., R. V. Ebert, M. Finland, and A. S. Relman, eds. 1974. Controversy in Internal Medi- cine II. Philadelphia: W. B. Saunders. Jordan, H. S., J. F. Burke, H. V. Fineberg, and J. S. Hanley. 1983. Diffusion of innovations in burn care: Selected findings. Burns 9:271-279. Lambert, E. C. 1978. Modern Medical Mistakes. Bloomington: Indiana University Press. Leff, A. R., D. Herskowitz, J. Gilbert, and A. Brewin. 1979. Tuberculosis chemoprophylaxis prac- tices in metropolitan clinics. Am. Rev. Respir. Dis. 119: 161-164. Leff, A. R., and A. Brewin. 1981. Tuberculosis chemotherapy practices in major metropolitan health departments in the United States. Am. Rev. Respir. Dis. 123:176-180. Manning P. R., and T. A. Denson. 1979. How car- diologists learn about echocardiography: A reminder for medical educators and legislators. Ann. Intern. Med. 91:496-471. Manning, P. R., and T. A. Denson. 1980. How in- ternists learned about cimetidine. Ann. Intern. Med. 92:690-692. Market Facts. 1982. The Impact of Two Clinical Trials on Physician Knowledge and Practice. Con- tract Nol-HV-82940. Washington, D.C.: National Heart, Lung, and Blood Institute. McPherson, K., and M. S. Fox. 1977. Treatment of breast cancer. Pp. 308-322 in Costs, Risks, and Bene- fits of Surgery, J. P. Bunker, B. A. Barnes, and F. Mosteller, eds. New York: Oxford University Press. Miao, L. L. 1977. Gastric freezing: An example of the evaluation of medical therapy by randomized clinical trials. Pp. 198-211 in Costs, Risks, and Bene- fits of Surgery, J. P. Bunker, B. A. Barnes, and F. Mosteller, eds. New York: Oxford University Press. Moskowitz, G., T. C. Chalmers, H. S. Sacks, R. M. Fagerstrom, and H. Smith, Jr. 1981. Deficiencies of clinical trials of alcohol withdrawal. Alcoholism Clin. Exp. Res. 5:162-166. Office of Technology Assessment, U.S. Congress. 1978. Assessing the Efficacy and Safety of Medical Technologies. Stock No. 052-003-00593-0. Washing- ton~ D.C.: U.S. Government Printing Office. Office of Technology Assessment, U. S. Congress. 1981. Policy Implications of the Computed Tomo- graphy (CT) Scanner: An update. OTA-BP-H-8. Washington, D.C.: U.S. Government Printing Of- fice. Office of Technology Assessment, U.S. Congress. 1983. The Impact of Randomized Clinical Trials on Health Policy and Medical Practice. OTA-BP-H-22. Washington, D.C.: U.S. Government Printing Of- fice. Omenn, G. S. 1978. Prenatal diagnosis of genetic disorders. Science 200:952-958. 209 Peterson, W. L., R. A. L. Sturdevant, H. D. Frankl, C. T. Richardson, J. I. Isenberg, J. D. Elashoff, J. Q. Sones, R. A. Gross, R. W. McCallum, and J. S. Fordtran. 1977. Healing of duodenal ulcer with an antacid regimen. N. Engl. J. Med. 297:341- 345. Pryor, D. B., M. C. Hindman, G. S. Wagner, R. M. Califf, M. K. Rhoads, and R. A. Rosati. 1983. Early discharge after acute myocardial infarction. Ann. Intern. Med. 99:528-538. The Rand Corporation. 1983. Submission to the Office of Management and Budget of Supporting Statement and Data Collection Instruments for As- sessing the Effectiveness of the NIH Consensus Devel- opment Program (Contract No.: NO1-D-2-2128~. Ray, W. A., C. F. Federspiel, and W. Schaffner. 1977. Prescribing of tetracycline to children less than 8 years old: A two-year epidemiologic study among ambulatory Tennessee Medicaid recipients. J. Am. Med. Assoc. 237:2068-2074. Rockette, H. E., C. K. Redmond, and B. Fisher. Participating NSABP investigators. 1982. Impact of randomized clinical trials on therapy of primary breast cancer: The NSABP overview. Controlled Clinical Trials 3:209-25. Rogers, E. M., and F. F. Shoemaker. 1971. Com- munication of Innovations: A Cross-Cultural Ap- proach. New York: The Free Press. Russell, L. B. 1979. Technology in Hospitals: Med- ical Advances and Their Diffusion. Washington, D.C.: The Brookings Institution. Schade, R. R., and R. M. Donaldson, Jr. 1981. How physicians use cimetidine: A survey of hospital- ized patients and published cases. N. Engl. J. Med. 304: 1281-1284. Schaffner, W., W. A. Ray, C. F. Federspiel, and W. O. Miller. 1983. Improving antibiotic prescribing in office practice: A controlled trial of three educa- tional methods. J. Am. Med. Assoc. 250:1728-1732. Sherrington, A. M. 1965. An annotated bibliogra- phy of studies on the flow of medical information to practitioners. Method Inform. Med. 4:45-57. Stern, B. J. 1927. Social Factors in Medical Prog- ress. New York: Columbia University Press. Stross, J. K., and W. R. Harlan. 1979. The dissem- ination of new medical information. J. Am. Med. As- soc. 241 :2622-2624. Stross, J. D., and W. R. Harlan. 1981. Dissemina- tion of relevant information on hypertension. J. Am. Med. Assoc. 247:3231-3234. Sussman, E. J., W. G. Tsiaras, and K. A. Soper. 1982. Diagnosis of diabetic eye disease. J. Am. Med. Assoc. 247:3231-3234. Thomson, G. E., M. H. Alderman, S. Wassertheil- Smoller, J. F. Rafter, and R. Samet. 1981. High blood pressure diagnosis and treatment: Consensus recom- mendations vs. actual practice. Am. J. Public Health 71 :413-16.

210 Wagner, E. H., R. A. Truesdale, and J. T. Warner. 1981. Compliance, treatment practices and blood pressure control: Community survey findings. J. Chronic Dis. 34:519-25. Warner, K. 1975. A"desparation-reaction" model of medical diffusion. Health Service Res. 10:369-83. Warner, R., S. M. Wolfe, and R. Rich. 1978. Off Diabetes Pills: A Diabetics Guide to Longer Life. ASSESSING MEDICAL TECHNOLOGY Washington, D.C.: Public Citizen's Health Research Group. Williamson, P. M. 1975. The adoption of new drugs by doctors practicing in group and solo prac- tice. Social Sci. Med. 9:233-236. Wilson, N. I. L., P. A. Wright, and C. S. McArdle. 1982. Survey of antibiotic prophylaxis in gastrointesti- nal surgery in Scotland. Br. Med. J. 285:871-873.

Next: 5. Reimbursement and Technology Assessment »
Assessing Medical Technologies Get This Book
×
Buy Hardback | $90.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

New drugs, new devices, improved surgical techniques, and innovative diagnostic procedures and equipment emerge rapidly. But development of these technologies has outpaced evaluation of their safety, efficacy, cost-effectiveness, and ethical and social consequences. This volume, which is "strongly recommended" by The New England Journal of Medicine "to all those interested in the future of the practice of medicine," examines how new discoveries can be translated into better care, and how the current system's inefficiencies prevent effective health care delivery. In addition, the book offers detailed profiles of 20 organizations currently involved in medical technology assessment, and proposes ways to organize U.S. efforts and create a coordinated national system for evaluating new medical treatments and technology.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!