NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Committee on Technological Innovation in Medicine; Gelijns AC, editor. Modern Methods of Clinical Investigation: Medical Innovation at the Crossroads: Volume I. Washington (DC): National Academies Press (US); 1990.

Cover of Modern Methods of Clinical Investigation

Modern Methods of Clinical Investigation: Medical Innovation at the Crossroads: Volume I.

Show details

1Medical Technology Development: An Introduction to the Innovation-Evaluation Nexus


The increase in fundamental knowledge concerning human health and the mechanisms of disease has been so rapid during the second half of this century that we have often been described as living in a time of biological revolution. In the spirit of Francis Bacon, who observed that the true essence of progress is in the application of scientific knowledge for enhancing the human condition, our society for the past several decades has valued biomedical innovation and its promise of improving the management of health and disease. Rapid advances in biomedical research have indeed stimulated the development of numerous efficacious medical technologies, but their translation into clinical use has raised complex medical, economic, and social issues. The emergence of these issues—as illustrated by the development of new aquired immune deficiency syndrome (AIDS) drugs—is spurring new interest in medical innovation: how it occurs, what can be expected of it, and how it might be improved.

Technological innovation in medicine covers the wide range of events by which a new medical technology is discovered or invented, developed, and disseminated into health care. One of the most vulnerable links in this innovation chain today is the development phase, the “D” of R&D, in which research findings are brought into clinical practice. More specifically, medical technology development can be defined as a multi-stage process through which a new biological or chemical agent, prototype medical device, or clinical procedure is technically modified and clinically evaluated until it is considered ready for general use. Although this definition suggests an organized and systematic process, much developmental activity actually occurs in a non-orderly fashion in everyday clinical practice.

Among the many factors influencing development, the criteria and methods of clinical evaluation have become increasingly important determinants of how—and indeed whether—new medical technologies are developed. This first volume of the Institute of Medicine (IOM) Committee on Technological Innovation in Medicine focuses on the interplay between strategies for clinical evaluation and the development of new drugs, devices, and clinical procedures.


Two major considerations influenced the selection of the theme of this volume. The first is the emergence of widespread concern over the way in which new medical technologies are evaluated clinically during the development process. 1 For example, the development of drugs for life-threatening diseases has become the subject of extensive reporting in the professional literature and the daily press, as well as a matter of serious policy debate. A key issue is whether the pre-marketing evaluative requirements governing drug development are sufficiently flexible or are interpreted flexibly enough in the case of drugs for fatal diseases such as cancer or AIDS. For example, one might question whether and when intermediate endpoints, instead of survival, should be evaluated in pre-approval trials. The Food, Drug, and Cosmetic Act allows considerable latitude for subjective interpretation of the terms “safety” and “effectiveness” in determining the acceptable risk-benefit ratio for a marketing approval decision. 2 But because of social and political pressures to reduce the risk to essentially zero, pre-marketing requirements have become increasingly detailed over time. Although the resulting system has provided important information on the efficacy and safety of new drugs, it has also considerably lengthened the pre-marketing development process. Moreover, despite this increase, there are clearly no “zero-risk” approval decisions. For example, the detection of delayed or rare (less than 1:10,000) adverse effects would require extremely long periods of testing or the exposure of many thousands of patients. Furthermore, valuable therapeutic information on the risks and benefits of a new drug may emerge only after its diffusion into the often messy environment of general use. For instance, in the period 1982-1986, six newly approved drugs were withdrawn shortly after introduction and five others required substantial relabeling, despite rigorous pre-marketing evaluation (1). A classic example of side effects that may be hard to detect in the carefully controlled setting of pre-approval trials is the acute hypertension induced by the antidepressant tranylcypromine if the patient happens to eat a particular kind of cheese. The traditional response to the realization that taking drugs may be a risky business has been to increase pre-marketing requirements for clinical evaluation. It is now timely to ask whether this strategy will remain appropriate or whether a point of diminishing returns has been reached, and if a shift in emphasis toward obtaining information in the post-marketing clinical setting would not be more appropriate.

A different issue is concern about the adequacy of the evidence underlying development and dissemination of clinical procedures into health care (2). For example, extracranial-intracranial vascular bypass surgery for stroke was first tried in human beings in 1967; the procedure underwent rapid diffusion during the 1970s, but was only recently reported ineffective in preventing cerebral ischemia in patients with atherosclerotic disease of the carotid and middle cerebral arteries (3). At a national level, the considerable geographic variations in the use of certain clinical procedures may largely be explained by insufficient evidence about their diagnostic, therapeutic, and ultimate health effects (4). The consequences of such variations for the quality of medical care and the cost-effective use of resources hardly need further explanation, and an argument for more systematic evaluation of clinical procedures has been made repeatedly. Important questions, however, remain as to what evidence should be collected and by what methods during the various stages of the development process. For example, when during the development of a new surgical procedure should a randomized controlled clinical trial be initiated? What are the strengths and weaknesses of modern epidemiological methods during the evolution of new clinical procedures? Given the increasing importance of quality of life as an endpoint in medical care, how do we obtain a more systematic understanding of patient preferences about different health outcomes? And which policy and institutional mechanisms can assure that adequate clinical studies of new procedures are indeed undertaken? These issues, which concern the scientific basis for decisions during development, need to be addressed urgently.

The second consideration for focusing on the interplay between clinical evaluation and technology development concerns the rapid progress occurring in the art and science of clinical evaluation today. Since its inception in the early 1950s, the randomized controlled clinical trial (RCT) has been accepted as an extremely powerful tool for assessing the efficacy of new drugs and biologicals. However, it has also become clear that RCTs are not necessarily practical or feasible for answering all clinical questions. Therefore, a variety of other methods, such as non-randomized trials or observational methods, have been adopted to provide complementary information. Traditionally, these methods were regarded as weaker than RCTs for clinical evaluation. Recent methodological advances, such as the use of non-classical statistics and the ability to link large-scale automated data bases for analysis (e.g., those of health insurance networks and hospitals), are strengthening these approaches. In addition, methods for synthesizing the evidence that results both from experimental and observational studies are being improved. The IOM Committee on Technological Innovation in Medicine observed that these methods may well provide an opportunity to address some of the concerns mentioned above. Although these methods are conceptually appealing, there are important questions as to their strengths and weaknesses and the quality of the evidence they provide.

In view of these considerations, it seemed timely to publish a volume of papers analyzing the validity of these modern methods of clinical investigation and asking if and how their systematic application could improve the technology development process. Before addressing some of the points made by the various authors, a more complete picture is needed of current shortcomings in the clinical evaluation of new medical technologies. The following section will explore some of these shortcomings, using the development of specific pharmacological, surgical, and medical device technologies for the treatment of stable angina pectoris as a case example.



In the late 1950s, Slater and Powell at Eli Lilly serendipitously discovered the pharmaceutical compound dichloroisoproterenol while developing long-acting bronchodilators (5). This compound was found to have beta-adrenergic blockade activity, but also had partial agonist (sympathomimetic) activity; its development was not pursued. At the same time James Black—a 1988 Nobel laureate for physiology or medicine—hypothesized that blocking the beta-adrenergic receptors would diminish the heart's demand for oxygen, providing relief for angina sufferers. He saw the clinical potential of dichloroisoproterenol, and with his colleagues at Imperial Chemical Industries (ICI) started to synthesize its analogues. The first of these compounds to be tested in humans, pronethalol, had a beneficial effect on angina in Phase I trials (6). In a full-scale clinical trial, however, it induced such side effects as nausea, vomiting, and light-headedness. When long-term toxicity tests in animals revealed that it might also be carcinogenic, its development was discontinued. Subsequently, propranolol was synthesized and found to be free of both the agonist activity of dichloroisoproterenol and the side effects of pronethalol (7). 3 It became the first beta-adrenergic antagonist to be marketed in the United Kingdom in 1965 (see Figure 1.1).

FIGURE 1.1. Small chemical differences but large clinical differences.


Small chemical differences but large clinical differences.

In subsequent years, structural analogues of propranolol were introduced on the basis of systematic animal testing and clinical evaluation. These early beta-blockers acted on all beta-adrenergic receptors, which was troublesome for asthmatics. In 1966, Dunlop and Shanks of ICI discovered an analogue that acted selectively on heart receptors (8). This compound was marketed in 1970 in the United Kingdom as practolol, for use by asthmatic patients. In spite of rigorous pre-marketing evaluation, practolol was found to cause very serious side effects, including blindness, in day-to-day clinical practice. Although the incidence of these events was high—l:500—and the events emerged shortly after widespread use began, it took a year or more, during which 100,000 or more patients were treated, before the first voluntary reports reached the Committee on Safety of Medicines and the drug was withdrawn.

As a result of the practolol incident, there was a growing awareness that the system of adverse effect reporting alone, however valuable for the detection of very rare effects, was insufficient for optimal clinical and regulatory decision making. In the United Kingdom, Inman established the Prescription-Event Monitoring Scheme, which tracks the performance of all new chemical entities in clinical practice, to speed the early detection and analysis of adverse events (see Chapter 6). Such monitoring also can facilitate the earlier detection and analysis of benefits; following their introduction into practice, beta-blockers were found to be of potential value in a wide variety of cardiac and non-cardiac conditions. They now are used for more than 20 medical conditions, including hypertension, myocardial infarction, anxiety, and alcoholism (9). Because drugs, once marketed, are subject to empirical innovation and the regulatory system is designed not to interfere with the practice of medicine, the clinical evidence supporting drug use for specific conditions can be quite variable. By 1987, for example, the Food and Drug Administration (FDA) had approved only eight of the many conditions for which beta-blockers are used. Although industrial, governmental, and academic investment in post-marketing pharmaceutical research is increasing, this area remains relatively underdeveloped.

Coronary Artery Bypass Grafting

The development of surgical techniques for angina pectoris presents quite a different picture. The evolution of such surgery can be traced to the turn of the century when cardiac denervation was proposed as a treatment for the crippling pain associated with the disease (10). In the decades preceding the first clinical application of coronary artery bypass grafting (CABG), many new surgical techniques were developed by surgical schools in a variety of countries. Often these procedures coexisted for years, only to be discarded later because of inadequate efficacy or unacceptable side effects. As Effler argues, the earliest surgical development was based on a bad premise: treatment preceded diagnosis (11). It is only with the introduction of Mason Sones's arteriography in 1958 that the success of surgery in terms of graft patency could be validated objectively, and rational patient selection criteria established. Rene Favaloro at the Cleveland Clinic is generally credited with the first report on coronary artery bypass surgery using a saphenous vein graft in 1968 (12). Following the initial discussion of the new procedure at conferences and in the literature, it underwent rapid diffusion and further incremental development. Clinical circumstances favored swift acceptance of the operation: the condition is life-threatening and decreases quality of life, especially for those unresponsive to drug treatment; the operation made sense anatomically and physiologically; and from the outset it seemed very effective in the relief of disabling angina (13). The feeling that the procedure was rational and the fact that the technical aspects of the procedure were still evolving led to a situation in which randomized studies were not carried out; the surgical innovators and those who followed them felt it was too early for an RCT. In the first years there were many publications on graft patency, mortality, and relief of angina, all on the basis of uncontrolled clinical series. With increasing surgical experience and incremental improvements in surgical technique, mortality rates decreased considerably. By 1972-1973, many felt CABG had become the treatment of choice for patients with severe stable angina, and that it was thus too late to carry out RCTs (13). Although there was no dispute about the new procedure's efficacy in relieving the pain of angina, doubt remained about its effect on survival. Three large multicenter RCTs were initiated during the 1970s to analyze the effect on life expectancy: the Veterans Administration (VA) trial, the European Cooperative Surgery Study, and the Coronary Artery Surgery Study (CASS) (14,15 and 16). At the end of the 1970s these trials provided valuable evidence on the safety and efficacy of CABG in specific patient groups, and follow-up results on long-term safety and efficacy were published during the 1980s (17,18).

Although these trials made an important contribution to our knowledge base, two major questions emerge from the above pattern of innovation and evaluation. The trials provided their initial information on safety and efficacy 10 years after the procedure had first been used in clinical practice. During that decade, clinical decision making had to depend to a large extent on anecdotal evidence. As Preston remarks when he argues for encouragement of surgical innovation but questions the process of development itself: “Can the profession afford yet another cycle of unrecognized experimentation, widespread application without validation of benefit, immense economic and professional gratification, gradual disillusionment, and ultimate abandonment in favor of the next ‘new' operation?” (10). In other words, the question is whether establishing a mechanism to systematically initiate and coordinate surgical trials on the basis of early clinical experience (analogous to Phase I drug trials) could have expedited the design and implementation of CABG trials. 4

The other question is whether trial results carried out a decade ago can still be considered valid today. During these years, the indications for CABG have widened to include unstable angina, myocardial infarction, and minimal angina pectoris. Hlatky et al., for example, compared the patient population in the cardiovascular disease data base at Duke University with the patients enrolled in the above-mentioned RCTs (19). They found that only 13 percent met the criteria for the VA trial, 8 percent met the eligibility criteria for the European study, and 4 percent met those for the CASS. In addition to such changes in patient indications, surgical techniques have also undergone further development. For example, internal mammary arteries have recently been found to have a much higher long-term patency rate than saphenous vein grafts (20). In the three RCTs, however, internal mammary arteries were used in only a very small number of cases. These examples illustrate the need for long-term surveillance of new procedures as they evolve in everyday clinical practice.

PTCA Catheter Equipment

In 1977, Andreas Gruentzig at the University of Zurich performed the first clinical percutaneous transluminal coronary angioplasty (PTCA) procedure as an alternative to coronary artery bypass surgery (21). With the firm Schneider-Medintag, he developed a flexible double-lumen dilation catheter with a balloon that could be inflated to compress the deposits that block an artery. In 1979, Gruentzig reported on his first 50 patients in The New England Journal of Medicine and concluded that his results were “preliminary.” More information and follow-up data are needed before coronary angioplasty can be accepted as one form of treatment for coronary-artery disease. However, the results in patients with single-vessel disease are sufficiently good to make the procedure acceptable for prospective randomized trials. Such trials are clearly needed if we are to evaluate the efficacy of this new technique as compared with current medical and surgical techniques” (22). Among cardiologists, however, there was a strong feeling that comparative trials of PTCA and medical or surgical therapy should be delayed until the technology had evolved and the learning curves were established. Thus, the National Heart, Lung, and Blood Institute established an international voluntary registry in 1979 to monitor the safety and effectiveness of PTCA.

Under the newly established medical device amendments to the Food, Drug, and Cosmetics Law, the first balloon dilation catheter was approved for marketing in the United States by the FDA in 1980 (23). To date, nine dilation catheter systems have undergone full pre-marketing safety and efficacy review by the FDA. All were approved not on the basis of RCTs, but on the basis of comparing the results of clinical series with those of other marketed PTCA devices or registry data. Because the PTCA market is very competitive, new modifications emerge almost every month and any product can be outdated within 6 to 12 months (24). These incremental improvements do not require full FDA review but are approved under so-called supplemental pre-marketing approval decisions. In addition to rapid technological change, patient selection criteria are also changing considerably. PTCA was initially used predominantly in discrete noncalcified single-vessel lesions, but it is now being applied in disease affecting multiple vessels and where there are multiple lesions in the same vessel, as well as in unstable angina and acute infarction. The National Institutes of Health (NIH) registry data have been extremely valuable in monitoring these changes in technology and application, as well as their effects on effectiveness and safety. Despite these data, however, there is still no conclusive evidence on the comparative efficacy and safety of PTCA versus medical treatment in single vessel disease, and of PTCA versus CABG in multivessel disease. Randomized controlled clinical trials are clearly overdue. In 1987, the NIH and the VA decided to support three such clinical trials; their results, however, are not expected until the early 1990s to mid-1990s.

Evaluative Shortcomings in Technology Development

The example of stable angina pectoris refutes a popular belief, which holds technology development to be a linear progression from bench to bedside. Surgical innovation often occurs in a decentralized environment with numerous surgical schools trying to find a solution to a particular problem in day-to-day practice. Drugs and devices are also subject to further development in clinical practice. New indications can be revealed in practice, as illustrated by the off-label use of beta-blockers. Also, early clinical experience with a new product may provide impetus to the development of improved products. For example, due to such feedback PTCA catheters have been miniaturized, made more flexible, and given improved angiographic visibility. A more realistic picture of technology development, in which development and diffusion are highly interactive and partially overlap, is the basis for discussing shortcomings in today's strategies for clinical evaluation.

The often inadequate conceptualization in health sciences policy of innovation as linear and sequential has contributed to a system of clinical investigation with major emphasis on providing safety and efficacy information prior to a technology's diffusion. However, as the angina pectoris case illustrates, certain information on the risks and benefits of a technology may emerge only after its diffusion into general use. Furthermore, much developmental activity occurs not before but during everyday practice; consider, for instance, changes in surgical technique or in patient indications. Evaluative strategies, however, have rarely attempted to provide information on the effectiveness and long-term safety of technologies as they evolve in normal, uncontrolled, daily medical life.

In addition, the angina pectoris example reveals a remarkable asymmetry in the existing strategies for providing safety and efficacy information: drugs undergo rigorous clinical testing before their introduction into general use, clinical procedures are still assessed mainly in an ad hoc fashion, and evaluations of new medical devices are somewhere in between. For example, a randomized trial was initiated a few weeks after the initial testing of a beta-blocker in humans, but it took five years before the first RCT was initiated for CABG. From a historical perspective, differences in the nature of innovation among drugs, devices, and procedures have contributed to different types of regulatory approaches, which in turn have contributed to this imbalance in safety and efficacy information (see Appendix A). Clinical and other health care decisions, however, require comparable information first on the safety and efficacy of a new technology, and then on its effectiveness. Moreover, because the management of clinical conditions such as stable angina increasingly requires choices among alternative diagnostic and therapeutic options, information is also needed on the relative effectiveness and safety of all the various technological alternatives. There are few assessments that provide this kind of information, and these shortcomings in evaluative strategies have been detrimental to a rational and efficient transfer of biomedical research findings into clinical practice.


A major premise of this volume is that we need a more balanced assessment strategy that depends on an adequate model of the development phase within the innovation continuum. The papers in this volume deal with the design and implementation of such a strategy, and address three major issues: (1) What kinds of clinical evidence or endpoints should be evaluated during what stage of the development process? (2) What is the role of observational methods relative to experimental methods (including RCTs) in providing this evidence, and what is the role of methods for synthesizing primary clinical data? (3) What policy mechanisms would ensure that adequate clinical evidence is a major decision-making factor during the development phase of the innovation process?

The Selection of Endpoints in Evaluative Research

A spectrum of relevant endpoints, ranging from physiological or anatomical parameters to mortality, morbidity, health status, functional status, and quality of life, can be evaluated during the development process. The notion of what constitutes valid endpoints is in continual flux. Because many therapeutic agents for today's chronic degenerative diseases treat only symptoms, improvements in functional status, health status, and quality of life are increasingly important endpoints in clinical evaluation. However, Marilyn Bergner in this volume asserts that the inclusion of health status or quality of life considerations in clinical trials is often an afterthought. She argues for a broader approach, especially regarding quality of life, and the inclusion of measures that are reliable and well-validated in clinical trials.

Kenneth Melmon contends that the different participants in the development process—those in industry, regulatory agencies, and clinical research and practice—require different kinds of evidence as a basis for their decision making. This is well illustrated, for example, by the differences in information needed for regulatory decisions as distinct from clinical decisions. The marketing approval decision requires evidence of a new technology's safety and efficacy, but post-marketing regulatory decisions require evidence on its long-term safety in everyday clinical practice. Clinical decisions, however, also require information on effectiveness, and if various technological alternatives are involved in the management of a clinical condition, on relative effectiveness. Furthermore, insight is needed into patient preferences for the health benefits and risks associated with these options.

In the context of regulatory approval decisions, considerable uncertainty exists over the role of intermediate endpoints as surrogates for such clinical endpoints as mortality, morbidity, disability, and quality of life. In some cases the FDA has accepted intermediate endpoints, such as lowered blood pressure with the use of anti-hypertensives. But the value of surrogate endpoints is in dispute for matters such as tissue plasminogen activator, erythropoietin, and cancer chemotherapy. As John Bunker illustrates, the acceptability of these endpoints is affected by such factors as the lethality of the disease, the availability of alternative technologies, the length of time before clinical results will be known, and the strength of the relationship between intermediate endpoints and the patient outcomes of disease treatment. In those cases where intermediate endpoints are appropriate, regulatory acceptance can be increased by systematic follow-up of clinical endpoints in the post-marketing setting.

Several authors in this volume emphasize the need to improve monitoring of outcomes in “real world” clinical practice. Chapter 2 underlines the need to include all-cause outcomes, in addition to disease-specific outcomes, in these studies. For example, some have questioned whether the decrease in cardiac mortality associated with lowering blood cholesterol may be offset by an increase in cancer mortality. To date, the concept of offsetting risks and benefits in innovation remains weak and often is not taken sufficiently into account.

The Selection of Methods for Clinical Investigation

A variety of experimental and observational methods can provide the needed evidence. As mentioned, the RCT is generally regarded as the statistically most powerful method for determining pharmaceutical efficacy in pre-marketing evaluations. During the development of devices and clinical procedures, some real conceptual, practical, and ethical difficulties may exist regarding the use of RCTs, and efficacy evaluation will need to depend on other adequately controlled study designs. John Wennberg, for example, argues that randomization may be unethical when alternative treatment modalities are being developed to increase quality of life, if different interventions are associated with very variable risks and benefits. In this situation, assignment according to patient preferences may be an ethically unavoidable imperative. The value of patient preference trials depends on our ability to distinguish therapeutic effects from effects of preference, placebo, and compliance. Today this understanding is not available, but an innovative research proposal to start disentangling these effects is described in Chapter 4.

Following randomized or otherwise well-controlled safety and efficacy trials, long-term surveillance should be undertaken of the safety and effectiveness of new technologies in actual use. The emphasis in this volume is on the strengths and weaknesses of observational methods, and their role in providing such information. With regard to drugs, William Inman discusses the United Kingdom's Prescription-Event Monitoring System. Using prescription-based cohorts as a starting point, this system actively solicits responses from physicians about patient events (which are very different from suspected adverse effects). In essence, this system links pharmacy records with medical record data bases. Similarly, the FDA, industry, and academia are increasingly investing in the use of Medicaid and other medical record linkage data bases for pharmaco-epidemiological research. Given the increased availability of large-scale automated data bases, the possibilities of inexpensive monitoring of health outcomes are appealing. Leslie and Noralou Roos, Fisher, and Bubolz describe the strengths and weaknesses of health insurance data bases, and discuss how combining administrative and clinical data bases could compensate for some weaknesses. The discussion of the benign prostatic hyperplasia assessment, which compares different surgical techniques and watchful waiting, exemplifies the complementary role of observational methods and experimental methods during the development process.

In addition to methods for primary data analysis, this volume discusses methods for synthesizing existing data and the opportunity they may provide for improving regulatory, industrial, and clinical decision making. If we are to improve clinical decision making, decision analysis is an important tool. As Albert Mulley explains, its value is in the synthesis of the results of both experimental and observational studies, and the distinction it makes between matters of fact—as provided by evaluative research—and value judgments inherent in the use of a technology (for instance, variability in patients' preferences). As such, decision analysis defines uncertainties and demonstrates specific needs for further clinical investigation. Meta-analysis is becoming an important new tool for improving the aggregation of experimental and observational information for decision making purposes, including regulatory decisions. In this respect one will read with interest Stephen Thacker's discussion of meta-analysis techniques based on classical statistics, and David Eddy's discussion of Bayesian statistics. Eddy reviews the existing spectrum of methods, ranging from anecdotal evidence to large-scale RCTs, that can provide clinical evidence during the development process. He asserts that all these methods provide information on the magnitude of risks and benefits, and on the extent of uncertainty in these estimates. The logistics, costs, and time needed for the various study designs differ considerably. In addition, each of these methods is subject to different types of bias that affect its internal and external validity. Because of the complexity of choosing acceptable methods for particular kinds of decisions, decision makers generally apply simple heuristics to determine if a particular study design is acceptable or not. However, these heuristics often do not take into account that different study designs may provide complementary evidence. Furthermore, in view of widespread use of the weaker methods of evaluation and recognizing that decision making often depends on less than perfect information, efforts to improve these methods can be expected to have a substantial impact on enhancing the transfer of biomedical research findings into practice. Eddy describes a methodological approach that identifies the biases inherent in particular studies, estimates their magnitude, and adjusts the results for these biases. Implementation of this approach would enhance the reliability of various evaluative methods that form the basis of developmental decision making.

Policy Mechanisms for Improving Developmental Decision Making

In the aggregate, this volume reflects on the evaluative shortcomings in the present-day development of drugs, devices, and clinical procedures and argues for a more balanced assessment strategy that provides comparable information on the relevant outcomes for all technologies. Recent advances in the art and science of clinical evaluation open up new opportunities for providing this evidence. The major question now remains how to ensure their appropriate application without unduly hampering innovation.

What incentives would encourage increased support of post-marketing research for drugs and devices? This research could provide information on their effectiveness and long-term safety for approved indications, as well as a means for monitoring the emergence of new indications of use. In our opinion, such a change can be effected without modification of the Food, Drug, and Cosmetics Act. Powerful demand and supply factors are stimulating investment in this kind of evaluative research. In today's health care environment, for example, there is an increasing demand for relative effectiveness and long-term safety information by health care professionals and third-party payers, and a growing recognition—from an economic point of view—of the marketing advantages that may accrue if such benefits can be demonstrated. On the supply side, rapid advances in methods for clinical investigation are allowing this information to be provided more reliably and efficiently. This is important in the case of drugs, because the effective patent life for new drugs has decreased considerably over time and the industry is not likely to invest in post-marketing research that provides outcomes information only after the drug has turned generic. The industrial incentive to invest in systematic Phase IV outcomes research would, of course, increase if such investment meant that the time spent in pre-approval evaluations could be shortened.

With regard to procedures, a systematic approach toward providing both “pre-marketing” and “post-marketing” information is needed. We do not wish to imply that the establishment of a federal regulatory system governing the development of procedures is needed or probably would even be effective, especially in view of the decentralized and incremental nature of development. One appealing non-regulatory model for improvement of the innovation-evaluation nexus can be found in the outcomes initiative. It tends to focus on clinical conditions instead of individual technologies, and it provides comparative assessment information on the various technological alternatives. It also includes a diverse spectrum of endpoints, and employs both experimental and observational methods. This initiative would provide a means for early identification of the (incremental) development of procedures in a decentralized environment. On the basis of such information, clinical trials could then be initiated as appropriate. The systematic use of observational methods for monitoring actual performance of new procedures in clinical practice would also allow earlier detection of their long-term safety and effectiveness in everyday use. Moreover, as the focus is on the management of clinical conditions, this initiative will at the same time monitor the long-term effectiveness and safety of the drugs or devices involved.

Federal support for this kind of evaluative research has recently increased. For example, support of outcomes research is a critical part of the congressional mandate to the newly established Agency for Health Care Policy and Research. Drug and device manufacturers can also be expected to take interest in helping fund this initiative as a way of providing relative safety and effectiveness information on their new products. However, if the stronger financial sectors of our health care system (the drug industry, for instance, invests roughly $6.5 billion in R&D in the United States) were to share the financial burden of performing evaluations of clinical procedures, their involvement could pose conflicts of interest. It therefore seems timely to explore acceptable models of privatepublic cooperation in funding this kind of clinical investigation.

In conclusion, a more rational and efficient development stage in the innovation process will require stronger and new kinds of alliances in evaluative research among the various participants: those who develop new technologies; those who improve and apply the science and tools of evaluation; and those who use the resulting information for regulatory approval, reimbursement, or clinical decisions. It will also require a willingness to explore and debate the often complementary value of various evaluative methods for improving developmental decision making. We hope this volume, the first in a series on issues in medical innovation, will contribute to such a debate.


Wiener H. Problems in the assessment of side effects of new drugs. (photocopy). New York.
Eddy DM, Billings J. The quality of medical evidence: Implications for quality of care. Health Affairs 1988; Spring:20–32. [PubMed: 3360391]
EC/IC Bypass Study Group. Failure of extracranial-intracranial bypass to reduce the risk of ischemic stroke. New England Journal of Medicine 1985; 313:1191–1200. [PubMed: 2865674]
Wennberg JE. Improving the medical decision making process. Health Affairs 1988; Spring:99–106. [PubMed: 3283013]
Powell EE, Slater IH. Blocking inhibitory adrenergic receptors by a dichloro analog of isoproterenol. Journal of Pharmacology and Experimental Therapeutics 1958; 122:480–488. [PubMed: 13539775]
Black JW, Stephenson JS. Pharmacology of a new adrenergic beta-receptor-blocking compound (netholide). Lancet 1962; 2:311–314. [PubMed: 13869657]
Black JW, Crowther AF, Shanks RG, Smith LH, Domhorst AC. A new adrenergic beta receptor antagonist. Lancet 1964; 1:1080–1081. [PubMed: 14132613]
Sneader W. Drag Discovery: The Evolution of Modern Medicines. Chichester, U.K.: John Wiley & Sons, 1985.
Frishman WH. Clinical differences between beta-adrenergic agents: Implications for therapeutic substitution. American Heart Journal 1987; 113:1190–1198. [PubMed: 2883867]
Preston TA. Historical development of operations for coronary artery disease. Chapter 2 in Coronary Artery Surgery: A Critical Review. New York: Raven Press, 1977.
Effler DB. Myocardial revascularization surgery since 1945 A.D.: Its evolution and its impact. The Journal of Thoracic and Cardiovascular Surgery 1972; 72:823–828. [PubMed: 792576]
Favaloro RG. Saphenous vein autograft replacement of severe segmental coronary artery occlusion: Operative technique. Annals of Thoracic Surgery 1968; 5:334–339. [PubMed: 5647919]
Bunker JP, Hinkley D, McDermott WV. Surgical innovation and its evaluation. Science 1978; 200:937–941. [PubMed: 347581]
Murphy ML, Hultgren HN, Detrec Thomsen J, Taharo T, and participants of the Veterans Administration Cooperative Study. Treatment of chronic stable angina. A preliminary report of survival data of the randomized Veterans Administration cooperative study. New England Journal of Medicine 1977; 297:621–627. [PubMed: 331107]
European Coronary Surgery Study Group. Long-term results of prospective randomized study of coronary artery bypass surgery in stable angina pectoris. Lancet 1982; 2:1173–1180. [PubMed: 6128492]
Coronary Artery Surgery Study (CASS) principal investigators and their associates. A randomized trial of coronary artery bypass surgery: Survival data. Circulation 1983; 68:939–950. [PubMed: 6137292]
The Veterans Administration Coronary Artery Bypass Surgery Cooperative Study Group. Eleven-year survival in the Veterans Administration randomized trial of coronary bypass surgery for stable angina. New England Journal of Medicine 1984; 311:1333–1339. [PubMed: 6333636]
Varnauskas E and the European Coronary Surgery Study Group. Twelve-year follow-up of survival in the randomized European coronary surgery study. New England Journal of Medicine 1988; 319:332–337. [PubMed: 3260659]
Hlatky MA, Leek L, Harrel FE, Califf RM, Pryor DB, Marck DB, Rosatti RA. Tying clinical research to patient care by use of an observational database. Statistics in Medicine 1984; 3:375–384. [PubMed: 6396793]
Loop FD, Lytle BW, Cosgrove PM et al. Influence of the internal mammary artery graft on 10-year survival and other cardiac events. New England Journal of Medicine 1986; 314:1–6. [PubMed: 3484393]
Gruentzig A. Transluminal dilation of coronary-artery stenosis. Lancet 1978; 1:263. [PubMed: 74678]
Gruentzig A, Senning A, Siegenthaler WE. Non-operative dilatation of coronaryartery stenosis: Per-cutaneous transluminal coronary angioplasty. New England Journal of Medicine 1979; 301:61–68. [PubMed: 449946]
USCI Division of CR Bard, Inc. Summary of safety and effectiveness for the USCI Gruentzig Dilaca Coronary Artery Balloon Catheter. Rockville 1980.
Topol EJ, Myler RK, Stertzer SH. Selection of dilatation catheter hardware for PTCA-1985. Catheterization and Cardiovascular Diagnosis 1985; 11:629–637. [PubMed: 2936462]


1 This concern is also evident regarding the economic evaluation of new technologies during their development. This issue will be the subject of a subsequent publication, and thus will not be further discussed in this volume.

2 Effectiveness refers to the probability of benefits under average conditions of use, and efficacy refers to this under ideal conditions of use. Although the law uses the term effectiveness, the approval decision is made on the basis of efficacy information. This paper will therefore use the term efficacy in the context of pre-marketing clinical investigations, that is, to refer to testing under ideal conditions of use.

3 Koppe of Boehringer Ingelheim synthesized propanolol shortly before pronethalol was discovered. However, its clinical potential was not recognized at the time, and no patent was filed.

4 For example, although the use of CABG in humans was first reported in 1968, the VA trial in 1972 originally set out to evaluate the much earlier developed Vineberg procedure. Only after some time did it shift its resources to CABG. If there had been a mechanism to monitor surgical development, this delay could perhaps have been prevented.

Copyright © 1990 by the National Academy of Sciences.
Bookshelf ID: NBK235486


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.3M)

Related information

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...