A brief history of evidence-based medicine

| INTRODUCTION: Evidence-based medicine is one of the most widespread trends in contemporaneous medical education. It proposes a scientific framework not only for medical training, but also for medical research and practice. However, knowledge about EBM roots and historical developments is not usual within Brazilian medical community. Indeed most common publications for non specialists medical readers, like handbooks and tutorials papers on EBM, are not sufficiently rich for providing historical knowledge. OBJECTIVES: To present a brief narrative of the historical development of evidence-based medicine. METHODS: Historiographical review essay. MATERIALS: Primary and secondary sources on the history of EBM. RESULTS: As EBM founder, David Sackett stated against clinical decisions based solely on physicians authority and intuition achieved by long term clinical experience and pathophysiological knowledge. Paradoxically, in his retiring letter, Sackett alleged that his own prestige and authority could retard the scientific advance of EBM. CONCLUSION: Since the early 2000’s, critical appraisal, systematic reviews and clinical practice guidelines has merged in a unified approach that characterizes current practice in EBM.

In 1964, a report from the Canadian government recommended the creation of a new medical school at McMaster University, Ontario, that should introduce a new approach to medical education, current medical school programs were evaluated out-of-date. This new approach was based in the introduction of clinical epidemiology and biostatistics in the medical program also the medical curriculum would be based on valid outcomes of medical research. David Sackett was the first director of the Department of Clinical Epidemiology and Biostatistics at the Medical School established in 1967. He claimed that clinicians should be trained to develop skills needed to ask epidemiological questions relevant to solve practical clinical problems. Since then, debates about Sackett's proposal have target the meaning and relevance of epidemiological knowledge in clinical practice, as well as the uncertainty of medical judgments based on medical authority 1 . The general scientific problem with which we are primarily concerned is that of testing a hypothesis that a certain treatment alters the natural history of a disease for the better. The particular problem is the value of various types of evidence in testing the hypothesis. The oldest, and probably still the commonest form of evidence proffered, is clinical opinion. This varies in value with the ability of the clinician and the width of his experience, but it value must be rated low, because there is no quantitative measurement, no attempt to discover what would have happened if the patients had had no treatment, and ever possibility of bias affecting the assessement of the result. It could be described as the simplest (worst) type of observational evidence 3 .
At that time a dilemma emerged: What are the rules of evidence that should be adopted as the basis for the clinical management of patients? Should only RCT -validated evidence be used to prevent or minimize the use of therapeutic resources innocuous or harmful to patients? Or should clinicians experiences also be admitted as a basis for maximizing the potential patients health benefits 8 ?
Yet in 1976, at Copenhagen, Hendrik Wulff had detailed the logical and probabilistic aspects involved in the application of RCT outcomes in clinical practice 9 .
He drew attention to the difference between therapeutic efficacy as measured by the statistical likelihood obtained from RCT and clinical effectiveness as measured by the subjective likelihood of the physician's belief in the cure of a particular patient, calculated by applying the Bayes theorem. Even patients with the same disease differ in a number of ways, so that it is not always rationally certain that the physician will base his belief (subjective probability) on the overall experience of a group of patients (statistical probability). In addition, the physician should assess to what extent it is appropriate to apply group experience to the individual 10 .
In the 1980s, clinical epidemiology spread internationally in medical education curricula, despite the difficulties faced due to the required mathematical and statistical knowledge and skills 11 . The initial stimulus for this diffusion came from the Rockefeller Foundation, which in 1978 funded the establishment of the International Clinical However, according to David Eddy 19 , there were actually two "evidence-based" approaches: one aimed developing guidelines (EBG) and another target the individual development of physicians (EBID). The latter was designed, developed, and disseminated by Sackett and his partners, while the other was his own work, following a line of research on health care costs that began in the 1980s with RAND Corporation: "In the 1980s a group at RAND began publishing studies showing that large proportions of procedures being performed by physicians were considered inappropriate even by the standards of their own experts 19 ." The RAND corporation was a center of health services expertise by the late 1970s, when neoliberal policies was introduced by Ronald Reagan in the United States 20 . This time, the focus of the political debate on health had shifted from poor people's access to health services to managing the costs of health services: In the Medicare program, as in American health care more generally, the concerns of policymakers soon shifted from access to cost (…) In time, then, the field of health services research turned its attention to the technical issues and quality concerns related to cost containment. An important point in this process was the RAND health insurance experiment 20 .
An important point in this process was the Rand health insurance experiment 20 . RAND had developed an institutional expertise in the application of quantitative methods to the education and health.
The impetus for the health insurance experiment came not so much from clinicians concerned about health outcomes, as from economists who focused their attention on the relationship between the cost of medical care and its consumption. By the time, the experiment was intended to anticipate the creation of a national health insurance, but it failed in its purpose. Nonetheless, outcomes of the experiment contributed to the big increase of cost sharing that occurred in the 1980s, and this tended to reduce services in an indiscriminate fashion with adverse effects on the health of vulnerable: (...) it tended to reduce services in an indiscriminate fashion -the good along with the bad. Furthermore, the experiment showed that cost-sharing had adverse effects on the health of vulnerable groups, such as low-income children, "just a catastrophic drop in the use of services, clearly services that were needed as well as services that weren't that you didn't see so much for kids with a higher income 20 ." David Eddy himself has been linked to the private health insurance industry from 1984 to 2005, as chief scientist for the Technology and Coverage Program and the Medical Advisory Panel of Blue Cross Blue Shield, a federation of health insurers, which serves more than 100 million Americans 21, 22 . In this context that Eddy wrote the first evidence-based guidelines for the American Cancer Society: First, there must be good evidence that each test or procedure recommended is medically effective in reducing morbidity or mortality; second, the medical benefits must outweigh the risks; third, the cost of each test or procedure must be reasonable compared to its expected benefits; and finally, the recommended actions must be practical and feasible 23 .
At that time there was indeed a concern with the relationship between the search for good evidence for the clinical effectiveness of the procedures with their cost and benefit efficiency. In his 2005 paper, David Eddy asked that, since the current definition of EBM includes EBID but not EBG, whether the definition of EBM should be expanded to include evidence-based guidelines and its related branches, instead of focusing only on physicians and their individual decisions, comprising a set of principles and methods to ensure that medical decisions, protocols, guidelines and other types of health policy would be based and consistent with good evidence of effectiveness and benefit 19 .
Indeed, in 1997, in the same year that David Sackett published his EBM handbook 24 , Muir Gray also published another EBM handbook by the same publisher, Churchill Livingstone, of the Elsevier (Elsevier Science) group 25 the first focused on individualized clinical practice and the second on health policies. In his book, Muir Gray includes rising costs and "delayed implementation of research results in practice" in the list of major convergent and common problems to the delivery of healthcare world-wide, so that the same solutions should be adopted, either in the post-industrial northern countries or in the "third world" countries, whose health systems should be restructured. In short, the solutions highlighted by Muir Gray focused on aspects such as cost control, healthcare purchasing, and clinical practice management: In 2017, in his commemorative paper on EBM's 25th Anniversary, Gordon Guyatt acknowledged the seminal roles played by David Sackett, Archie Chocrane and David Eddy in the early days of EBM, when they argued for critical appraisal, development of systematic reviews, and clinical practice guidelines, three domains that merged in the 2000s to characterize the current practice of EBM 28 .

Author contributions
Lapa TG is the first author of the article which is an extract from her thesis. Rocha MD supervised the research. Almeida N was coadvisor of the thesis, contributing mainly, but not exclusively, to the aspects of work related to Evidence-Based Medicine, including his vision as a model of contemporary education in the health sciences field. Mattedi A contributed to the methodological aspects related to the history of scientific controversies.

Conflicts of interests
No financial, legal or political competing interests with third parties (government, commercial, private foundation, etc.) were disclosed for any aspect of the submitted work (including but not limited to grants, data monitoring board, study design, manuscript preparation, statistical analysis, etc.).