Author + information
- Received December 14, 2015
- Revision received March 7, 2016
- Accepted March 22, 2016
- Published online July 12, 2016.
- aDivision of National Heart and Lung Institute, Imperial College, London, United Kingdom
- bNetherhall Gardens, London, United Kingdom
- ↵∗Reprint requests and correspondence:
Dr. Desmond J. Sheridan, C/O Catherine Enright, Division of National Heart and Lung Institute, Guy Scadding Building, Royal Brompton Campus, Imperial College, London SW3 6NP, United Kingdom.
Evidence-based medicine (EBM) has a long history, but was revived in the early 1990s by a campaign mounted by a movement that took its name. The EBM movement focused attention on the need for greater objectivity in medical decision-making and led to the Cochrane Collaboration, which provides reviews of evidence on the basis of comparative research. Important limitations of EBM’s effect on medicine have also emerged. Failure to acknowledge the limitations of clinical trials and systematic reviews has limited their applicability to individual patients’ circumstances. An almost exclusive focus on drugs and devices has left vast areas of health care in an evidence vacuum. An overdependence on commissions for its research may have limited its independence in selecting what it investigates. EBM needs to widen its scope beyond drugs and devices to address many areas that often lack evidence at present, notably, health policy, management, and reforms.
The concept of evidence-based medicine (EBM), in terms of words and meaning, is relatively straightforward: medicine practiced on the basis of evidence. However, this fails to capture what has come to be understood by evidence-based medicine today, following a campaign launched in the early 1990s by a group known as the EBM movement (1,2). For the purposes of this discussion, we use EBM to indicate the specific meaning advocated by this group and to distinguish it from a wider meaning of evidence-based medicine. Motivated to improve outcomes of diagnosis and treatment, the group launched a campaign to enhance the evidence on which medicine is practiced. Coining the phrase evidence-based medicine and claiming to have identified a new paradigm of medical practice were brilliant rhetorical devices that successfully captured the attention of the general and medical media (3,4). The group called for greater reliance on up-to-date published research, especially clinical trials, and proposed that the value and reliability of evidence should be considered in a hierarchy, with trial data trumping other forms of evidence. In particular, it argued that authoritative opinion, clinical judgment, and mechanistic reasoning are less reliable and should be relegated to a lower position. Ironically, these ideas were put forward on the basis of fictional clinical presentations (1,2).
Despite its immediate success, much of what the EBM movement originally proposed has long been discounted. For example, the idea that a new paradigm of medical practice was identified has been widely criticized for ignoring the long history of efforts to improve the evidence base of medicine (e.g., see Tröhler ) and is no longer defended. Furthermore, the concept that evidence required for clinical decision-making should be considered in a hierarchy of reliability, with clinical trial data higher than clinical evidence and mechanistic reasoning, has also been widely criticized (6). In practice, mechanistic reasoning plays a crucial role in several areas of medicine; clinicians must often use it when attempting to apply data from large, population-based studies to individual patients. It plays an essential role in managing patients for whom no specific trial data is available, and also in formulating hypotheses to justify all original research. For a more detailed discussion of the epistemology of mechanistic reasoning in clinical practice, see Loughlin et al. (7). The notion of a hierarchy of evidence was later abandoned by some members of the EBM movement in a modified definition that described EBM as “the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients” (8). This avoided any attempt to categorize evidence or to clarify what is meant by “judicious” and “best evidence,” and seemed to abandon much of the detail of what had been originally proposed. In reality, this could be applied to best practice at any time, including before the introduction of EBM or even clinical trials. Indeed, it may well have been in the mind of Sir William Osler when he said, “He who studies medicine without books sails an uncharted sea, but he who studies medicine without patients does not go to sea at all” (9). Despite this clarification, for some, the notion of a hierarchy persists, as exemplified in the Tamiflu case described later in the text. The result has been a lingering confusion about what EBM actually means, depending on the perspective of how health care is viewed, for example, by patients, clinicians, public health workers, or policy makers. EBM has made significant and important contributions to medicine. It has also had some negative consequences, both of which are explored here (Table 1). This review is written from a U.K. perspective, in part reflecting the experience of the authors, but it also seems appropriate because the Cochrane Collaboration, which is the most significant development of EBM, was sponsored and founded from the United Kingdom. In addition, health activism in general seems to have been more prominent in the United Kingdom than elsewhere in recent decades, which may explain why EBM expanded there after its launch from McMaster University in Canada. We welcome the response to this paper commissioned by the editors (10), and we hope it will stimulate further thought and discussion about EBM. For these authors, the “baby in the bath water” is clinical evidence related to individual patients, which is often at risk in an ocean of data of uncertain relevance.
Achievements of EBM
The most striking success of the EBM movement has been in capturing the agenda for deciding the future of medicine. Its presence is felt in several journals devoted to EBM and many others that use “evidence-based” as a prefix in the title. We have a web site for EBM, and academic departments, centers, and professors of EBM. Other activist groups have achieved notable success in capturing the health agenda (e.g., the acquired immune deficiency syndrome campaign and some famine relief campaigns); however, they have focused on single issues over limited time periods. In contrast, EBM has extended its influence over the past 2 decades.
The Cochrane Collaboration is the most prominent result of the EBM movement’s efforts. It currently harnesses the efforts of around 40,000 volunteers to produce over 400 systematic reviews of clinical trials/year related to a wide range of medical treatments (11). The Cochrane library, which makes these available, had almost 4 million downloads in 2010 (12).
The launch of EBM coincided with growing interest in clinical guidelines as a means to improve clinical practice (13), some of which had recently been published (14). Such guidelines seek to bring together the best available evidence on specific topics, and the systematic reviews published by the Cochrane Collaboration and other academic groups were important contributors to this development. Indeed, clinical guidelines are intimately connected to EBM in that they provide a means to disseminate what the EBM movement advocates.
Advocacy has been at the center of the EBM movement from the outset, and this has continued most recently in calls for full disclosure of clinical trial data. The issue arose from concerns that published systematic reviews of the effects of the anti-influenza drugs zanamivir and oseltamivir may have been unreliable due to lack of access to some trial data, and this led to a much wider campaign for full disclosure of all clinical trial data by the pharmaceutical industry, later backed by the World Health Organization (WHO) (15) and the European Union (16).
The EBM movement can also claim some credit for increasing awareness of “overdiagnosis,” not least as an exemplar of how to use campaign language to draw attention to an issue. It had been known for several decades that for some diseases, increased rates of diagnosis did not lead to reduced mortality, for example, thyroid cancer and prostate cancer (17,18). However awareness of overdiagnosis has increased dramatically since the 1990s; the word has appeared in 448 papers in PubMed since 1973, of which 424 have been published since 1990. Overdiagnosis has received the most attention in relation to prostate and breast cancer, but is increasingly seen as a problem facing almost all areas of medicine and has become the focus of new campaigns (19,20). Improving clinical outcomes in relation to need and cost is clearly important, and these campaigns will hopefully produce some benefit. However, as will be discussed later in the text, they also risk distorting and oversimplifying the discussion as a binary question of too much versus too little.
EBM also contributed to efforts to reform medical ethics and professionalism and to ensuring that practice using the best evidence is central to them (21,22). Timing of the launch of EBM coincided with growing concerns about medical professionalism. For example, the Tuskegee study in the United States had raised serious concerns about medical ethics and led to important reforms, set out in the Belmont Report in 1979 (23). In addition, many felt that health care had become too costly and that doctor-patient relationships were too patriarchal. In the United Kingdom, an inquiry into an unusually high number of perioperative deaths at the Bristol pediatric cardiac surgical unit between 1984 and 1995 identified a range of serious failures in the health service due to lack of funding, poor organization and leadership, and inadequate facilities (24). The launch of EBM also included criticism of medical practice as being inefficient, inaccurate, and wasteful and causing avoidable harm, and thus became another voice calling for reforms. In addition, the remarkable success of the EBM movement in capturing the attention of all concerned about health care ensured that EBM was high on the agenda as an essential standard to be considered in discussions about medical professionalism. This had the positive effect of ensuring that EBM is explicitly stated as an essential component of medical practice.
Limitations of EBM
A restricted view of evidence
The most troubling aspect of EBM, as advocated since the 1990s, has been the notion that evidence relevant to medicine can be categorized and prioritized in a hierarchical system. In the decades before this time, the volume of published medical research had increased dramatically, as indicated, for example, by a 7-fold increase in the weight of the Index Medicus between 1955 and 1977 (25). The randomized clinical trial had been introduced as a powerful tool for measuring the effectiveness and safety of treatments. The EBM movement argued that medical practice was too subjective and sought to promote greater reliance on published research. In doing so, it proposed that clinical judgment and mechanistic reasoning are less reliable forms of evidence in medicine (1,2). Although this view appears to have been abandoned by the EBM movement (8), it continues to exert influence, resulting in confusion and controversy. A recent example occurred when public health physicians in the United Kingdom recommended the prophylactic use of the antiviral drug oseltamivir in an effort to mitigate the spread of influenza (26). Family doctors objected, fearing that it might not be in the best interest of their patients because of uncertainties about the benefits and safety of these drugs, particularly in elderly people who did not have influenza, many of whom might have other illnesses and be receiving other treatments (27). The case exemplifies a degree of uncertainty that is often present in applying clinical trial data to individual cases, particularly in patients who are frail, have multimorbidities, are receiving several therapies, or may have difficulty in comprehending the risk and benefit of treatment. A letter from Public Health England to family doctors escalated the issue, arguing that its advice was on the basis of the best available evidence, and hinting that noncompliance might result in legal and regulatory consequences (28). The Medical Defence Union, which offers legal advice and support for doctors, intervened to clarify that, “guidelines inform practice but don’t dictate it. They do not replace the knowledge and skills of clinicians. Doctors are expected to be familiar with guidelines, but this does not mean they cannot depart from guidance when in the best interest of their patients. They must be prepared to explain and justify their decisions and actions in such cases” (29). Such conflicts could be avoided by an acknowledgment within guidelines that the clinical evidence needs to be considered in applying them to individual patients, that guidelines are not obligatory, and that they are not to be implemented as “rules.”
This conflict can be traced to the oversimplified and restricted view of evidence as originally advocated by the EBM movement. The recommendations by public health physicians were well motivated in seeking to mitigate a looming epidemic of influenza and were on the basis of what appeared to them to be the best available evidence. However, family doctors had to consider that evidence in the context of their individual patients and the clinical evidence derived from them. In essence, both parties believed they were pursuing EBM. However, in reality, their conclusions were on the basis of different assemblies of evidence. Evidence derived from individual patients would not have been available to public health physicians, but was essential to the clinical decisions made by family doctors. Thus, the evidence needed to make decisions in medicine differs depending on the questions being considered. It follows that the notion of a single assembly of evidence arranged in hierarchy, with clinical trials trumping clinical evidence and mechanistic reasoning, is misconceived.
Limitations of clinical trials
The introduction of randomized controlled clinical trials was 1 of the greatest achievements of medicine during the last century. Despite some early, unjustified claims to the contrary, they were enthusiastically adopted in cardiology and have formed the basis of most current treatments. The development of clinical trial designs and methods were derived from the specialty of epidemiology and this may, in part, explain why the evidence that they provide was given priority by the EBM movement, which was largely composed of members of that specialty (2). However, despite their strengths, the limitations of clinical trials are increasingly recognized in applying evidence in the practical world of clinical medicine. Although the limitations of clinical trials are not the fault of EBM, an over-reliance on them, at the expense of clinical evidence, can be misleading.
• Unrepresentativeness of trial subjects. Well-designed clinical trials have clearly defined inclusions and exclusions. These usually include age limits, the presence of other diseases, the concomitant use of drugs, and a range of clinical and biochemical criteria. This often means that only a minority of those with the condition under study is recruited. A high proportion of patients being treated today are elderly. Aronow (30) has drawn attention to the poor representation of older patients in clinical trials related to coronary disease carried out up to the end of the last century. For example, 37% of all patients admitted with myocardial infarction in the United States are over 75 years of age, but they accounted for only 2% of patients with acute coronary syndromes recruited to clinical trials from 1966 to 1990 and 9% from 1991 to 2000, with women similarly unrepresented (31).
• Long-term therapy. Millions of patients with cardiovascular diseases are prescribed drugs such as aspirin, statins, beta-blockers, and angiotensin-converting enzyme inhibitors for decades (and often for life), and yet, clinical trial evidence is only available for 5 to 10 years. Furthermore, they are frequently used in combination and continued into old age, for which there is no reliable evidence (for a review, see Rossello et al. ). Development of new therapies, such as primary coronary intervention for acute coronary syndromes, could potentially alter the natural history of disease with the result that trials carried out prior to their introduction may no longer be relevant to some patients. Thus, there is a major gap in our knowledge about the risks and benefits of long-term treatment. In particular, no studies have examined whether continuation of treatment long-term is beneficial or harmful compared with withdrawal. Until recently, there has been a tacit assumption that randomized controlled trials (RCT) using 1 drug to treat a single condition provide a “gold standard” for guiding treatment and creating clinical guidelines.
• Comorbidity. Multimorbidity is common in patients with chronic diseases, and especially in older patients. It has also often been raised as a potential confounding effect of treatment when several drugs may be prescribed for several reasons. First, patients with comorbid diseases are nearly always excluded from clinical trials (33), as are many patients receiving concurrent drug therapy (34). Therefore, it is often unclear whether data from most trials can be safely applied to such patients. A recent study followed 8,578 elderly patients for a mean of 24 months with 2 or more of the following comorbid conditions: atrial fibrillation, coronary artery disease, chronic kidney disease, depression, diabetes, heart failure, hyperlipidemia, hypertension, and thromboembolic disease. The study compared the effect on mortality of 9 commonly used drugs (beta-blockers, calcium-channel blockers, clopidogrel, metformin, renin-angiotensin system blockers, selective serotonin reuptake inhibitors and serotonin-norepinephrine reuptake inhibitors, statins, thiazides, and warfarin) with clinical trial results (35). The study found that for beta-blockers, calcium-channel blockers, renin-angiotensin system blockers, statins, and warfarin, the average effect on mortality was comparable to those reported in RCTs; however, the effects of beta-blockers and warfarin varied according to coexisting conditions, and clopidogrel, metformin, serotonin reuptake inhibitors, and serotonin-norepinephrine reuptake inhibitors showed no survival benefit. Thus, although this initial study may provide some reassurance in some areas of comorbidity and polypharmacy, it also confirms that for many patients, the applicability of clinical trial data remains uncertain and there is clearly a need for further research.
• Statistical, as opposed to clinical significance. There seems to be something magical about p < 0.05. It is common practice to consider p < 0.05 as “significant,” and a p value higher than this as nonsignificant. But significance cannot really be dichotomized in this way. Furthermore, a skilled statistician can, by introducing “corrections,” move a p value from 0.051 to 0.049, and this has been done. What a statistician or epidemiologist sees as “significant” may not seem so to the clinician. Because of limitations of size, few trials have enough patients in pre-defined subgroups to allow meaningful statistical analysis. Clinicians know that patients vary greatly from each other in their risk and in their responsiveness to treatment. This is well illustrated by the MIAMI (Metoprolol in Acute Myocardial Infarction) trial of metoprolol in acute myocardial infarction (AMI), which found no significant overall benefit from the treatment (36) (Table 2). Would it not be more rational to conclude, bearing in mind the MIAMI findings, that there are biologically plausible subgroups, albeit not statistically significant, that would merit beta-blocker therapy?
• Similarly, in the low-dose aspirin trial in people at cardiovascular risk, treatment significantly reduced the relative risk of cardiovascular deaths by 44% (37). This might suggest that the results should give practitioners the confidence to recommend low doses of aspirin for primary prevention in individuals who have 1 or more risk factors. However, clinicians, aware that this represents a very low reduction in absolute risk (number needed to treat = 6 of 1,000) and that treatment was associated with an increased risk of severe bleeding, might be reluctant to do so (Table 3).
• Misleading results. Clinical guidelines depend heavily on meta-analyses and systematic reviews, as exemplified by the Cochrane Reviews. These contribute data from large numbers of patients and may provide a legitimate basis for subgroup analysis. However, there are major concerns about their validity for a number of reasons. Bias in the reporting of clinical trials is well recognized and can arise for several reasons, such as inappropriate subject selection, poor study performance, or incorrect analysis of data. Despite evidence that the quality of published clinical trials has improved in recent years, the risk of bias remains significant (38). Bias is also known to occur due to journals’ reluctance to publish negative results, an example of which was witnessed by one of the authors (D.G.J.) when a positive trial of pre-hospital thrombolysis in 300 patients was published in the BMJ while, at the same time, a negative study of 5,500 similar patients was rejected by The Lancet. Furthermore, although it is recognized that analyses funded by pharmaceutical companies are highly suspect, conflicts of interest may also be involved if the authors’ careers or reputations might be adversely affected by the result. It is important that those carrying out such reviews are not influenced by such considerations. It is not uncommon for reviews carried out by different groups to come up with strikingly different conclusions. For example, in reviewing the use of beta-blockers in AMI, Maggioni et al. (39) concluded that, “Systematic review of 27,536 patients showed significant reduction in mortality. Subgroup analyses are untrustworthy. All patients with AMI, in the absence of specific contra-indications, should be treated with a beta-blocker within 24 hours from the onset of symptoms.” By contrast, Freemantle et al. (40) reported that a systematic review of 29,260 patients showed no significant reduction in mortality. If subgroup analysis is untrustworthy, one must conclude that no patients with AMI should be treated with beta-blockers.
Another important reason for concern about publication bias is exemplified by cardiac rehabilitation. Cardiac rehabilitation was introduced in the 1960s in an effort to restore patients to full health following a cardiac event, and has been included in clinical guidelines on the basis of evidence that it reduces all-cause mortality by about 20% (41,42). However, these guidelines were on the basis of systematic reviews of small trials conducted more than 30 years ago (43). The WHO multicenter trial, published in 1983, failed to show any reduction (44). A systematic review, published in 2000, reported a significant reduction in all-cause mortality in 12 trials that used only exercise rehabilitation, but no reduction in 28 trials that used a more comprehensive program (45). A further systematic review, published in 2004, that combined all of these trials reported a significant reduction (46). However, pooling of trials published after the 1983 WHO trials showed no reduction (47). A RCT of comprehensive rehabilitation in 1,813 patients following myocardial infarction between 1997 and 2000 also showed no effect on mortality or morbidity (48). Likely reasons for these conflicting results include the fact that early studies were small and vulnerable to publication bias, and that the prognosis of coronary disease changed dramatically after the 1960s, due to improvements in treatment and prevention. In addition, the absence of any rational explanation for the remarkable effect claimed for rehabilitation on mortality leaves many clinicians doubtful about them. Despite these uncertainties, advocacy of cardiac rehabilitation on the basis of mortality reduction remains strong among its long-term supporters (49). This appears to reflect a belief that accumulating numbers of patients included in ever-larger systematic reviews provide the most reliable evidence of benefit or harm, while ignoring potential confounding bias. This, in turn, goes to the heart of conflicts about applying evidence derived exclusively from systematic reviews to individual patients.
Some adverse effects of EBM on clinical practice
EBM has made valuable contributions to medicine through its emphasis on objective data from clinical trials and guidelines. However it has also had some negative indirect effects on clinical medicine, leading some to argue that EBM is in crisis and needs to be reformed (50).
• Bias of the easily measurable. A major aim of the EBM movement was to make available objective data to identify effective treatments and discard those that are ineffective or harmful (1,2), and this has been an important advance. But this also begs the questions: what data are measured, what can be measured, and how are the data used? Outcome measures, such as mortality, morbidity, and cost-effectiveness, are effective methods for this purpose and are relatively simple to assess. However, other, less tangible, but important elements of medicine are more difficult to quantify, such as quality of care, patients’ experience of illness, respect for human dignity, and contributions to the knowledge base. Benefits proven on the basis of objectively measured data from clinical trials will inevitably take precedence over aspects for which evidence is lacking. But proven ineffectiveness is not the same as lack of evidence. Thus, what we choose to measure can add a layer of bias in relation to what matters in medicine. A famous example of this was Archie Cochrane’s “battle” to stop the development of coronary care units (CCUs) (51). He proposed a trial of mortality outcomes on the basis of home versus hospital care, and described cardiologists who cautioned that randomization would be difficult as suffering from a heavy dose of the “god complex.” The trial went ahead, but in its final follow-up, only 23% of eligible patients were randomized (52). It showed no difference in mortality, but did show that hypotension was a predictor of higher mortality and that general practitioners may have been aware of this, in that they elected to admit more patients with hypotension. There was also a higher incidence of hypotension among patients randomized to hospital admission, raising concerns of possible bias. However, Cochrane hailed the trial as a great success, “If anyone had any doubts about the need for doing RCTs to evaluate therapy recent publications using this technique have given ample warning of how dangerous it is to assume that well-established therapies which have not been tested are always effective. Possibly the most striking example is Dr. Mather’s RCT in Bristol” (51). Cochrane’s opinion had a disastrous effect in England, where the Department of Health concluded that the development of CCUs should be discouraged. Ironically, CCUs provided an ideal milieu for research that advanced our understanding of coronary disease and for the mega-trials that transformed the practice in cardiology and the death rate from myocardial infarction. Clearly, measurement of mortality outcomes was an easier task than measurement of the contribution of CCUs to advances in medical science. Selecting what is easy to measure has the attraction of economy, but, in a study that aimed objectively to measure the value of CCUs, also carried a risk of bias by ignoring important benefits. However, Cochrane (51) had previously argued that original clinical science made no contribution to medical progress, was wasteful, and should be replaced with comparative research. Thus, he would have seen no reason to consider advances in knowledge as a possible value of CCUs and, therefore, was unaware of the bias implicit in the trial design.
• Bias in commissioned research. Throughout the 20th century, there has been tension about who should decide the direction of science. Should it be scientists who are engaged at the cutting edge seeking new knowledge, or should funders and sponsors of science commission research on the basis of what they regard as highest priority? Nowhere has this been more intense than in the United Kingdom. The concept that scientists should lead science was established there by the U.K. Machinery of Government Committee, set up and led by Viscount Haldane in 1917 (53). This remained the basis for publicly funded research, more or less unscathed, despite unease in government departments, until 1971. The Rothschild report (54) dismissed the idea of “science led by scientists,” proposing that publicly funded research should be done on a “customer-contractor basis” and that research should be commissioned and overseen by government departments that fund research. Although this policy was widely criticized and eventually reversed in 1981, it received further backing from a House of Lords Select Committee on Science and Technology inquiry into medical science in 1988 (55). This resulted in the Department of Health receiving a substantial new funding stream to sponsor applied and operational research in support of its policy aims, and led to the creation of the National Health Research Authority in 1991. This coincided with the launch of EBM, and the ideas advocated by the movement were almost perfectly aligned with the objectives of the National Health Research Authority (56). Thus, although the idea of EBM was launched from McMaster University in Canada, the Cochrane Collaboration, its most significant result, was founded in the United Kingdom with funding from the new National Health Service research and development program, which also recruited a senior member of the EBM movement to set up a new Centre for EBM at Oxford (57). This may also explain the direction that EBM has followed, principally as a source of evidence related to treatments and management of diseases, reflecting the interests of the health providers who commission its work. It has consistently excluded sponsorship from the pharmaceutical industry and for-profit industries, but not from government departments or for-profit insurance companies and health management organizations. The former has clear benefits; however, an over-reliance on the latter raises concerns about commissioning bias. EBM, as understood in a wider context of medicine and medical professionalism, cannot be just be about drugs and devices; it must be concerned with all aspects of health care and be prepared to challenge political power, if necessary, to ensure health service delivery on the basis of reliable evidence (58). The past 3 decades have been a period of unprecedented changes in health care policy, particularly in the United Kingdom, where initiatives have almost always been announced in an evidence vacuum, resulting in confusion and controversy (59–61). To date, the Cochrane Collaboration has resisted extension of its remit, arguing against mission creep on the basis of limited resources. As a result, EBM has been largely silent on these issues; indeed, the volume of its rhetoric on its chosen campaigns may have served to deflect attention from many areas in serious need of evidence. Thus, EBM, as currently implemented, is too narrow; its approach to evidence that matters in medicine is too restricted, and it fails to meet the requirements of EBM in a wider context of medical professionalism.
Comparative research and original science
Almost all research activity currently undertaken in support of EBM involves evaluations of existing evidence to determine the best available, usually in the form of systematic reviews, and this form of research has expanded greatly since the launch of EBM. This is quite different from original science, which seeks to explore and discover new knowledge. Conflation of these forms of research has been a consistent misconception in discussions of medical science in recent years. Examples of this are seen in several areas.
• Antibiotic resistance. The problem of antibiotic resistance is most often presented as one of overuse or misuse, and the need to conserve what we currently have through hand-washing, regulation, and oversight of prescriptions. As important as these are, microbiologists of the late 19th and early 20th centuries were well aware of microbial resistance and how this would limit the utility of antibiotics; indeed, both Fleming (62) and Florey (63) drew attention to it in their Nobel lectures in 1945. By inference, they foresaw a continued need for new antibiotics or new ways to counter infections. This will only be possible through new original science. However, in the United Kingdom, where EBM is strongest, the priority is overwhelmingly to conserve our declining stocks, whereas the urgency to develop new antibiotics through new scientific discovery is rarely mentioned or is thought to be possible by encouraging the pharmaceutical industry through manipulation of markets (64). In contrast, the U.S. approach seems more balanced in supporting conservation and new science (65).
• Overdiagnosis. As discussed earlier, EBM has contributed to increasing awareness of “overdiagnosis” and to the launch of several new campaigns seeking to control “overdiagnosis” and “overtreatment” of many diseases. These newly coined words well describe the results of comparative research and have proved to be powerful campaign language, but they are also deeply misleading in the context of clinical medicine and of what is needed to find solutions. The concept of overdiagnosis relies on diagnosis as a binary phenomenon (patients either have a disease or they do not) and on the idea that too many people are being diagnosed, and that the solution needed is to diagnose and treat fewer patients. But in clinical medicine, this is a grossly misleading oversimplification; here, diagnosis includes not only the illness, but its causation, stage, and rate of progress, complications that have arisen, and prognosis. Patients may be wrongly or inadequately diagnosed if any of these elements are incorrect or lacking. In the case of prostate cancer, we now have campaigns against “overdiagnosis,” “overtreatment,” and “too much medicine” (19,20). The campaign rhetoric of these is to “do less”; they fail to acknowledge that the primary need is for better tests to replace the hopelessly outdated and inadequate prostate-specific antigen test, and yet prostate cancer remains the most common malignancy in men and a major cause of cancer deaths. This pattern of seeking primarily to restrict effort is common to all of the overdiagnosis campaigns and seems to reflect fears that new knowledge and discoveries may threaten health care provision (66). This simplistic approach may seem attractive campaign rhetoric, but it ignores the wider causes of the problem (67) and distracts attention from what is needed to solve them. It is also contrary to the fundamental requirement of medical professionalism to seek new knowledge to reduce illness and improve health.
The Future of EBM
Efforts to improve the evidence on which medicine is based have a long history and will undoubtedly continue. The launch of EBM in the early 1990s gave this a powerful impetus, and the concept of EBM is now an explicit element of medical professionalism. EBM has brought greater awareness of the importance of reliable published evidence in reaching decisions in medicine and has contributed to a large body of reviews that can be easily accessed for this purpose. It has also been an important contributor to the introduction of clinical guidelines. The EBM movement has also highlighted the need for and successfully campaigned for full access to clinical trial data to improve the reliability of research results.
Despite its success, however, there has never been an agreed upon definition of EBM, reflecting, perhaps, its launch on a wave of powerful and successful rhetoric, rather than rational discourse. This has contributed to a restricted view of what evidence matters in medicine and tensions between different areas of health care. This will have to change if EBM is to have a meaningful place in the wider context of medicine. In particular, the information derived from individual patients about their illnesses and their wishes regarding treatment needs to be acknowledged as crucial evidence that must be integrated with other data in making clinical decisions.
EBM cannot be just about drugs and devices, leaving great swathes of evidence related to health care untouched, particularly in the fields of health policy and reforms. Neither can it be a system subject to the motives of research commissioners, be they commercial or political. Such an enterprise is likely to produce a biased and distorted view of the evidence that matters in medicine and health care. EBM needs to be and be seen as independent in the selection of topics it addresses. It must be willing to interrogate and challenge all aspects of medicine and health care, whether commercial, ideological, or political. EBM has shone a light on several important areas of medical practice, but to have a relevant and durable future, it must begin to illuminate the many areas of medicine and health care that operate in the shadows of evidence at present.
Prof. Sheridan’s book, “Evidence Based Medicine: Best Practice or Restrictive Dogma,” is published by Imperial College Press. Prof. Julian has no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute myocardial infarction
- coronary care unit
- evidence-based medicine
- Received December 14, 2015.
- Revision received March 7, 2016.
- Accepted March 22, 2016.
- American College of Cardiology Foundation
- Guyatt G.H.
- Hitt J.
- Dickersin K.,
- Straus S.E.,
- Bero L.A.
- Tröhler U.
- Sackett D.L.,
- Rosenberg W.M.,
- Gray J.A.,
- et al.
- Osler W.
- Brush J.E. Jr..,
- Halperin J.L.
- ↵Cochrane Collaboration. Annual Report & Financial Statements 2010/11. Available at: http://community-archive.cochrane.org/news/tags/authors/collaboration%E2%80%99s-201011-annual-report-financial-statements. Accessed May 4, 2016.
- Irving D.
- ↵(1990) Guidelines for management of asthma in adults: I. Chronic persistent asthma. Statement by the British Thoracic Society, Research Unit of the Royal College of Physicians of London, King's Fund Centre, National Asthma Campaign [published correction appears in BMJ 1990;301:924]. BMJ 301:651–653.
- ↵La Santos Quintana A. PM+: EU’s clinical trials regulation ‘a step in right direction.’ Parliament Magazine. 2014. Available at: https://www.theparliamentmagazine.eu/articles/sponsored_article/pm-eu%E2%80%99s-clinical-trials-regulation-%E2%80%98-step-right-direction%E2%80%99. Accessed May 4, 2016.
- Welch H.G.,
- Black W.C.
- ↵The BMJ. Too much medicine. Available at: http://www.bmj.com/too-much-medicine. Accessed May 4, 2016.
- Preventing Overdiagnosis Board. Preventing overdiagnosis. Available at: http://www.preventingoverdiagnosis.net. Accessed May 4, 2016.
- Royal College of Physicians. Doctors in society: medical professionalism in a changing world. Report of a Working Party of the Royal College of Physicians of London. 2005. Available at: https://cdn.shopify.com/s/files/1/0924/4392/files/doctors_in_society_reportweb.pdf?15745311214883953343. Accessed May 4, 2016.
- ↵Office of the Secretary, Department of Health, Education and Welfare, National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research. The Belmont Report: ethical principles and guidelines for the protection of human subjects of research. 1979. Available at: http://archive.hhs.gov/ohrp/humansubjects/guidance/belmont.htm. Accessed May 4, 2016.
- ↵The Bristol Royal Infirmary Inquiry. The report of the public inquiry into children’s heart surgery at the Bristol Royal Infirmary 1984–1995: learning from Bristol. July 2001. Available at: http://webarchive.nationalarchives.gov.uk/+/www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPolicyAndGuidance/DH_4005620. Accessed May 16, 2016.
- ↵Cosford P. The use of antivirals for the treatment and prophylaxis of influenza. Available at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/370676/Letter_to_clinicians.pdf. Accessed May 4, 2016.
- Cohen D.
- ↵Payne G, Grice D. Re: Tamiflu for the prophylaxis of influenza in nursing and care homes. Available at: http://margaretmccartney.com/wp-content/uploads/2015/01/Tamiflu-letter.pdf. Accessed May 4, 2016.
- Fryar C.
- Tinetti M.E.,
- McAvay G.,
- Trentalange M.,
- et al.
- Maggioni A.P.,
- Latini R.,
- Tognoni G.,
- et al.
- Freemantle N.,
- Cleland J.,
- Young P.,
- et al.
- Department of Health
- Giannuzzi P.,
- Saner H.,
- Björnstad H.,
- et al.
- World Health Organization (WHO)
- West R.R.
- West R.R.,
- Jones D.A.,
- Henderson A.H.
- Dalal H.M.,
- Doherty P.,
- Taylor R.S.
- Greenhalgh T.,
- Howick J.,
- Maskrey N.,
- Evidence Based Medicine Renaissance Group
- Cochrane A.
- Mather H.G.,
- Morgan D.C.,
- Pearson N.G.,
- et al.
- Ministry of Reconstruction
- ↵(1971) A Framework for Government Research and Development (The Rothschild Report), Cmnd 4814. The Organisation and Management of Government R&D (HMSO, London, UK).
- House of Lords Select Committee on Science and Technology
- Royal College of Physicians
- Tunis S.R.
- Kmietowicz Z.
- ↵Fleming A. Penicillin. Nobel Lecture, December 11, 1945. Available at: http://www.nobelprize.org/nobel_prizes/medicine/laureates/1945/fleming-lecture.pdf. Accessed May 5, 2016.
- ↵Florey HW. Penicillin. Nobel Lecture, December 11, 1945. Available at: http://www.nobelprize.org/nobel_prizes/medicine/laureates/1945/florey-lecture.pdf. Accessed May 5, 2016.
- ↵O’Neill J. Review on antimicrobial resistance: tackling a crisis for the health and wealth of nations. 2014. Available at: http://amr-review.org/sites/default/files/AMR%20Review%20Paper%20-%20Tackling%20a%20crisis%20for%20the%20health%20and%20wealth%20of%20nations_1.pdf. Accessed May 5, 2016.
- ↵The White House, Office of the Press Secretary. Fact sheet: President’s 2016 budget proposes historic investment to combat antibiotic-resistant bacteria to protect public health. January 27, 2015. Available at: https://www.whitehouse.gov/the-press-office/2015/01/27/fact-sheet-president-s-2016-budget-proposes-historic-investment-combat-a. Accessed May 5, 2016.
- Gabbey J.,
- Walley T.