Vol.4 No.1 Serial No.12 January 25th 2004¡¡

¡¾CLC number¡¿R-4  ¡¾Document code¡¿A  ¡¾Article ID¡¿1672-2531(2004)01-0003-07

Evidence-based Medicine:Past, Present and Future (II)


3 What is the Evidence Supporting EBM?


  Medical practice is intended to allow patients to live longer, and feel better. The ultimate test of EBM would therefore be a trial in which medical students would be randomized to conventional or EBM training and followed for a few years subsequent to graduation, with monitoring of the outcomes of their patients. Such a trial is, for a host of reasons, not feasible.

  The dictates of EBM suggest that when the ideal study is not available, one falls back on the strongest evidence that one can access. In some cases, physiological rationale provides the best guide to patient management. An analogous argument addresses the issue of the evidence supporting EBM.

   EBM suggests that, to the extent possible, clinicians should base their management decisions on findings from systematic reviews of randomized trials. When appropriate, these systematic reviews summarize their findings in meta-analyses that provide single best estimates of the magnitude of treatment impact on patient outcomes.

   Figures 1 and 2, adapted from a paper by Antman and colleagues, depict what is called cumulative meta-analyses of the evidence concerning the impact of thrombolytic therapy and prophylactic lidocaine on death rates in patients suffering from myocardial infarction[9]. The horizontal line in the centre of the left portion of Figure 1 represents an odds ratio of 1.0. An odds ratio of 1.0 denotes that thrombolytic therapy neither increases or decreases the likelihood of dying. Values to the left of the line indicate that treatment is beneficial and reduces death rates; to the right that treatment is harmful and increases death rates.
      The first randomized trial of thrombolytic therapy, designated A1" in the figure, conducted in the late 1950s, enrolled 23 patients. Results suggested that thrombolytic therapy could result in a 50% reduction in the odds of dying (the point estimates, the best estimate of treatment effect given the existing data, are depicted in the figure by dots). However, the 95% confidence interval around that point estimate (in the figure, the lines to either side of the dots) suggest that the results were also consistent with an odds of dying with thrombolytic therapy of almost 2.After investigators completed the second trial of thrombolytic therapy (designated A2") a total of 65 patients had participated in trials B 23 in the first trial, 42 in the second. The point estimate continued to suggest a 50% odds reduction. The confidence interval, while narrower, was still wide.

  As the data continued to accumulate, the point estimate suggested smaller odds reductions, closer to 25%, and the confidence interval began to narrow. By the early 1970s, when over 2 500 patients had been enrolled in 10 RCTs, the confidence interval no longer crossed 1.0, suggesting a real effect of thrombolytic therapy on decreasing death rates. Even a conservative interpretation of the results would have suggested the answer was in by the early 1980s when over 6 000 patients had been enrolled in 27 trials and the confidence interval indicated that odds reductions of less than 10% were very unlikely.

  Did these data mean an end to RCTs of thrombolytic therapy? No, it did not. After the answer was in, another 40 000 patients were enrolled in trials of thrombolytic therapy. Half these patients received placebo or standard care; they were denied the benefits of thrombolytic therapy, and some therefore died unnecessarily. By 1990 when over 48 000 patients had been enrolled in 70 RCTs, the confidence interval around the odds ratio was extraordinarily narrow.Why was it necessary to enrol over 40 000 patients after the answer was in? The right side of Figure 1 provides at least part of the explanation. This part of the figure categorizes experts¡¯ recommendations published in textbooks and journal articles as these data were accumulating. Recommendations offered at the same time varied between experts. For instance, in the latter 1980s, recommendations ranged from suggestions that thrombolysis be administered routinely to all myocardial infarction patients, to categorizations of thrombolysis as an experimental therapy, to not even mentioning thrombolysis.
The right side of Figure 1 suggests two conclusions. First, experts often disagree, offering contradictory recommendations. Second, recommendations may lag substantially behind the data. It was close to a decade between the time the answer was in, and when experts consistently recommended thrombolytic therapy for patients with myocardial infarction.

Figure 2 depicts the history of routine administration of prophylactic lidocaine to patients with myocardial infarction. In this case, there was never any RCT evidence supporting prophylactic lidocaine therapy. Did this stop the experts from recommending routine lidocaine administration? No, it did not. Indeed, up until 1989 when the first meta-analysis was published, the majority of experts continued to recommend prophylaxis as standard therapy for myocardial infarction patients. There was, however, the same sort of substantial variability in expert recommendations throughout the 1980s as we observed with thrombolysis.

Evidence-based recommendations were not available at the time these data were accumulating. Why not? The fundamental methodology of developing such recommendations, the systematic review, had not been applied to the data in question.

Had evidence-based recommendations been available, from the early 1980s they would have consistently recommended thrombolysis. Following the second key criterion for evidence-based recommendations, that values and preferences underlying the recommendations be made explicit, they would have been qualified with statements that the recommendations placed a relatively high value on preventing premature death, a readiness to incur additional health care costs to achieve that objective, and that patients who were extremely stroke averse might decline thrombolysis.

Throughout the 1980s, evidence-based recommendations would have consistently recommended against prophylactic lidocaine. Initially, the recommendations would have been qualified with the proviso that risk-taking patients might wish to receive the treatment. As the data accumulated, the recommendation against lidocaine would have become progressively stronger.

It is not difficult to offer other such examples of discrepancies between recommendations arrived at through traditional approaches to medical decision-making, and those developed using evidence-based approaches. Consider Hormone-replacement therapy (HRT). For over a decade, recommendations from a wide variety of august bodies strongly favoured the widespread use of HRT. These guidelines failed to acknowledge the weakness of the evidence from observational studies suggesting a decrease in cardiovascular risk, the resultant weakness of any recommendation in favour of long-term ingestion of HRT, and the acknowledgement that risk-averse women would want to stay away from the therapy. The subsequent publication of two randomized trials suggesting that HRT does not decrease, and may increase, cardiovascular risk, has reversed recommendations[10,11]Such examples highlight the limitation of traditional approaches of arriving at health care decisions. They provide the evidence that evidence-based approaches often yield courses of action more in accord with patients¡¯ long-term best interests.

4 The Evolution of EBM

In 1992, an article that one might describe as an EBM manifesto appeared in the JAMA[12]. The article emphasized four important differences between EBM and traditional medical practice. The authors suggested that 1) EBM places a higher value on systematically collected evidence and a lower value on unsystematic clinical observation; 2) EBM places a higher value on experiments that focus on outcomes that are important to patients, and a lower value on physiological rationale; 3) that interpreting the medical literature was a key skill for clinicians, and that a formal study of rules of evidence was necessary for a skilled interpretation; 4) finally, that EBM places a higher value on the independent assessment of the individual clinician and a lower value on authority.

Less than a decade later, the 25th and final instalment of the JAMA series of Users¡¯ Guides to the medical literature took a somewhat different approach in describing the principles of EBM[13]. This article described two key elements of the EBM approach. The first was a hierarchy of evidence going from the strongest (N of 1 randomized trials, followed by systematic reviews of randomized trials) to the weakest (physiological rationale and unsystematic clinical observation). The second key element suggested that evidence itself was insufficient for making health care decisions, and that values and preferences must be part of every decision.

Another major step in the evolution of thinking about EBM made the distinction between evidence-based practitioners and evidence-users[14]. This article acknowledged that only a minority of clinicians would attain a high level of expertise in interpreting the original medical literature. Most EBM practitioners, the authors argued, will understand the fundamental principles of differentiating strong from weak evidence, and will be able to understand estimates of the magnitude of risks and benefits of the management strategies they are considering. They will be able to identify evidence-based synopses and summaries of the literature, and evidence-based guidelines and recommendations that follow from the evidence, and will rely on these sources to guide their practice.

A mature presentation of the principles of EBM emphasizes how EBM fits in to a humanistic approach to the practice of medicine that fully acknowledges the clinicians¡¯ responsibility to the community, the need for a compassionate and empathetic practice, and the primacy of patient and societal values and preferences in medical decisions[15]. The discussion puts great emphasis on the limitations of evidence in making medical decisions, and the need for incorporating values and preferences. Key passages include the following description of the first principle of EBM, that evidence alone is never sufficient to establish the best course of action.

APicture a patient with chronic pain resulting from terminal cancer. She has come to terms with her condition, has resolved her affairs and said her goodbyes, and she wishes to receive only palliative therapy. The patient develops pneumococcal pneumonia. Now, evidence that antibiotictherapy reduces morbidity and mortality from pneumococcal pneumonia is strong. Almost all clinicians would agree, however, even evidence this convincing does not dictate that this particular patient should receive antibiotics. Despite the fact that antibiotics might reduce symptoms and prolong the patient's life, her values are such that she would prefer a rapid and natural death.

Now envision a second patient--an 85-year-old man with severe dementia, who is incontinent, contracted, and mute, without family or friends, who spends his days in apparent discomfort. This man develops pneumococcal pneumonia. Although many clinicians would argue that those responsible for this patient's care should not administer antibiotic therapy because of his circumstances, others, by contrast, would suggest that they should do so. Again, evidence of treatment effectiveness does not automatically imply that treatment should be administered. The management decision requires a judgment about the trade off between risks and benefits; and because values or preferences differ, the best course of action will vary from patient to patient and among clinicians.

By values and preferences, we mean the underlying processes we bring to bear in weighing what our patients and our society will gain--or lose--when we make a management decision. The explicit enumeration and balancing of benefits and risks that is central to EBM brings the underlying value judgments involved in making management decisions into bold relief.

The acknowledgement of the primacy of values and preferences has profound implications for the role of the clinician in the decision-making process. The discussion in the Users¡¯ Guides book puts it this way.

AThus, knowing the tools of evidence-based practice is necessary but not sufficient for delivering the highest quality patient care. In addition to clinical expertise, the clinician requires compassion, sensitive listening skills, and broad perspectives from the humanities and social sciences. These attributes allow understanding of patients' illnesses in the context of their experience, personalities, and cultures. The sensitive understanding of the patient links to evidence-based practice in a number of ways. For some patients, incorporation of patient values for major decisions will mean a full enumeration of the possible benefits, risks, and inconvenience associated with alternative management strategies that are relevant to the particular patient. For some of these patients and problems, this discussion should involve the patients' family. For other problems BB the discussion of screening with prostate-specific antigen with older male patients, for instance BB attempts to involve other family members might violate strong cultural norms.

Many patients are uncomfortable with an explicit discussion of benefits and risks, and they object to having what they perceive as excessive responsibility for decision making being placed on their shoulders. In such patients, who would tell us they want the physician to make the decision on their behalf, the physician's responsibility is to develop insight to ensure that choices will be consistent with patients' values and preferences. Understanding and implementing the sort of decision-making process patients desire and effectively communicating the information they need requires skills in understanding the patient's narrative and the person behind that narrative.@
This mature understanding of EBM places the patient, in one way, at the centre of the decision-making process. After all, only patients have a full and deep understanding of their circumstances, and their values and preferences. In another way, however, it places the physician at the centre of the process[16]. In most instances, the clinician will be the only party with a full understanding of the direct evidence that bears on the decision. Even with the exceptional patient who may have read and understood the relevant literature, the clinician will have an understanding of the medical context of the disease process not available to the patient. The clinician, moreover, must judge the manner in which the patient wants to be involved in the decision-making process. Finally, the clinician has brings an objectivity that may be crucial in achieving a satisfactory decision-making process, and that is often unavailable to the patient.

These considerations give rise to the model of evidence-based practice presented in Figure 3. The patients¡¯ circumstances define the problem. The clinician must make the diagnosis, and must have a knowledge of all reasonable management options that might follow from that diagnosis. The clinician must be aware of all the direct and indirect evidence that bears on the relative benefits, risks, inconvenience and costs associated with the alternative management strategies. The clinician must also ascertain the patient¡¯s desired style of decision-making, and how the patient¡¯s particular circumstances, and values and preferences, bear on the benefits and down sides of the alternative management strategies.

Ultimately, for one patient, the clinician may present the benefits and down sides of alternative course of actions to the patient, who will make the decision. For another patient, the clinician may present the benefits and down sides, but also offer advice and direction regarding the optimal choice. For a third patient, following a discussion of preferences and values, the clinician might recommend a specific course of action that she believes is in the patient¡¯s best interests.

5 The Future of EBM

Acknowledging the primacy of values and preferences, patients¡¯ differing wishes about their role in the decision-making process, and the respective roles of the clinician and patient highlights the tremendous gaps in our knowledge about how best to incorporate evidence in the decision-making process. Reaching a full understanding of benefits and down sides of the alternative courses of action is challenging for the expert, to say nothing of the front-line clinician. Our knowledge of how best to ascertain the patients¡¯ desired style of decision-making and (to the extent the patient wishes) communicate the magnitude of the benefits and risks, remains superficial. The more one considers both the complexity of the process, and the limited time available in the clinical encounter, the greater the temptation to throw up ones¡¯ hands in despair at the apparent impossibility of a satisfactory solution.

Nevertheless, as clinicians wanting to do the best for our patients, we cannot retreat from the challenge. As individuals, we make dozens, if not hundreds, of management decisions each day. Most are trivial, a minority are momentous for our patients. Key research and conceptual questions for evidence-based decision makers include the following. How should we decide which decisions warrant using the traditional short-circuit approach, and which merit additional time and energy? How can we best tailor our decision-making style to the patients¡¯ wishes? How do we best communicate risk and benefit information to our patients? How do we decide when the outcome of the decision-making process is optimal, and when it is not?

Some may find it ironic where the focus on rigour in making causal inferences, and the critical use of the medical literature in clinical practice, has led us B that is, to a detailed consideration of how to bring patient values and preferences to bear on individual decisions. Personally, I find it fascinating, and exciting. EBM will continue to evolve, possibly to further unexpected destinations.

(Received date£º2003-12-04  Revised date£º2003-12-14)
¡¾Key word¡¿
Evidence-based medicine; evidence-based medicine history; evidence-based decisionª²making
[1]  Guyatt G, Rennie D, eds. The Users¡¯Guides to the Medical Literature: A Manual for Evidenceª²Based Clinical Practice [M]. Yarmouth: AMA publications, 2002
[2]  Chalmers I, Trohler U. Helping physicians to keep abreast of the medical literature: Medical and Philosophical Commentaries, 1773-1795 [J]. Ann Intern Med, 2000; 133(3): 238-243
[3]  Medical Research Council: Streptomycin treatment of
pulmonary tuberculosis [J]. BMJ, 1948; 2: 769-782
[4]  Department of Clinical Epidemiology and Biostatistics,
McMaster University. Clinical disagreement I: how often it occurs and why [J]. Can Med Assoc J, 1980;123 (6):499-504
[5]Department of Clinical Epidemiology and Biostatistics,
McMaster University. How to read clinical journals: I. Why to read them and how to start reading them critically [J]. Can Med Assoc J, 1981; 124:555-558
[6]  Department of Clinical Epidemiology and Biostatistics,
McMaster University. The interpretation of diagnostic data: I. An introduction, and how to do it with pictures[J]. Can Med Assoc J, 1983;129: 429-432.
[7]  Guyatt GH. Evidenceª²Based Medicine[editorial]. ACP
Journal Club 1991:Aª²16. (Annals of Internal Medicine; vol. 114, suppl. 2)
[8]  Guyatt GH, Rennie D. Users¡¯Guides to the Medical Literature[J]. JAMA, 1993;270(7):2096-2097
[9]  Antman EM, Lau J, Kupelnick B, Mosteller F, Chalmers TC. A comparison of results of meta analyses of randomized control trials and recommendations of clinical experts. Treatments for myocardial infarction[J]. JAMA, 1992;268(2):240-248
[10] Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, Vittinghoff E. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women[J]. Heart and Estrogen/progestin Replacement Study (HERS) Research Group. JAMA, 1998; 280(7): 605-613
[11] Writing Group for the Women¡¯s Health Initiative. Risks and benefits of estrogen plus porgestin in health postmenopausal women[J]. JAMA, 2002; 288:321-333
[12] Evidenceª²based Medicine Working Group. Evidenceª²based
medicine: a new approach to teaching the practice of medicine\[J\]. JAMA, 1992; 268:2420-2425
[13] Guyatt GH, Haynes RB, Jaeschke RZ, Cook DJ, Green L, Naylor CD, Wilson MC, Richardson WS. For the Evidenceª²Based Medicine Working Group. Users¡¯Guides to the Medical Literature: XXV. Evidenceª²Based Medicine: Principles for Applying the Users¡¯Guides to Patient Care[J]. JAMA, 2000; 284(10):1290-1296
[14] Guyatt GH, Meade MO, Jaeschke RZ, Cook DJ, Haynes RB. Practitioners of evidence based care. Not all clinicians need to appraise evidence from scratch but all need some skills[J] BMJ, 2000;320(7240):954-955
[15] Guyatt G, Haynes RB, Jaeschke R, Cook D, Greenhalgh T, Meade M, Green L, Naylor CD, Wilson M, McCalister F, Richardson S. The philosophy of evidenceª²based medicine[M]. In: Guyatt G, Rennie D. The Users¡¯Guides to the Medical Literature: A Manual for Evidenceª²Based Clinical Practice. Yarmouth: AMA publications, 2002
[16] Haynes RB, Devereaux PJ, Guyatt GH. Physician' and
patients' choices in evidence based practice[J]. BMJ, 2002; 324(7350):1350


Gordon Guyatt

(Department of Medicine
& Clinical Epidemiology & BiostatisticsMcMaster University,
Hamilton, Ontario, Canada L8N 3Z5, 905-525-9140)

Biography: Gordon Gugatt, MD, Prof. of Medicine& Clinical Epidemiology & Biostatistics in McMaster Univesity, involve in Evidenceª²based clinical research and practising EBM for 20 years first developing¡°Evidenceª²based Medicine¡± in JAMA in 1991; this article specially for the EBM practioners in China.

ISSN 1672-2531
CN 51-1656/R
Copyright © 2004
All rights reserved.