Evidence-Based Medicine

Summary

Evidence-based medicine (EBM) is the idea that clinical decisions should be grounded in the best available research evidence, especially from randomized controlled trials and systematic reviews. Since the 1990s, EBM has become the dominant framework for evaluating whether medical treatments work. It grades evidence in hierarchies, placing large trials and meta-analyses at the top and case reports at the bottom. But the framework has significant problems. Research biases systematically make drugs look more effective and safer than they are. Statistical manipulations inflate apparent benefits. Clinical trials study average patients, not the particular person sitting in front of the doctor. And since roughly the year 2000, large trials show that many common interventions have very small average effects. Philosophers and clinicians have argued that EBM cannot replace the practical judgment physicians have always needed to care for individuals — it can only supplement it.

The EBM Framework

Medicine relies on concepts, theories, inferences, and policies that are philosophically complicated and controversial.(Stegenga, 2018) A seemingly straightforward question — are antidepressants effective for depression? — turns out to require a compelling view about what evidence is needed and insight into the social, legal, and financial nexus from which that evidence emerges.(Stegenga, 2018) Evidence-based medicine attempts to address this complexity by ranking research methods according to the potential for those methods to suffer from bias.(Stegenga, 2018)

The EBM movement draws its authority from the randomized clinical trial.(Haller, 2014) Ronald Fisher first formulated the principle of randomization for agricultural experiments, and Austin Bradford Hill applied it to medicine; Hill’s 1937 Principles of Medical Statistics became the standard reference for medical problem-solving and aimed to rule out both conscious and unconscious bias on the part of patient or investigator.(Haller, 2014) The first properly randomized clinical trial tested streptomycin for tuberculosis in 1948; the reason random-number allocation was preferred over simple alternation was that it was easier to conceal, not that it better abolished selection bias.(Haller, 2014)

Archie Cochrane’s 1972 book called for better-quality reasoning than what then passed as evidence-supported practice, leading to the Cochrane Collaboration, which by 2003 included fifty review groups and fifteen centers worldwide.(Haller, 2014)

The textbook that codified the field for working clinicians, Sackett, Straus, Richardson, Rosenberg, and Haynes’s Evidence-Based Medicine: How to Practice and Teach EBM (2nd ed., 2000), defines EBM as “the integration of best research evidence with clinical expertise and patient values.”(Sackett, David L. et al., 2000) By “best research evidence” the authors mean clinically relevant research, especially patient-centered work into the accuracy and precision of diagnostic tests, the power of prognostic markers, and the efficacy and safety of therapeutic regimens.(Sackett, David L. et al., 2000) The authors identify with expression of these ideas in post-revolutionary Paris, where clinicians like Pierre Louis rejected the pronouncements of authorities and sought the truth in systematic observation of patients, and note a colleague’s suggestion of an earlier origin in ancient Chinese medicine.(Sackett, David L. et al., 2000) Earlier antecedents also appear in the Islamic medical tradition: al-Majusi (died 994) stated in the Kitab al-Maliki that the best way to determine the effects of a drug is to test it on healthy people as well as the sick, and to keep accurate records of outcomes.(Saad Said, 2011) Avicenna articulated seven rules for drug testing in his Canon that included requirements for testing pure drugs on simple diseases, using multiple subjects, and confirming effects in humans rather than relying on animal models alone.(Saad Said, 2011) In the modern era, however, the ideas were “consolidated and named EBM in 1992 by a group led by Gordon Guyatt at McMaster University in Canada,” after which the literature on evidence-based practice grew from a single publication in 1992 to roughly a thousand by 1998.(Sackett, David L. et al., 2000)

Sackett et al. attribute EBM’s rapid spread to four shared realizations among clinicians: a daily need for valid information about diagnosis, prognosis, therapy, and prevention; the inadequacy of traditional sources (textbooks out of date, experts wrong, didactic continuing medical education ineffective, journals overwhelming and variable in validity); a disparity between diagnostic skills, which increase with experience, and up-to-date knowledge, which declines; and an inability to spend more than a few seconds per patient finding evidence.(Sackett, David L. et al., 2000) Five infrastructural developments turned this state of affairs around: methods for tracking down and appraising evidence; systematic reviews and the Cochrane Collaboration; evidence-based journals of secondary publication; information systems that bring evidence to the bedside in seconds; and strategies for lifelong learning.(Sackett, David L. et al., 2000) The “full-blown” practice of EBM, on this account, comprises five steps — converting an information need into an answerable question, tracking down the best evidence, critically appraising it for validity, impact, and applicability, integrating that appraisal with clinical expertise and patient values, and evaluating one’s own effectiveness.(Sackett, David L. et al., 2000)

Sackett et al. acknowledge that no randomized trial has shown EBM “works” to improve patient outcomes, citing problems of sample size, contamination, blinding, and follow-up, and the ethical concern of withholding evidence access from controls; they rely instead on observational outcomes research showing that patients who receive evidence-based therapies fare better than those who do not.(Sackett, David L. et al., 2000) They concede three EBM-specific limitations — the skill burden of searching and critical appraisal, the time pressures on busy clinicians, and the slow arrival of evidence that EBM works — but reject “pseudo-limitations” alleging that EBM denigrates clinical expertise, ignores patient values, or amounts to cookbook medicine.(Sackett, David L. et al., 2000) They are also explicit that EBM is not a cost-cutting tool: care directed toward maximizing quality of life often increases costs and “raises the ire of health economists.”(Sackett, David L. et al., 2000)

Evidence Hierarchies

EBM assesses the diversity of medical evidence with hierarchies that rank research methods by their vulnerability to bias.(Stegenga, 2018) Systematic reviews and meta-analyses sit at the top of these hierarchies. Randomized controlled trials are near the top. Non-randomized cohort and case-control studies are lower. Near the bottom are laboratory studies and anecdotal case reports.(Stegenga, 2018)

Two philosophical positions compete over what these hierarchies mean. The “black box thesis” holds that evidence from randomized trials is both necessary and sufficient for causal inference about medical interventions: if a well-conducted trial shows that a drug works, we are justified in believing it works, even without knowing how it works.(Stegenga, 2018) The “mechanista thesis” counters that mechanistic knowledge is additionally required — that knowing the physiological pathway matters for evaluating and applying trial results.(Stegenga, 2018)

The extrapolator’s circle illustrates the difficulty of moving between these positions. Knowing that the physiological mechanisms of an animal model are sufficiently similar to those of a human patient already requires knowing about the causal relation in the target population — the very thing the extrapolation is supposed to establish.(Stegenga, 2018)

Hill’s criteria — strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment, analogy — are not strict criteria but heuristics for causal inference in epidemiology, establishing that multiple lines of evidence can ground causal claims even without RCTs.(Stegenga, 2018)

The Problem of Bias

Biases in medical research systematically tend to generate evidence suggesting that medical interventions are more effective and safer than they actually are.(Stegenga, 2018)

Confirmation bias is compounded by the placebo effect and the natural course of diseases.(Stegenga, 2018) If a person with a fluctuating condition seeks treatment during a more severe period, her symptoms will decrease after intervention regardless of the treatment’s effectiveness — a pattern that makes single-case reports of medical intervention extremely unreliable.(Stegenga, 2018)

P-hacking — analyzing data in multiple unconstrained ways to find statistical signals — is a widespread form of design bias that inflates false positive rates.(Stegenga, 2018) A trial can make many different kinds of measurements, the data can be sorted in many ways, and many statistical tests can be applied to a single set of data; this flexibility increases the chance that researchers will find spurious signals.(Stegenga, 2018)

The pharmaceutical industry has commercialized the RCT in ways that further distort the evidence base. Privatized contract research organizations, publications skewed toward results favorable to drug sponsors, ghost authorship, and efforts by some companies to exclude placebos from their drug trials altogether have eroded trust in the system.(Haller, 2014)

The choice of outcome measure compounds the problem. Reporting results as “relative risk reduction” makes interventions appear far more impressive than reporting the same data as “risk difference.” Some studies suggest that physicians and patients systematically confuse these two measures, leading to gross overestimates of drug benefits.(Stegenga, 2018)

The frequentist inference framework used in most clinical research has its own vulnerabilities. Critics argue that a low p-value (the probability of the data given the null hypothesis) does not by itself justify rejecting the null hypothesis; what is ignored in the standard inference step is the prior probability of the hypothesis, and ignoring it commits the base-rate fallacy.(Stegenga, 2018) The retroactive intercessory prayer trial illustrates why Bayesian priors matter: no matter how low the p-values, if the prior probability that retroactive prayer works is essentially zero, the posterior probability should remain very low.(Stegenga, 2018)

Clinical Judgment and Phronesis

The most sustained philosophical objection to EBM concerns not its methods but its reach. Montgomery argues that we make a great, even dangerous mistake about medicine when we assume it is a science in the realist Newtonian sense — assuming that physicians’ knowledge is invariant, objective, and always replicable.(Montgomery, 2006)

Clinical knowing is first of all the interpretation of what is happening with a particular patient — still called “an opinion,” arrived at through “judgment” — and in this, physicians resemble lawyers and judges more than scientists.(Montgomery, 2006) Physicians share this narrative rationality with lawyers, moral reasoners, and detectives, all of whom must negotiate the fit between organizing principles and specific problematic situations; in each field the rational procedure that determines what a given case is a case of is neither induction nor deduction but C. S. Peirce’s “abduction.”(Montgomery, 2006) Aristotle’s Nicomachean Ethics distinguishes phronesis (practical reasoning) from episteme (scientific knowledge): inquiries into ethics and health are particular, circumstantial, and necessarily uncertain, and absolute answers are unobtainable.(Montgomery, 2006)

EBM’s answers are useless without a clearly asked clinical question, and formulating that question is the province of clinical judgment.(Montgomery, 2006) The understanding of a patient’s condition needed to ask the right question requires clinical experience, careful observation, a nuanced recognition of variation and anomaly, and the ability to synthesize all of this.(Montgomery, 2006) Clinical reasoning, as Montgomery describes it, necessarily moves between two opposing aspects: generalization (lumping) and particularization (splitting), alternating in tension as the reasoner negotiates between general rules and the individual patient in front of her.(Montgomery, 2006) EBM is designed to address the problem of particularization by enabling physicians to assess clinical research in light of their patients’ particular characteristics.(Montgomery, 2006)

Statistical knowledge fails to provide certainty for individual patients. No one survives 82 percent: survivors survive entirely, those who die are completely dead, and statistics say nothing about any one particular person.(Montgomery, 2006) Simple extrapolation from trial populations to clinical populations is the working assumption of EBM, but it is unwarranted because features of real patients — age, gender, severity, comorbidities — systematically differ from trial subjects.(Stegenga, 2018) Cassell, in The Nature of Clinical Medicine (2014), gave this problem a spatial image: on a scatter diagram, the regression line represents the statistical truth, but few if any actual data points lie on it. Each point is an individual. The clinician’s problem is always the point off the line, not the line itself.(Cassell, 2014)

Cassell also traced the institutional mechanism by which judgment was displaced. Post-World War II American medicine embraced positivism — the view that only scientifically verifiable facts should guide diagnosis and treatment. In this ideal, the individual doctor became merely the tool bringing scientific knowledge to bear; since scientific knowledge can be employed by anyone with training, the neophyte physician would be as effective as the experienced attending.(Cassell, 2014) Feinstein’s Clinical Judgment (1967) had attempted to correct this by attending to the qualitative nature of clinical data. But the field that grew from his work replaced “judgment” with “decision-making,” a shift that allowed quantitative rules to substitute for the necessarily qualitative reasoning that most clinical situations require. Feinstein himself was disappointed.(Cassell, 2014) Guidelines and the call for “evidence-based medicine” are, in Cassell’s analysis, the latest instances of a recurring attempt to bypass the necessity for individual physician judgment — an attempt he considers impossible given judgment’s inherently public and social character.(Cassell, 2014)

The error most dreaded in clinical medicine is generalizing from a single case, yet medicine differs from the physical and social sciences in a way that matters: anomalous patients cannot be discarded, because they are still patients in need of care.(Montgomery, 2006) Understanding the particulars, despite the inexact relevance of biological science and statistical epidemiology to the circumstances of one person’s illness, is medicine’s chief moral and intellectual task.(Montgomery, 2006)

Montgomery proposes that clinical medicine operates through a “phronesiology” — a theory of practical knowing expressed through counterweighted maxims rather than universal laws.(Montgomery, 2006) The therapeutic imperative (“always do everything for every patient forever”) is counterweighted by primum non nocere — together capturing the tug-of-war at the heart of every treatment decision.(Montgomery, 2006) The counterweighted tension about anecdotes — “avoid the anecdotal” versus “pay attention to stories” — will not disappear even when reference to Cochrane analyses becomes standard.(Montgomery, 2006)

The Placebo Problem

Henry Beecher’s 1955 article “The Powerful Placebo” announced that in clinical trials the effects of the placebo often exceeded those of a pharmaceutical drug for pain, headache, and nausea, challenging the entire edifice of conventional therapeutics.(Haller, 2014) The psychiatrist Arthur K. Shapiro went further, hypothesizing that of the 4,875 different remedies and 16,842 prescriptions used over the course of recorded history, only a handful were actually effective; most, he argued, were surrogates for the placebo effect.(Haller, 2014)

Placebos are best understood not as inert substances but as interventions that produce effects via expectation mechanisms, distinct from pharmacological mechanisms.(Stegenga, 2018) Recent research suggests that placebos can be effective even when patients know they are receiving a placebo (open-label placebos), which would allow their therapeutic use without deception.(Stegenga, 2018)

The placebo has served as both mediator and judge between conventional and unconventional medicine, raising unanswered questions about whether complementary and alternative medicine effects are reducible to the placebo response.(Haller, 2014) Iain Smith argued in the British Medical Journal that the reductionist approach of the RCT may fail to allow for the holistic effect that is central to the philosophy of most complementary therapies.(Haller, 2014) The doctor-patient relationship itself is a therapeutic factor with its own healing power, operating independently of specific medical or surgical treatments and evident across all cultures and throughout the history of medicine.(Cassell, 1991) Cassell argues that this power grows in importance as medical technology becomes more sophisticated, yet hospital medicine largely neglects or undervalues it.(Cassell, 1991)

Double-blind sham surgery trials in the 1950s for bilateral internal mammary artery ligation showed the surgery had no greater effect on angina than the placebo, which Beecher used to advocate for the RCT in his 1961 article “Surgery as Placebo.”(Haller, 2014)

The Declaration of Helsinki permits placebo use only where no proven intervention exists or for compelling scientific reasons, and requires that placebo recipients not face serious or irreversible harm.(Haller, 2014) Benjamin Freedman’s principle of clinical equipoise holds that the use of a placebo in an RCT is ethical only when the professional community has not yet reached consensus on a standardized treatment.(Haller, 2014)

Medical Nihilism

Since roughly the start of the twenty-first century, methods like randomized trials and meta-analyses show that medical interventions, on average, have very small effect sizes.(Stegenga, 2018) The case for what Stegenga calls “medical nihilism” rests on several converging arguments: these small average effects, the research biases that systematically overestimate them, industry conflicts of interest, and poor theoretical understanding of many treatment mechanisms.(Stegenga, 2018)

The statin case illustrates the problem concretely. Statins prescribed for primary prevention (patients with no history of heart disease) require treating approximately sixty people for five years to prevent a single nonfatal heart attack, and in this population they do not decrease mortality; conversely, statins cause significant harms, including muscle damage and diabetes.(Stegenga, 2018)

Merck’s MK-869 clinical trials revealed that patients receiving an inert placebo did almost as well as those receiving the new antidepressant, leading some psychiatrists to argue that many popular antidepressants are nothing more than expensive placebos.(Haller, 2014)

Before antibiotics and insulin, physicians widely held therapeutic nihilism — the belief that medicine had few genuinely effective treatments. Lewis Thomas recalled that, with a few exceptions regarded as anomalies, his generation was educated to be skeptical about the treatment of disease.(Stegenga, 2018) The antibiotic revolution overturned that skepticism, but Stegenga argues that the underlying concerns about most treatments remain valid.(Stegenga, 2018)

The definition of medical effectiveness itself depends on one’s theory of health and disease: a narrow view restricts effectiveness to disease elimination, while a broader view includes caring interventions that improve quality of life without curing anything.(Stegenga, 2018) How one answers that philosophical question determines what counts as evidence that medicine “works.”

Institutional History

The term ‘evidence-based medicine’ was consolidated and named in 1992 by a group led by Gordon Guyatt at McMaster University in Canada.(Sackett, David L. et al., 2000) Since then, the number of articles about evidence-based practice has grown exponentially from one publication in 1992 to about 1000 in 1998.(Sackett, David L. et al., 2000) Diffusion to the United Kingdom ran through Muir Gray, who introduced William Rosenberg to clinical epidemiology in 1992 and sent him to McMaster; on Rosenberg’s return, he and Oxford colleagues began running EBM training sessions; a year later David Sackett moved to Oxford, and the Centre for Evidence-Based Medicine was established at the John Radcliffe Hospital.(Sackett, David L. et al., 2000) Sackett spent his final five years of clinical practice testing the method on a 200-plus-admissions-per-month inpatient service, arguing that the result demonstrated “we can practice real EBM in real time on a busy in-patient service” rather than only as an academic exercise.(Sackett, David L. et al., 2000)

The Sackett group’s early sabbatical work shifted from clinical epidemiology toward medical informatics on the recognition that “the evidence being generated by medical researchers around the world wasn’t getting to practitioners and patients in a timely and dependable way.” That diagnosis produced the secondary-publication infrastructure — ACP Journal Club, Evidence-Based Medicine, Evidence-Based Nursing, Evidence-Based Mental Health, and Best Evidence — which selected the small fraction of published clinical articles meeting criteria for both validity and immediate clinical use.(Sackett, David L. et al., 2000) Alongside Best Evidence, the Cochrane Library systematically reviewed trials of healthcare interventions, while Best Evidence summarized individual studies and reviews from over a hundred medical journals using explicit criteria for scientific merit and clinical relevance.(Sackett, David L. et al., 2000) Audits of the new approach, conducted on inpatient general medicine, psychiatry, surgery, pediatrics, and general practice, documented that 82% of primary interventions for primary diagnoses were evidence-based — 53% supported by randomized trials or systematic reviews and a further 29% by convincing non-experimental evidence.(Sackett, David L. et al., 2000)

EBM textbooks by Sackett and colleagues were openly polemical about the older infrastructure of clinical knowledge. The 2000 edition of Evidence-Based Medicine opens its search-strategy chapter with Sydney Burwell’s epigraph from Harvard Medical School — “Half of what you are taught as medical students will in 10 years have been shown to be wrong, and the trouble is, none of your teachers knows which half” — and then advises clinicians to “burn your traditional textbooks” because there is no way to tell what is up to date and what is not.(Sackett, David L. et al., 2000)(Sackett, David L. et al., 2000) A dependable modern textbook, on Sackett’s account, must be revised at least yearly, heavily referenced for any claim about diagnosis or management, and select its evidence by explicit principles.(Sackett, David L. et al., 2000) The explicit tagging of recommendations with levels of evidence — the icons that now accompany guideline statements — was first systematized by the Canadian Task Force on the Periodic Health Examination in the 1970s, and has elaborated into successive grading schemes since.(Sackett, David L. et al., 2000)

Sackett’s group also distinguished three modes in which evidence-based practice actually occurs at the bedside. For common conditions, clinicians ideally operate in the “appraising” mode — full searches plus critical appraisal. For less common conditions, they shift to “searching” alone, relying on pre-appraised resources like Best Evidence or the Cochrane Library. For very infrequent problems, they “replicate” — following expert recommendations more or less blindly, the most epistemically risky mode because it cannot distinguish authoritative (evidence-based) advice from authoritarian (opinion-based, “resulting from pride and prejudice”) advice.(Sackett, David L. et al., 2000) Within the textbook itself, Sackett and colleagues acknowledged that the integration of qualitative research — phenomenology, ethnography, grounded theory, and the patient-experience domain on which much nursing scholarship is based — was “one of the major current challenges in EBM” and ceded the field to others, an unusual concession from a movement most known for its hierarchies.(Sackett, David L. et al., 2000)

The textbook’s prefaces preserve several origin stories worth noting. Brian Haynes traces his own EBM impulse to a second-year medical school lecture on Freud, when the psychiatrist, pressed for the evidence behind the theories, conceded that there was none, that he himself did not believe them, and that he had been asked by the head of the department “to give the talk.”(Sackett, David L. et al., 2000) Sackett also lays out four ingredients readers must add to the book for it to benefit patients: mastery of the clinical skills of interviewing, history-taking, and physical examination, without which neither the diagnostic hypothesis-generation that begins EBM nor the integration with patient values that ends it is possible; continuous lifelong self-directed learning; humility, “without which you will become immune both to self-improvement and to advances in medicine”; and “enthusiasm and irreverence” for the work.(Sackett, David L. et al., 2000) Within the search-strategy chapter, Sackett draws a foundational epistemic line between an “evidence resource,” which cites original studies or systematic reviews so claims can be verified, and a mere “information source,” which may or may not provide such grounding; the distinction is meant to discipline clinical reading rather than to add another tool to it.(Sackett, David L. et al., 2000) Sackett also frames the experience of not knowing as “cognitive dissonance” that can either motivate adaptive learning or trigger maladaptive responses (hiding ignorance, anxiety, guilt, shame), with a darker condition still being ignorance one is not aware of, requiring regular review of current best evidence.(Sackett, David L. et al., 2000)

Guidelines occupy an uneasy place in the EBM scheme. Sackett accepts the standard definition of clinical practice guidelines as “systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances,” but insists that any guideline be evaluated as having two distinct components: the evidence summary on the average effect of an intervention, and the detailed instructions for applying that evidence to a particular patient.(Sackett, David L. et al., 2000) The reviews underlying these two components differ in kind. Free-standing systematic reviews tend to be evidence-driven, appearing only when high-quality evidence exists, while reviews supplying the evidence component of a guideline are necessity-driven, synthesizing the best evidence available (even if it is of shaky quality) to guide a decision that has to be made tonight. That asymmetry is one reason explicit levels-of-evidence tagging matters in guidelines but not always in standalone reviews.(Sackett, David L. et al., 2000)

Acupuncture and Chinese Herbal Medicine in RCT Research

Chinese medicine has accumulated a substantial body of clinical research since the early 1970s, providing a case study in the difficulties of applying RCT methodology to a complex traditional system. Since the early 1970s, more than 500 randomized controlled trials of acupuncture have been conducted in North America, Europe, Australia, and New Zealand, with most focusing on pain conditions; other well-studied areas include emesis, ophthalmology, otolaryngology, substance abuse, cerebrovascular illness, neurological problems, gynecological complaints, asthma, and gastrointestinal problems.[kap00-app-e-003] The 1997 NIH Consensus Development Panel on Acupuncture concluded from this evidence base that needle acupuncture is efficacious for adult post-operative and chemotherapy nausea and vomiting, probably efficacious for nausea of pregnancy, and efficacious for postoperative dental pain.[kap00-app-e-002]

The primary methodological problem is the sham control. Needling at non-acupuncture sites may have physiological effects through counter-irritation and neurotransmitter release, making it not a true inert placebo; random needling may have its own neurological and endocrine effects beyond ordinary nonspecific placebo outcomes, which could modulate pain and produce effects beyond what would be expected from an inert sham.[kap00-app-e-004] One explanation for positive results in emesis trials — using a single acupuncture point (Pericardium 6, Nei Guan) — is that RCT methodology most easily detects efficacy when an acupuncture intervention is well-defined and discrete, creating a simple “drug-like” replicable model.[kap00-app-e-005] Despite this relatively favorable domain, Western acupuncture researchers as a group have not had the resources and skills to conduct trials considered methodologically conclusive by mainstream standards, and Kaptchuk concludes that the research remains methodologically weak.[kap00-app-e-006]

Basic science research supports plausible mechanisms. Acupuncture analgesia may be initiated by stimulation of high-threshold small-diameter nerves in the muscles that send messages to the spinal cord and activate three centers — the spinal cord, brainstem (periaqueductal gray), and hypothalamic neurons — to stimulate endogenous opioid mechanisms including endorphins, enkephalins, and ACTH. Evidence also supports neuropeptide gene expression as one mechanism, and fMRI has begun to show point-specific, quantifiable effects on relevant brain structures.[kap00-app-e-007] Six major systematic reviews of acupuncture adverse events agree that complications are rare given the frequency of application, and that acupuncture can generally be considered safe; most adverse events result from inadequately trained practitioners.[kap00-app-e-008]

Chinese herbal medicine has also been tested in RCTs. The first important trial addressed recalcitrant adult atopic dermatitis in London, using a crossover design in which 31 patients alternated between standardized herbal formula and dummy treatment; the trial showed highly significant improvement in erythema, surface damage, itching, and sleep during the herbal treatment phase, and was preceded by an equally impressive RCT in 47 children with nonexudative atopic eczema.[kap00-app-e-009] A Sydney trial of 116 IBS patients assigned to placebo, a standardized Chinese herbal formula, or an individualized prescription adjusted by a traditional Chinese herbalist found that both active groups outperformed placebo during the 16-week treatment phase; at 14-week follow-up, only those treated with individualized herbs maintained improvement.[kap00-app-e-010]

Kaptchuk argues that Western quantitative assessment has contributed to East Asian medicine’s acceptance and self-awareness in the West, and that cooperation between the two traditions should expand while rhetorical debates over market shares and medical resources should be avoided.[kap00-app-e-016] By the second edition of The Web That Has No Weaver (2000), Kaptchuk held a full-time appointment at Harvard Medical School and served on the NCCAM National Advisory Council at the NIH, positioning him as a rare figure contributing to both traditional medicine scholarship and the mainstream scientific evaluation of that medicine.(Kaptchuk, Ted J., 2000)

EBM and Alternative Medicine

When alternative medicine practices like homeopathy are tested by randomized trials, they generally perform no better than placebo.(Stegenga, 2018) Defenders respond that the setting of randomized trials is excessively artificial and cannot incorporate the full range of activities and expertise that constitutes alternative medicine. The basic thrust of this defense is to appeal to the holistic nature of these interventions — an appeal that, whatever its merits, exposes a genuine limitation of the RCT as a universal evaluative instrument.(Stegenga, 2018)

Pierre Charles Alexandre Louis’s la methode numerique in the early nineteenth century was a distant precursor of EBM: his emphasis on observation, classification, and numerical analysis helped dismantle the rationalistic beliefs behind bleeding and the heroic uses of calomel and tartar emetic, leading to a more expectant regimen of healing.(Haller, 2014) Still earlier, the 1789 French royal commission on Mesmer’s animal magnetism — which included Benjamin Franklin — concluded that the presumed cures could be better explained as products of imagination, representing an early use of masked assessment to test medical claims.(Haller, 2014)

Cartesian dualism gave birth to two polar traditions in Western medicine: one grounded in reductionist biochemical processes, the other explaining healing in terms of the soul’s effect on the body.(Haller, 2014) [GAP: Claim that the tension between these traditions has never been resolved is unsupported by cited cards.] The doctor-patient relationship affects the patient’s physiological processes through mechanisms not yet understood, beyond the placebo effect.(Cassell, 1991) [GAP: Claim that EBM measures the reductionist strand with increasing precision, but the other strand remains largely outside its reach is unsupported by cited cards.]

See Also

• clinical-judgment
• empiricism-in-medicine
• placebo-effect
• medical-nihilism
• randomized-controlled-trial
• therapeutic-nihilism

Sources

Editorial Notes

Gaps the encyclopaedia compiler flagged for future evidence work, collected from inline markers in the body and frontmatter.

Streptomycin trial (1948) and Cochrane Collaboration history