Evidence-Based Medicine (EBM)

Evidence-based medicine (EBM) represents a relatively new phenomenon that has influenced the way in which health care professionals practice and the way a new generation of students learns medicine.   The term itself seems somewhat intuitive—what else would medicine be based on if not evidence? Until relatively recently, however, the basis for medical care rested primarily on three pillars: the biological sciences, which led to theories about how to diagnose and treat diseases; anecdotes such as clinical case reports of successful treatments; and tradition. The advent of evidence-based medicine brought with it the recognition that all aspects of medical practice, whether based on the biological sciences, tradition, or even common sense require scrutiny. The idea with EBM is that unless a given treatment poses no potential harm to patients and negligible costs to society, it is imperative to confirm that any given treatment’s effectiveness outweighs any potential harms and costs. 

This science for evaluating the effectiveness of clinical practice, sometimes called clinical epidemiology,¹ consists of the tools for designing and interpreting clinical trials, cohort studies, and other techniques for studying disease diagnosis, prognosis, and treatment. Thus, while the biologic sciences such as physiology, genetics, and molecular biology provide the basis for developing new medical technologies, EBM provides the tools for evaluating the extent to which these new technologies actually benefit patients.

These tools have been fruitfully applied to all aspects of clinical practice, ranging from the manoeuvres that make up the physical examination ² (e.g., To what extent does listening to the heart work in detecting important cardiac conditions? ^3-6 Can examination of the abdomen adequately detect major conditions like inflammation of the gall bladder or appendicitis?^{7, 8}) to the evaluation of drugs,^9-11 surgeries,¹² medical devices,¹³ and even complementary and alternative medicine.^14-16 In all of these cases, EBM has as its goal the evaluation of care independent of the reasons for thinking that these treatments or tests are a good idea. In other words, the scientific framework for evaluating these medical technologies stands outside the theories that give rise to them. One need not, for instance, ascribe to the philosophy that underlies acupuncture in order to determine whether or not it confers benefit for specific conditions.^{15, 16}       

The term “evidence–based medicine” (EBM) is usually attributed to a commentary written in 1991 by Gordon Guyatt, an internist and clinical epidemiologist at McMaster University in Canada,^{17, 18} though the term “evidence-based” had been introduced in 1990 by David Eddy, an American surgeon and healthcare economist.¹⁹   A widely cited definition describes EBM as “the conscientious, explicit and judicious use of current best evidence in making decisions about the care of individual patients.”²⁰ It contributes to the field of medicine through scientific studies that evaluate tests and treatments, but also equips health professionals with tools to interpret such studies in order to inform decisions in their practices.  Finally, though, the emphasis in discussions of EBM inevitably focuses on existing evidence and its appraisal, advocates of EBM have clearly articulated the importance of integrating external evidence, individual professional expertise, and patient preferences (see discussion below in “Critiques of EBM”) .

The basis for medical decision making (or, rather, the surprising lack of a scientific base for many medical decisions) provides a useful lens through which to view the rise of EBM and appreciate its importance to clinical practice.²¹ Until about 40 years ago, medical decisions were seldom questioned. The combination of intensive training, individual experience, and sound professional judgment was thought to produce decisions that were generally correct and therefore rarely scrutinized. Several developments changed this uncritical attitude towards the decisions of medical professionals.

Beginning in the early 1970s, John Wennberg and colleagues at Dartmouth University produced multiple studies that demonstrated dramatic regional variations in how often physicians recommended tonsillectomy, hysterectomy, admitted patients to the hospital, or prescribed antibiotics for the same condition.^{22, 23} Given these dramatic variations in the rates at which physicians recommended diagnostic tests and various types of therapies, the obvious question became: what was the right rate? Was there a basis for judging some tonsillectomies or hysterectomies as medically appropriate and some as inappropriate?  Investigators at RAND in Santa Monica, California developed a methodology for combining expert opinion with the best available evidence to characterize specific clinical scenarios in terms of the appropriateness of care. ^24-28 In addition to identifying large numbers of inappropriate (and therefore potentially unnecessary) medical treatments, these studies underscored the degree to which much medical care was based on expert opinion (often divergent) rather than scientific evidence.

As these two areas of research underscored the gaping need for an ‘evidence base’ to much of medical practice, the tools to provide such evidence were being developed by other investigators.  These tools included the randomized controlled trial, meta-analyses of multiple clinical trials addressing the same clinical question, the application of Bayesian reasoning (see below) to the interpretation of diagnostic tests, and more sophisticated mathematical techniques for analyzing large clinical databases.  Finally, economic factors played a role in encouraging the dissemination of EBM. As healthcare costs sky rocketed, demonstrating that some aspects of medical care had no scientific basis appealed to insurers, governments, and other healthcare payers.  

More than a particular theory, EBM represents a paradigm for clinical medicine. As with any paradigm shift, the EBM movement involved the work of many individuals, but some of them were particularly influential.

Archie Cochrane ²⁹ was a Scottish physician who championed the randomized controlled trial as the ‘gold standard’ for evidence in medicine and encouraged the use of evidence summaries called systematic reviews (see below).  Alvin Feinstein was a physician at Yale who pioneered the development of objective methods for measuring and assessing clinical data ^{30, 31}and trained a generation of clinical researchers. Finally, McMaster University, seen by many as the historic epicenter of EBM, fostered the work of physicians such as David Sackett, Gordon Guyatt, and Brian Haynes who disseminated critical appraisal skills for doctors attempting to apply the results of clinical research to their practices^32-38 and developed many of the methodologies of EBM.

Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology : a basic science for clinical medicine. Boston: Little, Brown; 1985. This book on ‘clinical epidemiology,’ written by four physicians from McMaster University in Canada, represents one of the founding texts of evidence-based medicine. 

Emphasis on clinical endpoints over “surrogate outcomes”

Many treatments in the past were based solely on studies that showed a benefit on a “surrogate outcome” (e.g., blood  pressure or cholesterol levels) without any evidence that the treatment in question conferred the true benefit of interest (e.g., a reduction in heart attacks and strokes or, even more fundamentally, an increase in life expectancy).  Two famous examples of discrepancies between the impacts of treatments on surrogate outcomes versus clinical endpoints include:

o       a class of cardiac drugs for preventing arrhythmias did in fact suppress rhythm problems, but turned out to increase the risk of dying ³⁹

o       hormone replacement therapy for post-menopausal women improved various surrogate outcomes, such as cholesterol levels, but conferred no benefit on clinical outcomes and, in fact, caused an excess of cardiovascular complications, such blood clots⁹

Patient Centered Outcomes

Although proponents of EBM advocated for an emphasis on clinical endpoints, rather than surrogate outcomes, they also recognized that many so-called clinical endpoints did not equate with outcomes that patients would regard as important. For instance, heart attacks are defined by the results of blood tests, electrocardiography, and other more sophisticated investigations, without any reference to symptoms or the impacts on patients’ lives. Some non-fatal heart attacks can be quite mild, while others leave patients practically incapacitated. A more patient centered outcome, therefore, would capture patients’ functional capacity or quality of life in general. A number of advocates of EBM have championed the development and validation of instruments for measuring the impacts of health conditions on patients’ quality of life.^40-42     

“Bayesian thinking” for the interpretation of diagnostic tests

Theoretically, if one knows how well a given diagnostic test performs, one can calculate the probability that a test will detect the condition of interest in people who have it and the probability it will incorrectly identify it in those who do not (“false positives”). In practice, however, we do not know who has the condition; we only know who has a positive test result. Thomas Bayes, an 18^th century British mathematician and Presbyterian minister, developed a theorem for calculating the “inverse probabilities” that allow one to work back from a test result to the probability of having a disease.  Bayes’ theorem allows one to calculate how likely it is that a given person has the condition of interest based on knowing the test’s performance characteristics and one other key piece of information: how common the condition is in the population being tested.

Though Bayes’ theorem had existed for some 200 years, physicians did not apply it to the interpretation of diagnostic tests until the last quarter of the twentieth century. Proponents of EBM pioneered the interpretation of diagnostic tests within a formal framework for combining the sensitivity and specificity of a test with the base (or “pre-test”) probability that a given patient has the condition of interest.

A classic application of Bayesian thinking to diagnostic testing involves the evaluation of patients who may have developed a blood clot in their lungs, known as a pulmonary embolism, which typically arises from blood clots in the veins of the legs.  Patients with pulmonary embolism can present with a variety of symptoms including sharp chest pain that worsens with coughing or taking a deep breath, shortness of breath, and a fast heart rate.  Because pulmonary embolism can progress rapidly to serious and potentially fatal compromise of the lungs and circulatory system, diagnosis must proceed expeditiously and accurately. Patients who have pulmonary embolism must be identified promptly so they can receive appropriate treatment (with anticoagulants or “blood thinners”) while patients who do not have this condition must also be identified quickly so that diagnostic evaluation can focus on other possible causes of shortness of breath or chest pain.

Until relatively recently, the main diagnostic test used to diagnose pulmonary embolism was something called a ventilation-perfusion scan. This test uses a radioactive tracer to show the degree to which blood flow and aeration of the lungs are mismatched. An area with a blood clot will have normal ventilation but decreased blood flow (‘perfusion’). Unfortunately, the pictures produced by this test, patterns of radioactive tracer that show up as clusters of dark spots of varying density, do not yield clear-cut results. The test is therefore interpreted in terms of probabilities of having a blood clot, with certain patterns reported as “high probability,” others as “low probability,” and still others as “indeterminate.”   Given the probabilistic nature of these test results, recommendations for the interpretation of ventilation-perfusion scans were explicitly Bayesian. Clinicians were instructed to formulate a “pre-test probability” that captured their suspicion of pulmonary embolism in a given patient and then interpret the result of the test in light of this clinical suspicion. A high “pre-test” probability followed by a “high probability” test result indicated a sufficiently high likelihood of pulmonary embolism that treatment could proceed without further testing. Conversely, a low pre-test probability followed by a low probability ventilation-perfusion scan safely ruled out pulmonary embolism. However, other combinations of clinical suspicion and test results required further investigation.

Importantly, the Bayesian perspective applies even to tests with more apparently definitive results.  Computed tomography (CT) scans of the lung have become sufficiently advanced that they can visualize the interruption of blood flow caused by blood clots. Even with this apparently “black or white” answer (blood clot seen or not seen on CT), studies have shown that the likelihood of pulmonary embolism still depends on the “pre-test probability.” For instance, in a landmark study, 40% of CT scans that indicated the presence of a blood clot were “falsely positive” if the clinical suspicion before the test was low.⁴³ Equally important, 40% of CT scans that detected no blood clot were “falsely negative’ if the clinical suspicion prior to the test was high. 

These results and the Bayesian reasoning that underlies them have tremendous importance in practice. For a patient who presented with symptoms clearly suggestive of a blood clot but then had a “negative” CT scan, clinicians can have a firm basis for continuing to pursue this diagnosis with further investigation rather than prematurely casting it aside. Conversely, for a patient who seemed unlikely to have a blood clot but for whom a CT was ordered “just to be safe,” the unexpected “positive” test result can be interpreted with appropriate suspicion and pursued further. Confirmatory tests can be obtained before exposing the patient to the potential dangers of long-term treatment with blood thinners. 

This concept of combining the clinical suspicion of a condition with the result of a diagnostic test underlies all contemporary medical diagnosis. If a physician obtains a test in a patient very unlikely to have a given condition (e.g., an HIV test in a patient with no risk factors for contracting HIV), then a positive result is more likely to represent a false positive than a true indication of disease. Conversely, if a patient seems very likely to have a given condition, then an initial negative test result may represent a false negative and requires further investigation with more definitive tests.   

Guidelines for Care

In addition to evaluating tests and treatments, EBM has also been touted as a way to identify standards of care, from screening for target diseases (e.g., as part of the “annual check-up”) to detailed recommendations about screening for the treatment of various patient populations and their varied health problems. The US Preventive Services Task Force (http://www.ahrq.gov/clinic/uspstfix.htm) has for years produced evidence-based recommendations on the age at which people should have their blood pressure checked routinely, when women should start undergoing mammography to screen for breast cancer, whether asymptomatic smokers should have annual chest x-rays to try to catch lung cancer, and a variety of other specific preventive measures.

At first it might appear that more screening is always a good thing: why wouldn’t one want to check everyone’s blood pressure, screen all women for breast cancer, all men for prostate cancer, and so on? Importantly, the issue is not simply one of costs. Rather, the issue involves the balance of true positive test results to false positives. As discussed in connection with the Bayesian interpretation of diagnostic tests, the meaning of a given test result depends on the base probability that a given person could have had the condition of interest. For women under 40 years of age, in the absence of a family history of breast cancer, the chance of having breast cancer is so low that any positive result on a screening test (such as mammography) far more likely represents a false positive result than a true positive one. These false positive results lead to further invasive testing and considerable anxiety as women wait for the test results to see if they in fact have cancer. While individuals may differ on the precise balance of false positives to true positives (e.g., while everyone agrees women over aged 50 should undergo yearly mammograms, controversy surrounds the use of routine mammography in women aged 40-49), the contribution of EBM lies in the view that informed decision-making both by medical professionals and by patients requires data about the rates of true and false positive diagnoses.

Recommendations regarding screening for prostate cancer have also attracted attention and controversy with regard to EBM. For instance, the US Preventive Services Task Force found insufficient evidence to recommend for or against routine screening for prostate cancer, despite recommendations from other organizations (e.g., the American Cancer Society) that physicians offer prostate cancer screening beginning at age 50 (age 45 for African Americans and any men who have a father, brother, or son who was diagnosed with prostate cancer before age 65). The reason for this difference in recommendations relates to the uncertain benefit of screening. No one questions that screening tests such as the PSA (for ‘prostate specific antigen’) can detect early-stage prostate cancer. The question comes with the benefit of such early detection. With some cancers, early detection does not translate to longer survival, just longer periods of time knowing about the cancer. Screening also brings frequent false-positive results, which in turn lead to unnecessary anxiety, biopsies, and potential complications of treatment of some cancers that may never have affected a patient's health.

Importantly, EBM is not associated with a specific recommendation about screening for prostate cancer (or breast cancer or any other such recommendation). Rather, EBM provides the framework in which such recommendations can be derived. In other words, EBM lays out the evidence that a given screening strategy confers some benefit (such as years of life saved), the human and financial costs incurred to obtain this benefit, and an explicit statement of the uncertainty surrounding these estimates. Crucial to this product is an explicit framework or grading system for rating the strength of evidence supporting candidate screening or treatment strategies. Many variations of such frameworks exist, but representative examples include the system used by the US Preventive Services Task Force (available at http://www.ahrq.gov/clinic/3rduspstf/ratings.htm ) and one developed by the Centre for Evidence-based Medicine at Oxford University (available at: http://www.cebm.net/index.aspx?o=1025).

Critical Reading Skills for Physicians

While EBM has contributed greatly to the scientific conduct of clinical research, it has also equipped physicians with the tools to appraise clinical research in order to make appropriate decisions in practice.  For the majority of physicians, reading clinical research presents a daunting challenge, one for which medical school, which emphasizes what is already known about the diagnosis and treatment of disease, does little to prepare them. Thus, many clinicians found great value in a series of articles published in the Journal of the American Medical Association, called the Users Guide to the Medical Literature, written by many of the leaders in EBM.¹⁸ This series of articles (now available in book form³⁸)  takes readers through the approach to appraising the validity and applicability of published studies about new drug treatments, diagnostic tests, prognostic tools, and a variety of other important topics.^{37, 44, 45}  

Cookbook Medicine

One common criticism of EBM depicts its practitioners as relying solely on preset diagnostic and treatment algorithms and other recipe-like protocols to guide patient management.⁴⁵ However, leaders in the field of EBM have emphasized the need for combining the best available external evidence with the clinical judgement and experience of individual physicians: “Good doctors use both individual clinical expertise and the best available external evidence, and neither alone is enough. Without clinical expertise, practice risks becoming tyrannised by evidence, for even excellent external evidence may be inapplicable to or inappropriate for an individual patient. Without current best evidence, practice risks becoming rapidly out of date, to the detriment of patients.” ²⁰ 

To take a concrete example, if a middle aged man sees his physician over a several month period for lower back pain, then headaches, and finally difficulty sleeping, an astute physician will consider the diagnosis of depression as underlying these various somatic complaints and tactfully ask questions relevant to confirming this diagnosis. Here, the art of medicine plays out in the recognition that a single, subtle diagnosis unifies a series of apparently mundane and unrelated problems. Once the diagnosis is made, however, the patient would clearly benefit from the physician consulting a guideline that summarizes the latest recommendations on which patients benefit from what type of therapy (e.g., antidepressants, psychological counseling, or possibly both), and, if medications are chosen, which of the many available antidepressants represent “first-line” therapies.

Even with such a guideline in hand, however, the EBM perspective acknowledges that, while evidence can inform physicians, it can never replace clinical expertise. ²⁰  Each patient has individual values and preferences, not to mention other medical conditions that may affect therapeutic decisions, and that physicians must help patients integrate these various factors with the recommendations contained in guidelines. Without incorporating current best evidence, physicians’ practices become out of date and deprive patients of important benefits. But, without individual clinical expertise, the best evidence can be applied inappropriately. 

Merely a Cost-Cutting Method

Other critics of EBM have argued that the goal of standardizing approaches to diagnosis and treatment merely reflects a desire to cut costs. Under this view EBM represents a tool for administrators to save money at the expense of patient care.  Some defenders of EBM might grant this perspective, perhaps quibbling only with the “merely” a method to cut costs. All countries, whether they have single payer, government subsidized, or private insurer-based healthcare must face the reality of limited resources. Millions of dollars wasted on unnecessary tests or procedures due to a lack of standardization represent millions of dollars unavailable to provide proven tests, procedures, and medications to patients in need of them. Moreover, the financial consequences of applying EBM do not always consist of “doing less.”²⁰ Sometimes considering the most up-to-date and rigorous evidence validates an expansion of indications for new or expensive treatments, so that overall health care costs rise in order to benefit patient care. In fact, although “cost-effectiveness analyses” have been advocated as a tool to help contain health care costs, such analyses have more commonly resulted in supporting new technologies that require additional expenditures rather than a low-cost alternative.⁴⁶ The crucial point is that EBM does not dictate how much to spend on a given treatment. Rather, EBM provides a framework within which we can decide if the gains in health conferred by a given treatment are sufficient to justify the expenditure required to achieve it. It is also worth noting that many of the major contributions of EBM have had nothing to do with costs, but rather with assessing intrinsic benefits and harms: most treatments are cast aside because they are ineffective or have too many side effects, not because they cost too much.

The “Tyranny” of Evidence-Based Medicine

EBM has recently come under fire recently for exerting a totalitarian like dominance of medicine.^{47, 48} Not only does this critique misrepresent the reality of contemporary medicine—faculties of medicine at major universities remain overwhelmingly dominated by physician scientists, not clinical epidemiologists or other champions of EBM—it misses the spirit of EBM entirely, which is very much that of the minority voice. Proponents of EBM have historically argued for the critical appraisal of aspects of care that have been handed down by tradition, championed by scientists (who typically have much larger research budgets than clinical epidemiologists), or been bank-rolled by large pharmaceutical companies. If anything the proponents of EBM are typically underdogs, not hegemonic rulers of departments and institutions.

We have seen how evidence based medicine contributes to the clinical science of medicine itself while also providing tools to individual physicians to make decisions in daily practice. Proponents of EBM have also directed substantial efforts to improve methods for optimally delivering evidence to physicians in a manner that permits easy and timely retrieval. Dr. Brian Haynes, a physician and researcher at McMaster University, has articulated this broader vision of EBM and the continuum of resources that it encompasses in terms of a “5-S Model.” ^48,49

The first “S” in the 5-tiered pyramid refers to “studies”—individual research articles about diagnostic tests and treatments. In the early days of EBM, proponents focused on disseminating the critical skills physicians needed to appraise individual studies to determine whether their conclusions warranted adoption in daily practice. While these skills still serve a purpose (as obviously do the studies themselves), the subsequent phase of EBM emphasized the importance of the next level in the 5-S hierarchy, namely “syntheses.”

Systematic reviews, such as those produced by the Cochrane Collaboration (named after Archie Cochrane; link to http://www.cochrane.org/ ), clinical practice guidelines, and other syntheses of the totality of available evidence on a given topic, offer greater value than individual studies. The authors of such syntheses have followed explicit methodologies to identify all relevant evidence, assess its quality, and arrive at a judgment about the overall benefit or harm of a given treatment, as well as an explicit statement of the degree of certainty about this judgment. Of note, many journals and online resources provide summaries of these syntheses for patients or other non-professional users, as many of the topics addressed by these reviews are of direct interest to patients. For instance, the Cochrane Library includes reviews of the evidence on the effectiveness of cranberries in preventing urinary tract infections ⁵⁰ and interventions for preventing the pain and discomfort of screening mammography,⁵¹ among many other topics.   

Valuable as these evidence syntheses are, physicians in busy practices often do not have time to digest them. These syntheses are themselves pieces of research, often involving statistical pooling of data from many clinical trials, so absorbing the information they present can take time. Moreover, new research can emerge quickly and overturn even quite recent syntheses.⁵² Consequently, the amount of reading required for clinicians to keep up-to-date can be overwhelming, even if they focus just on syntheses. Consequently, the next level, “synopses”, has emerged. Synopses refer to brief (e.g., 250-500 word) structured abstracts that alert physicians or other health professionals to important new evidence in readily digestible formats. Examples include the American College of Physician’s ACP Journal Club (http://www.acpjc.org ), Evidence-Based Medicine (http://ebm.bmj.com ), Evidence-Based Nursing (http://ebn.bmj.com), and InfoPOEMS (for Patient-Oriented Evidence that Matters,  http://infopoems.com ).    

Synopses have value precisely because they present the result of a primary study or synthesis that addressed a very focused clinical question, such as how well does a particular text diagnose X or how effective is such-and-such a medication for the treatment of X. In practice, however, clinicians often do not have a single focused question. Rather, they have a patient who they suspect might have X (say, depression) and they want to consult a concise resource that summarizes up-to-date approaches to the diagnosis and treatment of X. Here, the fourth level of the 5-S hierarchy, “summaries”, shows its value.  UptoDate, an online textbook published by the Massachusetts Medical Society and updated on a quarterly basis, and Clinical Evidence, published by the British Medical Journal, present concise but thorough and evidence-based overviews of all aspects of patient care for a broad range of conditions. It is increasingly common for practicing clinicians and physicians in training to consult such resources “on the fly” in the course of seeing patients.

The final and most advanced level of the 5-S hierarchy refers to “systems” for delivering evidence at the time it is needed, regardless of whether or not the recipient recognizes the need.  For example, if a 57 year old man comes for a yearly check-up, his physician might review the studies looking at the most appropriate age to start recommending screening colonoscopy to look for polyps or other early signs of colon cancer. Or, if the physician didn’t think to look up that specific topic, she might consult a synopsis on general preventive measures appropriate to men of this age and come across the same information. But, if the physician did not think to consult either type of resource, or, as commonly happens, if the patient were coming for another reason, say back pain, so that the physician had no reason to think of general preventive care, then a valuable opportunity to discuss screening for colon cancer would be missed. The “systems” in the 5-S model deliver such recommendations to physicians during their interaction with the patient (or, more accurately, during the physician’s interaction with the patient’s medical record.) For instance, in an office with an electronic medical record, regardless of why the patient came to the office, the physician could receive a screen prompt to consider referring the patient for colonoscopy based on the patient’s age and other elements of his medical profile. 

It is easy to imagine a future in which computers seamlessly offer diagnostic and treatment suggestions to physicians. In reality, however, such suggestions can quickly become a nuisance, either because they interrupt too many tasks or because they frequently give inappropriate advice. However, when substantial evidence exists to support a large benefit from a particular diagnostic test or therapy, then the natural evolution of the EBM perspective is to develop a system for delivering that evidence into the hands of physicians at the time they need it, rather than relying on their remembering to seek it out. 

EBM resources of interest to clinicians include frameworks for rating evidence according to strength or quality, calculators for determining the probability of a given diagnosis based on the “pre-test probability” and the accuracy of the test, templates for critically appraising research studies, and other aides for quantifying or interpreting statements of effectiveness. The Oxford Centre for Evidence-Based Medicine makes many such resources publicly available on its website (http://www.cebm.net ). The Dartmouth Biomedical Library provides a list of various web-based EBM resources at http://www.dartmouth.edu/~biomed/resources.htmld/guides/ebm_az_list.shtml 

Despite the overwhelming amount of scientific studies (generating the need for syntheses, synopses, summaries, and systems), there remains an important role for searching the biomedical literature for individual studies or systematic reviews and other syntheses of multiple studies addressing the same topic. Physicians have long had resources to facilitate the retrieval of published studies from the vast number of biomedical journals. However, the internet in general, and U.S. National Library of Medicine in particular, have revolutionized this process. For over a decade now, the National Library of Medicine has made its MEDLINE database freely available online to the public in the form of PubMed (http://www.ncbi.nlm.nih.gov/sites/entrez?db=pubmed ).

The main PubMed search page, the freely available search engine for the biomedical literature maintained by the US National Library of Medicine. Click here to link to the search page. 

The MEDLINE database includes over 17 million citations from biomedical and other life science journals from the 1950s to the present. PubMed has the capacity to support sophisticated searches by trained users, but also allows surprisingly effective free text searches by users with no knowledge of the technical taxonomy of subject headings or the logical connectors embedded in complex searches. Moreover, PubMed includes various “search filters” (developed by Dr. Brian Haynes and colleagues at McMaster University) that allow users to hone their free text searches. Clicking on the Clinical Queries heading in the left-hand column of the main PubMed search page (see figure) takes the user to a page (figure below) that allows searches that focus on diagnosis, treatment, prognosis, or etiology (the cause of a condition). For instance, simply typing “cell phones cancer” and selecting etiology retrieves articles on the possible link between cell phone use and various forms of cancer. To link to these built-in searches, click here.

Below the search filters for etiology, diagnosis, therapy, and others, is a search filter for “systematic reviews.” Here users can retrieve any syntheses of studies that address a topic suggested by free text terms. For instance, simply typing cranberries into the search window for systematic reviews retrieves the previously mentioned Cochrane review of the effectiveness of cranberries in preventing urinary tract infections.⁵⁰ (To link to the systematic review search window in PubMed, click here.)

The National Library of Medicine also includes resources for patients (e.g., http://medlineplus.gov ). However, there are more general search engines for health professionals and non-professionals that sort search results based on the type of resource. Two examples include the TRIP database (“Translating Research Into Practice” at www.tripdatabase.com) and a meta-search engine developed at the University of Texas Health Science Center called SUMsearch (http://sumsearch.uthscsa.edu/). For instance, typing “cranberries urinary tract” into the search window of the TRIP database provides links to 3 systematic reviews, 9 evidence-based synopses, 6 clinical practice guidelines, and 17 electronic textbooks, as well as numerous patient information leaflets and articles addressing the role of cranberries in preventing urinary tract infections. 

In practice, evidence based medicine (EBM) is a phenomenon involving a partnership between individual clinical expertise and the use of the latest clinical studies to improve patient care.  EBM equips researchers with the tools to rigorously evaluate the effectiveness of all aspects of medical care and provides clinicians with the tools to critically appraise such evaluations and integrate this information as appropriate as they care for patients. From the latest cancer treatments, to the basic elements of the time honoured physical examination, EBM provides a framework for assessing the benefits and harms of everything medical professionals do for patients. Far from taking the art out of medicine, EBM highlights the interface of the art and science of taking care of patients: it identifies what we know and what remains unknown, and encourages adapting evidence from the latest available studies to patients’ individual characteristics and preferences.

1.         Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology : a basic science for clinical medicine. Boston: Little, Brown; 1985.

2.         Simel D, Rennie D. Rational Clinical Examination : Evidence-Based Guidance: McGraw-Hill Professional; 2008.

3.         Choudhry NK, Etchells EE. The rational clinical examination. Does this patient have aortic regurgitation? JAMA 1999;281:2231-2238.

4.         Etchells E, Bell C, Robb K. Does this patient have an abnormal systolic murmur? JAMA 1997;277:564-571.

5.         Marcus GM, Vessey J, Jordan MV, et al. Relationship between accurate auscultation of a clinically useful third heart sound and level of experience. Arch Intern Med 2006;166:617-622.

6.         Vukanovic-Criley JM, Criley S, Warde CM, et al. Competency in cardiac examination skills in medical students, trainees, physicians, and faculty: a multicenter study. Arch Intern Med 2006;166:610-616.

7.         Wagner JM, McKinney WP, Carpenter JL. Does this patient have appendicitis? JAMA 1996;276:1589-1594.

8.         Trowbridge RL, Rutkowski NK, Shojania KG. Does this patient have acute cholecystitis? JAMA 2003;289:80-86.

9.         Hulley S, Grady D, Bush T, et al. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. Heart and Estrogen/progestin Replacement Study (HERS) Research Group. JAMA 1998;280:605-613.

10.       Guay LA, Musoke P, Fleming T, et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 1999;354:795-802.

11.       Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994;344:1383-1389.

12.       Khan NE, De Souza A, Mister R, et al. A randomized comparison of off-pump and on-pump multivessel coronary-artery bypass surgery. N Engl J Med 2004;350:21-28.

13.       Kadish A, Dyer A, Daubert JP, et al. Prophylactic defibrillator implantation in patients with nonischemic dilated cardiomyopathy. N Engl J Med 2004;350:2151-2158.

14.       Bent S, Kane C, Shinohara K, et al. Saw palmetto for benign prostatic hyperplasia. N Engl J Med 2006;354:557-566.

15.       Linde K, Streng A, Jurgens S, et al. Acupuncture for patients with migraine: a randomized controlled trial. JAMA 2005;293:2118-2125.

16.       Witt C, Brinkhaus B, Jena S, et al. Acupuncture in patients with osteoarthritis of the knee: a randomised trial. Lancet 2005;366:136-143.

17.       Guyatt G. Evidence-based medicine. ACP J Club (Ann Intern Med) 1991;14 A-16

18.       Guyatt GH, Rennie D. Users' guides to the medical literature. JAMA 1993;270:2096-2097.

19.       Eddy DM. Practice policies: where do they come from? JAMA 1990;263:1265, 1269, 1272 passim.

20.       Sackett DL, Rosenberg WM, Gray JA, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn't. BMJ 1996;312:71-72.

21.       Eddy DM. Evidence-based medicine: a unified approach. Health Aff (Millwood) 2005;24:9-17.

22.       Wennberg J, Gittelsohn. Small area variations in health care delivery. Science 1973;182:1102-1108.

23.       Wennberg J, Gittelsohn A. Variations in medical care among small areas. Sci Am 1982;246:120-134.

24.       Winslow CM, Kosecoff JB, Chassin M, Kanouse DE, Brook RH. The appropriateness of performing coronary artery bypass surgery. JAMA 1988;260:505-509.

25.       McGlynn EA, Naylor CD, Anderson GM, et al. Comparison of the appropriateness of coronary angiography and coronary artery bypass graft surgery between Canada and New York State. JAMA 1994;272:934-940.

26.       Bernstein SJ, Hilborne LH, Leape LL, et al. The appropriateness of use of coronary angiography in New York State. JAMA 1993;269:766-769.

27.       Leape LL, Hilborne LH, Park RE, et al. The appropriateness of use of coronary artery bypass graft surgery in New York State. JAMA 1993;269:753-760.

28.       Kleinman LC, Kosecoff J, Dubois RW, Brook RH. The medical appropriateness of tympanostomy tubes proposed for children younger than 16 years in the United States. JAMA 1994;271:1250-1255.

29.       Cochrane AL. Effectiveness and Efficiency. Random Reflections on Health Services London: Nuffield Provincial Hospitals Trust; 1972.

30.       Feinstein AR. An additional basic science for clinical medicine: IV. The development of clinimetrics. Ann Intern Med 1983;99:843-848.

31.       Feinstein AR. Clinimetrics. New Haven: Yale University Press; 1987.

32.       Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: I. Why try to keep up and how to get started. Ann Intern Med 1986;105:149-153.

33.       Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: IV. Using the literature to solve clinical problems. Ann Intern Med 1986;105:636-640.

34.       Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: II. Deciding which journals to read regularly. Ann Intern Med 1986;105:309-312.

35.       Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical epidemiology: a basic science for clinical medicine. Boston: Little, Brown and Company; 1991.

36.       Guyatt GH, Sackett DL, Cook DJ. Users' guides to the medical literature. II. How to use an article about therapy or prevention. A. Are the results of the study valid? Evidence-Based Medicine Working Group. JAMA 1993;270:2598-2601.

37.       Guyatt GH, Sackett DL, Cook DJ. Users' guides to the medical literature. II. How to use an article about therapy or prevention. B. What were the results and will they help me in caring for my patients? Evidence-Based Medicine Working Group. JAMA 1994;271:59-63.

38.       Guyatt G, Drummond R, eds. Users' Guides to the Medical Literature: A Manual for Evidence-Based Clinical Practice. Chicago: American Medical Association; 2002.

39.       Preliminary report: effect of encainide and flecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infarction. The Cardiac Arrhythmia Suppression Trial (CAST) Investigators. N Engl J Med 1989;321:406-412.

40.       Guyatt GH, Feeny DH, Patrick DL. Measuring health-related quality of life. Ann Intern Med 1993;118:622-629.

41.       Gill TM, Feinstein AR. A critical appraisal of the quality of quality-of-life measurements. JAMA 1994;272:619-626.

42.       Guyatt GH, Cook DJ. Health status, quality of life, and the individual. JAMA 1994;272:630-631.

43.       Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med 2006;354:2317-2327.

44.       Jaeschke R, Guyatt G, Sackett DL. Users' guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? Evidence-Based Medicine Working Group. JAMA 1994;271:389-391.

45.       Laupacis A, Wells G, Richardson WS, Tugwell P. Users' guides to the medical literature. V. How to use an article about prognosis. Evidence-Based Medicine Working Group. JAMA 1994;272:234-237.

46.       Azimi NA, Welch HG. The effectiveness of cost-effectiveness analysis in containing costs. J Gen Intern Med 1998;13:664-669.

47.       Healy B. Who says what's best? US News World Rep 2006;141:75.

48.       Holmes D, Murray SJ, Perron A, Rail G. Deconstructing the evidence-based discourse in health sciences: truth, power and fascism. Int J Evid Based Healthc 2006;4:180–186.

49.       Haynes RB. Of studies, syntheses, synopses, summaries, and systems: the "5S" evolution of information services for evidence-based health care decisions. ACP J Club 2006;145:A8.

50.       Jepson R, Craig J. Cranberries for preventing urinary tract infections. Cochrane Database Syst Rev 2008:CD001321.

51.       Miller D, Livingstone V, Herbison P. Interventions for relieving the pain and discomfort of screening mammography. Cochrane Database Syst Rev 2008:CD002942.

52.       Shojania KG, Sampson M, Ansari MT, Ji J, Doucette S, Moher D. How quickly do systematic reviews go out of date? A survival analysis. Ann Intern Med 2007;147:224-233.