Users' Guides to the Medical Literature

THE MEDICAL LITERATURE

Users’ Guides to the Medical Literature XXV. Evidence-Based Medicine: Principles for Applying the Users’ Guides to Patient Care

Gordon H. Guyatt, MD, MSc This series provides clinicians with strategies and tools to interpret and in- R. Brian Haynes, MD, PhD tegrate evidence from published research in their care of patients. The 2 key Roman Z. Jaeschke, MD, MSc principles for applying all the articles in this series to patient care relate to the value-laden nature of clinical decisions and to the hierarchy of evidence Deborah J. Cook, MD, MSc postulated by evidence-based medicine. Clinicians need to be able to dis- Lee Green, MD, MPH tinguish high from low quality in primary studies, systematic reviews, prac- C. David Naylor, MD, PhD tice guidelines, and other integrative research focused on management recommendations. An evidence-based practitioner must also understand the Mark C. Wilson, MD, MPH patient’s circumstances or predicament; identify knowledge gaps and frame W. Scott Richardson, MD questions to fill those gaps; conduct an efficient literature search; critically for the Evidence-Based Medicine appraise the research evidence; and apply that evidence to patient care. How- Working Group ever, treatment judgments often reflect clinician or societal values concerning whether intervention benefits are worth the cost. Many unanswered ques- CLINICAL SCENARIO tions concerning how to elicit preferences and how to incorporate them in A senior resident, a junior attending, a clinical encounters constitute an enormously challenging frontier for evidence- senior attending, and an emeritus pro- based medicine. Time limitation remains the biggest obstacle to evidence- fessor were discussing evidence-based based practice but clinicians should seek evidence from as high in the ap- medicine (EBM) over lunch in the hos- propriate hierarchy of evidence as possible, and every clinical decision should pital cafeteria. be geared toward the particular circumstances of the patient. “EBM,” announced the resident with JAMA. 2000;284:1290-1296 www.jama.com some passion, “is a revolutionary devel- opment in medical practice.” She went “their positions are diametrically op- INTRODUCTION on to describe EBM’s fundamental in- posed. They can’t both be right.” Evidence-based medicine, the ap- novations in solving patient problems. The emeritus professor looked proach to clinical care that underlies the “A compelling exposition,” re- thoughtfully at the puzzled physician 24 Users’ Guides to the Medical Litera- marked the emeritus professor. and, with the barest hint of a smile, re- ture, which JAMA has published dur- “Wait a minute,” the junior attend- plied, “Come to think of it, you’re right ing the last 8 years,1 is about solving ing exclaimed, also with some heat, and too.” clinical problems. The Users’ Guides presented an alternative position stat- ing that EBM merely provided a set of Author Affiliations: Departments of Clinical Epide- The original list of members with affiliations ap- additional tools for traditional ap- miology and Biostatistics (Drs Guyatt, Haynes, and pears in the first article of the series ( JAMA. 1993; Cook) and Medicine (Drs Haynes and Jaeschke), 270:2093-2095). A list of new members appears in proaches to patient care. McMaster University, Hamilton, Ontario; Depart- the 10th article of the series ( JAMA. 1996;275: “You make a strong and convincing ment of Medicine and Office of the Dean, Faculty of 1435-1439). The following member of the Evidence- case,” the emeritus professor com- Medicine, University of Toronto, Ontario (Dr Nay- Based Medicine Working Group contributed to this lor); Department of Family Medicine, University of article: Anne Holbrook, MD, MSc. mented. Michigan, Ann Arbor (Dr Green); Department of Corresponding Author and Reprints: Gordon H. Guy- “Wait a minute,” the senior attend- Medicine, Wake-Forest University School of Medi- att, MD, MSc, Department of Clinical Epidemiology cine, Winston-Salem, NC (Dr Wilson); and Depart- and Biostatistics, Room 2C12, 1200 Main St W, ing exclaimed to her older colleague, ments of Ambulatory Care and Research, South McMaster University Faculty of Health Sciences, Hamil- Texas Veterans Health Care System and Medicine, ton, Ontario, Canada L8N 3Z5. See also Patient Page. University of Texas Health Sciences Center, San Users’ Guides to the Medical Literature Section Edi- Antonio (Dr Richardson). tor: Drummond Rennie, MD, Deputy Editor.

1290 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved. USERS’ GUIDES TO THE MEDICAL LITERATURE provide clinicians with strategies and those gaps; to conduct an efficient lit- evidence of treatment effectiveness does tools to interpret and integrate evi- erature search; to critically appraise the not automatically imply that treat- dence from published research in their research evidence; and to apply that evi- ment be administered. The manage- patient care. As we developed the Us- dence to patient care.3 The Users’ ment decision requires a judgment ers’ Guides, our understanding of EBM Guides have dealt with the framing of about the trade-off between risks and has evolved. In this article, since we are the question in the scenarios, with benefits, and because values or prefer- addressing physicians, we use the term searching the literature,4 with apprais- ences differ, the best course of action EBM but what we report applies to all ing the literature in the “Validity” sec- will vary between patients and be- clinical care provisions and the rubric tion, and with applying the evidence in tween clinicians. “evidence-based health care” is equally the “Results” and “Applicability” sec- Picture a third patient, a healthy 30- appropriate. tions. Underlying these steps are 2 fun- year-old mother of 2 children who de- In 1992, in an article that provided damental principles. One, relating pri- velops pneumococcal pneumonia. No a background to the Users’ Guides, we marily to the assessment of validity, clinician would have any doubt about described EBM as a shift in medical posits a hierarchy of evidence to guide the wisdom of administering antibi- paradigms.2 In contrast to the tradi- clinical decision making. Another, re- otic therapy to this patient. This does tional paradigm, EBM acknowledges lating primarily to the application of evi- not mean that an underlying value judg- that intuition, unsystematic clinical ex- dence, suggests that decision makers ment has been unnecessary. Rather, our perience, and pathophysiologic ratio- must always trade off the benefits and values are sufficiently concordant, and nale are insufficient grounds for clini- risks, inconvenience, and costs associ- the benefits so overwhelm the risks that cal decision making, and stresses the ated with alternative management strat- the underlying value judgment is un- examination of evidence from clinical egies, and in doing so consider the pa- apparent. research. The philosophy underlying tient’s values.5 In the sections that In current health care practice, judg- EBM suggests that a formal set of rules follow, we will discuss these 2 prin- ments often reflect clinician or soci- must complement medical training and ciples in detail. etal values concerning whether inter- common sense for clinicians to effec- vention benefits are worth the cost. tively interpret the results of clinical re- Clinical Decision Making: Consider the decisions regarding ad- search. Finally, EBM places a lower Evidence Is Never Enough ministration of tissue-type plasmino- value on authority than the traditional Picture a patient with chronic pain due gen activator vs streptokinase to pa- paradigm of medical practice. to terminal cancer who has come to tients with acute myocardial infarction, While we continue to find the para- terms with her condition, has re- or clopidogrel vs aspirin to patients with digm shift a valid way of conceptual- solved her affairs and said her good- transient ischemic attack. In both cases, izing EBM, as the scenario suggests, the byes, and wishes only palliative therapy. evidence from large randomized con- world is often complex enough to in- The patient develops pneumococcal trolled trials (RCTs) suggests the more vite more than 1 useful way of think- pneumonia. The evidence that antibi- expensive agents are, for many pa- ing about an idea or a phenomenon. In otic therapy reduces morbidity and tients, more effective. In both cases, this article, we describe the 2 key prin- mortality due to pneumococcal pneu- many authoritative bodies recom- ciples that clinicians must grasp to be monia is strong. Almost all clinicians mend first-line treatment with the less effective practitioners of EBM. One of would agree that this strong evidence effective drug, presumably because they these relates to the value-laden nature does not dictate that this patient re- believe society’s resources would be bet- of clinical decisions; the other to the hi- ceive antibiotics. Despite the fact that ter used in other ways. Implicitly, they erarchy of evidence postulated by EBM. antibiotics might reduce symptoms and are making a value or preference judg- We will also comment on additional prolong the patient’s life, her values are ment about the trade-off between deaths skills necessary for optimal clinical such that she would prefer a rapid and and strokes prevented, and resources practice and we conclude with a dis- natural passing. spent. cussion of the challenges facing EBM Picture a second patient, an 85-year- By values and preferences, we mean in the new millennium. old severely demented man, inconti- the underlying processes we bring to nent, contracted and mute, without bear in weighing what our patients and TWO FUNDAMENTAL family or friends, who spends his day our society will gain or lose when we PRINCIPLES OF EBM in apparent discomfort. This man de- make a management decision. A num- An evidence-based practitioner must be velops pneumococcal pneumonia. ber of the Users’ Guides focus on how able to understand the patient’s cir- While many clinicians would argue that clinicians can use research results to cumstances or predicament (includ- those responsible for this patient’s care clearly understand the magnitude of po- ing issues such as social supports and should not administer antibiotic therapy tential benefits and risks associated with financial resources); to identify knowl- because of his circumstances, others alternative management strategies.6-10 edge gaps, and frame questions to fill would suggest they should. Once again, Three Users’ Guides focused on the pro-

ments are usually right, but occasion- tinues until both the patient and clini- Table 1. A Hierarchy of Strength of Evidence for Treatment Decisions ally disastrously wrong. Recent ex- cian conclude that the patient is, or is N of 1 randomized trial amples include an increase in mortality not, obtaining benefit from the target in- Systematic reviews of randomized trials with administration of growth hor- tervention. N of 1 RCTs are unsuitable Single randomized trial 18 Systematic review of observational studies mone in critically ill patients ; of com- for short-term problems; for therapies addressing patient-important outcomes bined vasodilators and inotropes ib- that cure (such as surgical proce- Single observational study addressing opamine19 and epoprostonol20 in dures); for therapies that act over long patient-important outcomes Physiologic studies patients with congestive heart failure periods of time or prevent rare or unique Unsystematic clinical observations (CHF); and of beta-carotene in pa- events (such as stroke, myocardial in- tients with previous myocardial infarction, or death); and are possible only cess of balancing those benefits and farction,21 as well as the mortality- when patients and clinicians have the in- risks when using treatment recommen- reducing effect of ␤-blockers22 despite terest and time required. However, when dations11,12 and in making individual long-held beliefs that their negative ino- the conditions are right, N of 1 RCTs are treatment decisions.13 The explicit enu- tropic action would harm CHF pa- feasible,24,25 can provide definitive evi- meration and balancing of benefits and tients. Observational studies are inevi- dence of treatment effectiveness in risks brings the underlying value judg- tably limited by the possibility that individual patients, and may lead to ments involved in making manage- apparent differences in treatment ef- long-term differences in treatment ad- ment decisions into bold relief. fect are really due to differences in pa- ministration.26 Acknowledging that values play a role tients’ prognosis in the treatment and When considering any source of evi- in every important patient care deci- control groups. dence about treatment other than N of sion highlights our limited understand- Given the limitations of unsystem- 1 RCTs, clinicians are generalizing from ing of eliciting and incorporating soci- atic clinical observations and physi- results in other people to their pa- etal and individual values. Health ologic rationale, EBM suggests a hier- tients, inevitably weakening infer- economists have played a major role in archy of evidence. TABLE 1 presents a ences about treatment impact and in- developing a science of measuring pa- hierarchy of study designs for issues of troducing complex issues of how trial tient preferences.14,15 Some decision aids treatment. Different hierarchies are nec- results apply to individuals. Infer- are based on the assumption that if pa- essary for issues of diagnosis or prog- ences may nevertheless be strong if re- tients truly understand the potential risks nosis. Clinical research goes beyond un- sults come from a systematic review of and benefits, their decisions will reflect systematic clinical observation in methodologically strong RCTs with their preferences.16 These develop- providing strategies that avoid or at- consistent results and are generally ments constitute a promising start. Nev- tenuate the spurious results. Because somewhat weaker if we are dealing with ertheless, many unanswered questions few, if any, interventions are effective only a single RCT unless it is large and concerning how to elicit preferences, and in all patients, we would ideally test a has enrolled a diverse patient popula- how to incorporate them in clinical en- treatment in the patient to whom we tion (Table 1). Because observational counters already subject to crushing time would like to apply it. Numerous fac- studies may underestimate or more pressures, remain. Addressing these is- tors can lead clinicians astray as they typically overestimate treatment ef- sues constitutes an enormously chal- try to interpret the results of conven- fects in an unpredictable fashion,27,28 lenging frontier for EBM. tional open trials of therapy, which in- their results are far less trustworthy than clude natural history, placebo effects, those of RCTs. Physiologic studies and A Hierarchy of Evidence patient and health worker expecta- unsystematic clinical observations pro- What is the nature of the evidence in tions, and the patient’s desire to please. vide the weakest inferences about treat- EBM? We suggest a broad definition: The same strategies that minimize bias ment effects. The Users’ Guides have any empirical observation about the ap- in conventional trials of therapy involv- summarized how clinicians can fully parent relationship between events con- ing multiple patients can guard against evaluate each of these types of stud- stitutes potential evidence. Thus, the misleading results in studies involving ies.29-31 unsystematic observations of the indi- single patients.23 In the N of 1 RCT, pa- This hierarchy is not absolute. If treat- vidual clinician constitute one source tients undertake pairs of treatment pe- ment effects are sufficiently large and of evidence, and physiologic experi- riods in which they receive a target treat- consistent, for instance, observational ments another. Unsystematic clinical ment in 1 period of each pair, and a studies may provide more compelling observations are limited by small sample placebo or alternative in the other. Pa- evidence than most RCTs. Observa- size and, more importantly, by limita- tients and clinicians are blind to alloca- tional studies have allowed extremely tions in human processes of making in- tion, the order of the target and control strong inferences about the efficacy of ferences.17 Predictions about interven- are randomized, and patients make insulin in diabetic ketoacidosis or hip tion effects on clinically important quantitative ratings of their symptoms replacement in patients with debilitat- outcomes from physiologic experi- during each period. The N of 1 RCT con- ing hip osteoarthritis. At the same time,

1292 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved. USERS’ GUIDES TO THE MEDICAL LITERATURE instances in which RCT results contra- immediately spent 2 hours searching for help clinicians define the general is- dict consistent results from observa- the highest-quality evidence and re- sues that they need to consider when tional studies reinforce the need for cau- viewing the available RCTs. When he advising the individual patient,34 noth- tion. A recent striking example comes began to discuss his remaining uncer- ing can substitute for clinical exper- from a large, well-conducted RCT of tainty with his partner, an experi- tise in determining the specific consid- hormone replacement therapy as sec- enced dentist, she quickly cut short the erations relevant to that person. ondary prevention of coronary artery discussion by exclaiming, “But, my Thus, knowing the tools of evidence- disease in postmenopausal women. dear, that isn’t herpes!” based practice is necessary but not suf- While the dramatically positive re- This story illustrates the necessity of ficient for delivering the highest- sults of a number of observational stud- obtaining the correct diagnosis before quality patient care. In addition to ies had suggested the investigators seeking and applying research evi- clinical expertise, the clinician re- would find a large reduction in risk of dence in practice, the value of exten- quires compassion, sensitive listening coronary events with hormone replace- sive clinical experience, and the falli- skills, and broad perspectives from the ment therapy, the treated patients did bility of clinical judgment. The essential humanities and social sciences. These at- no better than the control group.32 De- skills of obtaining a history and con- tributes allow understanding of pa- fining the extent to which clinicians ducting a physical examination and the tients’ illnesses in the context of their ex- should temper the strength of their in- astute formulation of the clinical prob- perience, personalities, and cultures. ferences when only observational stud- lem come only with thorough back- The sensitive understanding of the ies are available remains one of the im- ground training and clinical experi- patient links to evidence-based prac- portant challenges for EBM. The ence. The clinician makes use of tice in a number of ways. For some pa- challenge is particularly important given evidence-based reasoning by applying tients, incorporation of patient values that much of the evidence regarding the the likelihood ratios associated with for major decisions will mean a full enu- harmful effects of our therapies comes positive or negative physical findings meration of the possible benefits, risks, from observational studies. to interpret the results of the history and and inconvenience associated with al- The hierarchy implies a clear course physical examination.33 Clinical exper- ternative management strategies that are of action for physicians addressing pa- tise is further required to define the rel- relevant to the particular patient. For tient problems—they should look for evant treatment options before exam- some of these patients and problems, the highest available evidence from the ining the evidence regarding their this discussion should involve the pa- hierarchy. The hierarchy makes it clear expected benefits and risks. tients’ family. For other problems, such that any statement to the effect that Finally, clinicians rely on their ex- as the discussion of screening with pros- there is no evidence addressing the ef- pertise to define features that affect the tate-specific antigen in older male pa- fect of a particular treatment is a non generalizability of the results to the in- tients, attempts to involve other fam- sequitur. The evidence may be ex- dividual patient. We have noted that, ily members might violate strong tremely weak—the unsystematic ob- except when clinicians have con- cultural norms. servation of a single clinician, or gen- ducted N of 1 RCTs, they are attempt- Many patients would be uncomfort- eralization from only indirectly related ing to generalize (or, one might say, par- able with an explicit discussion of ben- physiologic studies—but there is al- ticularize) results obtained in other efits and risks, and object to having ways evidence. Having described the patients to the individual before them. what they experience as excessive re- fundamental principles of EBM, we will The clinician must judge the extent to sponsibility for decision making placed briefly comment on additional skills which differences in the treatment (lo- on their shoulders.35 In such patients, that clinicians must master for opti- cal surgical expertise, or the possibil- who would tell us they want the phy- mal patient care, and their relation- ity of patient noncompliance, for in- sician to make the decision on their be- ship to EBM. stance), the availability of monitoring, half, the physician’s responsibility is to or patient characteristics such as age, develop insight to ensure that choices CLINICAL SKILLS, HUMANISM, comorbidity, or concomitant treat- will be consistent with patients’ val- SOCIAL RESPONSIBILITY, ment may affect estimates of benefit and ues and preferences. Understanding and AND EBM risk that come from the published lit- implementing the sort of decision mak- The evidence-based process of resolv- erature. The clinician must further con- ing process patients desire and effec- ing a clinical question will be fruitful sider if the available studies have mea- tively communicating the informa- only if the problem is appropriately for- sured all important outcomes, if patients tion they need requires skills in mulated. One of us, a secondary care were followed up for a sufficient length understanding the patient’s narrative, internist, developed a lesion on his lip of time, and if experimental treatment and the person behind that narra- shortly before an important presenta- was compared with the most compel- tive.36,37 tion. He was quite concerned and, won- ling alternatives. While our Users’ Ideally, the technical skills and hu- dering if he should take acyclovir. He Guide on treatment applicability will mane perspective of evidence-based

ues, and the best ways of incorporating of therapy, published systematic re- Table 2. A Hierarchy of Preprocessed Evidence them in the clinical decision making pro- views, including the Cochrane Col- Primary studies cess, constitutes 1 important challenge laboration database, provide a rapidly Preprocessing involves selecting only for EBM. growing repository of clinically useful studies that are both highly relevant and with study designs that minimize bias Time limitation remains the biggest summaries. and thus permit a high strength of obstacle to evidence-based practice. For- Clinicians often seek answers to inference tunately, new resources to assist clini- questions about a whole process of care Summaries Systematic reviews provide clinicians with cians are available, and the pace of in- rather than a focused clinical ques- an overview of all the evidence novation is rapid. One can consider a tion. Rather than “What is the impact addressing a focused clinical question Synopses classification of information sources that of digoxin on my CHF patient’s lon- Synopses of individual studies or of comes with the mnemonic 4S: (1) the gevity?” the clinician may ask “Can I systematic reviews encapsulate the key methodologic details and results individual study, (2) the systematic re- prolong my CHF patient’s life?” or even required to apply the evidence to view of all the available studies on a given “How can I optimize the management individual patient care problem, (3) a synopsis of that sum- of my CHF patient?” Increasingly, cli- Systems Practice guidelines, clinical pathways, or mary, and (4) systems of information. nicians asking these sort of questions evidence-based textbook summaries of By systems we mean summaries that link can look to high-quality evidence- a clinical area provide the clinician with much of the information needed to guide a number of synopses related to the care based practice guidelines or clinical the care of individual patients of a particular patient problem (acute up- pathways to provide, in effect, a series per gastrointestinal tract bleeding) or of synopses that summarize available physicians will lead them to become ef- type of patient (the diabetic outpa- evidence. The best systems use com- fective advocates for their patients both tient) (TABLE 2). puter technology to match the patient in the direct context of the health sys- Evidence-based selection and sum- or problem characteristics with an evi- tem in which they work and in broader marization is becoming increasingly dence-based knowledge repository and health policy issues. This advocacy may available at each level. Secondary jour- provide patient-specific recommenda- involve changing the system to facili- nals such as ACP Journal Club and Evi- tions. Evidence suggests that these com- tate evidence-based practice; for ex- dence-based Medicine review a large puterized decision support systems may ample, improving infrastructure for ac- number of primary journals and in- change clinician behavior and im- cess to high-quality information to clude only articles that are both rel- prove patient outcome.38 At the same guide clinicians at the bedside. A con- evant and have passed a methodologi- time, we must remember that recom- tinuing challenge for EBM, and for cal filter. Clinicians can therefore be mendations can be made only for av- medicine in general, will be to better confident that any data they gather from erage patients, and the circumstances integrate the new science of clinical these sources is already high on the hi- and values of the patient before us may medicine with the time-honored craft erarchy of evidence in Table 1. These differ. One way of dealing with this of caring for the sick. secondary journals not only restrict might be to bring the tools of decision themselves to studies of superior de- analysis to the bedside. Whatever the ADDITIONAL CHALLENGES sign, but present the information as ultimate solution, this exploration re- FOR EBM structured abstracts that provide a syn- mains a frontier for EBM. In 1992, we identified skills necessary opsis of the individual studies and sys- These developments emphasize that for evidence-based practice. These in- tematic reviews from the primary jour- evidence-based practice involves not cluded the ability to precisely define a nals. The structure of the abstract is only being able to distinguish high from patient problem, and what information crucial: evidence-based synopses pro- low quality in primary studies, but also is required to resolve the problem, con- vide critical information about a study in systematic reviews, practice guide- duct an efficient search of the litera- that are necessary for determining va- lines, and other integrative research fo- ture, select the best of the relevant stud- lidity and for applying results to indi- cused on management recommenda- ies, apply rules of evidence to determine vidual patients. While not always the tions. That is the reason the Users’ their validity, and to extract the clini- case, these synopses often provide most Guides have included articles that show cal message and apply it to the patient of the information clinicians need to in- clinicians how to use systematic re- problem.1 To these we would now add corporate the results of a new study into views,26 decision analyses,4,39 practice an understanding of how the patient’s their clinical practice. guidelines,5,40 economic analyses,6,10 and values affect the balance between ad- If there is any chance it may be avail- any articles that make treatment rec- vantages and disadvantages of the available, clinicians whose priority is effi- ommendations.8 The summary tables able management options, and the abil- cient evidence-based practice should from each Users’ Guide provide a ity to appropriately involve the patient seek a high-quality systematic review checklist that clinicians can use to en- in the decision. Studying the process of rather than the primary studies address- sure that synopses of each type of study eliciting and understanding patient val- ing their clinical question. For issues include the key information required

1294 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved. USERS’ GUIDES TO THE MEDICAL LITERATURE to assess both validity and applicabil- at the heart of evidence-based health Haynes RB. Practitioners of evidence-based care. BMJ. 2000;320:945-955. ity to their practice. policy making, but inevitably has im- 4. Hunt DL, Jaeschke R, McKibbon KA, for the Evi- The last decade has seen publica- plications for decision making at the in- dence-Based Medicine Working Group. Users’ guides to the medical literature, XXI: using electronic health tion of a plethora of high-quality sys- dividual patient level. information resources in evidence-based practice. tematic reviews and there is no slow- JAMA. 2000;283:1875-1879. ing in sight. Most practice guidelines, CONCLUSION 5. Haynes RB, Sackett DL, Gray JM, Cook DJ, Guyatt GH. Transferring evidence from research into prac- however, remain methodologically The Users’ Guides to the Medical Lit- tice, 1: the role of clinical care research evidence in weak.41 Evidence-based systems have erature provide clinicians with the tools clinical decisions. ACP J Club. 1996;125:A14-A16. 6. Guyatt GH, Sackett DL, Cook DJ, for the Evidence- great potential, and are beginning to ap- to distinguish stronger from weaker evi- Based Medicine Working Group. Users’ guides to the pear. Efficient production of evidence- dence, stronger from weaker synthe- medical literature, II: how to use an article about based systems of information, increas- ses, and stronger from weaker recom- therapy or prevention, part B: what were the results and will they help me in caring for my patients? JAMA. ingly user-friendly synopses, and mendations for moving from evidence 1994;271:59-63. further advances in easy electronic ac- to action. Much of the Users’ Guides are 7. Jaeschke R, Guyatt GH, Sackett DL, for the Evi- dence-Based Medicine Working Group. Users’ guides cess to all levels of evidence-based re- devoted to helping clinicians under- to the medical literature, III: how to use an article about sources should dramatically increase the stand study results and enumerate the a diagnostic test, part B: what are the results and will they help me in caring for my patients? JAMA. 1994; feasibility of evidence-based practice in benefits, adverse effects, toxic effects, in- 271:703-707. the next decade. convenience, and costs of treatment op- 8. Richardson WS, Detsky AS, for the Evidence- This article, and indeed the Users’ tions, both for patients in general and Based Medicine Working Group. Users’ guides to the medical literature, VII: how to use a clinical decision Guides as a whole, have dealt primar- for individual patients under their care. analysis, part B: what are the results and will they help ily with decision making at the level of A clear understanding of the principles me in caring for my patients? JAMA. 1995;273:1610- 1613. the individual patient. Evidence- underlying evidence-based practice will 9. Wilson MC, Hayward RS, Tunis SR, Bass EB, Guy- based approaches can also inform aid clinicians in applying the Users’ att G, for the Evidence-Based Medicine Working 42 Group. Users’ guides to the medical literature, VIII: how health policy making, day-to-day de- Guides to facilitate their patient care. to use clinical practice guidelines, part B: what are the cisions in public health, and systems Foremost among these principles are recommendations and will they help you in caring for level decisions such as those facing that value judgments underlie every your patients? JAMA. 1995;274:1630-1632. 10. O’Brien BJ, Heyland D, Richardson WS, Levine M, managers at the hospital level. In each clinical decision, that clinicians should Drummond MF, for the Evidence-Based Medicine of these arenas, EBM can support the seek evidence from as high in the ap- Working Group. Users’ guides to the medical literature, XIII: how to use an article on economic analysis appropriate goal of gaining the great- propriate hierarchy as possible, and that of clinical practice, part B: what are the results and will est health benefit from limited re- every clinical decision demands atten- they help me in caring for my patients? JAMA. 1997: sources. On the other hand, evidence tion to the particular circumstances of 277:1802-1806. 11. Guyatt GH, Sackett DL, Sinclair JC, Hayward R, as an ideology, rather than a focus for the patient. Clinicians facile in using the Cook DJ, Cook RJ, for the Evidence-Based Medicine reasoned debate, has been used as a jus- Users’ Guides will complete a review of Working Group. Users’ guides to the medical literature, IX: a method for grading health care recommen- tification for many agendas in health the evidence regarding a clinical prob- dations. JAMA. 1995;274:1800-1804. care, ranging from crude cost-cutting lem with the best estimate of benefits and 12. Guyatt G, Sinclair J, Cook D, Glasziou P. Users’ guides to the medical literature, XVI: how to use a treat- to the promotion of extremely expen- risks of management options and a good ment recommendation. JAMA. 1999;281:1836-1843. sive technologies with minimal mar- sense of the strength of inference con- 13. McAlister FA, Straus SE, Guyatt GH, Haynes RB. ginal returns. In the policy arena, deal- cerning those benefits and risks. This Users’ guides to the medical literature, XX: integrat- ing research evidence with the care of the individual ing with differing values poses even leaves clinicians in an excellent posi- patient. JAMA. 2000;283:2829-2836. more challenges than in the arena of in- tion for the final—and still inad- 14. Drummond MF, Richardson WS, O’Brien BJ, Levine M, Heyland D, for the Evidence-Based Medi- dividual patient care. Should we re- equately explored—steps in providing cine Working Group. Users’ guides to the medical lit- strict ourselves to alternative resource evidence-based care, which is consid- erature, XIII: how to use an article on economic analy- allocation within a fixed pool of health eration of the individual patient’s cir- sis of clinical practice, part A: are the results of the study valid? JAMA. 1997;277:1552-1557. care resources, or be trading off health cumstances and values. 15. Feeny D, Furlong W, Boyle M, Torrance GW. care services against, for instance, lower Multi-attribute health status classification systems: Acknowledgment: We thank Deborah Maddock for Health Utilities Index. Pharmacoeconomics. 1995;7: tax rates for individuals or lower health her extraordinarily skillful and dedicated manage- 490-502. care costs for corporations? How should ment of the administrative aspects of preparing this 16. O’Connor AM, Rostom A, Fiset V, et al. Deci- article, and the Users’ Guides series as a whole. We sion aids for patients facing health treatment or screen- we deal with the large body of obser- thank Sam Auron, MD, FRCPC, for telling us the old ing decisions: systematic review. BMJ. 1999;319:731- vational studies suggesting that social rabbinical joke on which we based the Clinical Sce- 734. and economic factors may have a larger nario. 17. Nisbett R, Ross L. Human Inference. Englewood Cliffs, NJ: Prentice-Hall International Inc; 1980. impact on the health of populations 18. Takala J, Ruokonen E, Webster NR, et al. In- REFERENCES than health care delivery? How should creased mortality associated with growth hormone 1. Guyatt GH, Rennie D. Users’ guides to the medi- treatment in critically ill adults. N Engl J Med. 1999; we deal with the tension between what cal literature. JAMA. 1993;270:2096-2097. 341:785-792. may be best for an individual, or for the 2. Evidence-Based Medicine Working Group. Evidence- 19. Hampton JR, van Veldhuisen DJ, Kleber FX, et al, based medicine: a new approach to teaching the prac- for the Second Prospective Randomized Study of iboap- society to which that individual be- tice of medicine. JAMA. 1992;268:2420-2425. mine on Mortality and Efficacy (PRIME II) Investiga- longs? The debate about such issues is 3. Guyatt GH, Meade MO, Jaeschke RJ, Cook DJ, tors. Randomised study of effect of Ibopamine on sur-

©2000 American Medical Association. All rights reserved. (Reprinted) JAMA, September 13, 2000—Vol 284, No. 10 1295 USERS’ GUIDES TO THE MEDICAL LITERATURE vival in patients with advanced severe heart failure. 28. Kunz R. Oxman AD. The unpredictability para- 35. Sutherland HJ, Llewellyn-Thomas HA, Lock- Lancet. 1997;349:971-977. dox: review of empirical comparisons of randomised wood GA, Tritchler DL, Till JE. Cancer patients: their 20. Califf RM, Adams KF, McKenna WJ, et al. A ran- and non-randomised clinical trials. BMJ. 1998;317: desire for information and participation in treatment domized controlled trial of epoprostenol therapy for 1185-1190. decisions. J R Soc Med. 1989;82:260-263. severe congestive heart fialure: the Flolan Interna- 29. Guyatt GH, Sackett DL, Cook DJ, for the Evidence- 36. Greenhalgh T. Narrative based medicine: narra- tional Randomized Survival Trial (FIRST). Am Heart J. Based Medicine Working Group. Users’ guides to the tive based medicine in an evidence based world. BMJ. 1997;134:44-54. medical literature, II: how to use an article about 1999;318:323-325. 21. Rapola JM, Virtamo J, Ripatti S, et al. Ran- therapy or prevention, part A: are the results of the 37. Greenhalgh T, Hurwitz B. Narrative based medi- domised trial of ␣-tocopherol and ␤-carotene supple- study valid? JAMA. 1993;270:2598-2601. cine: why study narrative? BMJ. 1999;318:48-50. ments on incidence of major coronary events in men 30. Levine M, Walter S, Lee H, Haines T, Holbrook 38. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects with previous myocardial infarction. Lancet. 1997; A, Moyer V, for the Evidence-Based Medicine Work- of computer-based clinical decision support systems 349:1715-1720. ing Group. Users’ guides to the medical literature, IV: on physician performance and patient outcomes: a sys- 22. CIBIS-II Investigators and Committees. The Car- how to use an article about harm. JAMA. 1994;271: tematic review. JAMA. 1998;280:1339-1346. diac Insufficiency Bisoprolol Study II (CIBIS- II): a ran- 1615-1619. 39. Richardson WS, Detsky AS, for the Evidence- domised trial. Lancet. 1999;353:9-13. 31. Oxman AD, Cook DJ, Guyatt GH, for the Evi- Based Medicine Working Group. Users’ guides to the 23. Guyatt GH, Sackett DL, Taylor DW, et al. Deter- dence-Based Medicine Working Group. Users’ guides medical literature, VII: how to use a clinical decision mining optimal therapy: randomized trials in indi- to the medical literature, VI: how to use an overview. analysis, part A: are the results of the study valid? vidual patients. N Engl J Med. 1986;314:889-892. JAMA. 1994;272:1367-1371. JAMA. 1995;273:1292-1295. 24. Guyatt GH, Keller JL, Jaeschke R, et al. Clinical 32. Hulley S, Grady D, Bush T, et al, for the Heart and 40. Hayward R, Wilson MC, Tunis SR, Bass EB, Guy- usefulness of N of 1 randomized control trials: three Estrogen/progestin Replacement Study (HERS) Re- att GH, for the Evidence-Based Medicine Working year experience. Ann Intern Med. 1990;112:293- search Group. Randomized trial of estrogen plus pro- Group. Users’ guides to the medical literature, VIII: how 299. gestin for secondary prevention of coronary heart dis- to use clinical practice guidelines, part A: are the rec- 25. Larson EB, Ellsworth AJ, Oas J. Randomized clini- ease in postmenopausal women. JAMA. 1998;280: ommendations valid? JAMA. 1995;274:570-574. cal trials in single patients during a 2-year period. JAMA. 605-613. 41. Shaneyfelt TM, Mayo-Smith MF, Rothwangl J. Are 1993;270:2708-2712. 33. Sackett DL. A primer on the precision and accu- guidelines following guidelines? the methodological 26. Mahon J, Laupacis A, Donner A, Wood T. Ran- racy of the clinical examination. JAMA. 1992;267: quality of clinical practice guidelines in the peer- domised study of N of 1 trials versus standard prac- 2638-2644. reviewed medical literature. JAMA. 1999;281:1900- tice. BMJ. 1996;312:1069-1074. 34. Dans AL, Dans LF, Guyatt GH, Richardson S, for 1905. 27. Guyatt GH, DiCenso A, Farewell V, Willan A, Grif- the Evidence-Based Medicine Working Group. Us- 42. Haynes RB, Sackett DL, Guyatt GH, Cook DJ, Gray fith L. Randomized trials versus observational studies ers’ guides to the medical literature, XIV: how to de- JA. Transferring evidence from research into prac- in adolescent pregnancy prevention. J Clin Epide- cide on the applicability of clinical trial results to your tice, part 4: overcoming barriers to application. ACP miol. 2000;53:167-174. patient. JAMA. 1998;279:545-549. J Club. 1997;126:A14-A15.

There are trivial truths and the great truths. The opposite of a trivial truth is plainly false. The opposite of a great truth is also true. —Niels Bohr (1885-1962)