repression inflicted by the previous regime, or perhaps 2 Benatar SR. Medicine and health care in South Africa-five years later. NEngIlMed 1991;325:30-6. orchestrated by radical union leaders. Whatever the 3 Van Rensburg HCJ, Benatar SR The legacy of apartheid in health and health explanation, they show that the maturity displayed by care. SouthAfricanJoumal ofSociology 1993;24:99-1 11. the South Africans at the time of the elections has 4 Department of National Health and Population Development. Health trends in South Africa 1993. Pretoria: The Department, 1994:19. not been shown by its striking workers. The honey- 5 Blumson D. The pathology of poverty. In: Huddle K, Dubb A, eds. moon phase between South African politicians and Baragwanath Hospital, 50 years: a medical miscellany. Bertscham South Africa: Baragwanath Hospital, 1994:7-11. their electorate is becoming tumultuous and the honey- 6 World Bank. World development report 1993. Investing in heakh. Oxford: Oxford moon is being threatened. University Press, 1993:238-9. 7 Vital signs. Asiaweek. 1994 December 7:47-88. 1 Benatar SR. Medicine and health care in South Africa. N Engl J Med 8 Van der Linde 1. Strike at King Edward: important lessons. South African 1986;315:527-32. MedicalJournal 1994;84:15. Evidence based medicine: an approach to clinical problem-solving William Rosenberg, Anna Donald _ Doctors within the NHS are confronting major electronic databases and widespread computer literacy changes at work. While we endeavour to improve the now give doctors access to enormous amounts of data. quality of health care, junior doctors' hours have Evidence based medicine is about asking questions, been reduced and the emphasis on continuing finding and appraising the relevant data, and harness- medical education has increased. We are confronted ing that information for everyday clinical practice. by a growing body ofinformation, much ofit invalid Most readers will recognise that the ideas underlying or irrelevant to clinical practice. This article dis- evidence based medicine are not new. Clinicians cusses evidence based medicine, a process of identify the questions raised in caring for their patients turning clinical problems into questions and then and consult the literature at least occasionally, if systematically locating, appraising, and using con- not routinely. The difference with using an explicit, temporaneous research findings as the basis for evidence based medicine framework is twofold: it clinical decisions. The computerisation of biblio- can make consulting and evaluating the literature a graphies and the development of software that relatively simple, routine procedure, and it can make permits the rapid location ofrelevant evidence have this process workable for clinical teams, as well as made it easier for busy clinicians to make best use of for individual clinicians. The term "evidence based the published literature. Critical appraisal can be medicine" was coined at McMaster Medical School in used to determine the validity and applicability ofthe Canada in the 1980s to label this clinical learning evidence, which is then used to inform clinical strategy, which people at the school had been decisions. Evidence based medicine can be taught developing for over a decade.5 to, and practised by, clinicians at all levels of seniority and can be used to close the gulf between good clinical research and clinical practice. In Evidence based medicine in practice addition it can help to promote selfdirected learning Evidence based medicine can be practised in any and teamwork and produce faster and better situation where there is doubt about an aspect of doctors. clinical diagnosis, prognosis, or management. Doctors must cope with a rapidly changing body of Four relevant evidence and maximise the quality of medical steps in evidence based medicine care despite the reduction in junior doctors' working * Formulate a clear clinical question from a patient's hours and scarce resources. We are deluged with problem information, and although much of it is either invalid * Search the literature for relevant clinical articles or irrelevant to clinical practice, an increasing amount * Evaluate (critically appraise) the evidence for its comes from powerful investigations such as random- validity and usefulness ised controlled trials. Yet we continue to base our * Implement useful findings in clinical practice clinical decisions on increasingly out of date primary training or the overinterpretation of experiences with individual patients,' and even dramatically positive SETTING THE QUESMON results from rigorous clinical studies remain largely A 77 year old woman living alone is admitted with Nuffield Department of unapplied.' Doctors need new skills to track down the non-rheumatic atrial fibrillation and her first bout of Clinical Medicine, John new types of strong and useful evidence, distinguish it mild left ventricular failure, and she responds to Radcliffe Hospital, Oxford from weak and irrelevant evidence, and put it into digoxin and diuretics. She has a history of well OX3 9DU In this we William Rosenberg, clinical practice. paper discuss evidence based controlled hypertension. An echocardiogram shows tutor in medicine medicine, a new framework for clinical problem moderately impaired left ventricular function. She is solving which may help clinicians to meet these an active person and anxious to maintain her indepen- Public Health and Health challenges. dence. During the ward round on the following day a Policy, Anglia and Oxford debate ensues about the risks and benefits of offering Regional Health her long term anticoagulation with warfarin, and Authority, Oxford What is evidence based medicine? rather than defer to seniority or abdicate responsibility OX3 7LF Evidence based medicine is the process of system- to consensus by committee, team members convert the Anna Donald, senior house atically finding, appraising, and using contem- debate into a question: "How does her risk of embolic officer poraneous research findings as the basis for clinical stroke, if we don't give her anticoagulant drugs, Correspondence to: decisions. For decades people have been aware of the compare with her risk of serious haemorrhage and Dr Rosenberg. gaps between research evidence and clinical practice, stroke ifwe do?" and the consequences in terms of expensive, ineffec- The questions that initiate evidence based medicine BMJ1995;310:1 122-6 tive, or even harmful decision making.34 Inexpensive can relate to diagnosis, prognosis, treatment, iatro- 1122 BMJ VOLUME 310 29APRIL1995 genic harm, quality of care, or health economics. In analyses in the future, the ability to appraise critically any event, they should be as specific as possible, publications of all types will remain an invaluable skill. including the type of patient, the clinical intervention, Searches may fail to uncover well conducted and and the clinical outcome of interest. In this example relevant meta-analyses and often it will be impractical two questions are prepared for a literature search. One for a busy clinician to conduct an independent question relates to prognosis and her susceptibility: systematic review of the literature each time a clinical "How great is the annual risk of embolic stroke in a 77 question is generated. On these occasions the most year old woman with non-rheumatic atrial fibrillation, effective strategy will be to seek out the best of hypertension, and moderate left ventricular enlarge- the available literature and to appraise critically the ment if she is not given anticoagulants?" The other evidence by using skills that can readily be learnt. question concerns treatment and asks, "What is the risk reduction for stroke from warfarin therapy in such APPRAISING THE EVIDENCE a patient, and what is the risk of harming her with this The third step is to evaluate, or appraise, the therapy?" evidence for its validity and clinical usefulness. This step is crucial because it lets the clinician decide FINDING THE EVIDENCE whether an article can be relied on to give useful The second step is a search for the best available guidance. Unfortunately, a large proportion of pub- evidence. To conduct searches on a regular basis, lished medical research lacks either relevance or clinicians need effective searching skills and easy access sufficient methodological rigour to be reliable enough to bibliographic databases. Increasingly the access can for answering clinical questions.'7 To overcome this, be proved by ward or surgery based computers, a structured but simple method, named "critical complemented by assistance in obtaining hard copies appraisal," developed by several teams working in of articles, and enabled by librarians who teach search- North America and the United Kingdom, enables ing skills and guide the unwary through the 25000 individuals without research expertise to evaluate biomedical journals now in print.67 clinical articles. Mastering critical appraisal entails Two sorts of electronic databases are available. The learning how to ask a few key questions about the first sort is bibliographic and permits users to identify validity of the evidence and its relevance to a particular relevant citations in the clinical literature, using varia- patient or group of patients. Its fundamentals can be tions ofMedline. The second sort of database takes the learnt within a few hours in small tutorials, workshops, user directly to primary or secondary publications of interactive lectures, and at the bedside