Today, countries worldwide face a common problem in healthcare service: the needs, demands and costs for healthcare are increasing at a rate that is greater than the rate at which resources are made available. The cost pressure has three inter-related causes: (i) population ageing, (ii) rising patient expectations, and (iii) new technology and new knowledge. Evidence-based medicine (EBM)^1-3 has opened up new opportunities to tackle this cost-pressure and is radically changing decision-making in healthcare which used to be largely opinion-based. In evidence-based healthcare, decisions should be based on the best evidence available.

Due to the collective ignorance of research evidence, many ineffective interventions have crept into medical practice and caused widespread use. Modern examples include lidocaine for treatment of ventricular fibrillation in acute myocardial infarction,⁴ and antioxidant vitamins⁵ and postmenopausal hormone replacement therapy⁶ for prevention of cardiovascular disease. EBM can facilitate the application of interventions that are safe, effective and value for money, reduce the number of ineffective interventions already in use, and prevent introduction of new but ineffective technologies. Quality improvement measures can be effective only when the interventions provided are effective. Thus worldwide, EBM is considered an important measure for slowing down the increase of healthcare expenditure and improving the quality and efficiency of healthcare.

EBM is about how to make evidence-based decisions and about how to apply research findings to medical activities. It does not apply only to clinical practice; all health and healthcare decisions should be based on evidence. These decisions may include legislation and regulations on health issues, healthcare policies and management, medical insurance, approval of new drugs, introduction of new diagnostic technologies, clinical guides and pathways, care of individual patients and public health practice, patients' choice, and malpractice litigation.

Evidence that can directly guide medical practice and decision-making comes from epidemiological and other research in human subjects. Findings from animal and in vitro laboratory studies cannot be directly generalized to humans and thus should not be used to inform healthcare decisions when quality epidemiological evidence is available. Major epidemiological studies include clinical trials, cohort studies, case control studies, cross-sectional studies, and systematic reviews. Each is most suitable for assessing particular questions in fields of healthcare needs, aetiology, diagnosis, treatment and prognosis.

For example, randomized controlled trials are best for evaluating the effectiveness of interventions. Case-control studies could be the only practical method for detecting a rare chronic adverse effect. Cross-sectional studies are best used to assess healthcare needs and accuracy of diagnostic tests. Cohort studies are most appropriate for investigating aetiology and prognosis. The systematic review is a secondary study of results from primary studies which include all the rest of study designs. For any healthcare question, quality systematic reviews of quality primary studies provide the best evidence. The Cochrane Library, developed by the World Cochrane Collaboration, is the best source for high quality systematic reviews.

The Cochrane Collaboration, an international organization founded in 1993 and named after the British epidemiologist, Archie Cochrane, facilitates EBM by making up-to-date, accurate information about the effects of healthcare readily available worldwide. It produces and disseminates systematic reviews of healthcare interventions and promotes the search for evidence in the form of clinical trials and other studies of interventions. The importance of the Collaboration for healthcare is likened to that of the Human Genome Project for basic biomedical research. The Hong Kong Cochrane Branch is based at The Chinese University of Hong Kong and promotes the development of EBM in this region.

High quality evidence can be relied upon for making decisions while poor quality evidence can mislead and lead to disastrous decisions. Thus, evidence must be appraised for its quality before it is used to inform decisions and much of the early developments in EBM is about critical appraisal of evidence. The quality of evidence, commonly known as internal validity, is the inverse of bias and expresses how close the observed findings are to the truth. It is determined by the type of study design and the methodological quality of the study. In critical appraisal of the quality of research findings, the first question is thus "Is this the best study design for answering the question?" The second question is about the methodological aspects of the study in hand. Regarding effectiveness of interventions, the randomized controlled trial is the best study design and its methodological quality is enhanced by randomization, allocation concealment, blinding, high follow-up rate, intention-to-treat analysis, appropriate clinical outcome, sufficient time of following-up, and so on.

Furthermore, clinically useful interventions must be capable to do more good than harm and the benefit must be sufficiently large so as to be worthwhile. A drug that can lower systolic blood pressure, say, by 0.5 mmHg may not be of any clinical value. The estimate of the benefit must also be statistically sufficiently precise. The precision is expressed with the confidence interval within which the truth lies. A confidence interval between 9-11 mmHg for the reduction in blood pressure by a drug is much more precise and informative for making decisions than, for example, a confidence interval from -2 to 22 mmHg. In the latter case, it is uncertain whether the drug could (i) increase blood pressure, (ii) have no clinically meaningful effect, or (iii) reduce blood pressure substantively.

However, interventions shown to be effective in high quality studies are not necessarily applicable to local patients. The local applicability of evidence, commonly known as external validity, is a complex issue and determined by many factors besides internal validity. Two general rules about the applicability are important to note: 1) the more similar the local situation including type of patients and therapeutic environments to that in the studies, the more applicable the research results are to the local population, and 2) the more heterogeneous the results are among different studies, the more cautious one should be about their local applicability.

To a population, the adoption of an intervention may be determined by factors, such as, the commonness of the problem, resources available, the importance of the benefit valued by the community, the standard of service including diagnosis and doctors' competence, and patients' compliance.

It is important to note that decisions should be made based on "best available" rather than "best possible" evidence as the latter may often be unavailable. Randomized controlled trials provide the best possible evidence for effectiveness but decisions still have to be made when such evidence is lacking. In addition, randomized controlled trials do not provide the best possible evidence for all clinical issues. For example, it is practically impossible to conduct a clinical trial to study a rare chronic adverse effect of a drug; in such circumstances one would often be looking for case-control studies.

Clinical experience also plays an important role in making evidence-based decisions. Clinical experience has three different components: (i) that related to conducting investigations and providing interventions such as history taking and communication skills, (ii) that related to integrating information and making a decision, (iii) and that about effectiveness of interventions. The importance of the first two components in practising EBM is obvious. It is the last component that is now given less importance than evidence from organized research. However, research evidence is always limited and needs to be complemented and supplemented by clinical experience so that average findings from research can be used to make sensible decisions for specific individuals.

EBM emphasizes the importance of evidence in healthcare decision making but evidence itself is not decisions; it only informs decisions. The evidence is international, and the techniques used to manage the evidence are the same in all countries. Decisions about healthcare are however not universal. Economics and ethics also influence decisions. Sometimes, evidence may be the primary determinant of a decision especially when the intervention is ineffective. On the other hand, an effective intervention may not be adopted because of lack of resources, or futile interventions may be used when the decision is dominated by values. For example, a son may decide to spend all his savings on clinically meaningless treatments for his mother in terminal illness purely due to love and gratitude to the mother.

When resources are limited, the healthcare service cannot cover all available interventions and rationing is inevitable. In order to maximize benefits, interventions that are most cost-effective or most value for money should be given the highest priority. However, this economic approach may deprive some patients of necessary treatments which are not sufficiently cost-effective. This is ethically an undesirable and unacceptable consequence in which resources are distributed unfairly among members of the community. Thus, evidence and economics are still not sufficient for making healthcare decisions. Any evidence-based decisions, whether in clinical practice, healthcare management or public health, are also influenced by the needs and values of the population. These values determine what is important and what is fair, which in turn influence the prioritization or rationing of healthcare services.

Due to the fact that the external validity of research evidence is determined by many factors beyond research and the influence of values and economics in decision making, different decisions about an intervention are often made in different places and for different populations even though they are all based on the same evidence. The contrast in the strategy regarding breast cancer screening in young women in Canada and the United Sates is a case in point.⁷

Knowledge is power. EBM is about transferring knowledge into effectiveness and efficiency in healthcare. Evidence-based decisions are influenced by evidence, economics and ethics.