Evidence-based practice: What does it really mean?

John W. Bennett*, Paul Glasziou

*Corresponding author for this work

Research output: Contribution to journalArticleResearchpeer-review

10 Citations (Scopus)


Evidence-based medicine is the process of finding and applying the best available clinical research evidence to the management of individual patients. This first requires posing an appropriate answerable clinical question about diagnostic possibilities, prognosis, risks, diagnostic tests or treatment. We then need to rapidly and efficiently identify the best clinical research evidence to answer these questions. This may involve using either 'predigested' sources of evidence, such as systematic reviews or evidence-based guidelines, or, when these are not available, finding the primary evidence ourselves. Finally, this research evidence needs to be weighed, appraised and integrated with our clinical expertise to apply it to the individual patient problem. Recent studies have suggested that 82% of treatment decisions in a general medical ward and 69% of decisions in ambulatory care can be based on convincing evidence, more than half of which comes from randomised controlled trials. The key problem then is finding and integrating such evidence into routine practice. The recently established Cochrane Collaboration is beginning to provide a systematic summary of randomised trial evidence of the effects of treatment, but will take 5 to 10 years to provide comprehensive coverage. However, a number of other summary sources, together with skills in using 'Medline' efficiently and effectively, will provide practitioners with ready access to the best evidence. We illustrate these methods with several cases from ambulatory practice.

Original languageEnglish
Pages (from-to)277-285
Number of pages9
JournalDisease Management and Health Outcomes
Issue number6
Publication statusPublished - Jun 1997
Externally publishedYes


Dive into the research topics of 'Evidence-based practice: What does it really mean?'. Together they form a unique fingerprint.

Cite this