LIBRARY SERVICES Evaluating the evidence Paula Funnell Senior Academic Liaison Librarian (Medicine and Dentistry) i
Why? To weigh up how valid and useful the research will be
Why? – to save time In order to keep up to date, clinicians would have to read 17 articles a day, 365 days a year Research is of variable quality Only an estimated 1% is judged clinically relevant Need to find the 1%
Publication bias Papers with “interesting” results are more likely to be: Submitted and accepted for publication Published in a major journal Published in English Quoted by authors Quoted in newspapers
Brainstorm What factors should you be bearing in mind when reading an article? Think about the research described how it is reported
Brainstorm the research described how it is reported
RCT checklist
How are the results presented? Number needed to treat (NNT) Odds Ratio Relative risk
Odds and risk Odds of winning 1:9 You versus the rest Risk of winning 1:10 You versus all the runners 10 horses running, you bet on 1 horse
Forest plots more than 1less than 1 1
Forest plots more than 1less than 1 1 Line of no effect
Forest plots more than 1less than 1 1 Line of no effect
Forest plots more than 1less than 1 1 Line of no effect Best estimate
Forest plots more than 1less than 1 1 Line of no effect Best estimate Confidence interval
Forest plots more than 1less than 1 1 Line of no effect Best estimate Confidence interval
Forest plots more than 1less than 1 1 Line of no effect Best estimate Confidence interval Pooled result
P-value Could the result have occurred by chance? p = (1 in 1000) p = 0.2 (1 in 5) A p-value of less than 0.05 (1 in 20) is considered to be statistically significant
How it works Involves answering a short questionnaire We use the CASP questionnaires at workforce/resources/critical-appraisals-skills- programme The questionnaires were devised by doctors for doctors
Summary Validity Is it trustworthy? Results What does it say? Relevance Will it help?