Flashcards in Critical Appraisal - Systematic Reviews & Meta-Analysis Deck (18)
What is a systematic review?
A literature review focused on a single question with well defined criteria for review. It is the highest level of evidence.
What is a meta-analysis?
The statistical analysis aspect of systematic review - analyses combined data from multiple studies. Done using statistical software.
What are the purposes of a meta-analysis?
Increase power, resolve uncertainty, improve estimates, answer other questions (such as secondary outcomes)
What are the inclusion and exclusion criteria used to select studies for a systematic review?
The PICOT parameters plus other relevant criteria such as sample size
How might bias be introduced when selecting studies?
By selecting only studies which are in English, or by selecting only studies which have been published
What is the CONSORT checklist?
A checklist of items that should be included in a randomised trial
What is the Cochrane guidelines?
A guide which looks at various types of bias to judge whether or not the individual study accounted for it
What are the key statistical issues in a meta-analysis?
outcome, weighting and heterogeneity
What does the outcome look at?
The average of each of the studies - such as the relative risk. Can be either a relative or absolute measure.
What determines the weighting of each study?
The number of outcomes - which is usually an indirect measure of the sample size
Why is heterogeneity important?
Because only studies which are not heterogenous should be combined
How is the information presented in a meta-anaysis?
In a forest plot
What does the vertical line in the forest plot represent?
The line of null effect
What does the size of the square in the forest plot represent?
The size of each of the studies
What does the horizontal line in the forest plot represent?
The confidence interval
What does the diamond in the forest plot represent?
The pooled result of all the studies - the vertical line of the diamond shows the point estimate and the horizontal width of the diamond shows the confidence interval
How is statistical heterogeneity determined?
Using a p value. If the p value is greater than 0.05 then the studies are not heterogenous so can be combined.