Sytsematic Reviews Flashcards
What are the AAAA in systematic reviews
Assessing- a situation and asking good questions
Accessing- relevant research evidence
Appraising- the evidence
Act: applying the findings to clinical practise
What is the best study design for frequency i.e incidence/prevalence
Cross-sectional survey
What is the best study design for causes or risk factors (aetiology)
Case control and cohort studies
What is the best study design for a diagnosis
Diagnostic test study
What is the best study design for effectiveness
Randomised controlled trial
What are the advantages of systematic reviews
Best source of evidence
Assimilation of large amounts of research evidence
Provide reliable unbiased estimates of effect
Provide information about generalisability and consistency of effect
Identify what info is missing
Useful for making desicion
What tool can we use to critically appraise the systematic review
CASP tool
What is a meta-analysis
A statistical method to summarise the results of more than one primary study
How are results from a meta-analysis displayed
In a forest plot
What is heterogeneity
Difference between the studies
What is homogeneity
Similarity between the studies
What does the size of the black squares of individual studies represent
The weight of the study in the meta-analysis
What does it mean if a vertical line of a study crosses the no effect line
The result is not significant
What does the diamond shape at the bottom of a forest plot represent
The combined or pooled results of all the trials
What does the width the diamond shape represent
The 95% confidence interval
When is the pooled results significant
When the width of the diamond shape does not cross the no effect line
Why might statistical heterogeneity be caused
Clinical differences between studies
Methodological differences between studies
Bias between studies
What are the 3 ways in assessing heterogeneity of a forest plot
- Eyeball test: look at the overlapping confidence interval with the summary estimate
- a statistical test of heterogeneity e.g the Cochran chi-squared
- the i-squared statistic
What is the chi-squared
Whether observed differences were compatible with chance alone
When do we reject the null hypothesis ie no difference in the studies
When p is less than 0.1
What is the interpretation of i squared
Closer it is to 100% the higher the inconsistent that cannot be explained by chance alone
What are the types of bias found in studies
Selection bias Performance bias Attrition bias Detection bias Publication bias Lead time bias Over-diagnosis bias
