EBD1 - Meta analysis Flashcards
What is a systematic review?
- literature review that uses systematic methods to collect secondary data and critically appraise
- provides exhaustive summary of current evidence
- cheaper and more effective than a new study
How are SRs beneficial over single studies?
- save readers time
- reliable evidence
- resolves inconsistencies
- identify gaps in evidence
- identify when question has been answered and no further research required (unethical to conduct further studies)
- explores differences between studies ie dose
What are the key characteristics of a systematic review?
- well formulated question
- comprehensive data search
- unbiased selection and abstraction process
- assessment of papers
- synthesis of data
What are SRs used for?
- formulate policy
- develop guidelines
What authors are required for a SR?
- two or more
- topic expert
- methodological expert
What is the study protocol?
Need to set out in advance what is planned to do, so that you cannot deviate and introduce bias
What is the search strategy?
- multiple electronic databases
- published and unpublished data
- no language restrictions
- hand searching
- grey literature (not peer reviewed)
What are the different types of synthesis for SRs?
- qualitative (narrative)
- quantitative (meta analysis)
Give examples of critical appraisal tools.
- AMSTAR2
- ROBIS
What makes up a well formulated question?
P - population
I - intervention
C - comparison
O - outcome
What are the different types of bias?
- publication bias (positive results published more)
- time lag bias (exciting published sooner)
- language bias
- citation bias
How do you ensure unbiased selection and abstraction process?
- data extraction using predefined guidelines
- process conducted independently by at least two reviewers
- clear description for inclusion/exclusion
What are the different types of quality assessment tool?
- composite scale (overall score for a paper)
- component scale (methodological aspects are individually scored - preferred)
What factors are considered for risk of bias in RCTs?
- sequence generation
- allocation concealment
- blinding
- incomplete outcome data
- selective outcome reporting
- other
When is meta analysis inappropriate?
- data is sparse
- heterogeneity exists
What is meta analysis?
- statistical methods to combine results of different studies
- estimates common treatment effect across studies
- improves precision of point estimate
Define dichotomous data.
- not a scale, binary
- illness or not / death or not / birth or not
Define continuous data.
- scale
- blood pressure / weight
What summary statistic would you use for dichotomous data?
- odds ratio
- risk ratio
- risk difference
- NNT
What summary statistic would you use for continuous data?
- weighted mean difference
- standardised mean difference
How do you calculate mean difference?
Intervention outcome mean - control outcome mean = mean difference
What does weighting a study mean?
- more weight given to studies with more information
- more participants, more events, lower variance
How are results of SRs displayed graphically?
Forest plot
What are the different types of heterogeneity?
- clinical (variation in PICOs)
- methodological (variation in methods)
- statistical (variation in results)
