Systematic review Flashcards
What are the four purposes of a systematic review?
- Collate and summarise evidence
- Permit evidence-based recommendations about useful treatments
- Identify gaps and shortcomings in our evidence
- Generate new hypotheses to be explored
What are 6 requirements in the making of a systematic review?
- Clearly stated set of objectives (questions)
- e.g. how effective is CBT for depression in adults compared to antidepressant medication? - Pre-defined eligibility criteria for studies
- Systematic literature search strategy
- identify relevant papers for review and analysis - Explicit, reproducible methodology
- Assessment of the quality of the study
- synthesise the evidence and arrive at answers to starting question
- higher quality study -> higher importance/weight - Systematic presentation and synthesis of studies and findings
- tables and graphics
Why does a systematic review requires explicit and reproducible methodology?
- Increases the reliability of the review process itself
- Minimizes bias on the part of the reviewer
What is the result of methodological shortcomings in RCTs?
Downgraded quality of the evidence
-> assess the quality before drawing conclusions
What is the GRADE of a study?
Grading of Recommendations, Assessment, Development and Evaluation (GRADE):
- 0 = study should be disregarded completely
- 1 = Very low quality (e.g. case series/report, downgraded observational study)
- 2 = Low (e.g. observational study, downgraded randomised trial)
- 3 = Moderate (e.g. upgraded observational study, downgraded randomised trial)
- 4 = High (e.g. double upgraded observational study, randomised trial)
Which limitations of a study can reduce (downgrade) its quality, following the GRADE?
- Lack of blinding
- Randomisation problems
- Small sample size, very variable results
- Poor participation retention
- Missing or incomplete data
- Subjective outcomes, selective reporting of results
Which factors of a study can increase (upgrade) its quality, following the GRADE?
- Designed to minimize bias
- Large effect sizes (treatment effect)
- Systematic relationships between ‘dose’ and response (clinical outcome)
-> strong supportive evidence
What are meta-analyses?
> Formal statistical approach
> Combine findings to draw an overall conclusion about treatments efficacy
-> together they provide clearer picture
> Overall estimate of the size of the treatment effect
- overall effect measure from various studies
(269 published between 2000-2011 for CBT for depression)
What is the purpose of standardised effect sizes?
Allow us to compare the results across studies using different outcomes
What are the two formulas for a standardised effect size?
- Raw mean difference
= Group B change - Group A change - Standardised mean difference
= mean difference/Standard deviation at baseline
How are results characterised with a standardised effect size?
> Zero = no difference following treatment
> Positive (+) = e.g. treatment favours group B
> Negative (-) = e.g. treatment favours group A
What are the statistic indexes used for a standardised effect size (standardized mean difference - SMD)?
> Cohen’s index
> Hedges g index
- 0.8 = large effect size
- 0.5 = moderate
- 0.2 = small
How do you standardize the effect size with discrete variables such as recovery or relapse?
Relative risk ratio - Risk Ratio (RR):
- relative likelihood that patient will show with events outcome in the active group, relative to the control gap
> Outcome defined in positive way: recovery
- effective treatment = Higher relative risk
> Outcome defined in negative way: relapse
- effective treatment = Lower relative risk
> No differential risk: RR = 1.0
In a meta-analysis table, what does the ‘weight’ refer to?
Proportion of the total number of participants in analysis contributed by each study
In a meta-analysis, what do the ‘events’ refer to?
The number of patients that show the outcome
