Systematic reviews and meta analysis: steps to follow
13 important questions on Systematic reviews and meta analysis: steps to follow
Aim/ research question of systematic reviews
- Should always depart from a clinical relevant and well defined research question
- Relevant from clinical perspective and well defined
Inclusion criteria (PICO)
I: interventions
C: controls
O: Outcomes
But also think about these important aspects:
- Timing of outcome measures (short or long term)
- Design of study (RCT/CCT)
- Time (e.g. New type of surgery introduced)
- Language
--> after defining inclusion criteria, you go look for literature that meet these criteria
To what extend should you also look for grey literature resources?
- Depends on research question whether or not it is important to include
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
How is the selection of studies done?
- Consensus meeting (with two independent reviewers and a third reviewer)
- First selection in hits in database
- Title, keywords, abstract
- Second selection based on full text papers
- Inclusion criteria
- When two reviewers disagree the third reviewer is consulted
What is the definition of Risk of Bias (RoB)
A systematic error, or deviation from the truth, in results or inferences potentially leading o underestimation or overestimation of the true intervention effect
What are some other important issues for the quality of a study?
- External validity
- Precision
- Statistical issues
- Others... (depending on topic)
Same as in RCT's, cause you usually include RCT's in sytematic reviews (unless there is a good reason not to do so)
- Selection, performance and attribution & detection bias
How do you assess risk of bias (RoB)
- Two independent reviewers
- Blind article for authors and journals
- Which items?
- Consensus meeting (third reviewer) Pilot
- Additional information from authors
Data extraction by two independent reviewers
- Study population (age, sex, setting)
- Disorder (type, duration, stage)
- Intervention (type, dose, frequency)
- Control group (no intervention, placebo, other intervention)
- Results (outcome measures, timing)
What is important when you do the analyses
- Define comparisons
- Clinical homogeneity
- Population and subgroups need to be the same
- Interventions, control groups need to be the same
- Outcome measures, instruments need to be the same
- Does it make sense to pool results?
- Are the results clinically relevant?
What is a meta analyses
- Statistically combined results of all studies
Prerequisite
- Clinical and methodological homogeneity
- Statistical homogeneity
- Sufficient data should be reported
Risk of bias and analyses how to combine
- Uses RoB as an inclusion criterion for review
- Descriptive
- Sensitivity analyses (preferred by most)
- High RoB vs low RoB
- Bias due to individual items
- Grade (or levels of evidence)
What are the levels of evidence methodology
- Strong evidence - when there are more high quality RCTs
- Moderate evidence - several low quality RCTS and/ or 1 high quality RCT
- Limited evidence - 1 RCT only (high or low)
- Conflicting evidence
Define 'more' 'severa', 'high quality', 'conflicting'
Three other types of biases that can be in systematic reviews
- Preliminary results (multiple papers -> 50 patients in 1st paper, 100 patients in 2nd paper -> 50 patients double counted)
- Salami tactic (1 paper on psychical outc. 1 paper on psychosocial. Or 1 paper on 1 year results, 1 paper on 3 years)
- Advanced salami (other authors in different papers on same study)
- Trials published in low or non-impact factor journals were more likely to report significant results than those published in high-impact mainstream medical journals
- Could be introduced in reviews exclusively based on English-language reports
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding
![Trustpilot Logo](/images/design_2020/lp/clp-trustpilot-en.png)