Systematic Review and Meta analysis

Question 1

Why are two independent reviewers important in a meta analaysis

Answer 1

1. It prevents data from being misses out thus increasing the thoroughness of the search
2. Two independent reviewers ensure that the data extraction is reproducible especially when there is high inter observer agreement
3. It minimises errors during the extraction of data
4. It ensures that one reviewers validity assessment of an article is not influenced by the other

Question 2

How do you increase the reproducibility of the meta analysis

Answer 2

1. Have two independent reviewers extract data, with a third serving as an adjudicator in the event of disagreement
2. Have a clearly specific search criteria and checklist for data extraction
3. Have the reviewers undergo formal training for data extraction

Question 3

Why is it important that the risk of bias is assessed in a SRMA

Answer 3

1. The results and conclusion of the SRMA are only as good as the internal validity of the individual articles. Articles with high risk of bias threaten internal validity and hence degrade results and conclusions of the SRMA
2. If studies are at a high risk of bias, authors may choose to exclude these studies from the meta analysis
3. Studies that are high risk of bias may contribute to heterogeneity in the results of the meta analysis

Question 4

What contributes to heterogeneity of the results of a meta analysis

Answer 4

1. Differences in clinical questions
2. Differences in the clinical setting where the study was conducted
3. Difference in study design and methodology of conducting the study
4. Differences in the baseline characteristics of the population
5. Differences in the intervention regime
6. C: Differences

Question 5

When should a random effects model be used over a fixed effect model

Answer 5

1. When there is heterogeneity of study results

Question 6

When should a fixed effect model be used over a random effect model

Answer 6

1. When there is a low number of studies (less than 5)
2. When there is a single large study, with good internal validity that can be trusted, amongst small studies of questionable internal validity