I want to do meta-analysis to estimate the pooled prevalence of diseases for a certain population. Some studies have reported their findings as a % with CI. However the majority have reported prevalence as number of cases(n) and I do have N. How do I combine both findings to create forest plots on STATA? I know about the “metaprop” command but I need further details. Thanks
Check this textbook:
A Gentle Introduction to Stata, Sixth Edition
Check these articles:
https://pubmed.ncbi.nlm.nih.gov/29420178/
Article Metaprop: a Stata command to perform meta-analysis of binomial data
As far as I know metanalysis can be conducted if the results of researches are similar in term of relative difference, e.g. all in Odds Ratio, Risk Ratio, Hazard Ratio etc.
Yu need to have all data in the same consistent format.
The best is to have a dataset with one lin eper study, and for each study one column with the number of events (n) and one column with the total population (N).
Then metaprop is very simple, the basic syntax for a random-effects meta-analysis is:
metaprop n N, random
kind regards
Jérémie
Hi Amani,
As Jérémie F Cohen said, the data needs to be in the same consistent format.
I am glad you mentioned STATA 16 because one major upgrade from 15 was to simplify meta-analysis and make it straightforward.
You should try what Jeremie suggested and follow the link below for details on different effect sizes you may want to analyse.
https://www.stata.com/new-in-stata/meta-analysis/
Hope this is helpful,
cheers,
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis