I have found 5 included studies for my meta-analysis with binary outcome. Two of them present the results as n/N (n: events; N: non_events patients), while the other 3 three as OR (95% CI). Is there any option to create a forest plot without the exact number of events and total sample size of the three included studies. I use STATA 14.2 and I did not found any possible solution.
Thank you in advance
Thank you Dimitris for your answer. It is exactly what I have been looking for.
Alexandros
Thank you for your reply. It is really a helpful tutorial for forest plots in R.
Alexandros
Thank you for your response!
I have already found the answer from the comment of dr. Dimitrios Giannis.
It is possible to perform wiythout knowing the exact size of number of vents but the statistical poewer is lower than if yoiu include them
It is a good way of graphical representation of the several studies is systematic reviews with metanalysis
Here's an alternative solution:
You can calculate all the ORs and 95% CIs for the studies reporting their data as n/N separately, then perform a meta-analysis of ORs in STATA as follows:
- Create an excel file with separate columns for the ORs, lower 95% CIs and upper 95% CIs ; let's assume the columns are named OR, LCI, and UCI, respectively.
- Import the data and run the following command:
gen lnor = ln(OR)
gen lnlci = ln(LCI)
gen lnuci = ln(UCI)
meta set lnor lnlci lnuci, random(dlaird) studylabel(Author) eslabel("Ln(OR)")
meta forestplot, random(dlaird) eform title(whatever title you want)
The meta command will give you forest plots of much better quality compared to the metan command.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis