Hi all,

I am performing an analysis using meta-analysis techniques, with raw data at patient level. As I am not performing a meta-analysis of clinical trials, I want to adjust my effect size estimates for some confounding factors, so I am using ipdmetan commands with Syntax 1.

I have mainly 3 big questions:

1) In previous meta-analysis of clinical trials, I always used the weights that the commands calculate automatically, and they are always based on the number of observations of each trial. In this case, the weights are calculated in a different way, and I would like to know how. Does it have sense to use user-defined weights based on observations in this case?

2) The command I am using is logit/logistic, and I would like to obtain the effect size in OR, but the output of ipdmetan is always the coefficient of the regression instead of the OR. Is there any way I can obtain OR?

3) By default, the meta-analysis where the data used has this format: ES seES, are performed with the inverse-variance method. Is there any option to go for a random effects analysis in my case?

This is an example of what I am trying to perform, with the example data of ipdmetan:

Code:
use "http://fmwww.bc.edu/repec/bocode/i/ipdmetan_example.dta", clear
ipdmetan, study(trialid) nograph : logistic try sex age
Many thanks!

Clara