To save Est values in Meta-analysis

[leili]

dear all,
I use the "meta" command for estimating weighted means of the fixed and random effects. Using "print" at the end of the command, I would have Est for each study. To get a summary of the estimated Ests, I need to save the new values on the spread sheet (insert the new variable to the list). Does anyone know how to save the generated Ests, as a variable?

[Joe Canner]

There is no meta command in Stata. Please specify exactly what command you are using (including all arguments, parameters, and options) and where the command came from.

[leili]

Generally the stata would not recognize the "meta" command, so you have to install packages through "ssc install metan". Then the command would be recognized. The stata example for this command is "meta logor selogor, eform graph(f) cline xline(1) xlab(.1,1,10) id(trialnam) b2title(Odds ratio) print". this command gives you weighted means for fixed and random effects. More Info is available at "http://www.biostat.jhsph.edu/~fdominic/teaching/bio656/software/meta.analysis.pdf".

[David Fisher]

Hi Leili,

I'm aware that meta is discussed in Sterne/Bradburn/Egger, but it's a very old command using Stata 7 graphics, and has been entirely superseded by metan, which does the same job.
The equivalent command would be something like:

metan logor selogor, eform xlab(.1,1,10) label(namevar=trialnam) xtitle(Odds ratio)

However, I'm a little unclear as to what value(s) you wish to save. If you mean the pooled estimates, these are stored in returned values r(ES) and r(seES), or the confidence limits in r(ci_low) and r(ci_upp), and may be viewed onscreen after running metan by typing return list. (Be careful when using eform, incidentally, as metan will return the exponentiated effect size and confidence limits, but the standard error of the log odds ratio.)

If you wish to run multiple meta-analyses and output the pooled results to a spreadsheet or new Stata dataset, there are various methods of achieving this, but I'm probably not the right person to suggest them (e.g. off the top of my head tabout, parmest, or simply writing to a log file).

Hope this helps -- as Joe Canner says, the more information you can give about your specific task, the better we can help.