meta-analysis of proportions: observational studies

[Tharani Loganathan]

Hi,

I'm a new Stata user.
I am doing a meta-analysis of observational studies. It will require pooling proportions (proportions of stool samples tested positive for rv). Most meta-analysis are for interventional trials, where 2 groups are compared. This meta-analysis, however will only combine one single group.

this is the data I have to pool:

	RV Positive	Total Tested	%
1	1130	2260	50%
2	820	1756	47%
3	445	912	49%
4	161	385	42%
5	87	273	32%

These are the basic steps that need to be performed:

• Each study estimate needs to be weighted by the inverse of its variance
• The 95% confidence interval of each RV detection proportion is calculated
• A pooled estimate for RV detection proportion is calculated using random effects model (unlikely use fixed effects model d/t heterogeneity)
• Forest plots generated
• Heterogeneity tests

How do I do this? Does anyone have the relevant syntax?

[Carlo Lazzaro]

tharani _I (as per FAQ, please, re-register with your name and surname via Contact us, right corner at the bottom of this page):
- help meta - can point you to useful user-written meta-analysis commands, references and links.

Kind regards,
Carlo

[Tharani Loganathan]

Thank you. Will do!

[Tharani Loganathan]

I do not think metan supports meta-analysis of proportions.

I found a post from 2005 that provides a solution for Stata 8.2.
However, running the solution of Stata 12 was unsuccessful.
Would the codes be different in the new versions?

[Svend Juul]

I ran the do-file successfully both in Stata 12 and Stata 13.1. What do you mean by unsuccessful?

[Carlo Lazzaro]

Tharani_I:

1) type -search metan, net -
2) click on the following link: sbe24_3 from http://www.stata-journal.com/software/sj9-2 that points you to an update for the user-written command - metan-
3) after Stata reports that -metan- is already installed on your computer, select the replace option to overwrite the existing files with the update;
4) re-run the do.file and everything should be fine.

Two asides:
- as already recommended, please re-register with your family name, too (as per FAQ), via the Contact us button at the bottom right of the screen;
- "being unsuccessful" in running a do.file is not enough informative. As Svend highlighed, please report exactly what you typed and what you got from Stata, so that everyone on the list can see when (and hopefully why) Stata chocked.

Kind regards,
Carlo