R: how to specify confidence intervals in metaprop of package meta

user_3391549 0
Viewed 266
library(meta)
event <- c(81, 15, 0, 1)
n <- c(263, 148, 20, 29)
#
m1 <- metaprop(event, n, sm="PLOGIT", method.ci="SA")

I'm interested in combining proportions in a meta-analysis. In the above example, I have 4 studies, each reporting a proportion. And it appears that metaprop calculates CIs for each of the 4 studies. However, since the 4 studies already report CI (in the original paper), is there a way to incorporate the actual, reported CIs into the meta-analysis calculations instead of having metaprop calculate them on its own?

I am also open to exploring other packages if they will allow me to specify the CIs.

statistics r
1 Answers

First, keep in mind that you might be making a huge mistake. I don't see a reason not to take the CI from the metaprop function. You should take the calculated CIs from the package, specially if you are using the random-effects model. The mean individual effects and CIs will be calculated differently from the fixed-effect one. Anyways, here it is:

You can find the lower confidence limits in:

> m1$lower
[1] 0.25219013 0.05273001 0.00000000 0.00000000

And the upper confidence limits in:

> m1$upper
[1] 0.3637795 0.1499727 0.0000000 0.1008922

You can also access the individual mean effects in TE:

> m1$TE
[1] -0.8095575 -2.1822989 -3.7135721 -3.3322045

Then, you could manually add your own calculated values in these vectors, and they will appear in your forest plot.

Let's say your own lower values are myLCL:

myLCL <- seq(-1, -.25, .25) # define lower confidence limits

m1$lower <- myLCL # assign them to the respective location
m1$lower
[1] -1.00 -0.75 -0.50 -0.25

And your upper confidence limits are myUCL:

myUCL <- seq(.25, 1, .25) # define upper confidence limits

m1$upper <- myUCL # assign them to the respective location
m1$upper
[1] 0.25 0.50 0.75 1.00
user_95101390
Related