[R-meta] can I do a meta analysis of one arm?

[Guido Schwarzer]

Am 07.05.20 um 19:56 schrieb Martin Lobo:

> Hello every one.
>
> I want to analyze studies with a technique evaluating successes and failures.
> the same technique is always used, but with two different protocols. I can do a meta analysis of a global and stratified arm by group

This sounds like a meta-analysis of single proportions.

You could do this, e.g., with metaprop() from R package meta or 
rma.uni() / rma.glmm() from R package metafor.

In order to conduct subgroup analyses for protocols, you can use 
argument 'byvar' in metaprop() or the 'mods' argument in the rma's.

This said, you must be aware that you cannot reliably compare the two 
protocols using information from these one-arm studies. All differences 
that you observe could be due to systematic differences in the 
individual studies.

Best wishes, Guido