[R-meta] Meta-regression of proportions using rma.glmm

[Viechtbauer, Wolfgang (SP)]

Dear Daniel,

No, studies with 0 events are not removed. To illustrate:

library(metafor)

xi <- c(0,1,0,0,1,2)
ni <- rep(10,6)

res <- rma.glmm(measure="PLO", xi=xi, ni=ni)
res
predict(res, transf=exp)

res <- rma.glmm(measure="PLO", xi=xi, ni=ni, subset=xi>0)
res
predict(res, transf=exp)

As you can see, in the second model, the estimate of the odds is much higher, since there I have explicitly thrown out the studies with 0 events.

Best,
Wolfgang

-----Original Message-----
From: Daniel Mønsted Shabanzadeh [mailto:dmshaban using gmail.com] 
Sent: Thursday, 10 October, 2019 10:11
To: Viechtbauer, Wolfgang (SP); r-sig-meta-analysis using r-project.org
Subject: Meta-regression of proportions using rma.glmm

Hey

I am performing af meta-regression of post-operative complications in single-arm studies. I am using the rma.glmm function on logit transformed data (measure ="PLO"). Many of the studies have 0 events - are these studie's proportions removed from the model? In that case, how can i include them?

Regards,
Daniel  

Daniel Mønsted Shabanzadeh
MD, PhD
Department of Gastroenterology, Surgical Unit
Hvidovre Hospital
Mobile +45 2546 5251