2
$\begingroup$

After searching,I didn't find any guidelines on conducting meta-regression on prevalence meta-analysis(no treatment group).

Where or how can i make it? here is a data set, after i made an overall estimate and there was heterogeneity(p<.001). How can I use meta-regression to track the source of this heterogeneity. In a priori,I suspected that the prevalence may change with the year,region and maybe high among pregnant women.

set.seed(100)
study<-paste("study",1:70)
events<-sample(1:100,70,replace=T)
studysize<-sample(500:5000,70,replace=T)
studyyear<-sample(1980:2015,70,replace=T)
studyregion<-rep(c("africa","america","asia","europe"),70/4)
studyincludedpregnants<-rep(c("yes","no"),70/2)



#overall estimate library(meta)
m1 <- metaprop(events,studysize)
$\endgroup$
2
$\begingroup$

I would use metafor for this as it has good facilities for meta-regression. First best to convert your proportions into effect sizes with their standard errors. metafor gives you a number of ways of doing this, look at escalc. Then you just use rma.uni with the mods parameter to specify your covariates (like year, region, pregnant). You can skip the first step and work directly in rma.uni with the numbers but I think it is cleaner to do it in two steps. Your mileage may vary.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you, Can you show me how to use rma.uni to do that? $\endgroup$ – Elmahy Jan 28 '16 at 22:23
  • $\begingroup$ I am not quite sure which part of that you did not understand. Can you elaborate? Perhaps show us the code you tried and what error message you got. $\endgroup$ – mdewey Jan 30 '16 at 16:41
  • $\begingroup$ I tried this > ess<-escalc(measure="PLO", xi=events,ni=studysize) > meta2<-rma(ess\$yi , ess\$vi , mods=~studyincludedpregnants,method="ML") $\endgroup$ – Elmahy Feb 3 '16 at 9:47
1
$\begingroup$

Elmahy! Just as you mentioned, the m-a of proportions is a little different than other types of meta-analysis- it includes studies that do not use controls. I recently made a tutorial on that on YouTube and shared my code on Github. This detailed, hands-on tutorial provides a step-by-step guide showing you how to conduct a full meta-analysis of proportions/prevalence, including meta-regression. My code allows you to conduct your analysis with either the logit transformation or double-arcsine transformation. You can also do it without transformation using my code. The R script shown in the video is readily adaptable for you to use for your own analyses.

Check out the tutorial here: https://youtu.be/2wbXTFvaRnM.

Download my code here: https://github.com/wnk4242/meta-analysis-of-proportions

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.