meta regression and bubble plot with metafor package in R -

i working on meta-regression on association of year , medication prevalence 'metafor' package.

the model used 'rma.glmm' mixed-effect model logit transformed 'metafor' package.

my r script below:

dat<-escalc(xi=a, ni=sample, measure="plo") print(dat) model_a<-rma.glmm(xi=a, ni=sample, measure="plo", mods=~year) print(model_a) 

i did significant results performed bubble plot model. found there no way perform bubble plot straight away 'ram.glmm' formula. did alternatively:

wi<-1/dat$vi plot(year, transf.ilogit(dat$yi), cex=wi) 

apparently got 'crazy' results, questions are:

1> how weight points in bubble plot study sample size? points in bubble plot should proportionate study weight. here, used 'wi<-dat$vi'. vi stands sampling variance, got 'escalc()'. doesn't seem right.

2> model correct investigate association between year , medication prevalence? tried 'rma' model got totally different results.

3> there alternative way perform bubble plot? tried:

percentage<-a/sample plot(year, percentage) 

the database below:

study       year    sample    study 1     2007    414      364 study 2     2010    142       99 study 3     1999    15         0 study 4     2000    17         0 study 5     2001    20         0 study 6     2002    22         5 study 7     2003    21         6 study 8     2004    24         7 study 9     1999    203        82 study 10    2009    647       436 study 11    2009    200       169 study 12    2010    156       128 study 13    2009    10753    6374 study 14    2007    143       109 study 15    2001    247        36 study 16    2004    318       184 study 17    2012    611       565 study 18    2013    180       167 study 19    2006    344       337 study 20    2007    209       103 study 21    2013    470       354 study 22    2010    180       146 study 23    2005    522       302 study 24    2000    62         30 study 25    2001    79         39 study 26    2002    85         43 study 27    2011    548       307 study 28    2009    218       216 study 29    2006    2901     2332 study 30    2008    464       259 study 31    2010    650       393 study 32    2008    2514      704