A problem for conducting network meta regression by STATA

[Rui Tang]

I'm conducting a network meta regression with package mvmeta, howwever, "too many variables sepcificed r(103)" was always reported. My code and data were followed:

foreach trt in A B C D E F {
if " `trt' "=="A" continue
gen y`trt' = log(r`trt'/(n`trt'-r`trt')) - log(rA/(nA-rA))
gen S`trt'`trt' = 1/r`trt' + 1/(n`trt'-r`trt') + 1/rA + 1/(nA-rA)
foreach trt2 in A B C D E F {
if " `trt2' "=="A" continue
if " `trt2' ">" `trt' " gen S`trt' `trt2' = 1/rA + 1/(nA-rA) if !mi(r`trt') & !mi(r`trt2')
}
}