Dear Stata Gurus!
My question today is on Network Meta-analysis. In brief, this is the statistical procedure by which we can estimate the (indirect) comparative effectiveness of two interventions (say, A and B) even though they have never been tested head-to-head, by borrowing information from 'indirect' sources (such as, A vs Placebo, and B vs Placebo etc.).
I know we can use the information on the sample size, mean, and SD from the interventions (A, B etc.), and Placebo to run the 'network meta' commands in Stata. However, as we know, there could be studies that only report the 'mean difference' comparing two treatment arms, instead of reporting individual estimates for Treatment A, or Placebo, for example. In R, we can use those 'mean difference' and the SE of the the 'mean difference' to conduct the network meta-analysis. I was wondering if we could do the same in Stata as well.
Note: I have tried to use the 'pairwise' data in network meta but got the following error message "The data contain multi-arm trials - this analysis in pairs format is wrong (Use force option to override - at your own risk)"
Any help will be much appreciated. Thanks guys!
Naz
My question today is on Network Meta-analysis. In brief, this is the statistical procedure by which we can estimate the (indirect) comparative effectiveness of two interventions (say, A and B) even though they have never been tested head-to-head, by borrowing information from 'indirect' sources (such as, A vs Placebo, and B vs Placebo etc.).
I know we can use the information on the sample size, mean, and SD from the interventions (A, B etc.), and Placebo to run the 'network meta' commands in Stata. However, as we know, there could be studies that only report the 'mean difference' comparing two treatment arms, instead of reporting individual estimates for Treatment A, or Placebo, for example. In R, we can use those 'mean difference' and the SE of the the 'mean difference' to conduct the network meta-analysis. I was wondering if we could do the same in Stata as well.
Note: I have tried to use the 'pairwise' data in network meta but got the following error message "The data contain multi-arm trials - this analysis in pairs format is wrong (Use force option to override - at your own risk)"
Any help will be much appreciated. Thanks guys!
Naz
Comment