[R-meta] Data for GeMTC (Network mtea-analysis)

Wu, Meng-Jia Mwu2 @end|ng |rom |uc@edu
Fri Feb 1 17:36:46 CET 2019

Hello everyone,

I have been working on a network meta-analysis project and recently I started to use the r package GeMTC. I found it's a very useful package. Even though the learning curve for using it is a little steep especially at the very beginning, I plan to explore it more as I can see the benefit of using it.

I have a question that I hope people with more experiences don't mind to share your thoughts. The project I am working on is related to the effectiveness of different learning modes (face-to-face (A) vs. blended (B) vs. fully online (C)). I have a small collection of studies that seem to meet the assumption of using network meta-analysis for comparing the effectiveness among A, B, and C groups. When setting up my dataset to be read to GeMTC, I have some studies report outcome mean/SD for each group separately (i.e., "arm-based data") and some studies I can only get contrast data (i.e., the effect size for A-B, B-C, or AC). My question is:

For the arm-base data, does the outcome have to be measured using the same scale in different studies? For example, one studies use the test with the range 0-100 points (with the outcome mean 87) while the another use the test with the range 200-800 (with the outcome mean 680). Can I import the mean/SD from these studies in their original scale to GeMTC? Or, should I calculate the ES between groups from different studies that report the arm-based data and just use the contrast data (I.e, the effect size  that shows the comparison between A, B, and C groups)?

Any thoughts would be much appreciated. Also, if you know any example that shows how to use GeMTC, please kindly share it. Otherwise, I will try to write one - once I figure most of the things out. :)



Meng-Jia Wu, Associate Professor

Research Methodology, Loyola University Chicago

	[[alternative HTML version deleted]]

More information about the R-sig-meta-analysis mailing list