Dear Ei Mon Phyo,

I'm not too familiar with Revman, but I'm assuming there is a way to enter pre and post means and have it calculate an effect size based on that.

If not, Meta-Essentials can help out (www.meta-essentials.com). It is a free tool for meta-analysis. For your data, you can use workbook 4 for 'Differences between dependent groups - continuous data'. You simply enter the pre and post-treatment data and Meta-Essentials will do all the calculations for you.

Good luck!

Henk

Dear Mr. Phyo,

are you sure you entered the correct settings (difference between means)? Those settings sounds for generic inverse of variance methods.

Thanks,

LH

Hi there,

I'm not much familiar with the MA, but you can actually pool the results at baseline and follow up separately and calculate the mean by summing up the readings devide it by total no. of participants, so that you can end up with single mean for baseline and single mean for follow up.

Dear Lucas Helal,

My outcome is change in weight from baseline after intervention. Some papers do not include change in weight mean(SD) value table. In calculation of change in mean, I subtracted final mean from baseline mean. But for change in SD, should I use the calculator in Revman showing “calculate based on several groups” to get pooled SD?

I am also not sure how to add 4 arms RCTs since Revman does not allow to add data for more than 2 arms RCTs. Should I choose only one as comparator or combine 4 arm values using that “calculate based on several groups” calculator?

Thank you very much

Ei Mon

Dear Ei Mon Phyo, sorry for the late reply.

Yes, you can calculate the pooled SD for a change-from-baseline (aggregated) data. The mean is correct. As for the SD, I would recommend to derivate it manually and it is quite tricky.

First, you may need to find a study that presents the SD pre, SD post and also the SD delta in similar interventions and outcomes of the studies you are intending to meta-analyze. Then, you will generate a correlation coefficient between the pre and post SDs. It is important because the formula that derivates the delta SD from pre SD and post SD may need to take SDs colinearity in account.

Having it in hands, the next step is easy. You will only need to input the observed pre SD, post SD and the correlation coefficient that you previously observed in the reference study.

Attached is the formula that I use for it.

Thanks,

LH