[R-meta] Help with Meta-analysis

Campana, Mattia M@tt|@@C@mp@n@ @end|ng |rom med@un|-muenchen@de
Thu May 14 12:58:55 CEST 2020


Dear R community,

I am doing a meta-analysis where I have got data from many studies where an intervention was added to a drug (Drug A) for the treatment of Schizophrenia. Let`s say I collected 100 studies. Now I can divide those studies in subgroups for example, 20 studies added psychotherapy to Drug A, 80 studies added Drug B to Drug A. Now I would like to compare some parameters between the 2 subgroups, for example the age of the patients. It is to mention that some of the studies have a lot of patients (200 or more) whereas other studies have little (around 10), and of course one subgroup has 20 studies, the other has 80. I would like to do a Wilcoxon test (data are not normally distributed) for the 2 independent groups. My problem is that I don�t know how to run a wilcox.test in R where the standard deviation and mean for every single study is accounted for. If I only use the mean age value of every study then the fact that some studies have 200 patients and other 10 patients is not accounted for. If, on the other hand, I only compare the two weighted age means (for psychotherapy and for Drug B) than the fact that one group has 20 studies and the other has 80 studies is not accounted for.

Is there a way, better a script :) where one can compare using a Wilcoxon test the weighted means with SD and Sample size of the 2 groups?

I hope my question was clear,
Thank you a lot for your support

Mattia


	[[alternative HTML version deleted]]



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