Meta-analysis one-arm with a continuous variable.

[Valeria Bustos]

Dear all,

My name is Valeria Bustos, and I am currently working in a meta-analysis for one-arm with a continuous variable. I want to run a meta-analysis to estimate overall satisfaction and a subgroup meta-analysis. However, the data that I have is based on a scale of 0-100, I just count with this information and the sample size, with no standard deviations (as many studies just reported an average of satisfaction without SD). For this, I thought I could use metaprop_one, however, this seems to be just for proportions data.

Does someone know what command I should use to run this type of meta-analysis and create my forest plot? Or what recommendation do you give me for this matter? This will be very helpful!

Thanks for your help,

Sincerely,

Valeria

[Tiago Pereira]

Dear Valeria,

You should be very careful, because meta-analysis of single-arm studies is very tricky.

In your case, I think that the best way is to perform a meta-analysis by using mean and standard error. E.g., metan mean se, random

But you should pay attention to the 95% CIs, since they can be outside 0-100.

The standard error for studies that did not report a standard deviation could be approximated using other studies (e.g., average SD from large well-conducted studies).

Hope this helps.

Tiago

[Valeria Bustos]

Dear Tiago, thanks for your reply, very much appreciate it. Yes, I wanted to do something like proportion metanalysis, but with continuous data. The idea is to assess overall satisfaction. Thank you very much for your help, I believe is what I needed. I will look which articles reported SD and, perhaps include just those in the meta-analysis.

Again thanks,

Valeria