Hello Dr. Geraldi,

Thank you for your question.

Well, I believe you can treat your continuous data as if they're from double-armed studies. I'll tell you how.

From your question, I understand that you have the mean, SD, and total number of participants BEFORE receiving the intervention. I also understand that you have the mean, SD, and total number of participants AFTER receiving the intervention.

Your aim is to conclude whether receiving this intervention may improve the mean & SD values, right?

Excellent. You're going to use the metacont function from the meta package in R studio.

First and foremost, make sure you prepare your data in a Microsoft Excel format appropriately. This means that you may want to have the left-hand column heading as "study_id" and each row under this column heading would represent each unique study. We need 6 more columns to the right of the "study_id" column to represent the data for the mean, SD, and total before AND after receiving the intervention.

The file I have attached below named "Set-up" may help you design the spreadsheet.

Once data from your spreadsheet has been placed in their respective cells, you're now ready to begin the analysis!

On your R studio, please ensure you already have the following packages installed and loaded:

install.packages("readxl")

install.packages("meta")

library(meta)

library(readxl)

Once done with that, you can type the following lines of code to analyze your data and obtain the forest plot:

#Feel free to name the stored data and meta-analysis in any format you want!

data_geraldi_researchgate

search it in github

hello Dr Mohamed I. Mohamed , Thank you for your kind answers. Can we use the metacont function in a single arm? In this single-arm the pre and post-intervention come from the same group, different than the usual metacont for event and control mean but from other groups, does this will give effect to the heterogenity analyzed?

Thank you for following-up.

Okay. This is a paired set of data. To the best of my knowledge, the method above would not be optimum as this would overinflate the variance value and double the sample size (due to the approach dealing with the group as non-paired data).

You may want to calculate the mean change and the change in the SD. As for the total number of participants, please report the smaller number (the one at the latest follow-up).

Try using this:

install.packages("meta")

install.packages("dplyr")

library(meta)

library(dplyr)

data_geraldi_researchgate

Thank you very much dr Mohamed I. Mohamed for answering Me, I've tried the coding you sent, but in R, the coding was error because the n.c is null?

The error you're encountering is likely because the n.c (control group sample size) argument is not defined in your dataset. Since you're working with single-arm studies (no control group), you don't need to specify n.c. Iskandar Purba Geraldi

I see, but when I tried meta cont coding in R, it still needed to type the n.c, mean.c, and sd.c, so how do I type that in the meta cont coding? Hassan El-Masry