If you are using R, you can use the "metaprop"" command of the R package "meta". You can then pool proportions across studies. However, you might encounter higher heterogeneity, so you might want to consider using some variance stabilizing transformations, like the arcsine transformation.

I am guessing that you are performing a meta analysis with binary outcome. If you are pooling the effect size of binary outcome then you should exclude the single group study design. You should have a case group and a comparative group.

Hello Mehreen,

Let's assume a given result, based on one batch of 92 cases, showed that 16 cases were classified as "positive" on some outcome of interest (the remaining 76 were "negative"). For this batch, the odds of a case being classified as positive are 16 / 76 = 0.211.

If a second batch, based on 65 cases, yielded 7 positive instances (and 58 negative cases), the odds for the second batch would be 7 / 58 = 0.121.

The odds ratio (of a case being positive), formed by batch 1 odds divided by batch 2 odds, would be: OR = 0.211 / 0.121 = 1.744. Batch 1 cases are 1.744 times more likely than batch 2 cases to be positive.

As Anees Bahji 's response indicates, there are meta-analytic software packages that can help with this conversion. As well, you could easily fix the formulas into a simple spreadsheet.

Good luck with your work!

Thank you Anees Bahji , Shakil Ahmed , David Morse for your kind responses. I am not actually looking to perform a meta-analysis due to so much variability in studies. I just wanted to calculate effect sizes of reported interventions to better understand which ones are effective using a standardised method. I could not figure out how to calculate effect size for single group pre-post test studies that reported proportions. I have just now come across the following article which I have found very helpful. They also provide excel sheet to help with calculations.

Article Relative Risk Reduction as a Metric to Standardize Effect Si...

Best regards