You can use the mean percentage difference between the two groups but you also need the standard deviation of that percentage difference in order to calculate an effect size.

Hi Dr. Gordon, 

Thank you for your comment. One more thing, does it matter if i use percentages for some of the studies and original concentrations from other studies? Can I calculate effect size for both in the same study objective?

Thanks

The measures of treatment effect are typically either the Odds Ratio, Relative Risk, Risk Difference when reporting dichotmous data. However, if you are including non-randomised studies the two groups (intervention and control) might not be in the same study so you might find yourself presenting percentages see the following paper where I have done exactly that.  Please note, the output is generated through a combination of Stata and R, so is not particularly intuitive

Mehanna H, Al-maqbili T, Carter B, Martin E, Campain N, Watkinson J, McCabe C, Boelart K, and Franklyn J. Differences in the recurrence and mortality outcomes rates of incidental and nonincidental papillary thyroid microcarcinoma – a systematic review and meta-analysis of 20,007 patient-year follow-up, JECM, 2014

Traditionally percentage data in meta-analysis (e.g. summarising the prevalence of a condition) is done through a transformation. Step 1: transform percentages; Step 2: perform meta-analysis; Step 3: back-transform to percentages. More often the Freeman-Tukey arcsin transformation is used, although alternatives exist. The methods has been implemented in Stata meta-analysis package metaan.

It is also appropriate to reveal raw scores as well as pecentages.

Any percentage is manipulable if its denominator is given. If it is at hand, you can convert proportions in odds ratios.