How can I measure standard deviation when I just have the mean and standard deviation of two groups of numbers?

I'm trying to do a meta-analysis, bur most of the studies don't report intra-individual data and the just mention the mean+SD. I need the net SD of two groups. How can I deal with this problem?