You can not do this unless you convert to a common effect size. There are many tools available to do this. See link below.

http://www.campbellcollaboration.org/resources/effect_size_input.php

Agreed with Dr Doi, without standardizing outcomes it will be something like comparison of lemons and mangoes a good example is discussed here 

http://trialsjournal.biomedcentral.com/articles/10.1186/1745-6215-8-39

Thanks Dr Doi, for your valuable suggestion. So, the better way is to convert them into effect size and then use for data pooling. I would appreciate if you add more on it.

Before converting to a common effect size, the fundamental question to be answered is related to heterogeneity.  Assuming there is little heterogeneity among the studies, then conversion to a common effect size OR using a meta-analysis software package that can comine various outcome measures together are the only two options.

Yes dru, heterogeneity has to assess before pooling the data. The question was conceptual and have to look before practical use.