.....

You are doing meta analysis... OR is what you want..
If they do not provide 2X2 information, you can filter out those in your selection criteria.
You dont need to include all the OR you found as some did not provide the enough information.

Using inverse variance method, you need OR and standard errors of OR calculated from 2X2 data. If you really need to use those OR without 2X2 information, and if those OR without 2X2 are not many, then you can impute standard errors from what you have.  

But make sure you have design the systematic review and meta analysis properly. It's not only about pooling the results using fancy statistical methods.