Zero events can be entered as 0.5 for ease of computation. Otherwise, you can use a "risk difference approach" instead on RR or OR.

Read more on:

http://handbook.cochrane.org/chapter_16/16_9_2_studies_with_zero_cell_counts.htm

Dear Dr

Thanks for your helpful answer but how can adding continuity (0.5) correction in binary data ?

For Event rate comparisons:

Say A is being compared to B

A has events n1/N1 and B has events n2/N2

say n1=0

Enter it as n1=0.5 and other values as such

Most software (including Revman) will do this automatically. This has some downsides, but is a generally acceptable means of analysis