I presume that the investigators used something like the t-test for a difference of means. You treat 1/2 the group and don't treat the other; if you selected t

I presume that the investigators used something like the t-test for a difference of means. You treat 1/2 the group and don't treat the other; if you selected these two groups randomly, then they should a priori have equal likelihood of infection. You then can create a confidence interval for the hypothesis: are the two means / rates different. See the second use here: http://en.wikipedia.org/wiki/Student%27s_t-test http://en.wikipedia.org/wiki/Student%27s_t-test