Using permutation tests and bootstrap confidence limits to analyze repeated events data from clinical trials