Failure rates among perfect users and during perfect use: a distinction that matters.

To make an informed decision when choosing a contraceptive, women and couples need to know how effective different methods are when used perfectly, where perfect use is defined as following the directions for use. In this article, we show that unbiased estimates of pregnancy rates during perfect use can be guaranteed only if information on consistency and correctness of use is available for each menstrual cycle. The estimated probability of pregnancy during a year of perfect use among the subset of women who always used a method perfectly will be biased upward.