Abstract
Accurate estimates of the effectiveness of hormonal emergency contraception (EC) are important for public policy and clinical application. Because there are no randomized trials of emergency contraception, observational studies with historical controls are employed to estimate effectiveness. There are many methodologic issues that can influence estimates of effectiveness, with multiple opportunities for bias. These include issues around use of EC and follow-up, selection of women for study, the historical reference population used for estimates of expected pregnancies, and the statistical approaches used to derive estimates of expected pregnancies. We discuss how the different sources of bias, including previously unrecognized sources, may affect existing estimates and make suggestions for future methodologic improvements to estimate effectiveness.
Keywords: hormonal emergency contraception (ec), pregnancy, ovulation, menstrual cycles, coitus
5