Retrospectively sampled time-to-pregnancy data may make age decreasing fecundity look increasing

The retrospective study based on pregnancy samples is the most commonly used method for studying time to pregnancy. Using a simple Monte Carlo simulation, this note focuses on a particular selection bias problem associated with this design: even if each woman's fecundity decreases with age, estimation of the effect of age may show the opposite trend. We recommend exercising increased caution in interpreting findings from pregnancy-based time to-pregnancy studies.