Seasonal variation in the time to pregnancy: avoiding bias by using the date of onset

To study seasonality in human fecundability, measured indirectly by time to the first pregnancy, we used data from 18,970 French-Canadian women who married for the first time during the 17th or 18th century. The time to pregnancy was approximated by the interval between marriage and first birth minus 38 weeks. We used the week of marriage and the week of conception as references to study seasonality. We found a minor seasonal pattern in time to pregnancy when using the week of marriage as a reference. The proportions of women with a short time to pregnancy were highest during December-January and June-July, indicating that these may be the most fecund periods. In contrast, we found an obvious seasonal pattern when using the date of conception as a reference. This pattern can be largely explained by a strong seasonal pattern in pregnancy planning (in this case, in marriages). When studying seasonal variation in the time to pregnancy, the date of onset of the time to pregnancy should be used as reference, not the date of conception. Otherwise, results will be biased owing to seasonality in pregnancy planning. The same is true for studies on seasonally bound exposures in relation to time to pregnancy.