Smoking, Sex, and Pregnancy-Reply

In Reply.—  Rosenberg and Spira raise three questions. First, does our method of counting cycles overestimate the effect of smoking on fertility? Counting cycles with sporadic birth control might result in either an overestimate or an underestimate. The direction of error (if any) depends simply on whether the risk of pregnancy in those cycles is more or less than half the usual risk. The most conservative approach is to exclude all cycles with sporadic use of birth control. In that case, substantially reduced fertility still remains in smokers (pregnancy rate for smokers is estimated to be 0.78 of that for nonsmokers using the Cox proportional hazard model).Second, does inclusion of women who were recently pregnant or nursing lead to an overestimate of the smoking effect? As stated in the original report, this factor was controlled in the effect measure we reported (smokers had an estimated pregnancy rate of 72%