Calculating risk ratios for spontaneous abortions: the problem of induced abortions.

The frequency of induced abortions varies greatly between and within populations, and the currently employed measures of the ratio of spontaneous abortions are distorted proportionally to the number of induced abortions. Corrections are therefore often necessary to avoid bias when risk ratios for spontaneous abortions are estimated. The potential magnitudes and direction of the bias associated with such risk ratios (under the null hypothesis) are given under certain model assumptions. Adjustment of the estimated risk ratios using an appropriate model for the given data is proposed, rather than using a fixed correction factor.