Critics have challenged previous comparisons of mortality from legal abortion and childbirth for containing biases in the crude data that spuriously favor the safety of abortion. To evaluate this concern, we reviewed the sources of mortality data on which these comparisons are based and examined the completeness of abortion mortality statistics, the completeness of childbirth mortality statistics, and the accuracy of the denominators for both these events. We found the evidence to be consistent in two directions: (1) abortion deaths appear to be more completely ascertained than childbirth deaths; (2) use of different denominator estimates has relatively little impact on the comparison. From this evidence, we conclude that the crude data are biased in a direction that overestimates the abortion risks for the woman relative to the risks of childbearing.
Cates W, Smith JC, Rochat RW, Grimes DA. Mortality From Abortion and ChildbirthAre the Statistics Biased?. JAMA. 1982;248(2):192-196. doi:10.1001/jama.1982.03330020036026