Summary

The analysis of intra-uterine mortality is made difficult by the interaction of many factors, some of them being pure artifacts resulting from the way in which the data are collected, or from the under-reporting of induced abortions. This paper deals with some ‘real factors’ of variation in the risk of spontaneous abortion (mother's age, pregnancy order, number of previous abortions), and with some of these ‘artifacts’ (inclusion of induced abortion, memory effect, differential continuation rates). Special attention is paid to the effect of heterogeneity of the risk of abortion. After a discussion of problems of observation, data from two different samples are analysed, first in a classic way: variation with age and pregnancy order, comparison between the rates of abortion for current and previous abortion. Next, detailed data on successive pregnancies are used to derive estimates of the distribution of risk between women. It is concluded that this distribution could and should be taken into account, and that its effects are different from those of age. The possibility of differential continuation rates by outcome of pregnancy is discussed briefly, in connection with previous points.