Unsafe abortion is another major cause of maternal mortality in poorer countries. In the first trimester the standard method of termination of pregnancy is surgical evacuation of the uterus, but this is expensive for countries with constrained health budgets. Medical termination of pregnancy therefore is likely to the preferred method, but only if the costs of this treatment can be kept to a minimum. It was probably this consideration which encouraged the World Health Organisation (pages 524–530) to perform a randomised trial of two doses of mifepristone given two days before misoprostol for termination of pregnancy in the first trimester. This was an equivalence trial, to test the hypothesis that the lower dose of mifepristone was as effective as the higher dose in achieving termination of pregnancy. The trial was placebo-controlled with randomisation from the centre in Geneva, and so two of the main sources of bias in randomised trials were minimised. Furthermore, seventeen centres in several continents participated in the trial, and so the results can be made general. There was no difference in the rates of abortion with the two doses of mifepristone. A secondary analysis showed that women who had a menstrual delay of greater than 21 days had an unacceptably low rate of abortion with either misoprostol, vaginal misoprostol or a more powerful prostaglandin.

As the World Health Organisation point out, rigorous follow up of medical termination of pregnancy is necessary in order to detect failures, bearing in mind the teratogenic effects of misoprostol. This is the topic of the case-control study performed by Iêda Orioli and colleagues (pages 519–523), who tested the hypothesis that failed termination of pregnancy with misoprostol bought over the counter was associated with fetal malformations. The authors used the information held by the Latin American Collaborative Study of Congenital Malformations, which has recorded all births in ten Latin American countries since 1967 and includes detailed descriptions of birth defects. There were over four thousand cases of congenital malformation which were each matched with the next normal infant of the same sex born in the same hospital. The limitations of this study are that use of misoprostol may not always be reported; the analysis is by defect and not by infant so that each infant may be counted more than once; statistical significance was sometimes achieved with only small numbers and so the confidence intervals are wide; and there are many comparisons such that statistical significance may have occurred by chance. Nevertheless, there was a strong association between taking misoprostol in pregnancy and constriction defects in the limbs, consistent with the results of smaller case series. The assumption is that misoprostol taken in early pregnancy causes tonic uterine contraction and ischaemia of the distal parts of the limbs. These studies indicate that it may be possible to reduce substantially maternal mortality from unsafe abortion by medical termination of pregnancy. However, since in a significant number of women medical termination may fail, there must be an effective system of follow up so that the harmful consequences of failed medical termination of pregnancy can be avoided.