Five papers in this month's issue are concerned with problems of labour and delivery. Although external cephalic version near term has been shown to reduce the rate of breech presentation at delivery and caesarean section, many of us are vexed at our inability to perform the procedure successfully, and on pages 798-802 Tze-Kin Lau and colleagues explain the reasons for our failure. The authors conducted a retrospective observational study of 243 breech presentations which were randomly assigned to two groups. Multiple logistic regression models were derived for each group and the validity of each model in predicting successful external cephalic version tested was tested in the other group. The results were remarkably similar, for each model found the same three factors which independently predicted unsuccessful external cephalic version-nulliparity, difficulty in palpation of the infant's head, and engagement of the breech. If none of these factors was present, external was virtually always successful but was successful in less than one in five cases if two were present and always failed if all three were present. This information will be useful in guiding women who have a breech presentation at term.

Induction of labour is one of the commonest and most controversial obstetric interventions. On pages 792-797 Jason Gardosi and his colleagues startle us by indicating the wide discrepancy between the definitions of prolonged pregnancy judged by menstrual dates and by ultrasound scan. The authors have studied their large database of nearly 25,000 pregnancies over seven years; the information is retrospective and the database was not designed primarily for the study of prolonged pregnancy, and furthermore more than one quarter of all the women were excluded from the analysis. Nevertheless, the study has yielded interesting and important information. The authors clearly indicate that length of pregnancy does not follow a normal distribution, but has a negative skew, and so the mean is not an appropriate measure of central tendency; probably the mode is best. This is important for the study shows that according to menstrual dates the modal length of pregnancy where labour is spontaneous is 283, not 280 days. But the main finding of the study is that by menstrual dates one tenth of all women will be thought to have prolonged pregnancy, compared with one fiftieth by ultrasound scan. Gardosi and colleagues advocate that induction of labour for prolonged pregnancy should be undertaken only on the basis of ultrasound estimation of gestational age.

If induction of labour is necessary, there are problems where the cervix is unfavourable. The standard treatment is vaginal or intracervical prostaglandin which brings about the physiological changes associated with ripening of the cervix but has the disadvantage of stimulating uterine contractions. The hormone relaxin induces changes in the connective tissue of some mammals which result in ripening of the cervix without causing uterine contractions. Does it work in humans? Not so, say Janet Brennand and her colleagues (pages 775-780) who report a randomised trial of recombinant human relaxin in women with an unfavourable cervix. There was no difference in the change in Bishop's score compared with placebo when relaxin was administered in a vaginal gel, or in the length of labour and method of delivery. The authors suggest that different results may be obtained with different dosage or route of administration.

There is also uncertainty over care of the perineum during childbirth, and on pages 787-79 1 Marion Shipman and her colleagues introduce us to antenatal perineal massage as a method of inducing relaxation of the perineum and preventing tears. The authors conducted a randomised trial in 861 nulliparous women, where the experimental group were asked to massage the perineum three or four times a week from 34 weeks of gestation. There were fewer perineal tears and fewer instrumental vaginal deliveries where massage had been performed, and these effects were particularly apparent in women over 30 years of age. The authors conclude that teaching of perineal massage should be part of routine antenatal care in nulliparae.

Surely there is no need for another randomised trial of routine administration of oxytocic drugs in the third stage of labour? Lennart Nordstrom and his colleagues disagree (pages 781-786). In Sweden there is a trend towards less intervention in labour, such that despite the evidence from randomised trials oxytocic drugs are not administered routinely in 34 out of 55 labour wards. The authors also point out that the magnitude of postpartum haemon-hage (500 mL) used as the primary outcome measure in the trials of routine oxytocics is unlikely to be clinically important in healthy women. Nordstrom and colleagues carried out a randomised trial of routine oxytocin in 1000 women undergoing vaginal delivery. The strengths of this trial are that oxytocin was compared with placebo and that the definition of postpartum haemorrhage (800 mL) was more clinically meaningful. Routine administration of oxytocin reduced the risk of postpartum haemorrhage by about one half. The case for giving oxytocic drugs routinely in the third stage of labour is irresistible.