Method of delivery for the preterm infant


Correspondence: Professor J. Lumley, Centre for the Study of Mothers' and Children's Health, La Trobe University, 251 Faraday St, Carlton, Vic 3053, Australia.


The principal issue concerning the mode of delivery is that while elective caesarean births may reduce the chances of fetal or neonatal death, this method of delivery might also increase maternal morbidity. Six trials with a total of 122 women investigated the effects of a policy of elective caesarean delivery (ECD) versus expectant management for small babies. A meta-analysis of these six trials also included additional unpublished data from several of the authors. Babies in the ‘elective’ group were less likely to develop respiratory distress syndrome and were more likely to have a low pH after birth. They were also less likely to have neonatal seizures and there were fewer perinatal deaths, although the studies were too small to detect even very large differences. However, there was significantly more serious maternal morbidity. There is insufficient evidence to evaluate a policy of ECD. All trials described major problems in recruitment and no subsequent trials have been identified. Changes in population-based data suggest that ECD of very preterm infants has increased markedly in the last 20 years. Given these changes, it seems unlikely that recruitment for future trials will be any easier than it was in the recent past.


The first four trials comparing elective caesarean section with expectant management for the small baby were published in the mid 1980s. As a contributor to this extremely small and unsatisfactory body of evidence, my recollections of the context are, that interest in the question (‘Is elective caesarean section better for the baby and the mother than expectant care?’) arose from monitoring the short- and long-term outcomes of neonatal intensive care for very preterm and very low birthweight infants. Collaborative follow-up across the State of Victoria, Australia showed that infants who received less than optimum care immediately after birth were more likely to die and more likely to have serious morbidity at two years1. Babies transferred to tertiary care after a delay, those not resuscitated immediately, and those whose temperature and blood pressure were outside strict physiological limits when the Newborn Emergency Transport System (NETS) arrived were at increased risk of adverse outcomes2,3. It was not surprising that questions arose as to whether there were modifiable factors, other than those already being implemented, such as regionalisation of perinatal care, and an integrated newborn transport system, which could be implemented prior to the arrival of the neonatologist in labour ward. One such possible strategy was elective delivery by caesarean section.

Competing Outcomes: Fetal and Maternal

The potential difficulties of this strategy were understood. From the perspective of the infant, increased survival at lower gestations might be associated with greater morbidity and longer-term disability4. The avoidance of trauma and possible asphyxia during spontaneous labour was a benefit5, to be weighed against the additional risks of maternal anaesthesia to the fetus, the difficult extraction occasionally encountered in abdominal delivery if labour progressed rapidly, and the increased risk of respiratory distress syndrome6.

From the perspective of the mother there were different competing risks. With elective caesarean delivery in labour, there were risks of anaesthesia, major abdominal surgery, the possibility that a classical caesarean incision would be necessary, blood loss and transfusion, post-operative morbidity, including thrombo-embolic complications and infection, possible longer-term effects on recovery and fertility, and possible longer-term effects on breast-feeding and depression7–12. With selective caesarean delivery there would be a much greater chance of spontaneous birth, including the possibility of spontaneous version from breech to cephalic presentation, time to give prophylactic corticosteroids13, some risk of an emergency caesarean and possibly a greater risk of pelvic floor damage10,11.

Issues in Trial Design and Completion

The discussion, which follows focuses on the author's personal experience of one trial14, though the experiences, which the research teams had in other very varied settings, suggest that all these trials had a great deal in common15–19. The process of planning and executing this trial was sobering. One sticking point was randomisation. The other, linked to randomisation, was the content of the consent form, which needed to be explicit to show that no-one knew whether one method of delivery was better than the other. The research team spent most of a year meeting, firstly with the hospital's Research and Ethics Committee and then with a specially convened subcommittee, to discuss consent.

Possible solutions, which would not involve individual randomisation, were raised by members of the subcommittee. One of these was alternating a six-month period with a policy of elective caesarean delivery with a six-month period of a policy of selective caesarean delivery. There were several reasons why this option was unsuccessful. The main reason was the perception that changes in neonatal intensive care were so rapid, that secular trends in better neonatal outcomes might confound the analysis. Another reason was the possibility that knowledge of what the policy was for care at any given time might lead to women receiving care outside the policy. This was not recognised as an important practical issue, though it would be nowadays. Another option, was to agree on a definition of ‘standard care’, to randomise eligible women and to seek informed consent only for the ‘non-standard care’ arm of the trial—the so-called Zelen option20, or post-randomisation consent. This was not taken up, though it was used in one of the other trials19.

The subcommittee was also concerned that discussing the possibility of preterm labour with women, before it happened, would be a source of unnecessary anxiety. What was finally agreed was that the trial would be discussed with all women at their first or second antenatal visit and they would be offered the information and consent form at this time for discussion with family members and for a provisional signature. Forms were to be reviewed with a senior registrar when women were admitted in preterm labour. Women transferred from other centres would be asked to participate on their admission to hospital. Two members of the research team, not directly involved in patient care, would be available to discuss queries about the trial with patients, their families or staff.

The trials were designed to be able to identify improved infant outcomes, specifically mortality, since this was the era of rapidly increasing survival with intensive care. There was limited specification of maternal morbidity outcomes. This was a design problem since allocation concealment was not possible. In our trial, the sample size was known to be too small to identify clinically important differences in disability among neonatal survivors. This decision was a trade-off between the need to complete the trial over an acceptable period of time, believed to be three years of recruitment plus two years of infant follow-up, and the number of eligible infants over three years. This number was calculated from the hospital's data on births from 1977 to 1979, as 240. It was agreed that any change in policy would require data on follow-up at two years unless there were unpredicted differences in short term outcomes, and that no change in policy would take place until the end of the trial. An informal approach was made at this stage in the planning to another teaching hospital in the same city in the hope that a collaborative study providing a total sample of 500 eligible infants in two years could be mounted. The request was turned down on the grounds that the other hospital had recently established a firm policy for the obstetric management of very low birthweight infants.

All six trials experienced major recruitment problems, and all stopped early. The research team for the trial described above were able to show that the main factor reducing recruitment was withdrawal of 40% of eligible women for elective caesarean delivery. Unwillingness to consent in the antenatal period was not an issue as 71% of new antenatal patients from July to December 1980 gave preliminary agreement to take part14.

In this trial, the reason why the trial had recruitment problems was clear once hospital birth data for 1971–1982 were retrieved and analysed. These data showed that infants born at 33–36 weeks had experienced caesarean birth at a similar level to term infants since at least 1971, but that from 1977 onwards they were more likely than term infants to have a caesarean delivery. Infants born before 33 weeks were much less likely than other preterm or term infants to have a caesarean birth up to and including 1975–1976. Within the next two years, their probability of caesarean section more than trebled to 30%, and infants below 33 weeks retained a high probability of caesarean delivery throughout the next four years.

Systematic Review of the Randomised Trials, Including Unpublished Data Provided by the Authors

Babies in the elective group were less likely to have respiratory distress syndrome (odds ratio (OR) 0.43, 95% confidence interval (CI) 0.18–1.06, although they were more likely to have a low cord pH immediately after delivery (OR 10.82, 95% CI 1.60–73.24). They were less likely to have neonatal seizures (0/39 versus 2/42) and there were fewer deaths (2/62 versus 6/60) but these differences did not reach statistical significance. However, their mothers were more likely to have serious morbidity (OR 6.44, 95% CI 1.48–27.89)21,22.

Grant and Glazener concluded that the implications for practice are severely limited by the small number of participants, but that serious complications for the baby associated with vaginal breech delivery, and serious complications for the mother associated with caesarean delivery were identified. As the numbers of participants were so low, any estimates of these adverse events are too imprecise to be helpful in decision-making. Thus the uncertainty, which had prompted the trials 20 years ago, remains substantially unchanged.

The systematic review did provide additional information showing that about one in six babies allocated to elective caesarean will be born vaginally and that one in six allocated to expectant care will be born by caesarean section. Thus the proportion of participants experiencing crossover from the randomised allocation is relatively high. The reasons why the allocated form of care was not always appropriate was reviewed from the detailed records made available to the reviewers. This post-hoc analysis demonstrated why the observational studies looking at outcome by actual method of delivery are prone to serious biases22,23. We cannot answer the question about the best method of delivery with newer cohort or case-control studies.

Recent Changes Influencing Policy and Practice

One recent factor influencing current practice is the marked improvement in healthy survival of extremely preterm infants, which has led to a discussion with parents of active resuscitation and ventilation of infants at gestations as low as 24 weeks, or even less. The perception that these infants are viable (Table 1) means that caesarean section is no longer seen as inappropriate.

Table 1.  Survival of extremely low-birthweight liveborn infants in Victoria, Australia, 1986–2000.
500–599 g7.112.920.037.143.5
600–699 g21.928.163.965.964.6
700–799 g50.062.369.280.080.5
800–899 g59.572.676.986.081.4
900–999 g72.978.393.689.597.8

The second factor is the findings of the Term Breech trial, where perinatal mortality, neonatal mortality, or serious neonatal morbidity was significantly lower for the planned caesarean section group than for the planned vaginal birth group (17 of 1039 [1.6%] versus 52 of 1039 [5.0%]; relative risk (RR) 0.33 [95% CI 0.19–0.56]; P < 0.0001). There were no differences between groups in terms of maternal mortality or serious maternal morbidity (41 of 1041 [3.9%] versus 33 of 1042 [3.2%]; 1.24 [0.79–1.95]; P= 0.35)24. A three month follow-up showed that women in the planned caesarean delivery group were less likely to report urinary incontinence than those in the planned vaginal birth group (36/798 [4.5%] versus 58/797 [7.3%]; RR 0.62; 95% CI: 0.41–0.93). Incontinence of flatus was not different between groups but was less of a problem in the planned caesarean delivery group when it occurred (P= 0.006). There were no differences between the groups in other outcomes, such as breastfeeding; infant health; ease of caring for infant and adjusting to being a new mother; sexual relations and relationship with husband/partner; pain, depression, or women's views regarding childbirth experience and study participation25.

Although the research group responsible for the Term Breech trial stated that the results should not be seen to apply to preterm breech infants, practice in some major centres has moved since the Term Breech trial to elective caesarean delivery of all preterm infants presenting by the breech, at the onset of labour or when delivery is indicated. In the same centres, this decision has not been extended to infants with a cephalic presentation, who are expected to have a safer passage, with less trauma, if allowed to deliver spontaneously (S. Higgins personal communication).

Table 2 summarises changes in the method of delivery from 1983 to 1997, within gestation groups, in Victoria, Australia. The most striking aspects of the table are the reduction in deliveries assisted by forceps, and the similar reductions in vaginal breech delivery, which have occurred at all gestations. Since the last year of data collection was 1997, these changes predate the Term Breech trial and cannot be attributed to its findings. The only other noticeable change is a more than doubling of the odds of elective caesarean delivery for births before 28 weeks and between 28 and 31 weeks, as also described by Kramer and colleagues26.

Table 2.  Changes in method of delivery, all first singleton births, Victoria, Australia, from 1983–1997.
Method of deliveryGestation (weeks)Mean (%) 1983–19971983–19851995–1997OR (relative to baseline)Chi square for linear trend
  1. aToo few births assisted by vacuum extraction before 32 weeks for analysis.

Vaginal breech<2831.731.927.30.800.28
Spontaneous cephalic<2850.650.
Elective caesarean<2810.05.612.72.4510.57
Emergency caesarean<
Vacuum extractiona32–361.

The development of a policy for caesarean delivery of very preterm infants presenting by the breech has other ramifications. What factors apart from malpresentation make it particularly difficult? Births at these gestations include a much higher proportion of multiple births, more women with a prior caesarean birth, more women with a history of infertility, more women with a prior preterm birth, more women with a prior perinatal death, older women and more women with other prior pregnancy losses. In tertiary-care settings, there will also be more women with major medical or surgical problems, and in inner cities women with substantial social problems, including some with little or no antenatal care. Thus there are plans in some hospitals for implementing caesarean section at these early gestations with highly skilled and experienced practitioners only, rather as the Term Breech trial planned to do at its participating centres for term breech delivery.

What Next?

Is it likely that anyone will be willing to take up the challenges27 of designing and completing a trial comparing elective delivery versus expectant care for the preterm breech? Grant and Glazener, after reviewing the old trials thought not. They were a little more optimistic that it may still be possible to investigate elective caesarean delivery in preterm infants with cephalic presentations22. Only time will tell.


Thanks to Dr Shane Higgins, Royal Women's Hospital, Melbourne, for a thoughtful discussion of current practice and issues.