Trial registration: www.clinicaltrials.gov, number NCT00835003
Elective caesarean section at 38 weeks versus 39 weeks: neonatal and maternal outcomes in a randomised controlled trial†
Article first published online: 20 MAY 2013
© 2013 RCOG
BJOG: An International Journal of Obstetrics & Gynaecology
Volume 120, Issue 9, pages 1123–1132, August 2013
How to Cite
Elective caesarean section at 38 weeks versus 39 weeks: neonatal and maternal outcomes in a randomised controlled trial. BJOG 2013;120:1123–1132., , , , , , , , , .
- Issue published online: 9 JUL 2013
- Article first published online: 20 MAY 2013
- Manuscript Accepted: 14 APR 2013
- The Danish Council for Independent Research Medical sciences
- Health Research Fund of Central Denmark Region
- The Philanthropic Foundation TrygFonden
- Aase and Ejnar Danielsen's Foundation
- Sophus Jacobsen's Foundation
- Dagmar Marshall's Foundation
- Marie Dorthea and Holger From, Haderslev's Foundation
- Elective caesarean section;
- maternal outcomes;
- neonatal intensive care unit admission;
- neonatal outcomes;
To investigate whether elective caesarean section before 39 completed weeks of gestation increases the risk of adverse neonatal or maternal outcomes.
Randomised controlled multicentre open-label trial.
Seven Danish tertiary hospitals from March 2009 to June 2011.
Women with uncomplicated pregnancies, a single fetus, and a date of delivery estimated by ultrasound scheduled for delivery by elective caesarean section.
Perinatal outcomes after elective caesarean section scheduled at a gestational age of 38 weeks and 3 days versus 39 weeks and 3 days (in both groups ±2 days).
Main outcome measures
The primary outcome was neonatal intensive care unit (NICU) admission within 48 hours of birth. Secondary outcomes were neonatal depression, NICU admission within 7 days, NICU length of stay, neonatal treatment, and maternal surgical or postpartum adverse events.
Among women scheduled for elective caesarean section at 38+3 weeks 88/635 neonates (13.9%) were admitted to the NICU, whereas in the 39+3 weeks group 76/637 neonates (11.9%) were admitted (relative risk [RR] 0.86, 95% confidence interval [95% CI] 0.65–1.15). Neonatal treatment with continuous oxygen for more than 1 day (RR 0.31; 95% CI 0.10–0.94) and maternal bleeding of more than 500 ml (RR 0.79; 95% CI 0.63–0.99) were less frequent in the 39 weeks group, but these findings were insignificant after adjustment for multiple comparisons. The risk of adverse neonatal or maternal outcomes, or a maternal composite outcome (RR 1.1; 95% CI 0.79–1.53) was similar in the two intervention groups.
This study found no significant reduction in neonatal admission rate after ECS scheduled at 39 weeks compared with 38 weeks of gestation.
British and American societies in obstetrics recommend elective caesarean section to be scheduled after 39 completed weeks of gestation.[1, 2] This recommendation is based on a subset of several observational studies suggesting a strong association between earlier gestational age at elective caesarean section delivery and risk of respiratory morbidity.[3-7] In addition, two recent, large cohort studies investigated timing of elective caesarean section and the incidence of a composite adverse neonatal outcome including neonatal death or any of a series of adverse events. Both showed a decreasing incidence of the composite outcome with increasing gestational age from 37 to 39 completed weeks of gestation.[8, 9] In contrast, any maternal benefit of postponing elective caesarean section to 39 completed weeks has not been shown, but knowledge is sparse when it comes to maternal consequences of elective caesarean section timing.[10, 11]
However, confounding by indication may impair results from observational studies. In this case, neonates with a higher risk of an adverse outcome may be over-represented in caesarean sections undertaken before 39 weeks for reasons other than the caesarean section per se. This methodological problem would only be solved in a randomised controlled trial. Furthermore, to fully investigate the neonatal and maternal benefits or adverse events associated with a certain elective caesarean section timing, outcomes from women with unscheduled procedures who intended to deliver by elective caesarean section should be included in the assessment.
We therefore conducted a randomised controlled trial of neonatal and maternal morbidity after elective caesarean section scheduled at 38+3 weeks compared with 39+3 weeks of gestation. The primary study objective was to investigate the risk of neonatal intensive care unit (NICU) admission associated with elective caesarean delivery scheduled before versus after 39+0 weeks of gestation.
We conducted a multicentre, open-label randomised controlled trial in seven Danish hospitals, all with an NICU within the hospital. Approvals were obtained from The Central Denmark Region Committees on Biomedical Research (ID M-20080142) and the Danish Data Protection Agency (ID 2008-41-2522). Registration at www.Clinicaltrials.gov was ensured with identification number NCT00835003. An independent trial steering committee was appointed for data monitoring, safety and advice during the trial.
Clinical management and data collection
Participants were randomly assigned in a 1:1 ratio to an elective caesarean section scheduled 11 days before the estimated due date (38+3 weeks; referred to as 38-week group) or to an elective caesarean section scheduled 4 days before the estimated due date (39+3 weeks; referred to as 39-week group). In both groups, the scheduled date was allowed to deviate ±2 days from this specific date.
Hospital staff and participants were all aware of the estimated gestational age at the elective caesarean section scheduled date. During antenatal consultations, participants and nonparticipants had a case report form assessing inclusion and exclusion criteria filled out by the responsible physician, and after written consent had been obtained, elective caesarean section was booked according to the randomisation. To limit the influence of subjective timing preferences, the responsible staff members were asked to book the elective caesarean section according to a prioritised gestational age sequence (38+3 > 38+2 > 38+4 > 38+5 > 38+1 versus 39+3 > 39+4 > 39+2 > 39+1 > 39+5).
Neonatal outcomes were assessed 30 days after delivery or at final discharge (with NICU length of stay >30 days) by a designated physician, research nurse, or midwife. Maternal outcomes were assessed 30 days postpartum. Data from all the participating hospitals were collected and checked for accuracy by the first author, before they were entered into a database (epidata software version 3.1).
Study population and setting
Singleton pregnant women with a gestational age determined by ultrasound before 15 weeks of gestation were eligible for inclusion, when a decision of delivery by elective caesarean section was made. Exclusion criteria were multiple gestations, age <18 years, language difficulties requiring an interpreter, or an estimated risk of the caesarean section being undertaken before 39+5 weeks of gestation. Accordingly, women with conditions such as placenta praevia, high blood pressure, or elevated levels of liver enzymes were not included. Women with type 1 diabetes or gestational diabetes, verified at the time of randomisation by an oral glucose tolerance test, were also excluded from participation. Any fetal condition warranting delivery before 39+5 weeks or necessitating NICU admission also excluded participation in the study.
In Denmark, elective caesarean section is only performed at public hospitals; hence, women assessed for eligibility in this study represented the entire population of the participating councils. All participating hospitals had one neonatal department (in our facilities named the neonatal intensive care unit/NICU) and in-house obstetrician, paediatrician and anaesthesiologists available day and night, with physicians typically working in 8-, 16- or 24-hour shifts. The participating units complied with the Danish National Obstetrics Guidelines for elective caesarean section including spinal anaesthesia as the preferred analgesia and administration of prophylactic intravenous antibiotics during the procedure. A midwife handled the neonate immediately after the birth.
Paediatricians were not routinely present in the operating theatre and were only called in if the neonate presented with clinical signs warranting paediatric assistance. At all the participating hospitals, NICU admission and separation of the mother and child were avoided if possible, and all decisions related to NICU admission were on a case-to-case basis. Although the paediatricians responsible for the decision to admit the neonate to the NICU were not blinded to gestational age at delivery, only neonates with a gestational age <35 weeks were routinely admitted to the NICU. If paediatricians were called to assess the newborn, they were unaware that a trial was being conducted.
Participants were randomised using a computer-generated voice response telephone randomisation system in a permuted block design with block sizes of two, four and six in random order, stratified by previous caesarean and site.
The primary outcome was admission to the NICU, regardless of cause, within 48 hours of birth, which was assessed by information from individual hospital records. Secondary short-term neonatal outcomes were admission to the NICU within 7 days of birth, NICU length of stay, type and duration of treatment (mechanical ventilation, continuous oxygen supplementation, continuous positive airway pressure [CPAP], intravenous antibiotics, treated hypoglycaemia), Apgar score, umbilical artery pH and standard base excess values. We defined any respiratory morbidity as mechanical ventilation, continuous oxygen supplementation, or CPAP during NICU admission. Serious respiratory morbidity was defined as mechanical ventilation, or three or more initiated days of treatment with oxygen or CPAP during NICU admission. Furthermore, birthweight and gender were registered. If a neonate was transferred to another NICU, data were registered continuously throughout the entire admissions, and the total admission period was registered until discharge from the final NICU.
The following maternal outcomes were registered: maternal death, hysterectomy, or serious thromboembolic complications (deep venous thrombosis or pulmonary embolism). Surgical adverse outcomes such as injury to bowel or bladder or a uterine lateral tear were registered if they were specifically mentioned in the procedure description. Likewise, procedural complications such as adherences, difficulties in delivery (prolonged, described as difficult, use of forceps or vacuum), or the presence of total uterine rupture or a noted dehiscence (intact serosa) were registered if mentioned in the operation description. Anaesthetic complications included serious cases of hypotension or hypertension, spinal headache, or the need for general anaesthesia. A re-operation was registered if any of the following procedures were performed after delivery: evacuation, wound opened because of suspected infection or bleeding in the abdomen, or other operations such as the placement of a JJ-catheter. Treatment for excessive postoperative bleeding was registered if a blood transfusion was administered, or if uterotonic medication in addition to prophylactic treatment was instituted at the operating theatre. Total blood loss volume was defined as the volume registered in the operating theatre during the caesarean section procedure, or in vaginal deliveries until transfer from the delivery room occurred. Any antibiotic treatment and the suspected source of infection (endometritis, wound infection, mastitis, or any other infection attributed to the delivery) were registered from the patient record. The diagnosis of endometritis was based on clinical criteria with fever, elevated biochemical inflammatory markers and vaginal discharge. Finally, we constructed a composite outcome with inclusion of at least one of the following adverse outcomes: maternal death, hysterectomy, pulmonary embolism or deep venous thrombosis, uterine incision lateral tear, injury to bowel or bladder, procedural or anaesthesia complications, rupture or dehiscence, bleeding requiring treatment, additional operations, or antibiotic treatment within 30 days of delivery.
With an α (two-sided) of 0.05 and a power of 0.80, an estimated sample size of 1010 participants was calculated based on primary outcome proportions of 8% (39-week group) and 14% (38-week group), which were numbers available from a large cohort (women with diabetes and fetal malformations excluded) at the beginning of the trial in 2008. After delivery of 750 participants, approximately 9% of the participants had not been delivered within the two 5-day randomisation group intervals for reasons such as the mother's request for another delivery date, miscalculations of the gestational age, or postponing of the procedure for logistic problems in the operating theatre. To account for the higher than expected noncompliance, the steering committee recalculated the sample size, without knowledge of the results until then, and decided to increase the sample size to a total of 1270 participants.
Basic demographic data are presented with counts and percentages for categorical variables, with mean and standard deviation for continuous Gaussian variables and with median and interquartile range (IQR) for continuous non-Gaussian variables. We analysed all data according to the intention-to-treat principle.
In the outcome analyses, relative risks and absolute risk differences with 95% confidence intervals (95% CI) were calculated for dichotomous outcomes using Fisher's exact test. Non-Gaussian continuous variables were compared using the Wilcoxon rank-sum test, and birthweight was compared using Student's t test. For the primary outcome and the maternal composite outcome a potential centre effect was evaluated by testing the hypothesis of no treatment by centre interaction in a multiple logistic regression model. Adjustment for multiple comparisons (Bonferroni) was made in evaluating the secondary outcomes.
All data analyses were performed using the statistical software program stata v.11 (2009).13
From March 2009 to June 2011, a total of 1274 pregnant women were enrolled in this study (Figure 1). Seven study sites recruited between 119 and 329 participants each, representing approximately 20–34% of all eligible women. Baseline characteristics (Table 1) were similar in the two intervention groups at trial entry; however, slightly more women with two or more previous caesarean sections were randomised to the 38-week group whereas more women in the 39-week group had a maternal request caesarean. Four women who were randomised did not fulfil the inclusion criteria: two of these had no ultrasound-validated gestational age, one woman was only 17 years old, and one fetus had a planned NICU stay after delivery (39-week group). One neonate (39-week group) died from congenital pulmonary lymphangiectasia 11 days postpartum. The intention-to-treat analysis included these participants. Only data from two participants (one in each group) were not included in the analyses of neonatal outcomes. Both had induced vaginal delivery because of stillbirth.
|Elective caesarean section at 38+3 weeks||Elective caesarean section at 39+3 weeks|
|Maternal age in years, mean (SD)||32.1 (4.4)||31.6 (4.6)|
|Maternal height in cm, mean (SD)||167.7 (6.5)||167.6 (6.5)|
|Maternal pre-pregnant weight in kg, median (IQR)||68 (50–86)||68 (49–87)|
|Body mass index ≥30||107 (16.8)||111 (17.4)|
|Maternal smoking||65 (10)||83 (13)|
|Gestational age validated by ultrasound before 15 weeks||636 (100)||636 (99.7)|
|Nulliparous||126 (19.8)||117 (18.3)|
|Previous caesarean section births|
|0||253 (39.8)||255 (40.0)|
|≥1||383 (60.2)||383 (60.0)|
|Previous vaginal births|
|0||426 (67.0)||421 (66.0)|
|≥1||210 (33.0)||217 (34.0)|
|Gestational age at randomisation|
|<32+0 weeks of gestation||279 (43.9)||290 (45.5)|
|≥32+0 weeks of gestation||357 (56.1)||348 (54.5)|
|Skin closure type a|
|Suture||63/628 (10.0)||64/621 (10.3)|
|Staples||565/628 (90.0)||557/621 (89.7)|
|Indication for caesarean section b|
|Two or more previous caesarean sections||128 (20.1)||102 (16.0)|
|Maternal disease/caesarean section advised by the physician||54 (8.5)||57 (8.9)|
|Breech or transverse lie presentation||114 (17.9)||90 (14.1)|
|Previously complicated birth||78 (12.3)||83 (13.0)|
|Maternal request||262 (41.2)||306 (48.0)|
A total of 1097 participants (571 in 38-week group versus 526 in 39-week group) had an elective caesarean delivery and 156 participants (59 in 38-week group versus 97 in 39-week group) had an unscheduled caesarean (within 8 hours of the decision). Among these, 49 women in 38-week group (7.7%) and 82 women in 39-week group (12.9%) had spontaneous onset of labour. Twenty-one participants delivered vaginally (five in 38-week group versus 16 in 39-week group). Apart from one vaginal stillbirth in each group, five out of 16 vaginal deliveries among women in 39-week group occurred because of imminent delivery, whereas the remaining ten women in 39-week group and four women in 38-week group decided to deliver vaginally because of spontaneous reversion to an occipital presentation or change of preferred mode of delivery after inclusion.
Systematic registration of nonparticipants during the inclusion period showed baseline characteristics similar to participants in terms of age, body mass index, smoking habits and previous vaginal births. Eligible nonparticipants were more likely than participants to be nulliparous (27% versus 19%), with no previous caesarean section (49% versus 40%), and to have a breech or transverse lie presentation (24% versus 16%).
Primary and secondary outcomes
The median difference in gestational age at delivery was 6 days (38+3; IQR 38+1–38+5 versus 39+2; IQR 38+6–39+5). Gestational ages at delivery in the two groups are illustrated in Figures 2 and 3. There was no significant difference in the primary outcome proportion as 88 of 635 neonates (13.9%) randomised to elective caesarean section at 38 weeks of gestation were admitted to the NICU compared with 76 of 637 neonates (11.9%) at 39 weeks of gestation (Table 2). The relative risk (RR) of NICU admission was 0.86 (95% CI 0.65–1.15) for elective caesarean section at 39 weeks compared with 38 weeks, and the risk difference was − 1.9% (95% CI − 5.6 to 1.8). No significant treatment by centre interaction was found for the primary outcome (P = 0.57).
|ECS 38 weeks (n = 635)||ECS 39 weeks (n = 637)||Relative risk (95% CI)||Unadjusted P value|
|NICU admission within 48 hours||88 (13.9)||76 (11.9)||0.86 (0.65–1.15)||0.3136|
|NICU admission within 7 days of delivery||101 (15.9)||89 (14.0)||0.88 (0.68–1.14)||0.3460|
|NICU length of stay ≥2 days||39 (6.1)||38 (6.0)||0.97 (0.63–1.50)||0.9070|
|NICU length of stay in hours, median (IQR)||44.5 (42.2–46.8)||47.9 (45.6–50.2)||NA||0.6995|
|Birth weight in g, mean (SD)||3369 (449)||3509 (511)||NA||<0.0001|
|Neonate male gender||324 (51.0)||329 (51.6)||NA||0.8237|
|Apgar score at 1 minutes ≤7||20 (3.1)||24 (3.8)||1.20 (0.67–2.14)||0.6458|
|Apgar score at 5 minutes ≤7||6 (1.0)||12 (1.9)||1.99 (0.75–5.28)||0.2345|
|Umbilical artery pH at birth <7.10||5 (1.0)||6 (1.2)||1.24 (0.38–4.03)||0.7700|
|Standard base excess ≤−10||3 (0.6)||2 (0.4)||0.69 (0.12–4.13)||1.0000|
|Mechanical ventilator treatmentb||1 (0.2)||4 (0.6)||3.99 (0.45–35.6)||0.3740|
|CPAP treatment||55 (8.7)||42 (6.6)||0.76 (0.52–1.12)||0.1711|
|CPAP treatment ≥1 day||17 (2.7)||9 (1.4)||0.53 (0.24–1.18)||0.1178|
|Continuous oxygen supplementation||23 (3.6)||14 (2.2)||0.61 (0.32–1.17)||0.1307|
|Continuous oxygen supplementation ≥1 day||13 (2.0)||4 (0.6)||0.31 (0.10–0.94)||0.0296c|
|Any respiratory morbidityd||57 (9.0)||43 (6.8)||0.75 (0.51–1.10)||0.1459|
|Serious respiratory morbiditye||11 (1.7)||9 (1.4)||0.82 (0.34–1.95)||0.6607|
|Intravenous antibiotic treatment||17 (2.7)||14 (2.2)||0.82 (0.41–1.65)||0.5916|
|Intravenous antibiotic treatment ≥5 days||7 (1.1)||5 (0.8)||0.71 (0.23–2.23)||0.5788|
|Treated hypoglycaemia||5 (0.8)||5 (0.8)||1.0 (0.29–3.43)||1.0000|
With respect to secondary neonatal outcomes (Table 2), duration of treatment with continuous oxygen for more than 1 day was significantly less frequent in the 39-week group, with an RR of 0.31 (95% CI 0.10–0.94) and a risk difference of − 1.4% (95% CI −2.7 to −0.2). After adjustment for multiple comparisons, this association was no longer significant (adjusted level of significance 0.003). There were no significant differences in the frequency of NICU admission within 7 days of birth, respiratory morbidity, or treatment with intravenous antibiotics. Similar proportions were found in the two groups with respect to NICU length of stay of ≥2 days (Figure 4).
Among the neonates who did not receive respiratory support (e.g. mechanical ventilation, nasal CPAP, oxygen supplementation, or intravenous antibiotics), 12 neonates in the 38-week group versus 11 in the 39-week group had transitory tachypnoea or respiratory distress as their primary diagnosis. Apart from prematurity (five neonates in the 38-week group versus three in the 39-week group), the remaining reasons for NICU admission comprised a broad range of categories in both groups, such as asphyxia (one versus two), hypoglycaemia (three versus none), or various suspected endocrine or haematological disorders. Two neonates in the 39-week group compared with none in the 38-week group were admitted because of congenital malformations detected at birth.
Participants were defined as compliant if elective caesarean section was performed within the randomisation group dates or at any other date because of labour or complications in pregnancy, as any rescheduling among the last-mentioned groups of women would be based strictly on professional criteria. Accordingly, we performed a per protocol analysis with exclusion of 117 non-compliant and four not eligible participants and found an RR of the primary outcome of 0.90 (95% CI 0.67–1.21). Finally, if vaginal deliveries were excluded from the analysis, the RR was 0.92 (95% CI 0.68–1.23).
With respect to secondary maternal outcomes no cases of maternal death, hysterectomy, or serious thromboembolic events were registered (Table 3). The risk of maternal bleeding >500 ml was significantly lower in the 39-week group (17.1%) compared with the 38-week group (21.7%) with an RR of 0.79 (95% CI 0.63–0.99), but this finding was insignificant after Bonferroni adjustment. The median bleeding volume was 300 ml in each group. The number of maternal adverse events was low in the majority of the complications evaluated. We registered three versus one case of uterine rupture or dehiscence (RR 3.04, 95% CI 0.31–29.13) and one versus four cases of injury to the bowel or bladder (RR 0.25; 95% CI 0.03–2.26) in the 39-week group versus the 38-week group, respectively. We also found a similar risk of uterine incision lateral tear, procedural difficulties, and bleeding needing further treatment in the two groups. The risk of a composite maternal outcome was similar in the two groups with proportions of 66/638 women (10.3%) in the 39-week group and 60/636 women (9.4%) in the 38-week group (RR 1.1; 95% CI 0.79–1.53). No significant treatment by centre interaction was found for the maternal composite outcome (P = 0.81).
|ECS 38 weeks (n = 636)||ECS 39 weeks (n = 638)||Relative risk (95% CI)||Unadjusted P value|
|Uterine incision lateral tear||17 (2.7)||16 (2.5)||0.94 (0.48–1.84)||0.8622|
|Bowel or bladder injury||4 (0.6)||1 (0.2)||0.25 (0.03–2.22)||0.2173|
|Adherences, procedural or delivery complications||14 (2.2)||16 (2.5)||1.14 (0.56–2.31)||0.8538|
|Anaesthesia complications||2 (0.3)||7 (1.1)||3.49 (0.73–16.7)||0.1781|
|Rupture or dehiscence||1 (0.2)||3 (0.5)||3.0 (0.31– 28.7)||0.6244|
|Bleeding needing treatment or transfusion||9 (1.4)||12 (1.9)||1.33 (0.56–3.13)||0.6610|
|Reoperationa||10 (1.6)||6 (0.9)||0.60 (0.22–1.64)||0.3286|
|Antibiotic treatment within 30 days of deliveryb||15 (2.4)||14 (2.2)||0.93 (0.45–1.91)||0.8537|
|Blood loss at delivery >500 ml||138 (21.7)||109 (17.1)||0.79 (0.63–0.99)||0.0399c|
|Composite maternal outcomed||60 (9.4)||66 (10.3)||1.1 (0.79–1.53)||0.6391|
In this randomised controlled trial including 1274 pregnant women, scheduling elective caesarean section at 39+3 weeks compared with 38+3 weeks did not result in a significant decrease in neonatal admission within 2 days of birth. In addition, no secondary neonatal or maternal outcomes improved significantly with late scheduling.
Strengths and weaknesses of the study
The major strength of this trial was the randomised design, which has, to the best of our knowledge, not previously been used for investigating the association between gestational age at elective caesarean section and adverse perinatal outcomes. With an intention-to-treat analysis, the data presented reflect the whole spectrum of consequences of timing of elective caesarean from booking of the procedure to birth of the child. The trial population was homogeneous, with approximately 95% being of white European descent which in this case should be considered a strength, as neonates delivered by White women in particular may be vulnerable to performing the elective caesarean section before rather than after 39 completed weeks of gestation. We had a 100% short-term follow-up rate, which among other circumstances was related to the fact that if a neonate required admission within 7 days postpartum, the neonate was admitted to the NICU in the same hospital as the delivery. Furthermore, a major advantage compared with previous studies was a due date validated by ultrasound in 99.8% of the participants.
With this trial we wanted to be able to assess the balance of potential adverse events or benefits of elective caesarean section timing to the neonate and mother according to the two timing interventions. Hence, NICU admission that reflects a composite outcome as the primary outcome should be considered an advantage over more specific measures. Although the decision to admit a neonate relies on the physician, this decision is taken on the basis of objective symptoms or signs necessitating further observation or treatment, and no neonate would be hospitalised without a clear indication. At the same time, admission has implications for the families involved, because it may lead to separation of mother and child, anxiety and worries.
Some limitations to this study may apply. In terms of serious events such as stillbirth, hysterectomy, thromboembolism, or death, the study was not powered to evaluate the influence of elective caesarean section timing. In a recent study, Rosenstein et al. found a similar risk of fetal and neonatal death associated with delivery and expectant management at 38 weeks, whereas at 39 weeks, delivery carried a lower risk of neonatal death than expectant management. Hence, large observational studies would be of value in this field. Another limitation to the current study may be that the outcomes evaluated were all short-term. In fact, delivery at 38 weeks may lead to increased healthcare use in early childhood or increased risk of special educational school needs compared with delivery at 39 weeks of gestation, but this association has not been investigated in elective deliveries only.[16, 17]
Before applying the results to other populations and settings, several factors have to be considered. Overall, the study population had a low body mass index and was very homogeneous, and the study aimed to only include healthy women with no a priori risks related to the neonate. In addition, the eligible non-participating women may have had a lower risk of an adverse neonatal outcome, because they were probably booked closer to their due date (more breech presentations) and fewer women were to have repeat procedures.
The NICU admission rate was quite high compared with those of other studies.[8, 9] This may be a consequence of all neonatal admissions being included, whereas studies conducted in facilities with differentiated levels of neonatal care may have excluded neonatal admissions that were not to the NICU. Comparing the incidence of a composite adverse neonatal outcome after caesarean delivery at 38 and 39 weeks of gestation, Tita et al. [8, 9] found an odds ratio of 0.67 and Wilmink et al. found an odds ratio of 0.71 in favour of delivery at 39 weeks of gestation, which are both estimates that lie within our calculated 95% CI of 0.65–1.15. Our primary outcome may compare with these composites, which in our setting in each case would probably translate into NICU admission. Some of the excess morbidity in our trial may be explained by prematurity or unscheduled procedures because of complications or labour. Still, these studies reported significant benefit of elective caesarean section at 39+0–6 weeks compared with 1 week earlier, in opposition to the smaller (insignificant) differences observed in our trial.[8, 9]
In terms of maternal morbidity, the most interesting aspect before the trial was whether more women with spontaneous onset of labour in the late ECS group would result in an increase in the number of women with infection or intraoperative complications because of the urgency of the procedure. Apart from maternal intensive care admission not being evaluated in our sample, the individual adverse outcomes otherwise assessed were comparable with the composite outcome of Tita et al. We found a higher proportion of composite maternal outcome in both groups (9.4% versus Tita et al.  7.5% at 38 weeks and 10.3% versus Tita et al. 6.6% at 39 weeks), but similarly, no difference was found between the 2 weeks of gestation. This may reflect the fact that unscheduled caesarean sections were included in our population.
When this study was initiated, a substantial number of elective caesarean sections in Denmark were scheduled before 39 weeks of gestation. Recently published literature on the timing of elective caesarean section almost uniformly advises against elective caesarean section before 39 completed weeks.[4, 8, 9, 16-22] The risk estimates from this trial are currently the only estimates available from a randomised trial, and they could be interpreted in different ways. If the power calculations overestimated the difference in NICU admission between the two groups, this study would be underpowered to detect a statistically significant difference. Assuming a valid power calculation, our study has an 80% chance that the findings were correct. We predicted (and found) that 14% of the 38-week group neonates would be admitted to the NICU, whereas the risk of NICU admission in the 39-week group was almost 50% higher than expected. This difference could be explained by an increased risk of an adverse outcome in women who were delivered before their scheduled date because of complications or labour. Interestingly, the proportion of neonates with an NICU length of stay >2 days was almost identical in the two groups. This may indicate that even if more neonates were admitted with elective caesarean section at 38 weeks, there was a similar risk of severe neonatal morbidity in the two groups.
Our results could suggest that scheduling elective caesarean section after the prevailing cut-off at 39+0 weeks of gestation may have less impact on short-term neonatal morbidity than previously anticipated. In the mothers, timing of elective caesarean section at 38 or 39 weeks seems to carry a similar risk of adverse events. Accordingly, scheduling of elective caesarean section 3–5 days before 39+0 completed weeks may be an acceptable option in women in whom an acute caesarean section delivery should be avoided. Otherwise, scheduling after 39 completed weeks may continue to be the best option, until further trials support the evidence from this study, and possible long-term consequences of elective caesarean section timing are clarified.
In conclusion, elective caesarean section scheduled after 39 weeks of gestation versus before 39 weeks carried a similar risk of neonatal special care admission. The incidence of other adverse neonatal and maternal outcomes was similar in the two groups. Any long-term consequences to the neonate or mother of elective caesarean timing at term are uncertain and need further investigation.
Disclosure of interests
The authors have no conflicts of interest.
Contribution to authorship
JG, SFK, NU and TBH had the idea for the study and were responsible for the data analysis. JG was the principle investigator and wrote the first draft of the protocol and the manuscript. JG, SFK, NU and TBH contributed to the gathering of data from the University Hospital of Aarhus and were responsible for the overall supervision of the trial management. MK was the responsible investigator at the Hospital in Kolding. AMM was the responsible investigator at Aalborg University Hospital. BBM was the responsible investigator at the Regional Hospital of Viborg. OBR was the responsible investigator at the Regional Hospital of Randers. JTC was the responsible investigator at the Regional Hospital of Herning. JSJ was the responsible investigator at Odense University Hospital. Guarantors for the study were JG, NU and TBH. All authors had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors reviewed and approved the final version of the paper.
Details of ethics approval
The study was approved by The Central Denmark Region Committees on Biomedical Research (ID M-20080142).
The study was supported by grants from The Danish Council for Independent Research Medical sciences, Health Research Fund of Central Denmark Region, The Philanthropic Foundation TrygFonden, Aase and Ejnar Danielsen's Foundation, Sophus Jacobsen's Foundation, Dagmar Marshall's Foundation, Marie Dorthea and Holger From, Haderslev's Foundation. The funders of the study had no role in study design, data collection and analysis, or preparation of the manuscript.
We wish to thank all the families who participated in this trial, the dedicated staff from all participating departments, and our trial steering committee members.
- 1NICE. Caesarean section. 2011. [http://guidance.nice.org.uk/CG132]. Accessed 15 February 2012.
- 2American College of Obstetricians and Gynecologists. Cesarean delivery on maternal request. Committee opinion no. 559. Obstet Gynecol 2013;121:904–7.
- 12EpiData Entry (3.1). A comprehensive tool for validated entry and documentation of data. Odense, Denmark: The EpiData Association; 2012., .
- 13StataCorp. Stata Statistical Software: Release 11. College Station, TX, USA: StataCorp LP; 2009.