Dr MK Dhanjal, Women’s and Children’s Services, Imperial College Healthcare NHS Trust, Queen Charlotte’s and Chelsea Hospital, Du Cane Road, London W12 0HS, UK. Email email@example.com
Objective To estimate the incidence of multiple repeat caesarean section (MRCS) (five or more) in the UK and to describe the outcomes for women and their babies relative to women having fewer repeat caesarean sections.
Design A national population-based prospective cohort study using the UK Obstetric Surveillance System (UKOSS).
Setting All UK hospitals with consultant-led maternity units.
Population Ninety-four women having their fifth or greater MRCS between January 2009 and December 2009, and 175 comparison women having their second to fourth caesarean section.
Methods Prospective cohort and comparison identification through the UKOSS monthly mailing system.
Main outcome measures Incidence, maternal and neonatal complications. Relative risk, unadjusted (OR) and adjusted (aOR) odds ratio estimates.
Results The estimated UK incidence of MRCS was 1.20 per 10 000 maternities [95% confidence interval (CI), 0.97–1.47]. Women with MRCS had significantly more major obstetric haemorrhages (>1500 ml) (aOR, 18.6; 95% CI, 3.89–88.8), visceral damage (aOR, 17.6; 95% CI, 1.85–167.1) and critical care admissions (aOR, 15.5; 95% CI, 3.16–76.0), than women with lower order repeat caesarean sections. These risks were greatest in the 18% of women with MRCS who also had placenta praevia or accreta. Neonates of mothers having MRCS were significantly more likely to be born prior to 37 weeks of gestation (OR, 6.15; 95% CI, 2.56–15.78) and therefore had higher rates of complications and admissions.
Conclusions MRCS is associated with greater maternal and neonatal morbidity than fewer caesarean sections. The associated maternal morbidity is largely secondary to placenta praevia and accreta, whereas higher rates of preterm delivery are most likely a response to antepartum haemorrhage.
The incidence of primary caesarean section is rising throughout the world and the UK also demonstrates this trend.1 This is thought to be a result of the introduction of fetal monitoring in labour,2 maternal preference, maternal obesity and, possibly, defensive obstetric practice. In addition, the National Institute of Health and Clinical Excellence (NICE) in the UK recently advised that delivery via caesarean section without a medical indication is acceptable within the National Health Service.3 It remains to be seen how this will impact on primary and, eventually, repeat caesarean rates. UK guidelines recommend that, after having three caesarean sections, women should be advised to undergo repeated elective caesarean section in any subsequent pregnancies, rather than attempt a vaginal delivery.4 This avoidance of vaginal delivery is thought to reduce the risk of uterine rupture which can be life-threatening for both the mother and baby. All caesarean section procedures have other associated risks, including infection, haemorrhage and thrombosis, which are leading causes of maternal mortality,5 and damage to the bladder, ureters or bowel. Repeated caesarean sections are also associated with placenta praevia, placental invasion into the myometrium and peripartum hysterectomy.6 Babies born via caesarean section are more likely to experience breathing difficulties7 and to require admission to a specialist neonatal unit.
Data on the management of women having repeated caesarean sections and the maternal–fetal outcomes have previously focused on women undergoing up to their fourth caesarean section. Higher order caesarean section outcome data have come mainly from hospital-based retrospective case analyses outside of the UK.8–11 One cohort study12 examined the outcomes of caesarean section relative to the number of previous caesareans. This study only included elective caesarean sections undertaken in selected tertiary units and, as such, may represent a biased sample. Complication rates are variously reported as not significantly different from those with fewer previous caesarean sections, or increased.
No population-wide prospective incidence studies of caesarean order have been undertaken previously and the UK incidence is unknown.
The aim of this study was to use the UK Obstetric Surveillance System (UKOSS)13 to identify all women in the UK undergoing multiple repeat caesarean section (MRCS), defined as the fifth order or higher caesarean section. This article describes the reported cases, management and outcomes for both women and babies, and draws comparisons with women delivering in the same units with fewer previous caesareans.
A cohort study was undertaken. It was not possible to accurately predict the number of cases that would be reported because of the limited nature of existing data; we therefore planned to undertake a study over 1 year and reviewed the power of the study before ceasing data collection. We identified 94 cohort women and 175 comparison women; a study of this size would be able to detect odds ratios between 2.9 and 8.3 with 80% power at the 5% level of statistical significance, for a prevalence range of the potential risk factors of between 1% and 13%.
The outcomes of the study were defined as postpartum haemorrhage (PPH), in which the blood loss exceeds 500 ml, major obstetric haemorrhage (MOH), in which the blood loss exceeds 1500 ml, major or minor placenta praevia, in which the placenta covers the internal os or implants into the lower uterine segment, respectively, placenta accreta, in which the placenta penetrates through the decidua basalis, and preterm delivery, when delivery occurs prior to 37 weeks of gestation.
The cohort was defined as all women in the UK giving birth between January 2009 and December 2009 via elective or emergency caesarean section who had previously undergone four or more other caesarean section procedures, i.e. this was their fifth or greater caesarean section.
Comparison group identification
Comparison women were obtained from the same hospitals as those undergoing MRCS, matched for date and time of birth and intention to deliver via elective caesarean section. The clinician reporting each cohort woman was asked to supply data for two comparison women who delivered by caesarean section immediately before the case and who had one to three previous caesarean deliveries. Identical data were collected from both women having multiple repeat caesareans and those having lower order caesareans.
We identified women in our cohort who were also diagnosed with either placenta praevia or placenta accreta, and compared the obstetric outcomes in this subgroup with outcomes in the comparison group. We also compared outcomes in the remaining cohort women (i.e. women who had MRCS but did not have either placenta praevia or placenta accreta) with outcomes in the comparison group.
We identified cases on a national basis through the monthly mailings of the UKOSS between January 2009 and January 2010. The methods have been described in detail elsewhere.13 In brief, nominated clinicians reported cohort women through a monthly card mailing system. When a card was received indicating that there had been a woman with MRCS delivered in the unit, the reporting clinician was sent a data collection form requesting further details of potential risk factors, management and outcomes. The reporting clinician was also asked to identify two comparison women for each case and to complete data collection forms for each. Data collection sheets for cohort and comparison women were checked to confirm that they met the cohort and comparison identification criteria. Duplicate reports were identified by comparing the woman’s year of birth, hospital and expected date of delivery. All data requested were anonymous. Up to five reminders were sent if completed forms were not returned.
Incidence rates with 95% confidence intervals (CIs) were calculated using the available national maternity data for 2009. Relative risks (RRs) were calculated with 95% CIs using the ratio of proportions between cohort and comparison women. Odds ratios (ORs) and adjusted odds ratios (aORs) were estimated throughout using unconditional logistic regression. To allow for the nonindependence of infants from multiple births, models included an option to specify that the calculated standard errors allow for within-group correlation. All analyses were carried out using STATA version 11 software (StataCorp LP, USA).
Two hundred and twenty-six UK hospitals with consultant-led maternity units contributed data to UKOSS, giving 100% participation. Data collection was complete for 87% of cases. Ninety-four confirmed cases of MRCS were reported through UKOSS in a reported 781 377 maternities, giving an incidence of 1.20 per 10 000 maternities (95% CI, 0.97–1.47) (Figure 1).
Data on comparison women were not received for three women in the cohort, and information on only one comparison woman was received for a further three cohort women. Therefore, 175 comparison women in total were identified. The frequency of the number of previous caesarean deliveries within each group is described in Table 1.
Table 1. Number of previous caesarean sections (CSs) in comparison and multiple repeat caesarean section (MRCS) cohorts
Number of previous CSs
Comparison group (n = 175)
MRCS cohort group (n = 94)
Full data on the index pregnancy were not available for two of the 94 women and we were therefore unable to include them any further in our analysis.
Four women in the MRCS group and four women in the comparison group had twin pregnancies. Therefore, 96 babies were born to mothers in the MRCS group and 179 babies to mothers in the comparison group.
Women in the MRCS group were significantly older and more likely to smoke during their pregnancy than those in the lower order caesarean group (see Table 2). There were no significant differences in the rates of obesity [cohort median body mass index (BMI) of 28.4 kg/m2 versus comparison median BMI of 27.3 kg/m2], professional/managerial occupational status or nonwhite ethnicity between the groups. The ethnicities of the MRCS group were as follows: white European (66%), black (15%), Asian (13%), unknown (5%); in the comparison group, the ethnicities were as follows: white European (69%), black (11%), Asian (18%) and unknown (5%).
Table 2. Demographic information, maternal complications and gestation at delivery for multiple repeat caesarean section (MRCS) and comparison groups
MRCS (n = 75)
Comparison group (n = 175)
RR (95% CI)
CI, confidence interval; RR, relative risk.
*Pre-eclampsia, postpartum haemorrhage, premature delivery, obstetric cholestasis and gestational diabetes.
**Asthma, epilepsy, HIV, diabetes and alcoholism.
***Pre-eclampsia, gestational diabetes, fetal growth restriction and vaginal bleeding.
Age >35 years
BMI > 30 kg/m2
Previous obstetric complications*
Past medical history**
Current obstetric complications***
Gestation at delivery
Table 2 outlines the rates of complications in the past obstetric and medical history, and also in the index pregnancy.
In the MRCS cohort group, 17 (18%) women had previous additional uterine surgery including surgical termination of pregnancy, evacuation of retained products of conception, and dilatation and curettage, relative to 15 (9%) women in the comparison group (RR, 1.66; 95% CI, 1.14–2.42). None of the women had had a previous myomectomy or known uterine perforation.
The median gestational age at delivery was 38 weeks (range, 24–43 weeks) in the MRCS group and 39 weeks (range, 26–41 weeks) in the comparison group (P < 0.00001) (see Table 2). The caesarean section was an emergency procedure in 13 (14%) women in the MRCS group and in nine (5%) women in the comparison group. A classical uterine incision was performed in four (4%) women in the MRCS group and in one (1%) woman in the comparison group (RR, 7.61; 95% CI, 0.86–67.1). More senior obstetricians and anaesthetists were involved in the delivery of the women having MRCS (RR, 2.75; 95% CI, 2.02–3.74) (Table 3).
Table 3. Seniority of obstetric and anaesthetic staff performing caesarean section
MRCS, multiple repeat caesarean section.
*Trainee with 3–7 years of experience in obstetrics and gynaecology.
**Trainee with 1–2 years of experience in obstetrics and gynaecology.
***Data missing in two MRCS and two comparison women.
Grade of obstetrician performing delivery
n = 92
n = 175
Grade of anaesthetist involved in delivery
n = 90***
n = 173***
Placenta praevia was diagnosed in 15 (16%) women undergoing MRCS (eight major, seven minor praevia) compared with three (2%) women in the comparison group (all major praevia). Two of the women with minor placenta praevia were diagnosed intra-operatively. Abnormal placental invasion occurred in 13 (14%) women in the MRCS cohort, but no women in the comparison cohort. Eight of these cases were diagnosed antenatally, six by ultrasound scan and two by magnetic resonance imaging.
The relationship between placenta praevia and placental invasion in the cohort group is summarised in Figure 2.
Maternal complications from delivery were significantly higher in women with MRCS, even after adjusting for differences in maternal characteristics between the groups (Table 4). Women in the MRCS group experienced PPH more frequently than women in the comparison group: 28 (30%) compared with 23 (13%) (aOR, 2.71; 95% CI, 1.35–5.44). The median estimated blood loss in women with PPH was 1550 ml (range, 500 ml–10.5 l) in the MRCS group, compared with 800 ml (range, 500–1500 ml) in the comparison group (P = 0.004). Seventeen (18%) women in the cohort group had an MOH, compared with only one (0.6%) woman in the comparison group (aOR, 18.6; 95% CI, 3.89–88.8). Sixteen (17%) women in the MRCS cohort group received a blood transfusion, whereas only two (1%) women in the comparison group were transfused (aOR, 25.3; 95% CI, 4.94–129.2). Women in the cohort group received a median of 7 units of packed cells (range, 1–24 units), whereas women in the comparison group received 2 and 3 units of packed cells, respectively. Twelve women in the cohort group required additional blood products, including fresh frozen plasma, platelets and cryoprecipitate.
**Three women required intubation and ventilation.
***Adjusted for maternal age, smoking status.
Massive obstetric haemorrhage (blood loss ≥1500 ml)
Damage to viscera*
Critical care admission
Nine (10%) women in the MRCS cohort group received further surgical or medical treatment to control their bleeding. The methods used included hysterectomy in eight (9%) women, intrauterine balloon catheter in two (2%) women, uterine packing in two (2%) women and the administration of factor VIIa in one (1%) woman. One (0.6%) woman in the comparison group required a B-Lynch suture. The incidence of uterine dehiscence or rupture was three (3%) in the MRCS cohort and two (1%) in the comparison group, which was not statistically significantly different (aOR, 5.72; 95% CI, 0.82–40.2). There were no maternal deaths in either group.
We performed a subgroup analysis of women in the MRCS cohort, comparing those MRCS women with placenta praevia and/or placenta accreta and those MRCS women with a normally sited, nonmorbidly adherent placenta, with the comparison group. Massive obstetric haemorrhage, the use of blood transfusion or critical care admission occurred in the majority of MRCS women with placenta praevia and/or placenta accreta. Notably, all eight women who had a hysterectomy were from this subgroup. However, women with MRCS and no placental complications were still more likely to have an MOH compared with women with fewer previous caesarean sections (RR, 2.56; 95% CI, 1.41–4.66), but did not have an increased requirement for blood transfusion or critical care (see Supporting Information in Table S1).
Forty-five (49%) women in the MRCS cohort underwent concurrent elective surgical sterilisation. Twelve (7%) women in the comparison group were sterilised. Future contraception or sterilisation was discussed with five (13%) of the women in the cohort group who had retained their fertility. The potential risks of further surgery were specifically documented in the notes of only one of the women undergoing their fifth or greater caesarean section.
Ninety-six babies were born to mothers in the MRCS cohort and 179 babies to mothers in the comparison group. One baby from each group was stillborn (P = 0.65), giving perinatal mortality rates of 10 per 1000 (95% CI, 2.75–590) in the cohort group and six per 1000 (95% CI, 1.45–314) in the comparison group. There were no neonatal deaths. There were significantly more children born prior to 37 weeks of gestation in the MRCS cohort: 23 of 96 (24%) compared with nine of 179 (5%) in the comparison group (OR, 6.15; 95% CI, 2.56–15.78). Neonatal admissions and complication rates for intraventricular haemorrhage, severe jaundice, severe infection and hypoxic ischaemic encephalopathy were higher in the MRCS cohort, although this was not statistically significant when adjusted for gestational age at delivery (Table 5).
*Intraventricular haemorrhage, severe jaundice, severe infection, hypoxic ischaemic encephalopathy.
**Adjusted for gestational age at birth.
Number of babies admitted to NNU
Length of NNU admission, days (range)
Significant medical complication*
Our data estimate a UK incidence of MRCS of 1.2 per 10 000 maternities (95% CI, 0.97–1.47). This equates to one case every 2 years in a unit delivering 5000 women. Although an unusual occurrence, MRCSs are not confined to tertiary centres and many district general hospitals manage such women.
The main findings of this study are that women undergoing their fifth or subsequent caesarean section have a significantly higher risk of maternal complications and preterm delivery than women having fewer caesarean sections, although, for most women, the outcomes are very good. We have quantified these risks and ascertained that the majority of maternal and neonatal complications occur in the subgroup of women who also have placenta praevia and/or placenta accreta. The majority of these women will have an MOH requiring a blood transfusion, nearly 50% will have a hysterectomy and two-thirds will require critical care. Obstetricians should counsel such women about the risks and coordinate multidisciplinary input from colleagues in anaesthetics, radiology, haematology and neonatology in order to optimise their management. This high-risk subgroup of women should ideally be delivered in a tertiary setting with access to cell salvage and interventional radiology. In the absence of placenta praevia and placenta accreta, there appears to be no greater risk of blood transfusion or critical care admission than that of lower order repeat caesarean sections.
Our study found a significant increase in the rates of placenta praevia (16%), placenta accreta (14%), MOH (18%), blood transfusion (17%), hysterectomy (9%), visceral damage (5%) and critical care admission (13%) in women undergoing MRCS compared with women having caesarean sections after fewer previous procedures. Other studies of women undergoing their fifth or greater caesarean section describe lower rates of placenta praevia (between 2% and 4%),9,12,14 placenta accreta (between 1% and 3.5%)9,12 and hysterectomy (between 0.8% and 5%).9,12,14 These differences may be accounted for by our study being the only prospective population, rather than a hospital-based cohort, and therefore encompassing deliveries of all complexities occurring in local and tertiary units. However, a limitation of this study is the relatively small cohort size, which may also account for discrepancies in comparisons with other studies. We did not identify an increase in uterine dehiscence or rupture in women having MRCSs, which is in keeping with previously published results.12
Pre-pregnancy counselling can inform women that there is an 18% chance of having a placental abnormality, but it is only in a subsequent pregnancy that placental location can be defined and more accurate complication rates discussed. Our results describe a strong association between placenta praevia and placental invasion; 73% of women with placenta praevia in our cohort also had placenta accreta. At present, making an antenatal diagnosis of placenta accreta is not reliable and a pragmatic approach for clinicians may be to regard any woman undergoing MRCS with an anterior placenta praevia as having a placenta accreta unless otherwise demonstrated. In our study, five cases (38%) were missed antenatally and diagnosed intra-operatively. These women all had ultrasound scanning of their placentas. Ultrasound scanning is available in all UK hospitals and is widely used for the diagnosis of placental invasion. However, its sensitivity is operator dependent and reduced in women with a high BMI. Recently, numerous investigators have assessed the role of MRI in the diagnosis of placenta accreta, and it has been suggested that it should be used in conjunction with ultrasound.15,16 Future studies establishing the sensitivity and specificity of MRI in the diagnosis of placental invasion may provide more guidance.
Our data describe an increased risk of delivery prior to 37 weeks of gestation in women having MRCS. We also observed higher rates of neonatal complications and admissions, although these were not statistically significant when adjusted for gestational age at delivery. Preterm deliveries occurred substantially in pregnancies complicated by placenta praevia or accreta (see Table S1), and we hypothesise that this may represent an iatrogenic response to (and fear of) antepartum haemorrhage in this very high-risk subgroup of women. Obstetricians may be anxious to deliver such cases electively in ‘office hours’ with senior staff available, even at preterm gestations. Women in this subgroup should be counselled that early delivery is a possible consequence of placental abnormalities, and that this may have implications for their baby.17
We identified that all women with MRCS were delivered by obstetricians and anaesthetists of ST3 grade or above (with at least 2 years of previous experience in obstetrics and gynaecology), and over three-quarters had consultants or associate specialists in obstetrics and anaesthesia involved in their delivery.
There are important economic implications that have emerged from this study. In women undergoing MRCS, there is a higher incidence of haemorrhage, blood transfusion and admission to a critical care setting. Although this study did not quantify the economic burden of MRCS, this would be an interesting area for further analysis.
The decision to collect data on women undergoing their fifth or more caesarean was pragmatic and enabled us to utilise the UKOSS methodology for rare case collection and provide robust data on the potential impact of an increasing number of repeat caesarean procedures. MRCS is an unusual procedure and the numbers are consequently small, which limits the power of our data. The comparison group of women undergoing fewer repeat caesareans was selected to provide a cohort which was relevant and meaningful to clinicians and women. It mainly comprised women undergoing their second caesarean, but also included those undergoing their third or fourth procedure; the group is thus heterogeneous.
This study shows that women having multiple repeat caesarean deliveries have an increased risk of visceral damage, MOH and hysterectomy, and that their babies are more likely to be delivered prematurely, compared with women having fewer caesarean sections. It does not provide evidence to make a recommendation to women on the maximum number of caesarean sections which should be performed.
The authors wish to thank the UK Obstetric Surveillance System (UKOSS) Steering Committee for their guidance on this study’s design and analysis.
Disclosure of interests
No conflicts of interest have been declared.
Contribution to authorship
The study arose from an original idea from MKD. All authors contributed to the study’s design. JRC wrote the first draft, and MK and MKD advised on the analysis. All authors contributed to the discussion and conclusion.
Details of ethics approval
The UK Obstetric Surveillance System (UKOSS) general methodology (04/MRE02/45) and this study (08/H0718/72) were approved by the London Multicentre Research Ethics Committee.
This study received funding from the Royal College of Obstetricians and Gynaecologists/UK Obstetric Surveillance System (RCOG/UKOSS) annual award 2008 from the Edgar Research Fellowship Fund.