To compare single- with double-layer closure of the uterus for the risk of uterine rupture in women attempting vaginal birth after one prior caesarean delivery.
To compare single- with double-layer closure of the uterus for the risk of uterine rupture in women attempting vaginal birth after one prior caesarean delivery.
From a total of 19 604 nulliparous women delivered by caesarean section in the years 2001–2007, 7683 women attempting vaginal birth in their second delivery were analysed.
Data from population-based registers were linked to hospital-based registers that held data from maternity and delivery records. Logistic regression was used to estimate the risk of uterine rupture after single- or double-layer closure of the uterus. Results are presented as odds ratios (ORs) with 95% confidence intervals (95% CIs).
Uterine rupture during labour occurred in 103 (1.3%) women. There was no increased risk of uterine rupture when single- was compared with double-layer closure of the uterus (OR 1.17; 95% CI 0.78–1.76). Maternal factors associated with uterine rupture were: age ≥35 years and height ≤160 cm. Factors from the first delivery associated with uterine rupture in a subsequent delivery were: infection and giving birth to an infant large for gestational age. Risk factors from the second delivery were induction of labour, use of epidural analgesia, and a birthweight of ≥4500 g.
There was no significant difference in the rate of uterine rupture when single-layer closure was compared with double -layer closure of the uterus.
Caesarean delivery is a common surgical procedure in modern obstetric care, and worldwide the number of caesarean deliveries is increasing. A scarred uterus from a previous caesarean section is the most important risk factor for uterine rupture in a subsequent delivery. The rate of uterine rupture in a trial of labour after one previous caesarean section has been reported to be 1.1% in a Swedish setting, although lower rates have been found in other studies.[4, 5] Uterine rupture is a severe complication, not only putting the mother but also the infant in danger.
The risk of uterine rupture in a trial of vaginal delivery after previous caesarean section may be dependent on the surgical technique used for uterine closure, i.e. closure with single or double layers of sutures. Traditionally the double-layer technique was used, but in the early 1990s the single-layer technique was introduced and was found to reduce pre- and postoperative morbidity for women. Uterine closure has been studied in respect to uterine rupture in a subsequent delivery, and results are conflicting about whether single-layer compared with double-layer closure increases the risk of uterine rupture, and odds ratios vary between 0.49 and 8.01.[7-10] Most studies are underpowered, however, and are retrospective.
According to recent randomised trials there is no difference in short-term complications for the woman with either technique, but results from long-term studies are lacking.[11, 12] Concern persists that single-layer closure of the uterus may increase the risk of uterine rupture, and more information is needed on the most appropriate surgical technique. Apart from the uterine closure technique used, a variety of factors, such as maternal age,[13, 14] pregnancy interval,[5, 15] infection related to previous labour, induction of labour,[5, 13, 17-20] and birthweight,[9, 14] have all been related to uterine rupture.
The primary aim of this study was to investigate whether single-layer compared with double-layer closure of the uterus increases the risk of uterine rupture in a subsequent labour after one previous caesarean section. The second aim was to identify other risk factors for uterine rupture.
From a cohort including 19 604 women who were delivered by caesarean section in their first delivery, 7683 attempted labour in a second pregnancy and comprised the study population. Eligible for the study were nulliparous women having their first delivery in a maternity unit that used the data record system Obstetrix® (Siemens Health Care). Inclusion started from the year 2001 when a data template with checkboxes for the reporting of caesarean section procedure was introduced. Twenty-three out of 46 maternity units in Sweden were using Obstetrix® and comprised the centres for recruitment. The included maternity units were university hospitals (six out of nine), county hospitals (eight out of 20), and district hospitals (nine out of 17), and managed approximately half of the deliveries in Sweden. The Regional Ethical Board in Umeå, Sweden, approved the study.
Data in the Obstetrix® database were linked with the Swedish Medical Birth Register (MBR), the Patient Register, and the Cause of Death register, with the aid of each woman's unique personal identity number. The research database thus comprised information from the women's first delivery during the years 2001–2007, and second delivery between 2001 and 2009, from Obstetrix® data records, and detailed information about the pregnancies, deliveries, and neonatal outcome from the National Health Registers. The study population had their second delivery in 42 different maternity units.
Cases with uterine rupture in the second delivery were identified in the MBR by the diagnosis codes O71.0 or O71.1, according to the tenth revision of the International Classification of Diseases. For identified cases with uterine rupture, full surgical reports were obtained from the maternity departments. The flowchart in Figure 1 describes the study population. Of 19 604 women, 12 369 had a second delivery during the study period. Patients with a pre-labour caesarean section (n = 4675) or having a corporal incision (n = 11) were excluded.
The MBR provided data about maternal age, categorised into <35 and ≥35 years, and maternal height, categorised into ≤160 and >160 cm. A height of 161 cm corresponds to the 25% percentile of women in Sweden. The body mass index (BMI) is calculated from height and weight (kg/m2) at the first antenatal visit, and was categorised into <30 and ≥30. Smoking habits were dichotomised into yes or no from data received in the standardised antenatal form at 32 weeks of gestation. Obstetric and labour characteristics were based on reported diagnosis of pre-gestational or gestational hypertensive disorders, diabetes mellitus, placental complications, induction of labour, or dystocia, all of which were dichotomised. Caesarean section was divided into: with labour, when performed after the onset of labour; and without labour, if performed before contractions.
Duration of the first labour in hours, from 3 cm of cervical dilatation to delivery, was calculated from the partograms and categorised into ≤12 hours and >12 hours. Protracted labour was defined as a duration of >12 hours. Information about the station of the fetal presenting part in the birth canal was grouped into above and below the ischial spinae. Failed operative vaginal delivery following vacuum extraction or forceps was also recorded.
Isthmus segment thickness at the index surgery was reported as thin or normal. Uterine closure was defined as single or double layers, regardless of continuous or interrupted technique. Suture materials were categorised into four groups: rapidly absorbable, fast absorbable, slowly absorbable, and non-absorbable. The use of prophylactic antibiotics in surgery was dichotomised into yes or no. Information about the duration of labour and surgical procedures was obtained from the Obstetrix® data record system.
A composite variable called ‘infection’ included intrapartum fever or postoperative uterine, wound, or urinary tract infection diagnosed in the MBR or the Patient Register.
Furthermore, the delivery interval was categorised into less than and more than 18 months, computed from the dates of first and second deliveries.[9, 15]
The MBR provided information about the second delivery. Onset of the second delivery was divided into spontaneous or induced. Reliable information about the induction method was not found in the register. Use of epidural analgesia during labour was recorded. We also obtained information about mode of second delivery, categorised into operative vaginal delivery or not, and occurrence of caesarean section.
Birthweight was categorised into <4500 and ≥4500 g, and large or small for gestational age was defined as more or less than ±2 standard deviations (SD) of mean birthweight.
Labour records were reviewed (by SH and MJ) to confirm a diagnosis of uterine rupture, and cases with a non-verified diagnosis were excluded (n = 5). Uterine rupture during labour was defined as a full thickness separation of the uterine wall or when a separation covered by the visceral or bladder peritoneum presented with clinical symptoms, such as vaginal bleeding, fetal heart rate abnormalities, abdominal pain, or palpable fetal part in the abdomen. The diagnosis was confirmed during caesarean section or during laparotomy performed immediately after vaginal delivery and, in one case, by abdominal ultrasound that revealed an abdominal haemorrhage and a scar defect of the uterus.
The associations between factors from the first and second delivery and uterine rupture were investigated using logistic regression, and were presented as crude and adjusted risk estimates (odds ratios, ORs, and 95% confidence intervals, 95% CIs). Multilevel logistic regression, with uterine rupture as the outcome measure and all the factors in Tables 2 and 3 as fixed factors nested under the random hospital factor, was performed in order to rule out heterogeneity between hospitals. The results did not differ substantially from the results found in Tables 2 and 3 (data not shown). SPSS 20 was used to perform the descriptive statistics and logistic regression analyses. The multilevel or mixed-effect models were retrieved using the r package lme4.[27, 28]
Hospitals were asked about their standard procedure of uterine closure, and approximately 50% had a policy of using double-layer sutures. In the case of an incidence of uterine rupture of 0.5 and 1.0%, respectively, and if single-layer closure increases the risk by four-fold, then this can be shown with an 80% power and a 95% CI using 1000 women in each group. If the relative increase of risk is instead doubled, then this can be shown using 5000 patients in each group.
Uterine rupture occurred in 103 (1.3%) of the 7683 women with a history of caesarean section in their first delivery. There were 70 anatomically complete ruptures and 33 incomplete. Single-layer closure of the uterus had been used in 38 women and double-layer closure had been used in 65 women with uterine rupture (Figure 1).
Information about the first delivery is presented in Table 1. Maternal height was ≤160 cm in approximately 21% of women. Most women in the cohort had been delivered by caesarean section after labour onset in their first delivery. Uterine incisions had been sutured with a double layer in 65% of women, and peritoneal closure was generally not performed. Prophylactic antibiotics had been administered in two-thirds of women, and intrapartum fever or postoperative uterine, wound, or urinary tract infection occurred in 4% of cases.
|Gestational length (weeks)|
|Maternal age (years)|
|Maternal height (cm)|
|Body mass index (kg/m 2 )|
|Parietal peritoneum closure|
|Visceral peritoneum closure|
Risk factors for uterine rupture from the first delivery are presented in Table 2. The risk of uterine rupture was not significantly increased when single-layer was compared with double-layer closure (OR 1.17; 95% CI 0.78–1.76). The type of suture material had no impact on the risk of uterine rupture (data not shown). Maternal age ≥35 years increased the risk of uterine rupture two-fold and maternal height ≤160 cm increased the risk with 69%. A body mass index of ≥30, smoking, pre-gestational or gestational diabetes, or dystocia were not associated with a risk of rupture in a subsequent labour. Delivery of an infant large for gestational age and infection related to labour increased the risk of uterine rupture. No increased risk was noted with placental complications, hypertensive disorders, induced labour, failed operative vaginal delivery, station of the fetal presenting part, or if prophylactic antibiotics had been omitted in the first delivery (data not shown).
|Study group n = 7683||Uterine rupture n = 103||Missing (%)||Crude OR||95% CI||Adjusted ORa||95% CI|
|Age ≥ 35 years||672||15||–||1.80||1.03–3.12||2.08||1.19–3.63|
|Maternal height ≤ 160 cm||1595||32||1.7||1.70||1.11–2.58||1.69||1.11–2.58|
|Body mass index ≥ 30||655||12||12.4||1.53||0.83–2.84|
|Large for gestational age||214||7||6.5||2.42||1.11–5.28||2.46||1.12–5.40|
|Delivery interval < 18 months||1146||10||–||0.61||0.32–1.18|
|Thin isthmus segment||331||7||–||1.59||0.73–3.45|
|Caesarean section with labour||5041||75||–||1.41||0.91–2.18|
A subanalysis of protracted first labours (>12 hours) was performed. The duration of labour was quantifiable in approximately two-thirds of the women delivered by caesarean section with labour. Of these, 25% had a labour duration >12 hours, which doubled the risk for uterine rupture in the subsequent delivery (OR 2.02; 95% CI 1.16–3.53).
In the second delivery attempted, vaginal delivery succeeded in 71% of cases, 6441 (83.8%) women had a spontaneous onset of labour, and 1242 (16.2%) were induced. Induction of labour resulted in an increased risk of uterine rupture. Record review of induced rupture cases (n = 27) showed that prostaglandin had been used in 19 women. Other risk factors for uterine rupture in the second delivery were maternal age ≥35 years, maternal height ≤160 cm, gestational diabetes, a birthweight of ≥4500 grams, and use of epidural analgesia (Table 3). Infant head circumference at birth was not associated with an increased risk of uterine rupture (data not shown).
|Study group (n = 7683)||Uterine rupture (n = 103)||Missing (%)||Crude OR||95% CI||Adjusted ORa||95% CI|
|One-layer closure (index)||2589||38||1.2||1.13||0.76–1.69||1.13||0.75–1.70|
|Age ≥ 35 years||1683||32||–||1.62||1.06–2.47||1.64||1.07–2.52|
|Maternal height ≤ 160 cm||1595||32||1.7||1.70||1.11–2.58||1.74||1.14–2.66|
|Body mass index ≥ 30||895||13||9.9||1.08||0.60–1.95|
|Gestational length ≥ 42 weeks||600||13||–||1.72||0.96–3.10|
|Birthweight ≥ 4500 g||314||9||0.1||2.31||1.15–4.61||2.11||1.05–4.26|
The risk of uterine rupture did not differ when single- was compared with double-layer uterine closure in women attempting a vaginal delivery after caesarean section. The strongest risk factor from the first labour was delivery of an infant large for gestational age, followed by maternal age ≥35 years, infection related to labour, and short maternal stature. In the second delivery, a birthweight of ≥4500 g, epidural analgesia, and induction of labour were associated with an increased risk of uterine rupture.
An important strength of the study is the size of the cohort. To the best of our knowledge, this is the largest study to date with regard to single- and double-layer closure of the uterus. The accuracy of coding for uterine rupture was checked by review of patient records and, as non-verified cases were excluded, an overestimation of the diagnosis was avoided. The included variables from the Swedish Medical Birth Register have been validated and have been shown to be reliable. Information from the standardised operation template, which was filled in by checkboxes for suture layers, was 98.8% complete. Furthermore, potential confounding effects of parity, previous vaginal delivery, and more than one caesarean section were eliminated by the study design. We also checked for heterogeneity between hospitals by controlling or conditioning the main analyses for the different hospitals. These results were omitted as no difference was shown from the results of the logistic regressions analyses.
The study was statistically underpowered to prove an increased risk of uterine rupture with single-layer closure if the risk was not at least two-fold, showing the difficulty of reaching sufficiently large study groups when the incidence of the outcome variable is low. The retrospective design of the study is also a limitation. Information about the technique used for uterine closure regarding locking or non-locking sutures was not available. According to a meta-analysis, single-layer closure with a locking but not with a non-locking suture technique is associated with a higher risk of uterine rupture than double-layer closure. There was no reliable information about induction method in the register and there was no information about oxytocin administered to augment labour. Furthermore, there was a selection bias for different lengths of follow-up dependent upon which year the woman had her first delivery.
The issue of uterine closure technique, i.e. single or double layer, is relevant as there is a potential relationship with uterine rupture, a greatly feared complication associated with significant neonatal and maternal morbidity and mortality.[4, 5, 13] Surgical technique is an influential factor and it is important to investigate its effects. The literature on a possible association is, however, sparse, and results are conflicting. Most studies are statistically underpowered, have a retrospective design, and derive from unselected populations, and randomised controlled trials are lacking. In a cohort study, 1980 women attempted vaginal delivery after one caesarean section, 23 ruptures occurred, and single-layer closure increased the risk of rupture four-fold compared with double-layer closure. In a smaller study on 768 women with a caesarean section in their first delivery, there was no association between single-layer closure and uterine rupture (n = 4) in a subsequent attempt of vaginal delivery. A recent meta-analysis, including 5810 women and 160 ruptures, did not find an increased risk of rupture with single-layer closure (OR 1.71; 95% CI 0.66–4.44). In our study, which is much larger than any previous single study, there was no significant difference in the risk of rupture when single- and double-layer closures were compared.
For women attempting vaginal birth after previous caesarean section the incidence of uterine rupture is approximately 0.6–1%.[2, 14, 20, 31] It is known that prior vaginal delivery reduces the risk of uterine rupture, and different definitions of uterine rupture have been used in different studies, which will have an effect on reported rates. The rate of uterine rupture (1.3%) in our study is higher compared with similar studies.[8, 20] In the study by Lydon-Rochell, where the rate of uterine rupture was 0.6%, the rate of inductions was about the same as in our study. Induction with prostaglandins is considered a strong risk factor for uterine rupture in women who attempt vaginal birth after caesarean section,[13, 17, 18, 20] and one reason for the higher rate of ruptures in the present study might be a possible liberal use of prostaglandins, indicated by the finding that among the uterine rupture cases prostaglandins were used in 19 out of 27 induced labours. Furthermore, the accuracy of coding of uterine rupture was not documented in the study by Lydon-Rochell, and the numbers in that study may therefore be underestimated. A validation study in a Norwegian birth register showed that the diagnosis of uterine rupture was under-reported by 37%.
Maternal age and height are factors to consider when a decision on delivery method is to be made. Our study is consistent with previous studies showing that high maternal age is a risk factor for uterine rupture,[13, 14] and is also related to failed vaginal delivery. There is an association between short stature and obstructed labour,[35, 36] and short women with a scarred uterus are vulnerable in situations with prolonged and augmented labour. It is possible to control oxytocin augmentation, and safe management may have preventive effects.[5, 19, 37]
High birthweight can also increase the risk of obstructed labour, and is associated with failed vaginal delivery and uterine rupture.[3, 9, 14] In the present study, an infant that is large for gestational age in the first delivery or with a birthweight of ≥4500 g in the second delivery doubled the risk of uterine rupture.
Women having epidural analgesia had a two-fold increased risk of uterine rupture in our study. The reason is unclear but is of concern. Initial symptoms could be masked or atypical, which in turn could delay diagnosis and proper action. The use of epidural analgesia could also be a surrogate marker of dystocia and prolonged, augmented labour. In Sweden, epidural analgesia for pain relief in labour is well established, and previous caesarean section is not considered a contraindication. Epidural analgesia associated with uterine rupture has been reported previously, and frequent dosing could be an indication of a protracted labour with impending risk of uterine rupture. This association needs to be investigated further.
The rate of caesarean sections is increasing, resulting in increasing numbers of women who are exposed to the risk of uterine rupture during labour. Rates of older women giving birth, inductions of labour, and macrosomia have also increased,[40-42] and are all risk factors to be considered when a trial of vaginal delivery is discussed. In the clinical situation, with a woman attempting vaginal birth after a prior caesarean section, knowledge of risk factors that can be influenced during labour, such as method of induction, management of protracted labour, use of oxytocin, and epidural analgesia, is of importance, and should be taken into consideration when assessing the health and safety of woman and child.
In conclusion, single-layer compared with double-layer closure did not significantly increase the risk of uterine rupture in women attempting vaginal delivery after one previous caesarean section. While awaiting results from randomised controlled trials, our results suggest that either technique can be used.
None to declare.
UH and MJ participated in the design of the study. SH, UH, and KE-S managed the data set and performed statistical analyses. All authors contributed to the interpretation of the results, as well as to the writing and editing of the article.
The Regional Ethical Board in Umeå, Sweden, approved the study: reference 2010-333-31M.
This work was supported by grants from the Center for Clinical Research Dalarna, the Swedish Research Council, Uppsala-Orebro Regional Research Council and from the Swedish Society of Medicine. The funders had no involvement in the collection, analysis and interpretation of these data or of the decision to submit the article for publication.
We thank Ellika Andolf, Charlotte Grunewald, and Håkan Lilja for contributing to the conception and design of this study.