Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Dr O Vikhareva Osser, Department of Obstetrics and Gynaecology, Malmö University Hospital, SE 205 02 Malmö, Sweden. Email email@example.com
Please cite this paper as: Vikhareva Osser O, Valentin L. Risk factors for incomplete healing of the uterine incision after caesarean section. BJOG 2010;117:1119–1126.
Objective To determine which factors increase the risk of large caesarean scar defects as assessed by transvaginal ultrasound.
Design Observational cross-sectional study.
Setting University Hospital.
Population One hundred and eight women who had undergone one caesarean section.
Methods Transvaginal ultrasound examination of the scar in the uterus 6–9 months after the caesarean. Published ultrasound definitions of large scar defects were used. Clinical information was obtained from medical records after all ultrasound images had been evaluated.
Main outcome measures Factors increasing the risk of large caesarean scar defects.
Results Twenty-two (20%) women had a large caesarean scar defect. The odds of a large defect increased with cervical dilatation at caesarean (0 cm, 1–4 cm, 5–7 cm, 8 cm or more; odds ratio [OR] 4.4 [95% CI 0.7–28.5]; 26.5 [4.3–161.8]; and 32.4 [6.1–171.0]; P < 0.001), station of the presenting part at caesarean below pelvic inlet (OR 14.1 [4.6–43.1]; P < 0.001), duration of labour at caesarean (0 hour, 1–4 hours, 5–9 hours, 10 hours or more; OR 2.0 [0.2–23.8]; 13.0 [2.2–76.6]; and 33.1 [6.6–166.9]; P < 0.001), oxytocin augmentation (OR 6.3, [2.3–17.3]; P < 0.001), retroflexed uterus at ultrasound examination (OR 2.9 [1.0–8.3]; P = 0.047). According to multivariate logistic regression no variable added information to cervical dilatation or the station of the presenting fetal part at caesarean.
Conclusions Caesarean in advanced labour is associated with increased risk of incomplete healing of the uterine incision as determined by transvaginal ultrasound.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The caesarean delivery rate is increasing worldwide.1 Several studies have shown that one caesarean section implies a high risk for caesarean section in the next pregnancy.2,3 Caesarean section, especially repeat caesarean section, is associated with an increased risk for uterine rupture, abnormal placental implantation, placental abruption and uterine scar dehiscence in subsequent pregnancies.4–8
In ultrasound studies of non-pregnant women, caesarean defects in the hysterotomy scar have been shown to be common.9–12 Large scar defects have been found in various proportions of women who have undergone caesarean section, the rate of large scar defects increasing with the number of caesarean deliveries.9,10,12 The clinical importance of large scar defects is not known. Possibly, they entail a greater risk of complications such as uterine rupture or placenta accreta in subsequent pregnancies than intact scars or scars with only small defects.
The aim of this study was to determine which factors increase the risk of large defects in caesarean scars as assessed by ultrasound in women who have undergone only one caesarean section.
The volunteers of this study are a subgroup of the 290 women included in another study on caesarean scars.12,13 The 108 women included in the current study are those who had undergone only one caesarean delivery, and these took place at our institution. The women were examined with ultrasound between October 2005 and December 2006. The Ethics Committee of the Medical Faculty of Lund University, Sweden, approved the study protocol and informed consent was obtained from all participants after the nature of the procedures had been fully explained. The design of the study has been described in detail in previous publications12,13 and is therefore only briefly outlined below.
A single examiner assessed all 108 women included in this study with transvaginal ultrasound 6–9 months after the caesarean section and also interpreted the results: 40 women were examined only with unenhanced ultrasound and 68 were also examined with hydrosonography (instillation of saline into the uterine cavity during scanning). The study was originally designed not to include hydrosonography. However, because we discovered by chance that hydrosonography facilitated delineation of the borders of a caesarean scar defect, we decided to supplement the examination with hydrosonography. After we had obtained ethical approval, all women were examined with unenhanced ultrasound and hydrosonography on the same day. Women who had only undergone unenhanced ultrasound were invited to come back for hydrosonography, but not all did so. Eighteen of the 68 women who were examined both with unenhanced ultrasound and hydrosonography underwent both examinations at the same visit. In the remaining 50 women hydrosonography was performed at a second visit at median 8 weeks (range 6–16 weeks) after the unenhanced ultrasound examination.
A standardised examination technique was used. Any indentation in the scar, however small, was classified as a defect, the word defect being used as a purely descriptive term of ultrasound findings. The technique of measuring scar defects and the distance between the scar and the internal cervical os has been described in previous publications.12,13 For the purpose of this study we also measured the distance between the inner cervical os and the lowest demarcation of any visible scar defect as described in Figure 1.
Predetermined definitions of large scar defects were used. These were based on the results of our previous studies.12,13 In these studies measurements were taken of the height, length and width of the defects, the thickness of the remaining myometrium over the defect, and the thickness of the myometrium adjacent to the defect. A ratio was also calculated between the thickness of the remaining myometrium over the defect and the thickness of the myometrium adjacent to the defect (‘ratio’). Using receiver operating characteristic curves we found that only the thickness of the remaining myometrium over the defect and the ‘ratio’ predicted whether a defect would be perceived to be large or not by the ultrasound examiner using subjective evaluation and these two measures had almost identical predictive performance. We decided to use the thickness of the remaining myometrium over the defect to define a large defect, because we find it logical to believe that the thinner the myometrium the greater the risk of complications such as rupture and dehiscence. This is in agreement with the results of the study by Rozenberg et al.14 showing that a thin uterine wall in the isthmus part of the uterus during the third trimester in women who have undergone caesarean section increases the risk of uterine rupture/dehiscence in the same pregnancy. Because the defects looked different at unenhanced ultrasound examination and hydrosonography, we chose different cutoffs to indicate a large defect for measurements taken at unenhanced ultrasound examination and hydrosonography.12,13 In women who only underwent unenhanced ultrasound examination a scar defect was defined as large if the thickness of the remaining myometrium over the defect was ≤2.2 mm,12 in women who underwent hydrosonography a scar defect was defined as large if the thickness of the remaining myometrium over the defect was ≤2.5 mm.13
Clinical information regarding the current pregnancy and delivery was obtained from medical records after all ultrasound examinations had been performed. The ultrasound images were evaluated and measurements were taken and noted in our standardised research protocol. The ultrasound examiner was blinded to the clinical information when performing the ultrasound examination as well as when evaluating the ultrasound images after the examination.
Statistical calculations were performed using the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA; version 12.2 and 16.0).
The statistical significance of the difference in distance between the internal cervical os and the lowest demarcation of the caesarean scar defect between women with small and large scar defects was determined using the Mann–Whitney U-test. To determine the statistical significance of differences between women who underwent caesarean at cervical dilatation 0–4 and 5–10 cm we used the Mann–Whitney test for continuous data and the chi-square test or Fisher’s exact test as appropriate for categorical data. A two-tailed P-value <0.05 was considered statistically significant.
To determine which factors predicted large caesarean scar defects as defined above we performed both univariate and multivariate logistic regression analysis with the likelihood ratio test. A two-tailed P-value <0.05 (likelihood ratio test) was considered statistically significant and was a prerequisite for including a variable in a logistic regression model. To avoid overfitting, only two variables were allowed in each logistic regression model.
To evaluate the ability of single continuous ultrasound variables and of logistic regression models to predict large scar defects, receiver operating characteristic (ROC) curves were drawn. The area under the ROC curve (AUC) and the 95% CI of this area were calculated. If the lower limit of the CI for the AUC was >0.5, the variable was considered to have discriminatory potential. The variable/model with the largest AUC was considered to be the best predictor of large scar defects.
The variables tested for their predictive ability are shown in Table 1. Immediate caesarean section was defined as a caesarean section carried out without delay, i.e. without inserting a bladder catheter and without scrubbing. Intraoperative blood loss was estimated in millilitres by the surgeon immediately after the operation. Intraoperative complications were defined as tears in the uterus or cervix, urinary bladder lesions, or other tissue damage that required extra suturing. Infection peri- or postpartum (6 weeks) was defined as chorioamnionitis, postpartum infection including wound infection, urinary tract infection, endometritis, or infection of unknown origin.
Table 1. Demographic background variables for all women and results of univariate logistic regression analysis
Women with large defect*
Women with no large defect
*In women who only underwent unenhanced ultrasound examination a scar defect was defined as large if the thickness of the remaining myometrium over the defect was <2.2 mm,12 in women who underwent hydrosonography a scar defect was defined as large if the thickness of the remaining myometrium over the defect was <2.5 mm.13
**Univariate logistic regression with likelihood ratio test.
***Number of women who had had at least one vaginal delivery before the caesarean. Time between the latest vaginal delivery and the caesarean section is median 3.05 years, range 1.0–11.8, 25th and 75th percentiles 2.08 and 4.29.
****The number of women is too small for these statistical calculations to be sensible.
*****Information on the haemoglobin level was not available in all women.
******The top of the caesarean scars was never located below the internal cervical os. The level of the internal os is denoted as 0 mm.
Distance between caesarean scar and the internal cervical os, % (n)
Of the 108 women included, 30 had intact caesarean scars, 56 had a small scar defect and 22 (20%) had a large scar defect. Demographic background data for the women included and results of univariate logistic regression analysis are shown in Table 1. In univariate analysis, the following variables predicted large scar defects: maternal age, duration of active labour (i.e. number of hours with regular contractions), cervical dilatation at caesarean, station of the presenting fetal part at caesarean, oxytocin augmentation during labour, delivery <37 gestational weeks, retroflexion of the uterus at ultrasound examination, and a scar located at the level of the internal cervical os as opposed to above the internal cervical os at ultrasound examination. The lowest demarcation of large scar defects (n = 22) was located lower in the uterus than the lowest demarcation of small scar defects (n = 56) (median distance −3.4 mm, range −9.3 to 8.6 for large defects versus 0 mm, range − 6.8 to 16.1 for small defects; P = 0.012). Sixteen percent (5/31) of the 31 women with an intact scar had had their hysterotomy closed with two suture layers compared with 18% (10/55) of those with a small defect and 9% (1/22) of those with a large defect.
The AUC for the distance between the caesarean scar and the internal cervical os was 0.633 (95% CI 0.510–0.757). Duration of active labour and cervical dilatation at caesarean section predicted a large scar defect with similar accuracy (AUC 0.839, 95% CI 0.754–0.933 and 0.835, 95% CI 0.740–0.929). The risk of a large scar defect increased if the caesarean section was carried out after 5 hours of active labour or at cervical dilatation ≥5 cm. Multivariate logistic regression showed that no variable added information to cervical dilatation at caesarean or to the station of the presenting fetal part at caesarean. However, the station of the presenting fetal part and the position of the uterus at ultrasound examination added information to the duration of active labour, with the risk of large defects increasing if the presenting part was below the pelvic inlet or if the uterus was in retroflexion at ultrasound examination (Table 2).
Table 2. Results of multivariate logistic regression and area under the receiver operating characteristic curve (AUC) of the logistic regression models
Odds ratio Point estimate
AUC (total model) Point estimate
Duration of labour
0 hour, reference
Station of presenting part
Duration of labour
0 hour, reference
0.186 to 28.803
Women who underwent caesarean section at cervical dilatation ≥5 cm were younger than those who were delivered earlier in labour, and if the caesarean section was carried out at cervical dilatation ≥5 cm, the woman was less likely to have had a previous vaginal delivery, and more likely to have received oxytocin augmentation and have a scar located low in the uterus (Table 3). This suggests co-variation of maternal age, parity, oxytocin augmentation and scar location with cervical dilatation at caesarean section. The changes in odds when adding these factors to cervical dilatation at caesarean section in multivariate logistic regression also support co-variation: substantial changes in odds ratios occurred when these variables were added to cervical dilatation at caesarean section.
Table 3. Association between cervical dilatation at caesarean section and other variables
Cervix dilated 0–4 cm
Cervix dilated ≥ 5 cm
*Mann–Whitney U test.
**Fisher’s exact test.
***Pearson chi-square test.
****The top of the caesarean scars was never located below the internal cervical os. The level of the internal os is denoted as 0 mm.
Age, years; median (range)
Ever delivered vaginally, % (n)
Oxytocin augmentation during labour, % (n)
Intraoperative complications, % (n)
Immediate caesarean, % (n)
Distance between the caesarean scar and the internal cervical os****% (n)
Preterm delivery (<37 weeks of gestation) % (n)
We have shown that the odds of developing a large caesarean scar defect, as assessed by transvaginal ultrasound 6–9 months after the caesarean section, increase the later in labour the caesarean section is performed, with the risk increasing dramatically if the duration of labour is ≥5 hours or cervical dilatation is ≥5 cm. We have also shown that large defects are located lower in the uterus than intact scars or scars with small defects, and that large defects are more common in uteri in retroflexion than in anteflexion.
The strength of our study is that it contributes information to an area that is poorly elucidated in the literature. We have found only three publications trying to determine which factors affect the appearance of caesarean scars at ultrasound examination.17–19
One limitation of our study is that the rather small sample size did not allow us to include more than two variables in multivariate logistic regression analysis, or to study interaction between variables. The small sample size also means that we may have been unable to demonstrate true effects as statistically significant, and that the magnitude of significant effects is uncertain (large confidence intervals for odds ratios). In this study we examined factors that we thought—on a theoretical basis—could affect the scar in the uterus, but we are aware that there are other factors than those that we have studied that may have affected the healing of the hysterotomy. One should also be aware that, theoretically, the thickness of the endometrium (cycle day, contraceptive pills) could affect the appearance of a caesarean scar at ultrasound examination and so the classification of defects as small or large. On the other hand, our definition of a large defect was based on the thickness of the remaining myometrium over the defect, and this measurement is unlikely to be affected by the thickness of the endometrium. In addition to our study being small, there is another limitation of our study: not all women underwent hydrosonography, or underwent hydrosongraphy after some time had elapsed. This is because the study did not originally include hydrosonography. After we discovered by chance that hydrosonography facilitated delineation of the borders of a caesarean scar defect, we decided to supplement the examination with hydrosonography prospectively; those who had already undergone unenhanced ultrasound were invited to return for hydrosonography. A re-analysis including only the 68 women who had undergone hydrosonography showed, in both univariate and multivariate logistic regression, essentially the same results as those presented in this paper (of course, the odds ratios changed slightly). Hence, even though we find hydrosonography to allow better delineation of scar defects, we do think that unenhanced ultrasound examination is also a good method for evaluating caesarean scars, and we believe it is justified to include the results of the 40 women who did not undergo hydrosonography. It might be seen as a third limitation of our study that we analysed women with a large defect but grouped women with small defects and intact scars together. This is because the group with small defects includes women with only a minimal indentation in the uterine wall. Even though we do not know the clinical importance of either an intact scar, a scar with a small defect or a scar with a large defect, we find it reasonable to believe that large defects reflect poorer healing than small defects and that therefore large defects are more ‘pathological’ than small defects.
Our results are in agreement with those of Armstrong et al.11 who found that scars with defects were seen at ultrasound examination only in women who were in labour at the time of caesarean section. This is perhaps not surprising because there is an association between the degree of cervical dilatation (in particular dilatation ≥9 cm) at caesarean and uterine or utero-cervical lacerations and extensive blood loss causing surgical difficulties.20 Such difficulties could theoretically affect the healing process. Another possibility is that changes in the myometrium induced by labour21 could affect healing negatively.
The fact that scars with large defects were located lower in the uterus than intact scars and scars with only a small defect might be an effect of the caesarean section having been carried out late in labour. Low incisions are likely to be more common if caesarean sections is performed late in labour than before the start of labour or early in labour, and our results confirm this (Table 3). The finding of Zimmer et al.22 that caesarean scars were located lower in the uterus if the caesarean had been performed in active labour also supports this. If the hysterotomy is low in the uterus, cervical tissue may be included in the closing sutures and the healing properties of the cervix might be less favourable than those of the myometrium in the isthmus or corpus uteri, explaining why large scar defects are more common in scars located low in the uterus.
Our findings that large scar defects are more common in uteri in retroflexion are in agreement with results reported by Ofili-Yebovi et al.10 They suggested that mechanical tension of the lower uterine segment in a retroflexed uterus might impair blood perfusion and oxygenation of the healing tissues, and that this could affect wound healing negatively. Tissue oxygenation is an important factor for wound healing.23
We have found only three publications examining which factors affect the appearance of caesarean scars at ultrasound examination.17–19 Two examined the importance of single-layer versus double-layer uterine closure. The results of these two studies are conflicting. In a small randomised trial including 30 women, the uterine closure technique had no effect on the ‘scar thickness’ as measured by ultrasound 48 hours, 2 weeks and 6 weeks after the caesarean section.18 In a nonrandomised study including 137 women, double-layer closure decreased the odds of finding a ‘wedge defect’ at least 5 mm in height in the scar at ultrasound examination 30–38 days after delivery.17 The third study,19 a randomised trial including 78 women, showed that a full-thickness suturing technique decreased the rate of incomplete healing of the uterine incision as assessed by ultrasound 40–42 days after the caesarean section. In our study, large scar defects were seen twice as often in women who had undergone one-layer uterine closure as two-layer uterine closure, but the difference was not statistically significant, and the suture technique was not an independent predictor of large scar defects in multivariate analysis. We cannot exclude the possibility that our inability to show an effect of uterine closure technique on the risk of large scar defects is because our study is under-powered for this specific purpose. For example, in multivariate analysis, uterine closure technique (one or two suture layers) added to cervical dilatation at caesarean section almost statistically significantly (P = 0.09; but this result is not shown in the Results section).
It is not known if caesarean scar defects detected at ultrasound examination in nonpregnant women are associated with greater risks of pregnancy complications in subsequent pregnancies than intact scars, or if large defects are associated with greater risks than small defects. However, a thin lower uterine segment during pregnancy as measured by ultrasound after caesarean has been shown to increase the risk of uterine rupture and dehiscence.14,24 It is reasonable to believe that large caesarean scar defects according to ultrasound examination reflect poor healing of the scar, at least they do reflect incomplete healing of a part of the hysterotomy. Whether this increases the risk of complications in subsequent pregnancies, e.g. scar pregnancy, placenta accreta, uterine dehiscence or rupture, needs to be elucidated in prospective studies. If such studies show that large defects as detected by ultrasound in nonpregnant women are indeed clinically relevant, studies should be undertaken to determine if the rate of large defects can be decreased by changing the surgical technique when caesarean section is carried out in advanced labour. Currently, the result of our study should have no influence on clinical practice.
Caesarean section in advanced labour is associated with increased risk of incomplete healing of the uterine incision as determined by transvaginal ultrasound.
Disclosure of interest
The authors have no potential conflicts of interest of a financial or other nature.
Contribution to authorship
Olga Vikhareva contributed to the protocol preparation, examination of all women, data collection, statistical analysis, interpretation of results and writing of the manuscript. Lil Valentin contributed to the conception and design of the study, protocol preparation, interpretation of results, statistical analysis and writing of the manuscript.
Details of ethics approval
The Ethics Committee of the Medical Faculty of Lund University, Sweden, approved the study protocol (Dnr 859/2004, approved 2004 12 22; Dnr 198/2006, approved 2006 05 08).
This study was supported by two governmental grants (Landstingsfinansierad regional forskning Region Skåne and ALF-medel), and funds administered by Malmö University Hospital. The funding bodies had no role in the design of the study, the analysis, the interpretation of results, the drafting of the report, or the decision to submit the paper for publication.