To evaluate whether women with a caesarean section at their first delivery have an increased risk of retained placenta at their second delivery.
To evaluate whether women with a caesarean section at their first delivery have an increased risk of retained placenta at their second delivery.
Population-based cohort study.
All women with their first and second singleton deliveries in Sweden during the years 1994–2006 (n = 258 608). Women with caesarean section or placental abruption in their second pregnancy were not included in the study population.
The risk of retained placenta at second delivery was estimated for women with a first delivery by caesarean section (n = 19 458), using women with a first vaginal delivery as reference (n = 239 150). Risks were calculated as odds ratios by unconditional logistic regression analysis with 95% confidence intervals (95%) after adjustments for maternal, delivery, and infant characteristics.
Retained placenta with normal (≤1000 ml) and heavy (>1000 ml) bleeding.
The overall rate of retained placenta was 2.07%. In women with a previous caesarean section and in women with previous vaginal delivery, the corresponding rates were 3.44% and 1.96%, respectively. Compared with women with a previous vaginal delivery, women with a previous caesarean section had an increased risk of retained placenta (adjusted OR 1.45; 95% CI 1.32–1.59), and the association was more pronounced for retained placenta with heavy bleeding (adjusted OR 1.61; 95% CI 1.44–1.79).
Our report shows an increased risk for retained placenta in women previously delivered by caesarean section, a finding that should be considered in discussions of mode of delivery.
The increasing rate of caesarean sections worldwide has been a major concern reported in the obstetric scientific literature of the last few years. In the USA as well as in Australia the rate has reached nearly 30%, which is well above the WHO recommendation of 15%.[1-3] As a consequence of this increasing rate, delivery complications related to previous caesarean sections, such as placenta praevia, placenta accreta, and uterine rupture, have increased.
Retained placenta after vaginal delivery is a potentially life-threatening complication because of its strong association with postpartum haemorrhage. Following uterine atony it is the second most common cause of postpartum haemorrhage and the most common indication for blood transfusion postpartum. Manual removal of the retained placenta has also been shown to increase the risk of postpartum endometritis. In the UK and Ireland the rate of retained placenta has increased during the last century.
The importance of identifying risk factors for retained placenta is highlighted in many studies. High maternal age, preterm labour, and a previous history of retained placenta are factors known to increase the risk at forthcoming deliveries.[8-11] Another potential risk factor for retained placenta is a previous caesarean section; however, results from earlier studies are contradictory. Some studies have shown that a previous caesarean section increases the risk for retained placenta,[12, 13] whereas others show no increase.[10, 11, 14] These previous studies have not distinguished between retained placenta with normal versus heavy bleeding.
In the present nationwide Swedish study, we included more than 250 000 women with their first and second deliveries between 1994 and 2006. The aim was to investigate whether women with a caesarean delivery at their first delivery have an increased risk of retained placenta at their second delivery.
The Swedish Medical Birth Register is a large national population-based database, established in 1973. It includes information on more than 98% of all births in Sweden, such as demographic data, reproductive history, pregnancy complications, and delivery and neonatal characteristics. In Sweden antenatal care is standardised and free of charge. During the first antenatal visit, usually taking place at the end of the first trimester, the mother is interviewed about her medical and reproductive history and smoking habits. The mother's height and weight are recorded. Complications during pregnancy and delivery are classified according to the International Classification of Diseases (ICD), as noted by the responsible doctor. The ninth version of the ICD (ICD-9) was used for the period 1994–1996, and thereafter the tenth version was used (ICD-10). Information on each pregnancy and delivery is forwarded to the birth register through copies of standardised antenatal, obstetric, and paediatric records. Individual record linkage between the birth register and other registers is possible through each individual's unique personal registration number, assigned to each Swedish resident.
During the years 1994–2006 approximately 1.3 million births were recorded in the birth register. During this period, 296 251 women had their first and second consecutive singleton pregnancies resulting in birth at 22 weeks of gestation or later. We excluded 37 643 women who had a caesarean section or a placental abruption in their second pregnancy, as retained placenta is not a possible outcome for these women. The final study population included 258 608 women.
Exposure was defined as a first delivery by caesarean section. The mode of delivery (caesarean, vaginal instrumental, or vaginal spontaneous) is recorded in the birth register.
The outcome examined was a second vaginal delivery complicated by retained placenta. In Sweden, the recommended clinical practice is to perform manual removal of the placenta 30–60 minutes after delivery of the child according to the National Institute of Health and Clinical Excellence (NICE) guidelines in the UK. Retained placenta was identified by the ICD-9 codes 666A and 667A, and by the ICD-10 codes O720 and O730. Retained placenta with normal (≤1000 ml) as opposed to heavy (>1000 ml) bleeding was ascertained by separating ICD-9 code 667A and ICD-10 code O730 (normal bleeding) from ICD-9 code 666A and ICD-10 code O720 (heavy bleeding).
We considered maternal reproductive history, sociodemographic, and anthropometric characteristics, as well as delivery and infant characteristics as possible confounding factors. Information on maternal body mass index (BMI), height, smoking habits, cohabitation with the infant's father, number of previous miscarriages and in vitro fertilisation in the present pregnancy was collected from the first antenatal visit in the second pregnancy. We did not have access to data on previous terminations of pregnancy. At delivery, information on maternal age, premature rupture of membranes (ICD–9 code 658B and ICD-10 code O42), labour dystocia (ICD-9 code 661 and ICD-10 code O62), mode of delivery (vaginal instrumental or vaginal non-instrumental delivery), as well as gestational length, infant birthweight, and infant sex was collected. In Sweden, gestational age is assessed by ultrasound scans in 97% of pregnant women, usually in the early second trimester. If no information about ultrasound was available, gestational age was calculated from the last menstrual period. Information on the mother's educational level was obtained from the education registry from 2005. The Registry of Population and Population Changes provided information on the mother's country of birth. The interpregnancy interval was calculated as the number of years between the birth of the first child and the estimated date of conception of the second child. All variables were categorised according to Table 1.
|Number of births||Rate (%)||Crude OR (95% CI)|
|Maternal age (years)|
|<25||31 091||1.44||0.81 (0.73–0.90)|
|30–34||100 040||2.27||1.28 (1.20–1.36)|
|≥35||36 311||2.79||1.59 (1.47–1.72)|
|Maternal height (cm)|
|<162||70 906||1.85||0.90 (0.84–0.96)|
|≥172||42 867||2.41||1.17 (1.09–1.26)|
|Body mass index second birth|
|25.0–29.9||55 500||2.11||1.03 (0.97–1.11)|
|≥30.0||20 945||2.18||1.07 (0.97–1.18)|
|Yes||19 087||2.37||1.18 (1.06–1.30)|
|Living with the father for second birth|
|Yes||238 544||2.05||1.03 (0.86–1.24)|
|≤12||61 131||2.17||1.03 (0.96–1.09)|
|Mother's country of birth|
|Non-Nordic||31 642||1.68||0.79 (0.72–0.86)|
|Previous miscarriages (numbers)|
|1–2||49 479||2.62||1.38 (1.29–1.47)|
|In vitro fertilisation|
|Interpregnancy interval (years)|
|Data missing||47 962||1.98||0.97 (0.91–1.04)|
|4–6||22 071||2.32||1.14 (1.04–1.26)|
|Delivery and infant characteristics|
|Premature rupture of membranes|
|Induction of labour|
|Yes||20 138||3.45||1.79 (1.65–1.94)|
|Yes||14 266||3.29||1.67 (1.52–1.84)|
|Instrumental vaginal delivery|
|Yes||46 017||2.77||1.46 (1.37–1.56)|
|Gestational length (weeks)|
|≥42||32 868||2.53||1.33 (1.23–1.43)|
|Infant birthweight (g)|
|2000–2999||19 994||2.56||1.39 (1.26–1.52)|
|4000–4999||61 418||2.38||1.29 (1.21–1.37)|
|Girl||125 885||2.25||1.19 (1.13–1.26)|
|Total births||258 608||2.07|
The risk of retained placenta at second delivery was estimated for women with a first delivery by caesarean section, using women with a first vaginal delivery as reference. Risks were calculated as odds ratios by unconditional logistic regression analysis with 95% confidence intervals. Maternal characteristics, including age, height, smoking habits, country of birth, previous miscarriages, in vitro fertilisation, years of interpregnancy interval, maternal BMI, cohabitation with the infant's father, and years of formal education were recognised as possible confounding factors. The last three variables were not associated with our outcome in univariate analyses, and were therefore excluded from further analyses. In a first multiple logistic model, we calculated risk of retained placenta related to maternal characteristics, and risks were adjusted for the remaining maternal characteristics and for year of second delivery (with the latter categorised into 1994–1997, 1998–2002, and 2003–2006). Thereafter we created a second multiple logistic model by adding delivery and infant characteristics, such as prelabour rupture of membranes, induction of labour, labour dystocia, instrumental vaginal delivery, use of epidural anaesthesia, gestational length at delivery, infant birthweight, and infant sex. Maternal, delivery, and infant characteristics were categorised according to Table 1. All analyses were performed using sas 9.1 (SAS Institute Inc., Cary, NC, USA), and all variables adjusted for were independently statistically associated with our outcome at the level of 5%.
In this population, 2.07% of the women were diagnosed with retained placenta. Table 1 presents risk of retained placenta according to maternal, delivery, and infant characteristics. Older and taller mothers had increased risks compared with younger and shorter women. Women born outside the Nordic countries had a decreased risk compared with Nordic women. Smoking, previous miscarriages, and in vitro fertilisation increased the risk of retained placenta. Delivery characteristics, including prelabour rupture of membranes, induction of labour, labour dystocia, instrumental vaginal delivery, and epidural use increased the risk for retained placenta. The highest OR values for retained placenta were shown for preterm delivery and low infant birthweight. Maternal BMI, cohabitation with the infant's father, and years of formal education had no impact on the outcome (Table 1).
Compared with women who had a vaginal first delivery, women with a caesarean first delivery had a crude OR of retained placenta at their second delivery of 1.79 (95% CI 1.65–1.94; Table 2). Adjustments for maternal characteristics had only a minor affect on this association. Adjustments for delivery and infant characteristics from the second delivery attenuated the risk to some extent (Table 2). When separate analyses were made for retained placenta with normal compared with heavy bleeding, a previous caesarean delivery was only associated with retained placenta and heavy bleeding (adjusted OR 1.61; 95% CI 1.65–1.94; Table 3).
|Caesarean section at first delivery||Total numbers||Retained placenta at second delivery|
|No.||Rates (%)||OR (95% CI)|
|Crude OR||Adjusted OR model 1||Adjusted OR model 2|
|Yes||19 458||670||3.44||1.79 (1.65–1.94)||1.75 (1.61–1.91)||1.45 (1.32–1.59)|
|Caesarean section at first delivery||Retained placenta second delivery|
|Heavy bleeding||Normal bleeding|
|OR (95% CI)||OR (95% CI)|
|No.||Rates (%)||Crude OR||Adjusted ORa||No.||Rates (%)||Crude OR||Adjusted ORa|
|Yes||493||2.5||1.87 (1.70–2.06)||1.61 (1.44–1.79)||177||0.9||1.59 (1.36–1.87)||1.11 (0.92–1.33)|
In Table S1 the adjusted risks of retained placenta in second delivery are presented for all maternal, delivery, and infant characteristics presented in Tables 1 and 2. The highest adjusted odds ratio for retained placenta was seen for preterm delivery (<32 weeks of gestation): OR 3.31 (95% CI 2.02–5.42). The risk of retained placenta after a first delivery with caesarean section was at a similar level as in pregnancies with a mother of high age (≥35 years), three or more prior miscarriages, induced labour, and in deliveries of small (<2000 g) or large (≥5000 g) infants.
In this large population-based cohort study of women with both first and second singleton deliveries we found that women with a caesarean section at their first delivery were at increased risk for retained placenta at their next delivery, compared with women with a first vaginal delivery. Women with previous caesarean delivery had an adjusted OR of retained placenta with heavy bleeding (more than 1000 ml) of 1.61 (95% CI 1.44–1.79), but they had no association with retained placenta with normal bleeding (1000 ml or less). Retained placenta is associated with increased morbidity and mortality for the mother.[5, 6] Our findings highlight the importance to restrict caesarean sections to women in need of abdominal deliveries in order to optimise delivery care.
To our knowledge, our study is the first to separate retained placenta with heavy bleeding from retained placenta with normal bleeding. This distinction is important because of the increase in maternal morbidity and mortality associated with severe postpartum haemorrhage. One limitation is the difficulty in estimating blood loss during delivery, however, but this possible source of misclassification should be non-differential. Another major strength of the study is the large population-based design where data on exposure and confounding factors were collected prospectively, which limits the risk of recall bias. The standardised and free antenatal and delivery care in Sweden minimises the possibilities of residual confounding. Another strength was the determination of a number of possible confounding factors that could affect the association under study.[8-14] We have also identified other risk factors for retained placenta, such as in vitro fertilisation and interpregnancy interval, which were not accounted for in previous studies.
One limitation is that data on prior terminations of pregnancy were unavailable, as some studies report an increased incidence of retained placenta in women with a previous history of surgically induced abortions. However, in Sweden today 79% of induced terminations are performed before gestational week 9, and 89% of these are medically induced. As no association between retained placenta and medically-induced terminations has been shown, this limitation will probably have a marginal effect on the studied association. Another limitation is that there is no validation of the diagnosis of retained placenta in the Swedish birth register; however, the incidence of retained placenta found here is in agreement with the Swedish study by Endler et al., where the diagnosis was set after a review of medical records.
The association between retained placenta and previous delivery by caesarean section has been investigated in previous studies.[10-14] Two Scandinavian case–control studies found no increased risk of retained placenta after a caesarean section.[11, 14] These studies included both nulliparous and parous women, and none of the studies had previous caesarean sections as their main exposure. As these reports included just 165 and 400 cases, the statistical power was limited.[11, 14] A case-control study by Titiz et al. with 114 cases showed the same result. On the other hand, two reports from Saudi Arabia and from New South Wales, Australia, presented results well in line with ours.[12, 13] The latter found an adjusted odds ratio of 1.34 for manual removal of the placenta after a previous caesarean section.
The pathophysiology underlying retained placenta can either be a trapped placenta, where the placenta has been detached from the uterine wall but trapped behind a closed cervix, or a placenta adherent to the uterine wall. The fact that we recorded an increased risk for retained placenta, especially for cases with heavy bleeding, suggests that there is an enhanced risk for placental adherence after a previous caesarean section. The most severe form of adherence is placenta accreta, which has a well-known association with previous caesarean section.
The mechanism behind the increased risk of retained placenta by a previous caesarean section is unknown. It was recently suggested that a uterine scar following caesarean section creates localised hypoxia, with subsequent defective decidualisation and abnormal trophoblastic invasion. It is likely that this mechanism is similar for retained placenta. Another theory is a contractile failure in the retroplacental area as a result of the uterine scar. Using ultrasound, Herman et al. found that the myometrium on the placental site was thickened, from <1 to >2 cm, in normal labour, but stayed thin in cases with retained placenta. It is possible that a change in myometrial thickness is less likely to occur in the area of a uterine scar. These theories are based on a direct effect of the uterine scar on placental detachment. Thus, it would be interesting to study placental location in the uterus and its impact on the association between previous caesarean section and retained placenta.
The finding of this study indicates that a previous caesarean section increases the risk of retained placenta with heavy bleeding. Although we cannot exclude that this increased risk may result from unmeasured confounding factors, rather than the procedure itself, we suggest that this information should be accounted for when an individual risk assessment is performed for a pregnant woman. Awareness of complications in women previously delivered by caesarean sections is key for influencing the increasing number of caesarean sections worldwide. Our findings enhance the need for more studies on the long-term effects on reproductive health caused by caesarean sections.
All authors report no conflict of interests.
A-KW had the original idea for the study. SC contributed with database management and expertise in epidemiology. A-KW, JB, OA, KE, and AM-L contributed to the design of the study. A-KW and JB performed the analyses and wrote the first draft of the manuscript. SC, OA, KE, and AM-L made substantial contributions to the interpretation of the results and to manuscript revision.
The research ethics committee at the Karolinska Institutet, Stockholm, approved the study (ref. no. 2005/4863; 28 September 2005).
The research was funded by, Uppsala University and the county council of Uppsala Sweden.