Article Text
Abstract
Objective To explore if newborns in the second pregnancy following a previous caesarean delivery (CD) have higher risk of perinatal mortality or cerebral palsy than newborns in pregnancies following a previous vaginal delivery (VD).
Design Cohort study with information from the Medical Birth Registry of Norway and the Cerebral Palsy Registry of Norway.
Setting Births in Norway.
Participants 294 598 women with their first and second singleton delivery during 1996–2015.
Main outcome measures Stillbirth, perinatal mortality, neonatal mortality and cerebral palsy.
Results Among 294 598 included women, 42 962 (15%) had a CD in their first pregnancy while 251 636 (85%) had a VD. Compared with the second delivery of mothers with a previous VD, the adjusted OR (adjOR), for stillbirth in the second pregnancy following a previous CD was 1.45, 95% CI 1.22 to 1.73; for perinatal death the adjOR was 1.42 (1.22 to 1.73) and for neonatal death 1.13 (0.86 to 1.49). Among children who survived the neonatal period, the adjOR for cerebral palsy was 1.27 (0.99 to 1.64). Secondary outcomes, including small for gestational age, preterm and very preterm birth, uterine rupture and placental complications (eg, postpartum haemorrhage and pre-eclampsia) were more frequent in the subsequent pregnancy following a previous CD compared with a previous VD, in particular for uterine rupture adjOR 86.7 (48.2 to 156.1). Adjustment for potential confounders attenuated the ORs somewhat, but the excess risk in the second pregnancy persisted for all outcomes.
Conclusion A previous CD was in this study associated with increased risk for stillbirth and perinatal death compared with a previous VD. Although less robust, we also found that a previous CD was associated with a slightly increased risk of cerebral palsy among children surviving the neonatal period. The aetiology behind these associations needs further investigation.
- obstetrics
- epidemiology
- neonatology
- maternal medicine
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
Prospectively recording of high-quality data in the two national registers.
Large number of births.
Misclassification may occur for some of the clinical outcomes.
Limited ability to address explanatory factors.
Introduction
Between year 2000 and 2015, the number of caesarean deliveries (CD) increased from 16 to 29.7 million deliveries worldwide.1 Moreover, in 2008, WHO estimated that 6.2 million not medically indicated CDs were performed each year.2 The ‘ideal’ CD rate has been debated for many years, and different studies have argued that a higher rate than 15% would not improve maternal and neonatal outcomes,3–5 while others have suggested national CD rate up to 19%.6 However, many middle-income and high-income countries have CD rates of 30%–50%.7 Although the risk for severe acute complications of CD is low in these countries,8 concerns regarding long-term complications, both for the mother and her child have been raised, also in subsequent pregnancy.9 10
In Norway, a country with a CD rate of 16%, approximately 65% of the total number are emergency CDs.11 However, in deliveries when planned CD is recommended to prevent potentially adverse outcomes of vaginal delivery (VD), these advantages should be weighed against potential long-term complications, including those in later pregnancies. This is particularly important when the absolute risk for acute complications during VD is low, such as in breech VD.12 Studies have also reported higher risk for placental complications and uterine rupture in pregnancies following a previous CD.10 13 14 These complications are also associated with higher risk for delayed neurodevelopment and cerebral palsy in the offspring.15 16A balanced information on potential long-term consequences of CD is even more important when the mother requests her baby to be delivered by CD, but where medical evidence to support this choice is limited.17
On this background, the main aim of the present study was to explore if offspring in deliveries following a previous CD, have higher risk for perinatal death or cerebral palsy than offspring where the mother had a previous VD. As secondary outcomes, we also assessed the risks for placental complications, uterine rupture, preterm birth, fetal growth restriction and neonatal morbidity in the offspring following an earlier CD compared with those following a previous VD.
Methods
In this population-based cohort study, including all births in Norway between 1996 and 2015 and information regarding pregnancy, delivery and the neonatal period was retrieved from the Medical Birth Registry of Norway (MBRN). This information was combined with information on a diagnosis of cerebral palsy recorded in the Cerebral palsy Registry of Norway (CPRN). The unique 11-digit personal identification number for every Norwegian citizen was used to link information from the two registers and made it possible to follow the women throughout all her pregnancies. Registration in the MBRN has been compulsory since 1967 with prospectively recorded information during pregnancy and at birth, including demographic variables, information on maternal health before and during pregnancy, interventions and complications during pregnancies, delivery and neonatal outcomes (including stillbirth). The CPRN is an informed consent-based national quality register established in 2006 and aims to record detailed information on all children with cerebral palsy born in Norway since 1996. Paediatricians at the neurohabilitation centres in Norway complete the information of each child using a standardised form. A validation study indicated that ~80% of the children with cerebral palsy born in Norway in 1996–2007 had detailed information in the CPRN.18 Later validations have suggested that the completeness is around 90% in recent years (annual reports of the CPRN).
Eligible for inclusion in this population-based study was women who delivered their first and second singleton child in Norway during 1996–2015. We excluded offspring who were born before 22 completed weeks of gestation and had a birth weight below 500 g. The offsprings were linked to their mothers by means of the national identification numbers, yielding maternal sibship files with the mother as the observation unit.
Study variables
The exposure variable was CD in the first pregnancy, regardless of whether this was emergency or planned. The reference group comprised mothers who delivered their first child vaginally (VD).
The predefined main outcome measures were stillbirth (antepartum and intrapartum and those with unspecified timing of death), perinatal mortality (stillbirth and deaths during first week of life), early neonatal mortality (deaths during first week of life), neonatal mortality (deaths during first 28 days of life) and cerebral palsy. In the CPRN, the diagnosis cerebral palsy is confirmed at 5 years of age according to the definition and classification proposed by the Surveillance of Cerebral Palsy in Europe.19
Secondary outcomes were preterm birth, evidence of fetal growth restriction, uterine rupture and placental complications such as placenta previa, placental abruption, abnormal invasive placenta, pre-eclampsia and postpartum haemorrhage. Preterm birth was defined as a birth before 37 weeks, and very preterm birth as a birth before 32 completed weeks of gestation. Evidence of fetal growth restriction was assessed by the proxy small for gestational age (SGA) defined by a birth weight below—2 SD of the population mean weight for GA, adjusted for sex.20
Other variables extracted from the MBRN were maternal and pregnancy-related disorders such as pregestational diabetes, gestational diabetes and pre-eclampsia. For the purpose of the present study, we included gestational diabetes, and diabetes mellitus type 1 and 2 in the composite variable ‘diabetes’, while pre-eclampsia included mild, severe and unspecified pre-eclampsia, in addition to eclampsia and haemolysis, elevated liver enzymes, low platelets. Maternal age, birth year, smoking and congenital anomalies were also extracted. Information on smoking, recorded in the MBRN since December 1998, is based on information from ‘the antenatal chart’, a standardised form intended for clinical use, carried by the mother to each antenatal visit and completed by midwives and doctors responsible for the antenatal care. The mother is asked about smoking habits (daily smoking, occasionally or not at all) at first antenatal visit and at the end of the pregnancy. However, in contrast to all other variables registered in the MBRN, the mother may decline that her information on smoking is reported to the MBRN. Thus, information on smoking is missing for a subsample of mothers (around 12% in 1999–2000, nearly 20% in 2004–2007 and decreasing to less than 10% since 2015). For those with smoking information registered, we defined smokers as mothers who indicated that they smoked daily or occasionally either at the beginning or at the end of the pregnancy or both. Non-smokers were mothers who responded that they did not smoke during the pregnancy.
Gestational age was based on ultrasound examination before 20 weeks of gestation in 84% of the pregnant women, while in the remaining it was calculated from the last menstrual period.
Statistical analyses
Differences in proportions between groups were analysed using the Χ2 test, and differences in means for continuous variables were analysed using the Student’s t-test not assuming equal variances.
Logistic regression analyses were used to calculate crude and adjusted ORs (OR) with 95% CIs for adverse outcomes of the second delivery following a CD in the first, compared with the outcomes of second deliveries following a first VD.
Potential confounders included in the multivariable analyses were selected based on availability, a priori knowledge and directed acyclic graphs methodology.21 These variables included pregnancy-related disorders (such as diabetes and pre-eclampsia), maternal age, birth year, SGA and congenital anomalies all recorded at the first delivery.
Since there is a risk of recurrence of adverse outcomes, such as stillbirth, perinatal death, preterm birth, fetal growth restriction and pre-eclampsia from one delivery to the next,22–24 we included complications in the first pregnancy that were defined as an adverse outcome in the second, as potential confounders. These complications may also affect the choice of mode of delivery in the first pregnancy, leading to potential confounding by indication, partly adjusted for in the logistic regression analyses described above. In addition, we also assessed confounding by indication in stratified analyses within subgroups of pregnancies with high and low risk for adverse outcomes.25 The high-risk group comprised women with any complication in the first pregnancy that in the second pregnancy were considered as primary or secondary outcomes (stillbirth, perinatal mortality, neonatal mortality, SGA, placenta previa, postpartum haemorrhage, uterine rupture, placental abruption, preterm delivery and pre-eclampsia). The low-risk group included mothers without any of the specified primary or secondary outcomes in the first delivery, and where the first infant was born at term.
In separate analyses, maternal smoking was explored as a potential additional confounder in the population of mothers who had available information on smoking, with first deliveries from 1999.
Moreover, the birth intervals between pregnancies are likely to be shorter in mothers with a perinatal loss than in mothers whose first baby survived.26 Mothers with a first delivery towards the end of the study period will have shorter interpregnancy intervals, and therefore, be more likely to have a first perinatal loss, and thus at increased risk for a recurrent second loss. In order to assess this possible bias, we performed a sensitivity analysis where we excluded women who had their first birth in 2012 or later, since these women had interpregnancy interval less than 4 years. Analyses from the MBRN show that 95% of women with two or more births, have their second child within 7 years following the first.27 IBM SPSS Statistics V.25 was used for data analyses.
Patient involvement
Patients were not involved in setting the research question or the outcome measures nor were they involved in the design and implementation of the study. There are currently no plans to involve patients in dissemination.
Results
During the study period, 454 031 mothers gave birth to a singleton as their first birth, but 159 433 had only one delivery. Among the remaining 294 598 mothers, 42 962 (15%) had a CD in their first pregnancy while 251 636 (85%) had a VD (figure 1).
Characteristics of the mothers and their babies, as well as adverse outcomes in the first pregnancy when mothers had a caesarean or VD are shown in table 1. The table shows that intrapartum stillbirth and neonatal loss occurred more frequently, whereas antepartum and total stillbirth occurred less frequently in the group of mothers with CD than mothers with a VD. Perinatal mortality did not differ between the two groups. A higher proportion of mothers in the CD group than in the VD group had pre-eclampsia and diabetes and more children were SGA, had congenital anomalies and were later diagnosed with cerebral palsy (table 1).
In their second delivery, approximately half of the women with a previous CD also had CD in their subsequent pregnancy, compared with 6% among those with a previous VD (table 2). Gestational age and mean birth weight were slightly lower and congenital anomalies, Apgar scores below 7 after 5 min and being transferred to the NICU were more common in the group with a previous CD (table 2).
When adjusting for potential confounders, the second infant of mothers with a previous CD had higher odds for antepartum and intrapartum stillbirth, total stillbirth and perinatal death compared with the second infant of a mother with a previous VD (table 3). The adjusted OR for early neonatal mortality was 1.27 (95% CI 0.92 to 1.73), for neonatal mortality the OR was 1.13 (95% CI 0.86 to 1.49) and for cerebral palsy OR was 1.27 (95% CI 0.99 to 1.64) (table 3).
When these associations were studied among women with high (n=78 056) and low (n=211 313) risk for adverse outcomes, the results were essentially the same as in the total study population (data are not shown).
All secondary outcomes, except abnormal invasive placenta, were more frequent in the subsequent pregnancy of women with a previous CD than among those with a previous VD (table 4). Most of these associations were attenuated in multivariable analyses, but the main findings persisted.
In the subsample of 199 269 women where information about smoking was available (68% of the total study population), the results regarding antepartum and intrapartum stillbirth, overall stillbirth, perinatal early and neonatal mortality persisted, but were slightly attenuated in the multivariable analyses (see online supplementary table 1). The crude OR for cerebral palsy was also unchanged but was reduced in the multivariable analyses to 1.23 (95% CI 0.90 to 1.67). However, adding smoking as a potential confounder to the other confounders did not change the adjusted OR neither for primary nor for the secondary outcomes (see online supplementary tables 1 and 2)
Supplemental material
Supplemental material
In the sensitivity analyses excluding women who had their first birth in 2012 or later (leaving 274 041 women for analysis), the results were nearly identical to the results in the total population. In this subsample, 39 812 (15%) had a CD in their first pregnancy and 234 229 (85%) had a VD (see online supplementary tables 3 and 4).
Supplemental material
Supplemental material
Discussion
In this population-based cohort study in Norway, we found increased risk for stillbirth, perinatal death and cerebral palsy in the second pregnancy of mothers who had a CD in their first pregnancy, compared with mothers who had a VD in their first pregnancy. In contrast we did not find a higher risk for neonatal mortality, although there was a trend towards increased risk for early neonatal mortality. Also, growth restriction, preterm delivery, pre-eclampsia, placental complications, postpartum haemorrhage and especially uterine rupture were increased in pregnancies following a CD.
Strength and limitations
The strength of this study is the large number of births based on mandatory registration and thus minimal risk of selection bias. The prospectively collected data from the MBRN and the CPRN, and the ability to link births to their mothers by the identification numbers enabled complete analyses of the first and second births to all mothers giving birth in 1996–2015. We restricted the analyses to singletons, since both CD and the studied outcomes are more frequent in multiple pregnancies. Moreover, the main findings persisted when adjusted for potential confounders in multivariable analyses, and when we assessed confounding by indication in analyses restricted to mothers with high and low risk of adverse outcomes. In these latter analyses, the main findings persisted in both groups, suggesting that confounding by indication was less likely. Finally, in the subpopulation with information on smoking, the main results also persisted.
A potential selection bias of the study population may arise from the fact that women with a perinatal loss in their first pregnancy tend to proceed to their second pregnancy earlier than women without such an event, thereby artificially increasing the proportion of women with losses in their second pregnancy (due to the known recurrence risk of perinatal losses). Since the implications of this selection bias is unclear, we performed sensitivity analyses where we excluded women who had their first birth in 2012 or later, and hence had an interpregnancy interval shorter than 4 years. Moreover, in the population born after 2012 there may be some underreporting of cases with cerebral palsy, since this diagnosis is reported to the CPRN at age 5 years, and since there may be some delay in the registration. Nonetheless, all results were essentially unchanged.
Other limitations of this study are related to potential misclassification of both exposure and outcome. The data on mortality and delivery mode are variables with high quality and are unlikely to be misclassified.28 For the secondary outcomes, both pre-eclampsia, SGA and preterm delivery are variables with high quality,29 while the validity for placenta previa and placental abruption has not been studied. In addition, inconsistent definitions may have led to some misclassification of uterine rupture and abnormal invasive placenta.30
Further limitations are related to unmeasured confounders, as well as to potential confounders with limited available data, such as body mass index and weight change during the pregnancy.
Comparison with other studies
Our findings regarding excess risk for stillbirth in subsequent pregnancies after a previous CD is consistent with several other studies.31–34 In a retrospective cohort study and meta-analysis published in 2015, the authors found an association between previous CD and the risk for antepartum stillbirth in the second pregnancy with a pooled HR 1.40 (95% CI 1.10 to 1.77),35 similar to our results.
Another systematic review and a meta-analysis published in 2018 reported no clear association of mode of delivery with perinatal death comparing previous CD with previous VD,9 as opposed to the results of the present study. However, this meta-analysis included only two studies with considerably smaller populations than the present study and in one of them, the estimated OR for perinatal death was 1.29 (95% CI 0.94 to 1.79),32 more in line with our study. Moreover, an older study not included in the meta-analysis but with a population of the same size as our study, found increased risk for perinatal death among children born by a mother with a previous CD compared with previous VD (RR 1.33, 95% CI 1.10 to 1.62).36
We have not found other studies addressing the risk of cerebral palsy in deliveries following a previous CD.
Regarding the secondary outcomes, the increased risks of SGA, preterm birth, uterine rupture and several placental complications in the second delivery following a previous CD compared with a previous VD, are in line with other studies.10 13 32 37–42 Somewhat unexpected the proportion with abnormal invasive placenta did not differ between the two groups, which may partly be explained by the quality of this variable in the MBRN. Another explanation may be that our study was limited to adverse outcomes only in the second pregnancy, and studies have reported that abnormal invasive placenta is strongly associated with increasing numbers of CD performed.43 44
Interpretation
The findings regarding antepartum and total stillbirth, and perinatal death were all in the same direction, relatively precise and in line with previous literature, indicating that these findings are robust. The increased odds for cerebral palsy and the trend towards increased risk for early neonatal death were less robust. However, the number of infants with cerebral palsy was low, and the number of infants who died during delivery and in the neonatal period were even lower, and these findings should therefore be interpreted with caution.
The increased risk for any of the primary adverse outcomes in the second pregnancy may be due to placental complications, as indicated by the higher occurrence of infants born SGA, as well as a higher occurrence of placenta previa, placental abruption and pre-eclampsia compared with pregnancies without a previous CD. We speculate that the scar in the uterus after previous surgery may play a role in the development of such placental complications, for example, through suboptimal placental implantation in the following pregnancy. In addition, a scar in the uterus is a locus minoris resistentiae, increasing the risk for rupture of the uterus in the following pregnancies, as supported by the very high increased risk of uterine rupture associated with previous CD found in our study.
Moreover, it is well documented that placental pathology, placental abruption, pre-eclampsia, uterine rupture, preterm birth and being born SGA are all risk factors for cerebral palsy.45–47 These risk factors may be involved both in antenatal as well as in perinatal events, leading to an insult of the brain. The increased risk for cerebral palsy is also consistent with the lower Apgar score among infants born by mothers with a previous CD, compared with those whose mothers had a previous VD, found in our study.
The trend towards increased risk for early neonatal mortality may be consistent with increased risk for intrapartum death. However, the risk for neonatal death was only slightly different between the groups, which may be reasonable due to other causes of death later in the neonatal period rather than complications during the delivery. More studies are required to investigate if and how mode of delivery in the second pregnancy may have affected this outcome.
Implication
In this population-based study from Norway, the absolute risk for stillbirth, neonatal death and cerebral palsy in the second pregnancy was low, as were the absolute risk for the secondary outcomes. Nonetheless, the increased relative risk for these complications should be included in the discussion with parents regarding mode of delivery, both when the mother requests a CD in the absence of a clear medical indication as well as in cases when a VD is associated with increased relative, but low absolute risk for adverse outcome. If the increased risk was entirely related to the actual mode of delivery in the second pregnancy, it might be reasonable to recommend a planned CD to all women with a previous CD in their subsequent pregnancy. However, this would not abolish the risk for antepartum stillbirth, SGA and preterm birth. Therefore, the main prevention of these severe complications in subsequent pregnancy may be to take individual plans for further pregnancies into account when considering CD without a clear medical indication. On the other hand, it should be underscored that overall, the absolute risks for stillbirth, perinatal mortality and cerebral palsy after a previous CD are indeed low in the following pregnancy.
Conclusion
A previous CD was in this study associated with increased risk for stillbirth and perinatal death compared with a previous VD. Although less robust, we also found that a previous CD was associated with a slightly increased risk of cerebral palsy among children surviving the neonatal period. The aetiology behind these associations needs further investigation.
Acknowledgments
The authors thank the Cerebral Palsy Register of Norway (CPRN) and the Medical Birth Registry of Norway (MBRN) for the data. A special thanks to all the children and parents participating in the CPRN.
References
Footnotes
Contributors SB proposed the research question, analysed the data, contributed to study design and data interpretation and wrote the first draft of the manuscript. GLA was principally responsible for the data from the CPRN and revised the manuscript. LK contributed to the data interpretation and revised the manuscript. KK contributed to the study design, the analyses and interpretation of the data and revised the manuscript. SH contribute to the data interpretation and revised the manuscript. PRR contributed to the data interpretation and revised the manuscript. SL was responsible for the statistical methods, the interpretation of the results and revised the manuscript. TV contributed to the research question, the study design, the analyses and the interpretation of the data and the revision of the manuscript. All authors approved the final version of the submitted manuscript.
Funding Supported by a grant from The Liaison Committee between the Central Norway Regional Health Authority (RHA) and the Norwegian University of Science and Technology (NTNU).
Competing interests All authors have completed the ICMJE uniform disclosure form at www.icmj.org/cio_disclosure.pdf (available on request from the corresponding author) and declare: no financial support from any organisation for the submitted work; no financial relationship with any companies that might have an interest in the submitted work in the previous three years; and have no non-financial interests or relationships that may be relevant to the submitted work.
Patient consent for publication Not required.
Ethics approval The study was approved by the Regional Ethical Committee for Medical Research in Mid-Norway (ref 2018/2145). We intend to present results directly to the Association for persons with cerebral palsy. In addition, dissemination to the Norwegian population (which constitutes the study population) and the broader public will be achieved through media outreach or other public presentations.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data may be obtained from a third party and are not publicly available. No data are available. For question regarding the analyses of the data, please contact the corresponding author at: solveigbjellmo@gmail.com. The protocol is also available on request.