In Norway, the rate of cesarean delivery has increased from 1.8% in 1967 to 16.4% in 2006.1,2 Because previous cesarean delivery is a frequent indication for cesarean delivery, an increase in the rate of primary cesarean delivery will also increase the rate of repeat cesarean delivery. In Norway in 1999, previous cesarean delivery was the third most frequent indication for cesarean delivery (9%), preceded by fetal stress (22%) and failure to progress (21%).3 This, together with an increasing occurrence of cesarean delivery performed on maternal request, makes it important to assess possible consequences of a previous cesarean delivery for later pregnancies.
Potential consequences of a previous cesarean delivery are complications during pregnancy and delivery as well as various conditions in the newborn. Placental complications such as placenta previa, placenta accreta, and placental abruption often are reported to be associated with previous cesarean delivery.4–12 Placental dysfunction in terms of intrauterine growth restriction also has been suggested.13 Some studies have reported a dose-response effect, where the risk of complications increases with the number of previous cesarean deliveries, although there is a large degree of heterogeneity between the different studies.5 Rupture of the uterus is a rare but serious complication with an increased risk after previous cesarean delivery,14–17 both in a repeat cesarean delivery and, more so, during labor after previous cesarean delivery.18
Women who have had a cesarean delivery are, to a certain extent, selected according to maternal health, pregnancy complications, or complications at birth—conditions that may recur in subsequent pregnancies. Thus, confounding by indication is a methodological problem that will, in general, cause an overestimation of effects.
Our objective was to assess associations between previous cesarean delivery and selected adverse pregnancy outcomes.
MATERIALS AND METHODS
We used data on births registered in the Medical Birth Registry of Norway between 1967 and 2003. The registry is based on compulsory notification and comprises data on all live births as well as all stillbirths after 16 weeks of gestation, identified by the national identification number of the mother.19 Consequently, a woman can be followed throughout all her pregnancies. In the present study, the mother represented the unit of analysis.
The birth notification form contains data on the mother’s health before and during pregnancy, pregnancy complications, and data on the delivery and the newborn. Data on complications and maternal disease were entered into the notification form as free text until 1998 and in checkboxes or as free text from 1999 onward. Data on maternal disease are coded according to International Classification of Diseases (ICD) (ICD-8 from 1967 to 1998 and ICD-10 from 1999 onward). From 1967 to 1998, free text on complications was coded according to a system for the registry based on international definitions; it has been coded according to ICD-10 since 1999. Data on mode of delivery were entered as free text until 1998 and in checkboxes thereafter.
We analyzed the first two or the first three births of women who gave birth for the first time in 1967 or later. All analyses were based on sibships with only single births. We also excluded sibships with birth weights below 500 g or below 20 weeks of gestation. In the analysis of women with one previous birth (n=637,497), selected complications in second pregnancy were outcomes. Women with a first cesarean delivery formed the exposed group, and women with a first vaginal delivery formed the reference group. In the analysis of women with two previous births (n=242,812), selected complications in third pregnancy were outcomes. Women with a cesarean delivery in the first or the second birth formed the exposed groups, and women with a vaginal first and second delivery formed the reference group. To evaluate possible dose-response effects of cesarean delivery, we also analyzed women with a cesarean delivery in first and second birth.
In an additional set of analyses, we excluded women with the selected adverse outcome in any of their previous pregnancies. These analyses were performed to adjust for confounding by indication. The effect of such an exclusion will be strong for complications with a high recurrence risk and a high risk of cesarean delivery, ie, complications that are strongly associated with both the exposure (previous cesarean delivery) and the outcome (a recurrent complication).
Outcome variables were selected based on their associations with previous cesarean delivery as reported in the literature. In addition to placental complications such as placenta previa, placenta accreta (from 1969 onward), and placental abruption, we included outcomes that may be linked to the function of the placenta, such as preeclampsia, bleeding during pregnancy, gestational age, and fetal growth as measured by mean birth weight and small for gestational age (SGA). Small for gestational age was defined as a birth weight below the 10th percentile for gestational age in weeks. The sex-specific percentiles were calculated on the basis of single births between 28 and 42 weeks of gestation from 1967 to 1996. Because some of the complications may share similar underlying causes, we analyzed combinations of complications, eg, SGA with preeclampsia and placental abruption with preeclampsia.20 In the analysis of combined complications, we accounted for confounding by indication by removing births with any of the actual complications in previous pregnancies. Also, because the pathophysiology of early-onset preeclampsia may differ from that of late-onset preeclampsia, we performed analyses of preeclampsia, SGA and the two combined, stratified for gestational age at delivery (less than 34 weeks, 34 weeks or more).21
Because birth weight and gestational age are directly influenced by cesarean delivery, we compared birth weight and gestational age only for first and third births among women with vaginal first and third births. Data on birth weight were missing for 0.2% of the births and on gestational age for 6.2% of the births. Small for gestational age was calculated for 89.4% of the sibships with two births and for 88.9% of the sibships with three births.
The data were analyzed in cross-tables with Pearson χ2 statistics. For outcomes with a cell frequency less than 5, observed odds ratios (ORs) and Fisher’s exact P value were calculated. Maternal age and year of birth were considered as potential confounders and were included in an unconditional logistic regression analysis as categorical variables, with categories of 24 or younger, 25–29, 30–34, and 35 or older for maternal age and 1967–1978, 1979–1990, and 1991–2003 for year of birth. Effects are presented as adjusted ORs with 95% confidence intervals (CIs) and two-sided P values, as obtained from the logistic regression analysis. We also evaluated effect modification by period of birth or gestational age at delivery (less than 34 weeks, 34 weeks or more) by including an interaction term in the logistic regression model. We estimated the reduction in numbers of cesarean deliveries needed to prevent one case with a formula based on adjusted ORs and the disease rates in the unexposed group.22 For the continuous variables birth weight and gestational age, mean and difference in mean and corresponding P values were calculated with Student’s t statistics.
Among women with one previous birth, the cesarean delivery rate at second birth was 41.3 per 1,000 after a first vaginal delivery and 508.0 per 1,000 after a first cesarean delivery (Table 1). Among women with two previous births, the cesarean delivery rate at third birth was 46.8 per 1,000 after a first and second vaginal delivery, 280.2 per 1,000 after a vaginal first delivery and a cesarean second delivery, and 974.0 per 1,000 after repeated cesarean delivery.
At second birth, all complications were more frequent among women with a cesarean first delivery than among women with a vaginal first delivery (Table 2). The highest OR was observed for uterine rupture (OR 37.4, 95% CI 24.9–56.2). Other ORs ranged from 1.1 to 2.9. After excluding women with the actual outcome in the first pregnancy, ORs were, in general, reduced. After this exclusion, the analysis included 94% of the bleeding during pregnancy cases, 70% of the preeclampsia cases, 63% of the SGA cases, 99% of the placenta previa cases, 96% of the placenta accreta cases, 96% of the placental abruption cases, and 99% of the uterine rupture cases in the second pregnancies. Based on this exclusion, reduction in numbers of cesarean delivery needed to prevent one case was estimated to be 389 (bleeding in pregnancy), 114 (preeclampsia), 56 (SGA), 1,140 (placenta previa), 3,706 (placenta accreta), 300 (placental abruption), and 461 (uterine rupture).
Similar results were observed for complications in the third births after one or two previous cesarean deliveries compared with only vaginal deliveries (Table 3). After a first vaginal delivery and a second cesarean delivery, with only vaginal deliveries as reference, OR for uterine rupture was 24.7 (CI 10.7–54.5), whereas other ORs ranged from 1.3 to 2.6. In general, the effects were weaker after a first cesarean delivery and a second vaginal delivery. We found no pattern of dose-response effect of previous cesarean delivery, and ORs for placental abruption and uterine rupture were even lower after two previous cesarean deliveries compared with one previous cesarean delivery. Also, in the analysis of complications at third birth, ORs were, in general, reduced after excluding women with the actual outcome in any of their previous pregnancies. After this exclusion, the analysis included 89% of the bleeding during pregnancy cases, 70% of the preeclampsia cases, 43% of the SGA cases, 99% of the placenta previa cases, 100% of the placenta accreta cases, 92% of the placental abruption cases, and 100% of the uterine rupture cases in the third pregnancies.
For the combined outcomes, ie, SGA with preeclampsia, placental abruption with preeclampsia, as well as any combination of these, the results were similar to those obtained for the separate outcomes, although the effects associated with repeat cesarean delivery tended to be slightly stronger but still modest, with ORs between 1.8 and 2.4. Furthermore, inclusion of an interaction term with period of birth and previous cesarean delivery (all outcomes) or an interaction term with gestational age at delivery and previous cesarean delivery (preeclampsia, SGA, or the two combined) did not reveal any statistically significant interactions.
In women with vaginal first and third deliveries, mean birth weight was 83 g lower and mean gestational age was 0.9 days shorter at third birth if the woman had a cesarean second delivery as compared with vaginal second delivery (Table 4). However, this difference was even greater at first birth, 116 g and 2.3 days.
Except for a considerably increased risk of uterine rupture after one or two previous cesarean deliveries, with ORs ranging from 13 to 37, we observed moderately increased risks of complications after previous cesarean delivery, with ORs mainly below 2. In general, the effects were lower when we, to adjust for confounding by indication, removed from the analysis women who had experienced the actual outcome in any of their previous pregnancies. Results based on combinations of complications or stratification for gestational age at delivery did not differ from those based on single complications for all gestational ages.
A strength of our study is that it is population-based and based on a large number of births. Because of the completeness of the Medical Birth Registry of Norway, in which close to 100% of all births in the country have been recorded since 1967, we were able to obtain the full obstetric history of the women. An important aspect of our data is that previous cesarean delivery is not a very important indication for cesarean delivery in Norway. In 1999, previous cesarean delivery was the primary indication in only 9% of all cesarean deliveries in Norway.3 In our data, about half of the women with a previous cesarean delivery had a vaginal delivery at their next birth. This gave us the opportunity to study the occurrence of complications after a previous cesarean delivery in a population for whom vaginal birth after one previous cesarean delivery is common.
An increased risk of complications after a cesarean delivery may be caused by the cesarean delivery itself, or alternatively it may be a result of confounding by indication; persistent problems that represented the indication of the first cesarean delivery also may be present in subsequent pregnancies. A study from the population-based medical birth registry in Finland aimed to reduce confounding by indication by excluding women with defined chronic medical problems,9 but this exclusion did not notably change the findings compared with the analysis of all women. Having the same aim of reducing bias as the Finnish study, our analysis differs in that we excluded women who had experienced the same outcome in any of their previous pregnancies. After this exclusion, the risks associated with previous cesarean delivery were, in general, reduced. However, because this method may not represent a full accounting of indication for cesarean delivery, confounding by indication might still account for some of the increased risk after a previous cesarean delivery in our results.
We observed an increased risk of uterine rupture associated with previous cesarean delivery, with adjusted ORs ranging from 13.1 to 37.4. This considerable increase in risk is consistent with other population-based cohort studies, some of which report relative risks above 40,14,17 although other studies report relative risks below 20.15,16
An increased occurrence of placental complications after previous cesarean deliveries is well known from the literature.4–12 There is, however, a large degree of heterogeneity in the reported strength of such associations, possibly due to study design or characteristics of the populations studied. In a meta-analysis of placenta previa and previous cesarean delivery, Ananth et al found stronger effects in case-control studies than in cohort studies, with pooled ORs of 3.8 and 2.4, respectively.5 We found relatively modest effects, with ORs of 1.8 or lower for placenta previa and ORs of 2.4 or lower for placental abruption. Our results are in line with more recent cohort studies, where relative risks between 1.2 and 1.7 for placenta previa and placental abruption at second birth are reported.7,11,12 Hemminki et al found ORs around 2 for placental problems at second birth.9 We also found modest effects for placenta accreta, with ORs below 2, which is low compared with those previously reported.12
We also analyzed outcomes that might be related to the function of the placenta, such as bleeding during pregnancy, preeclampsia, and fetal growth measured as SGA. Both preeclampsia and SGA were significantly associated with previous cesarean delivery. In particular, for preeclampsia we saw a strong effect of removing from the analyses women with preeclampsia in any of their previous pregnancies. Preeclampsia is a strong indication for cesarean delivery, has a high risk of recurrence, and is relatively frequent. Hence, a significant effect of taking account of previous complications to reduce confounding by indication could be expected. Still, the association remained significant, with ORs between 1.2 and 1.7. We find no support in the literature for a mechanism causing such an association.
If the risk of adverse pregnancy outcome is increased after one cesarean delivery, one might infer that the risk is even higher after a repeat cesarean delivery. However, for uterine rupture and placental abruption the opposite was observed. This may be because of the fact that two consecutive cesarean deliveries is a strong indication for a cesarean delivery in a third pregnancy. In our study, as many as 97% of such deliveries were cesarean, compared with 28% after one previous cesarean delivery. Because cesarean delivery before labor may prevent placental abruption and uterine rupture, the lower rates of these conditions after repeat cesarean delivery may be due to the extremely high cesarean delivery rate in these births. On the other hand, cesarean delivery most likely cannot prevent the other adverse outcomes that were studied. For these conditions, no dose-response relationship was observed.
We evaluated whether mean birth weight and mean gestational age were influenced by a previous cesarean delivery. If a previous cesarean delivery creates suboptimal placental implantation and thereby suboptimal placental functioning, an adverse effect in terms of lower birth weight or shorter gestational age would be expected. For this purpose, we analyzed sibships where first and third births were vaginal and considered exposure in relation to mode of delivery at second birth. Already after a first vaginal birth, women with a cesarean delivery at second birth had lower mean birth weight and shorter mean gestational age, clearly demonstrating the effect of selection to cesarean delivery. However, because from first to third birth, women with a cesarean delivery at second birth had a higher increase in mean birth weight and a higher increase in mean gestational age than women with a vaginal delivery at second birth, our data do not support a hypothesis of reduced fetal growth or gestational age after cesarean delivery.
In general, the risk of complications in the third birth was lower after a cesarean first and vaginal second delivery compared with a vaginal first and cesarean second delivery. Because the selection to cesarean delivery may differ for first births and second births, this difference might be due to confounding by indication. An alternative explanation is that the risk is related to time since the cesarean delivery. Geetahun et al examined whether the risk of placental complications after cesarean delivery was modified by interpregnancy interval.10 They reported that the risk of placenta previa at second birth was not increased after a short interpregnancy interval among women whose first delivery was vaginal, but the risk was increased by 70% among women who had a cesarean first delivery, suggesting that time since the cesarean delivery affects the risk of complications in subsequent births.
Given an overall current cesarean delivery rate in Norway of 16%, the population-attributable fraction of previous cesarean delivery will be modest for all of the analyzed outcomes except uterine rupture. Suggesting this cesarean delivery rate at first birth, the population-attributable fraction at second birth for conditions with a relative risk of 1.5 will be 7.4%. A reduction of the cesarean delivery rate to 10%, if medically acceptable, would imply a corresponding population-attributable fraction of 4.8%, ie, a reduction of 2.6 percentage points.
In conclusion, this study suggests a moderately increased risk of placenta previa, placenta accreta, and placental abruption and a highly increased risk of uterine rupture after a cesarean delivery. These risks should be considered in the clinical judgment of whether to perform a primary cesarean delivery and in the clinical judgment of obstetric risk among women with one or more previous cesarean deliveries. The results suggest that bias due to confounding by indication plays a significant role in observational studies such as this. Thus, in analyses of outcomes after a previous cesarean delivery, indications for the cesarean delivery should be considered.
1. Medical Birth Registry of Norway. Births in Norway through 30 years. Bergen: University of Bergen; 1997.
3. Kolas T, Hofoss D, Daltveit AK, Nilsen ST, Henriksen T, Hager R, et al. Indications for cesarean deliveries in Norway. Am J Obstet Gynecol 2003;188:864–70.
4. Hemminki E. Impact of cesarean section on future pregnancy—a review of cohort studies. Paediatr Perinat Epidemiol 1996;10:366–79.
5. Ananth CV, Smulian JC, Vintzileos AM. The association of placenta previa with history of cesarean delivery and abortion: a metaanalysis. Am J Obstet Gynecol 1997;177:1071–8.
6. Rasmussen S, Albrechtsen S, Dalaker K. Obstetric history and the risk of placenta previa. Acta Obstet Gynecol Scand 2000;79:502–7.
7. Lydon-Rochelle M, Holt VL, Easterling TR, Martin DP. First-birth cesarean and placental abruption or previa at second birth. Obstet Gynecol 2001;97:765–9.
8. Gilliam M, Rosenberg D, Davis F. The likelihood of placenta previa with greater number of cesarean deliveries and higher parity. Obstet Gynecol 2002;99:976–80.
9. Hemminki E, Shelley J, Gissler M. Mode of delivery and problems in subsequent births: a register-based study from Finland. Am J Obstet Gynecol 2005;193:169–77.
10. Getahun D, Oyelese Y, Salihu HM, Ananth CV. Previous cesarean delivery and risks of placenta previa and placental abruption. Obstet Gynecol 2006;107:771–8.
11. Yang Q, Wen SW, Oppenheimer L, Chen XK, Black D, Gao J, et al. Association of caesarean delivery for first birth with placenta praevia and placental abruption in second pregnancy. BJOG 2007;114:609–13.
12. Kennare R, Tucker G, Heard A, Chan A. Risks of adverse outcomes in the next birth after a first cesarean delivery [published erratum appears in Obstet Gynecol 2007;109:1207]. Obstet Gynecol 2007;109:270–6.
13. Rasmussen S, Irgens LM, Dalaker K. A history of placental dysfunction and risk of placental abruption. Paediatr Perinat Epidemiol 1999;13:9–21.
14. Rageth JC, Juzi C, Grossenbacher H. Delivery after previous cesarean: a risk evaluation. Swiss Working Group of Obstetric and Gynecologic Institutions. Obstet Gynecol 1999;93:332–7.
15. Gregory KD, Korst LM, Cane P, Platt LD, Kahn K. Vaginal birth after cesarean and uterine rupture rates in California. Obstet Gynecol 1999;94:985–9.
16. Taylor LK, Simpson JM, Roberts CL, Olive EC, Henderson-Smart DJ. Risk of complications in a second pregnancy following caesarean section in the first pregnancy: a population-based study. Med J Aust 2005;183:515–9.
17. Kaczmarczyk M, Sparén P, Terry P, Cnattingius S. Risk factors for uterine rupture and neonatal consequences of uterine rupture: a population-based study of successive pregnancies in Sweden. BJOG 2007;114:1208–14.
18. Lydon-Rochelle M, Holt VL, Easterling TR, Martin DP. Risk of uterine rupture during labor among women with a prior cesarean delivery. N Engl J Med 2001;345:3–8.
19. Irgens LM. The Medical Birth Registry of Norway. Epidemiological research and surveillance throughout 30 years. Acta Obstet Gynecol Scand 2000;79:435–9.
20. Ananth CV, Peltier MR, Chavez MR, Kirby RS, Getahun D, Vintzileos AM. Recurrence of ischemic placental disease. Obstet Gynecol 2007;110:128–33.
21. Rasmussen S, Irgens LM. Fetal growth and body proportion in preeclampsia. Obstet Gynecol 2003;101:575–83.
© 2008 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
22. Bender R, Blettner M. Calculating the “number needed to be exposed” with adjustment for confounding variables in epidemiological studies. J Clin Epidemiol 2002;55: 525–30.