Neonatal Mortality Risk for Repeat Cesarean Compared to Vaginal Birth after Cesarean (VBAC) Deliveries in the United States, 1998–2002 Birth Cohorts
To examine trends in repeat cesarean delivery, the characteristics of women who have repeat cesareans, and the risk of neonatal mortality for repeat cesarean birth compared to vaginal birth after cesarean (VBAC). Trends and characteristics of repeat cesareans were examined for: the period 1998–2002 for  all births,  low-risk births (singleton, term, vertex births) and  “no indicated risk” (NIR) births (singleton, term, vertex presentation births with no reported medical risks or complications). For low-risk and NIR births, neonatal mortality rates for repeat cesareans and VBACs were compared. Multivariate logistic regression was used to examine the risk of neonatal mortality for repeat cesareans and VBACs, after controlling for demographic and health factors. In 2002 the repeat cesarean rate was 87.4%, and varied little by maternal risk status or by demographic and health characteristics. From 1998–2002 rates increased by 20% for low risk and by 21% for NIR births, respectively. For low-risk women for the 1998–2002 birth cohorts, the adjusted odds ratio for neonatal mortality associated with repeat cesarean delivery (compared with VBAC) was 1.36 (95% C.I. 1.20–1.55). For NIR women, the adjusted odds ratio was 1.24 (0.99–1.55). The experience of a prior cesarean has apparently become a major indication for a repeat cesarean. Regardless of maternal risk status, almost 90% of women with a prior cesarean have a subsequent (i.e., repeat) cesarean delivery. This is the case even if there was no other reported medical indication. Our findings do not support the widely-held belief that neonatal mortality risk is significantly lower for repeat cesarean compared to VBAC delivery.
KeywordsCesarean deliveryRepeat cesarean deliveryVaginal birth after cesarean (VBAC)Neonatal mortalityLow-risk births
Delivery practices shifted substantially during the years 1989–2004 for women giving birth who previously had a cesarean delivery: After a steady increase from 1989 (18.9%) to 1996 (28.3%) the vaginal birth after cesarean section (VBAC) rate plummeted to 12.6 in 2002 and 9.2% in 2004 . The growing use of VBAC in the early 1990s had generated considerable controversy in the obstetrical community, with numerous articles and editorials advocating restrictions on the use of VBAC, largely because of a fear of uterine rupture [2–7]. These restrictions were based on risk factors (e.g. indication for prior cesarean, number of prior cesareans, nature of uterine scar) , site of care (recommending limiting VBACs to larger hospitals) , and labor management (e.g. use of induction) . By 2001, there was a call for virtual universal use of repeat cesareans for women who had had a previous cesarean . The importance of this debate has increased, because the growing primary cesarean rate has resulted in a much larger population of women giving birth who have had a prior cesarean. In 2002 a record 468,668 or 11.7 of all US births were to women who had had a prior cesarean . This paper examines characteristics of women who have repeat cesareans, and neonatal mortality for repeat cesarean births, compared to VBACs. Although other poor outcomes, such as preterm birth, low birthweight, birth injury, and congenital anomalies may have serious long-term consequences, in this paper we focus on neonatal mortality.
Since 1989, data on method of delivery have been available through the U.S. Standard Certificate of a Live Birth. By 1991, all states and the District of Columbia were reporting information on this item. This paper examines the repeat cesarean section rate, defined here as the number of repeat cesareans per 100 live births to mothers who have had a previous cesarean and whose parity is known to be 2 or greater. We examine trends and characteristics of repeat cesareans to mothers in three groups: (1) all women; (2) those at low risk as defined by the American College of Obstetricians and Gynecologists (mothers with singleton, full term [37+ weeks] births in vertex presentation) , and (3) mothers at “no indicated risk”(NIR). A previous report used national U.S. birth certificate data to create this new category for analysis—mothers at “no indicated risk” and then examined the growth of primary cesareans in these cases from 1991–2001 . Births to women at NIR were singleton, term (37–41 weeks), vertex presentation births that were not reported to have any of 16 medical risk factors1 or 15 complications of labor/delivery2 listed on the 1989 revision of the U.S. Standard Certificate of Live Birth.
We then compared infant and neonatal mortality rates in repeat cesarean and vaginal birth after cesarean (VBAC) deliveries for low-risk and NIR births for the 1998–2002 birth cohorts, the latest national data available. To exclude post term deliveries which may have higher risks, in the mortality analysis (Tables 2, 3) infant gestational age of 37–41 weeks was used for births to low-risk women. After 2002, complete national data for the variables of interest were not available due to implementation of the 2003 Revision of the U.S. Standard Certificate of Live Birth.
Finally, multivariate logistic regression analysis was used to examine neonatal mortality for repeat cesareans and VBACs, after controlling for maternal age, race and Hispanic origin, parity, education, infant birthweight and gestational age. The parameters in the model were estimated by the maximum likelihood method using the LOGISTIC procedure of SAS, version 9.1.3 . Those records with missing data, (less than 4% of all records) were excluded from the models. All statements in the text were tested for statistical significance using the two-proportion z-test, and any differences noted as higher or lower were statistically significant.
Rates of repeat cesareana delivery for all women, low-risk women; and women at “no indicted risk,” (NIR) by selected characteristics: United States Live Births, 2002
Low risk womenb
‘No indicated risk’ womenc,d
American Indian/Alaskan native
Asian or Pacific islander
Trimester prenatal care began
Period of gestation
Under 2,500 gms
Under 12 years
13 years or more
Maternal smoking during pregnancyg
Neonatal mortality rates (per 1,000 live births) for VBAC and Repeat Cesarean births to women with low-risk and ‘no indicated risk’ by age at death, 1998–2002 birth cohort
Low risk womena
No indicated risk womenb
Ratio repeat C/VBAC
Ratio repeat C/VBAC
Early neonatal (<7 days)
Late neonatal (7–27 days)
Number of deaths
Early neonatal (<7 days)
Late neonatal (7–27 days)
Adjusted odds ratios and 95% confidence intervals for neonatal mortality (<28 days) for low risk and ‘no indicated risk’ births delivered by VBAC or repeat cesarean: United States, 1998–2002 birth cohort
Logistic regression model
Method of delivery
Adjusted odds ratioa
Model 1—low risk womenb
Model 2—no indicated risk womenc
There have been dramatic changes in method of delivery for women who had experienced a prior cesarean, with a rapid growth in vaginal birth after cesarean in the early 1990s followed by a pronounced decline beginning in 1996 . It has been suggested that convenience of patient and/or provider may be a factor, perhaps more so for repeat cesarean delivery [13, 14].
When we examined the subpopulation of women who had no reported medical indications for a cesarean than a prior cesarean, we found that their rates of repeat cesarean delivery paralleled that of the population at large. It appears that the experience of a prior cesarean may have become a major indication for a repeat cesarean for more than 90% of mothers, even if there was no other reported medical indication. Repeat cesarean rates varied somewhat by region of the U.S., but rates were high for all. Geographic differences in rates of repeat cesarean delivery may be due to many factors, including the capacity of rural hospitals to support VBAC delivery, variation in physician practice patterns and patient preferences [15, 16]. There was little difference in repeat cesarean rates across different demographic groups (i.e., maternal age, race, education).
The rapid decline in VBACs that began in the mid 1990s and continued to the present has been justified on the basis of greater safety for mothers and for infants. American College of Obstetricians and Gynecologists (ACOG) guidelines for VBAC recommend that VBAC should only be attempted if all personnel and equipment for an emergency C-Section are available within 30 minutes . Further, a well publicized editorial  by Greene in the New England Journal of Medicine concluded that if safety of the infant were the issue, the writer’s “unequivocal recommendation” would be repeat cesarean . Our research suggests that consideration of neonatal outcomes could lead to different conclusions. We found, even after controlling for a wide array of confounders, that for low-risk women, the adjusted odds ratio for neonatal mortality was 36% higher for repeat cesareans compared to VBAC. For NIR women the adjusted odds ratio of 1.24 for neonatal mortality was marginally non-significant. Even though five years of data are combined for the multivariate modeling, small numbers of events in some categories (i.e. only 96 births in the NIR VBAC category) may have led to a lack of statistical power in the NIR model, and thus the marginally non-significant result.
In a previous study comparing NIR primary cesareans and vaginal births to women who had not had a previous cesarean (i.e., not VBACs), we found a larger difference in outcomes than in the present study . In that paper, the neonatal mortality rate for “NIR” vaginal births (not VBAC) was 0.62 per 1,000 and the rate for NIR primary cesarean births was 1.77 per 1,000. Neonatal mortality rates for NIR vaginal births (not VBACs) are comparable to those associated with “NIR” VBACs (0.66 per 1,000). Women having a first cesarean comprise a population with different risks than women having a repeat cesarean or a VBAC. However, neonatal mortality rates for NIR repeat cesareans (0.79 per 1,000) in this study, although higher than for NIR VBACs, were lower than those for primary cesareans (1.77 per 1,000), in the previous study. Our findings of equal or greater neonatal mortality rates for repeat cesarean deliveries, when compared to VBACs differ from the recommendation of Greene  and others [2, 4, 6] who have concluded that infant safety (as measured by lowered risks of mortality and morbidity) is greater for repeat cesareans than for VBACs.
While a number of studies questioning the safety of VBACs appeared in the mid to late 1990s [2, 3, 20], more recent studies have raised cautions about the widespread use of repeat cesareans [21, 22]. Two issues have been identified in these analyses. One is the increasing concern with the consequences of multiple repeat cesareans, particularly in terms of placental difficulties . The odds of placental problems (previa, percreta, accreta, and abruption) begin to rise rapidly with repeat cesareans  and so while the consequences for the initial repeat cesarean may not be felt by the baby in the index birth, subsequent births are at higher risk. The second issue is that given the relatively small likelihood of symptomatic uterine rupture, it requires a very large number of repeat cesareans to avoid a single uterine rupture that might be associated with a VBAC .
Investigators have noted a lack of sufficient evidence on the benefits and risks of planned VBAC vs. planned repeat cesarean delivery [26, 27]. Zweifler, et.al, reported higher mortality only for very low birthweight infants delivered by VBAC (possibly due to precipitous delivery) . Several studies found that infants in planned VBAC deliveries had much lower rates of respiratory morbidity and NICU admission than infants in elective repeat cesarean deliveries. Neonatal mortality rates were not reported [28, 29]. Studies directly comparable to our study were not found in the literature.
The strengths of our study include the comprehensive population-based nature of the data set, which includes all births and over 98% of infant deaths in the United States from 1998 to 2002, together with the large number of socio-demographic and medical variables available for analysis. Less than 4% of the records examined were excluded from this study due to missing data. Because the neonatal mortality rates for both repeat cesarean and VBAC deliveries for this low-risk population are extremely low, differences in risk of neonatal mortality could only be detected in a very large data set. Most clinic or hospital-based studies would have insufficient power, in terms of sample size, to detect a statistically significant difference in mortality of less than 1 infant death per 1,000 live births.
Our study has several limitations. Underreporting of medical risk factors and complications of labor and/or delivery in national birth certificate data has been documented [30–32]. Although underreporting may affect the level of the NIR measure, trends in the cesarean rate for NIR women are consistent with those for all women. For the NIR measure, the problem of underreporting of any specific item is somewhat lessened by the fact that our analysis does not rest on a single item, but rather looks at cases where none of those items is reported. Although data from the National Hospital Discharge Survey has consistently reported similar rates of cesarean delivery, higher rates of VBAC estimated from hospital records in 2003, (14.8 vs. 10.6) [1, 33], suggest that VBAC may be underreported on the birth certificate.
Also, we could not control for individual hospital characteristics, such as the availability of staff and services that allow women who have had a previous cesarean delivery to attempt a VBAC. It is also possible that because of a condition that is not reported on the birth certificate, mothers who have a repeat cesarean delivery are at higher risk than those who deliver vaginally. However, when about 90% of mothers with a previous cesarean are experiencing a repeat cesarean, this suggests that perhaps a previous cesarean is the main force driving the decision to perform a repeat cesarean. We are unable to explain why repeat cesarean rates for NIR mothers are higher compared to those for all mothers and low-risk mothers for each demographic and behavioral risk factor examined in Table 1. This may be further evidence that whether a woman has a repeat cesarean section is not related to her risk status; or this may reflect either risks that are not reported or poorer reporting of risk factors in an emergency situation.
For low-risk women, in terms of neonatal mortality risks, our findings suggest that a VBAC may be at least as safe as or safer than a repeat cesarean. The neonatal mortality rate was higher for low-risk women who had a repeat cesarean delivery compared to those who delivered vaginally. The difference in the neonatal mortality rate for NIR women who had a repeat cesarean delivery compared to those who delivered vaginally was not significant.
There has been a shift in obstetrical practice since 1996, from a repeat cesarean rate of 72%, to close to 90% of mothers experiencing a repeat cesarean by 2002  even if there were no other medical indications reported. Our research joins that of other recent studies suggesting there is a need for better understanding of the absolute and relative risks of mortality and morbidity related to trial of labor and labor management practices with respect to vaginal birth after a previous cesarean delivery (VBAC). The 2003 Revision of the U.S. Standard Certificate of Live Birth collects data on whether a trial of labor was attempted prior to a cesarean delivery . Other information potentially available from the revised certificate may help clarify the possible role of medical infections and maternal morbidity in decisions on medical intervention, including repeat cesarean delivery. Continued research, utilizing large databases and careful case ascertainment can further identify ways to most safely manage birth among all women with a prior cesarean.
Anemia, cardiac disease, acute or chronic lung disease, diabetes, genital herpes, hydramnios/oligohydramnios, hemoglobinopathy, chronic hypertension, pregnancy associated hypertension, eclampsia, incompetent cervix, previous infant 4000 + grams, previous preterm or small small-for-gestational-age infant, renal disease, Rh sensitization, uterine bleeding.
Febrile, meconium moderate/heavy, premature rupture of membrane, abruptio placenta, placenta previa, other excessive bleeding, seizures during labor, precipitous labor, prolonged labor, dysfunctional labor, breech/malpresentation, cephalopelvic disproportion, cord prolapse, anesthetic complication, fetal distress.