Customize your JAMA Network experience by selecting one or more topics from the list below.
Smith GCS, Pell JP, Cameron AD, Dobbie R. Risk of Perinatal Death Associated With Labor After Previous Cesarean Delivery in Uncomplicated Term Pregnancies. JAMA. 2002;287(20):2684–2690. doi:10.1001/jama.287.20.2684
Author Affiliations: Department of Obstetrics and Gynaecology, Cambridge University, Cambridge, England (Dr Smith); Department of Public Health, Greater Glasgow Health Board (Dr Pell), and Department of Fetal Medicine, The Queen Mother's Hospital (Dr Cameron), Glasgow, Scotland; and Information and Statistics Division, Common Services Agency, Edinburgh, Scotland (Mr Dobbie).
Context Trial of labor after previous cesarean delivery is associated with increased
risk of uterine rupture. However, no reliable data exist on the effect of
a trial of labor on the risk of perinatal death in otherwise uncomplicated
Objective To determine the risk of intrapartum stillbirth or neonatal death not
related to congenital abnormality among women with uncomplicated term pregnancies
who had a trial of labor after previous cesarean delivery, compared with women
having a planned repeat cesarean delivery, and multiparous and nulliparous
women at term not delivered by planned cesarean method.
Design and Setting Population-based, retrospective cohort study of data from the linked
Scottish Morbidity Record and Stillbirth and Neonatal Death Enquiry encompassing
births in Scotland between January 1, 1992, and December 31, 1997.
Population A total of 313 238 singleton births between 37 and 43 weeks' gestational
age in which the fetus was in a cephalic presentation.
Main Outcome Measure Delivery-related perinatal death, defined as intrapartum stillbirth
or neonatal death unrelated to congenital anomaly, compared among the 4 groups.
Results Among women who had a trial of labor following previous cesarean delivery
(n = 15 515), the overall rate of delivery-related perinatal death was
12.9 (95% confidence interval [CI], 7.9-19.9) per 10 000 women. This
was approximately 11 times greater (odds ratio [OR], 11.6; 95% CI, 1.6-86.7)
than the risk associated with planned repeat cesarean delivery (n = 9014),
more than twice (OR, 2.2; 95% CI, 1.3-3.5) the risk associated with other
multiparous women in labor (n = 151 549), and similar to the risk among
nulliparous women in labor (n = 137 160; OR, 1.3; 95% CI, 0.8-2.1). The
associations were not explained by differences in maternal height, smoking
status, socioeconomic status, age, fetal growth, or week of gestation at delivery.
Among women having a trial of labor, the rate of death due to mechanical causes,
including uterine rupture, was 4.5 (95% CI, 1.8-9.3) per 10 000 women.
This was more than 8 times greater than other multiparous women (OR, 8.5;
95% CI, 3.2-22.3) and nulliparous women (OR, 8.8; 95% CI, 3.2-24.2).
Conclusions The absolute risk of perinatal death associated with trial of labor
following previous cesarean delivery is low. However, in our study, the risk
was significantly higher than that associated with planned repeat cesarean
delivery, and there was a marked excess of deaths due to uterine rupture compared
with other women in labor.
Increasing rates of cesarean delivery are a major cause for concern
in almost all developed countries.1 A number
of strategies have been proposed that aim to reduce the overall proportion
of cesarean deliveries, including trial of labor after previous cesarean deliveries.2 Observational studies3,4
suggest that trial of labor is associated with a significantly increased risk
of uterine rupture. A meta-analysis reported an increased risk of perinatal
death associated with trial of labor5 but included
premature births between 28 and 36 weeks' gestation and breech deliveries.6 There is no large-scale study of the relative and
absolute risks of perinatal death among women previously delivered by cesarean
method but with an uncomplicated pregnancy at term. In the present study,
we sought to address this by linking national registers of pregnancy discharge
data and perinatal deaths.
The analysis was designed to determine the risk of intrapartum stillbirth
or neonatal death unrelated to congenital abnormality among women with an
uncomplicated term pregnancy who had a trial of labor following at least 1
previous cesarean delivery. Trial of labor was defined as any vaginal or emergency
(unplanned) cesarean delivery occurring at or beyond 37 weeks' gestation in
a woman who had previously been delivered by cesarean method. These women
were compared with 3 groups: (1) women having planned repeat cesarean delivery,
(2) other multiparous women at term not delivered by planned cesarean method,
and (3) nulliparous women at term not delivered by planned cesarean method.
Nulliparous women were women with no previous pregnancies or whose previous
pregnancies all ended in abortion.
The Scottish Morbidity Record (SMR2) collects information on clinical
and demographic characteristics and outcomes for all patients discharged from
Scottish maternity hospitals. The register is subjected to regular quality
assurance checks and has been more than 99% complete since the late 1970s.7 An analysis of 1414 records from 1996 through 1997
demonstrated that the register was free of significant errors in more than
98% of records in all the fields used in the present analysis, with the exception
of postcode (94.0%), height (96.2%), estimated gestation (94.4%), and method
of induction of labor (93.6%) (Jim Chalmers, MBChB, Information and Statistics
Division, National Health Service, Edinburgh, Scotland, written communication,
April 2001). The SMR2 records were linked to records from the Scottish Stillbirth
and Neonatal Death Enquiry. This national register has routinely classified
all perinatal deaths in Scotland since 1983. It is almost 100% complete and
has been described in detail elsewhere.8
The study group consisted of all births in Scotland between January
1, 1992, and December 31, 1997. The exclusion criteria for the study group
were multiple pregnancy, noncephalic presentation, delivery outside the range
of 37 to 43 weeks' gestation, perinatal deaths due to congenital anomaly,
antepartum stillbirth not due to congenital anomaly, and deliveries by planned
cesarean method, except among women who had been delivered by cesarean method
in a previous pregnancy.
The main outcome of this study was delivery-related perinatal death,
defined as intrapartum stillbirth or neonatal death not caused by congenital
anomaly. Stillbirths were defined as newborns that showed no signs of life
following delivery. Stillbirths were subdivided into antepartum (deaths before
the onset of labor) and intrapartum (deaths during labor). Neonatal death
was defined as death during the first 4 weeks of life in a live newborn. Deaths
caused by congenital anomaly were defined as any structural or genetic defect
incompatible with life or potentially treatable but causing death.
In the comparison of risk of perinatal death among groups of women,
the following factors were considered as possible confounders: socioeconomic
deprivation, smoking, maternal age, maternal height, gestational age, and
birth weight. Postcode of residence was used to derive Carstairs socioeconomic
deprivation scores.9 These are based on 1991
census data on car ownership, unemployment, overcrowding, and social class
within postcode sectors of residence that contain, on average, approximately
1600 residents. The deprivation scores were then used to categorize women
into quintiles of socioeconomic deprivation within the study cohort. Higher
quintiles indicate a greater degree of deprivation. Smoking was defined as
the smoking status of the woman at the time of first attendance for antenatal
care. Maternal age was defined as the age of the mother at the time of birth.
Maternal height was measured in centimeters, and the value used was that documented
in each woman's clinical record. Height is generally measured using a free-standing
or wall-mounted height measure. Gestational age at birth was defined as completed
weeks of gestation on the basis of the estimated date of delivery in each
woman's clinical record. Gestational age has been confirmed by ultrasound
in the first half of pregnancy in more than 95% of women in the United Kingdom
since the early 1990s.10 Birth weight was categorized
into sex- and gestational age-specific deciles, using a method previously
described in detail.11 Low birth weight was
defined as birth weight of less than 2500 g.
The cause of perinatal death was classified according to a hierarchical
system that is described in detail elsewhere.12
Deaths were initially classified according to the following direct obstetric
causes (in order): toxemia, hemorrhage, mechanical, maternal, and none of
these obstetric causes. Mechanical was defined as death from uterine rupture,
cord compression (including prolapse), birth trauma, or asphyxia associated
with disproportion. In the absence of any of the listed direct obstetric causes,
the deaths were classified by the pediatric diagnoses and these were grouped
into intrapartum anoxia, other (pulmonary causes, intracranial hemorrhage,
infection, other hemorrhage, and miscellaneous), and unexplained. The hierarchy
dictates that a perinatal death where there was severe preeclampsia complicated
by abruption would be classified as being due to toxemia because toxemia is
above both hemorrhage and intrapartum anoxia in the hierarchy.
Continuous variables were summarized by the median and interquartile
range, and comparisons between groups were performed using the Mann-Whitney U test. Univariate comparisons of dichotomous data were
performed using the χ2 test (>5 observations in all cells)
or Fisher exact test (≤5 observations in ≥1 cells). Ordinal data were
compared using the χ2 test for trend. The P values for all hypothesis tests were 2-sided and .05 was the significance
level. Adjusted rates were obtained using direct standardization. Crude and
adjusted odds ratios (ORs) were obtained using logistic regression analysis.
Cases with missing values were excluded from the multivariate analysis. The
statistical significance of interaction terms was assessed using the likelihood
ratio test. Model goodness-of-fit was assessed using the Hosmer-Lemeshow test
based on deciles of probability.13 All statistical
analyses were performed using version 7.0 of the Stata software package (Stata
Corp, College Station, Tex).
Of the SMR2 records with a perinatal death documented, 97.8% could be
linked to a corresponding record in the Stillbirth and Neonatal Death Enquiry.
From 1992 through 1997, there were 356 958 records of singleton births
in Scotland in the SMR2 database. Among these records, there were 697 (0.2%)
with missing values for gestational age, 4873 (1.4%) with missing values for
presentation at delivery, and 8 (<0.1%) with missing values for number
of previous cesarean deliveries. A total of 5564 records had 1 or more missing
values, leaving 351 394 records. Among this group, there were 16 427
fetuses (4.7%) with a noncephalic presentation, 621 perinatal deaths (0.2%)
related to congenital anomaly, 1479 antepartum stillbirths (0.4%) unrelated
to congenital anomaly, 20 628 births (5.9%) outside the range of 37 to
43 weeks' gestational age, and 9827 births (2.8%) by planned cesarean method
where the women had not had a previous cesarean delivery. A total of 38 156
records had 1 or more exclusion criteria, resulting in a study group of 313 238
(87.8% of all singleton births during the study period).
The study group was subdivided into 15 515 women previously delivered
by cesarean method who had a trial of labor, 9014 women who had previously
been delivered by cesarean method who delivered by planned cesarean method
in the current pregnancy, 137 160 nulliparous women who were not delivered
by planned cesarean method, and 151 549 multiparous women who had not
had a previous cesarean delivery and did not deliver by planned cesarean method
in the current pregnancy. Within the study group, there were missing values
for maternal height in 26 825 women (8.6%), for smoking status in 29 730
(9.5%), for deprivation quintile in 4 975 (1.6%), for birth weight in
61 (<0.1%), and for 5-minute Apgar score in 36 (<0.1%).
The study group characteristics and basic outcome data are given in Table 1. The highest rate of perinatal
death (12.9 per 10 000 women) was seen among women having a trial of
labor (Table 2). The risk of a
delivery-related perinatal death among women having a trial of labor was more
than 11 times (OR, 11.6; 95% confidence interval [CI], 1.6-86.7) that of women
having a planned repeat cesarean delivery (Table 3). The risk of death associated with a trial of labor was
similar when compared with nulliparous women in labor (OR, 1.3; 95% CI, 0.8-2.1)
but was more than twice that of other multiparous women in labor (OR, 2.2;
95% CI, 1.3-3.5). When delivery-related perinatal deaths among all women who
had previously been delivered by cesarean method were analyzed, it was estimated
that 91% (95% CI, 36%-99%) could be attributed to the increased risk of death
associated with a trial of labor.
Statistical comparison with the planned repeat cesarean delivery group
was problematic because there was only a single death among the 9014 women.
All the records with missing values were among the 9013 survivors and, therefore,
excluding missing records resulted in a weaker association between trial of
labor and perinatal death compared with planned repeat cesarean delivery simply
by reducing the denominator in the latter group (Table 3). However, among the cases with nonmissing values, adjusting
for maternal age, smoking status, height, deprivation quintile, gestational
age at birth, and birth weight decile strengthened the association between
trial of labor and perinatal death when compared with elective repeat cesarean
delivery (OR, 11.7; 95% CI, 1.4-101.6). When women having a trial of labor
were compared with nulliparous and other multiparous women, adjusting for
maternal age, smoking status, height, deprivation quintile, gestational age
at birth, and birth weight decile had no effect on the ORs (Table 3).
When the analyses were confined to births at or after 40 weeks' gestation,
the results were similar (Table 2).
Although there were no deaths among the 1064 planned repeat cesarean deliveries
at or after 40 weeks' gestation, the numbers were too small to be statistically
significantly lower than the trial of labor group. Among births at or after
40 weeks' gestation, the risk of death associated with a trial of labor was
similar when compared with nulliparous women in labor (OR, 1.2; 95% CI, 0.6-2.2)
but higher when compared with other multiparous women in labor (OR, 2.7; 95%
Among women previously delivered by cesarean method, 369 (2.4%) of those
having a trial of labor had more than 1 previous cesarean delivery, whereas
2962 (32.9%) of those delivered by planned cesarean method had more than 1
previous cesarean delivery. Of the 20 perinatal deaths among women having
a trial of labor, 19 women (95%) had only 1 previous cesarean delivery. Among
women previously delivered by cesarean method, 5206 (33.6%) of those having
a trial of labor had previously had a vaginal birth, whereas 988 (11.0%) of
those delivered by planned cesarean method had previously had a vaginal birth.
Of the 20 perinatal deaths among women having a trial of labor, 5 women (25%)
had previously had a vaginal birth. Among the trial of labor group, there
were 12 perinatal deaths among 3945 neonates born by emergency cesarean delivery
and 8 deaths among 11 570 neonates delivered vaginally (P = .001).
The rates of perinatal death due to different causes differed among
the 4 groups (Table 4). Compared
with other multiparous women, women having a trial of labor had more than
8 times the risk of a perinatal death due to a mechanical cause (OR, 8.5;
95% CI, 3.2-22.3) and almost 3 times the risk of a perinatal death due to
intrapartum anoxia (OR, 2.8; 95% CI, 1.3-6.5). When compared with nulliparous
women, women having a trial of labor had an increased risk of perinatal death
due to mechanical causes alone (OR, 8.8; 95% CI, 3.2-24.2).
Women undergoing a trial of labor were less likely to have a prostaglandin
E2 (PGE2) induction, had shorter labors, had lower rates
of operative vaginal delivery, and had higher rates of emergency cesarean
delivery than nulliparous women (Table 5). Women undergoing a trial of labor were more likely to have a
PGE2 induction, had longer labors, and had higher rates of both
operative vaginal and cesarean delivery than other multiparous women. Among
women having a trial of labor, there were 3 perinatal deaths among 2395 women
induced with PGE2 and 17 deaths among 13 120 women not treated
with PGE2 (P>.99).
In addition to the 15 515 women who fulfilled the criteria for
trial of labor at term, there were 35 women who fulfilled the same criteria
except that the newborn was an antepartum stillbirth not caused by congenital
abnormality at term. Fifteen of these women delivered before 39 weeks, 10
delivered in the 39th week of gestation, and 10 delivered at or after 40 weeks'
The ideal means to determine the risks and benefits of trial of labor
vs planned repeat cesarean delivery would be a randomized controlled trial.
In practice, this would be difficult to perform because many women would be
unhappy to have such a decision made in a random manner and large numbers
of women would be required for a study powered to detect differences in such
rare outcomes as uterine rupture and perinatal death. In the absence of randomized
controlled trial evidence, analysis of observational data must be used to
estimate the risks of these rare outcomes. A recent observational study4 has reported on the relative and absolute risks of
uterine rupture associated with trial of labor. Herein, we report the risk
of perinatal death associated with trial of labor among women at term with
a singleton pregnancy in a cephalic presentation.
In the present study, the risk of delivery-related perinatal death among
women having a trial of labor was more than 11 times that of women having
a planned repeat cesarean delivery. Women having a planned repeat cesarean
delivery experienced the lowest death rate among any of the groups. The low
rate of death was explained by the absence of any risk of intrapartum stillbirth
and a significantly lower risk of neonatal death. The risk associated with
planned repeat cesarean delivery at term was so low that only a single death
occurred in Scotland during the 6 years of the study period. Multivariate
statistical comparison among groups is problematic when the number of events
is so small, and larger studies will be required to analyze adequate numbers
of deaths following planned cesarean delivery. However, selection bias is
unlikely to explain the lower risk of death among women having a planned repeat
cesarean delivery, because they are more likely to have medical and obstetric
complications than women offered a trial of labor.4
Consistent with this, adjusting for maternal age, smoking status, height,
deprivation quintile, gestational age at birth, and birth weight decile for
gestational age strengthened the association between trial of labor and perinatal
death when compared with women having a planned repeat cesarean delivery,
though the CIs were wide because of the small number of cases.
Our observed rate of perinatal death associated with planned repeat
cesarean delivery, 1.1 per 10 000 women, is much lower than previously
cited.5 However, previously published perinatal
mortality figures for both trial of labor and planned cesarean delivery did
not exclude breech presentations and preterm newborns delivered between 28
and 36 weeks' gestation.6 Since prematurity
and breech presentation are associated with an excess of perinatal mortality,14,15 these data are unhelpful in informing
women who reach term with a fetus presenting cephalically. This group accounted
for 84% of women with a previous cesarean delivery in our study.
When compared with other multiparous women in labor, women having a
trial of labor had approximately twice the rate of delivery-related perinatal
death. This finding was due to an increased risk of death due to mechanical
causes, including uterine rupture, and death due to intrapartum anoxia not
related to uterine rupture. The overall rate of delivery-related perinatal
death among women having a trial of labor was not significantly greater than
nulliparous women in labor. The increased number of deaths due to mechanical
causes among women having a trial of labor compared with nulliparous women
was offset by a lower rate of death due to other causes. The observation that
the level of risk of nonmechanical perinatal death among women having a trial
of labor was intermediate between other multiparous and nulliparous women
probably reflects the fact that approximately one third of women having a
trial of labor had previously had a vaginal birth. There are other possible
factors that could contribute to differences in outcome among the groups that
were not recorded in the database, such as epidural anesthesia, use of electronic
fetal monitoring, and details of maternal medical and obstetric complications.
Further studies will be required to determine whether systematic variation
in any of these variables may contribute to differences in the risk of delivery-related
perinatal death among these groups.
The data presented in this article are collected nationally and form
an extract of a larger data set, which is reported in detail elsewhere.12 Overall, the statistics were comparable with previous
analyses of perinatal deaths. The overall rate of intrapartum stillbirth unrelated
to congenital abnormality of 2.5 per 10 000 births was comparable with
previous studies from Scandinavia,16 and the
total proportion of all stillbirths that were classified as intrapartum was
11%, which is similar to national data from England.17
The number of neonatal deaths observed in our study was lower than a report
from Wales,18 which described 7.4 neonatal
deaths attributable to an intrapartum event per 10 000 births. However,
that study included neonates of all gestational ages, using a birth-weight
cutoff of more than 1499 g, which would have included a significant proportion
of preterm births.
Current recommendations are that planned cesarean delivery should be
performed in the 39th week of gestation to reduce the risk of neonatal respiratory
morbidity.19 It could be argued that uterine
rupture may occur in earlier weeks of gestation and that the apparent protective
effect of planned cesarean delivery is exaggerated. However, 85% of delivery-related
perinatal deaths at term among women having a trial of labor occurred at or
after 39 weeks' gestation. This is consistent with the observation that approximately
15% of multiparous women will undergo labor between the start of the 37th
week and the start of the 39th week of gestation.20
Therefore, it seems likely that most deaths could have been avoided by planned
cesarean delivery at the start of the 39th week of gestation. Moreover, planned
cesarean delivery at this time would also avoid exposure to the risk of antepartum
stillbirth while awaiting the onset of labor. There were 20 antepartum stillbirths
among women having a trial of labor at or after 39 weeks' gestation and a
proportion of these may also have been prevented by planned cesarean delivery
at the start of the 39th week of gestation. Therefore, the potential protective
effect on perinatal death of planned cesarean delivery over trial of labor
may be greater than estimated in the present study of intrapartum stillbirths
and neonatal deaths.
The definition of a trial of labor used in this study was that a woman
who had previously been delivered by cesarean method was delivered at term
by a method other than planned cesarean. However, it is likely that a small
proportion of these women were due to have a planned cesarean delivery but
presented in labor before their scheduled date and an emergency cesarean delivery
was performed in early labor. Moreover, the database did not include information
on the nature of the incision used in the previous cesarean delivery. For
this reason, we repeated the analysis confined to births at or after 40 weeks'
gestation. By this time, all women scheduled for planned cesarean delivery
should have had the procedure, including any women who had previously had
a classic cesarean delivery. The risks of perinatal death were virtually unchanged,
suggesting that our results are robust. The outcome of trial of labor was
comparable with previous studies: 75% had a vaginal delivery, which was almost
identical to an analysis of more than 17 000 trials of labor from Switzerland.6
Our data provide essential information for women to make an informed
choice about a trial of labor. Overall, the point estimate of the risk of
a perinatal death associated with a trial of labor is 1 in 775, and the 95%
CIs indicate that the risk is unlikely to be higher than 1 in 500. The point
estimate of the risk of a perinatal death due to uterine rupture associated
with a trial of labor is 1 in 2200, and the 95% CIs indicate that the risk
is unlikely to be higher than approximately 1 in 1000.
Considerable caution should be applied when extrapolating these data
to considering possible benefits of planned cesarean delivery among women
who have not previously had a cesarean birth. More than one third of delivery-related
perinatal deaths among multiparous women who had not previously been delivered
by cesarean method were observed before 39 weeks of gestation. Scheduling
planned cesarean delivery for the start of the 39th week of gestation may
have failed to prevent many of these deaths. Moreover, we present no data
on the risk of perinatal death following planned cesarean delivery among nulliparous
Obstetricians have faced pressure from government and health care insurers
to advocate vaginal birth after cesarean delivery as one strategy to reduce
the overall rate of cesarean delivery. However, this pressure has been exerted
in the absence of any reliable information on the risks to the newborn for
most women. This study is the first to our knowledge that is adequately powered
and analyzed to provide information on the risks of perinatal death associated
with the management of women with a history of cesarean delivery but an otherwise
uncomplicated pregnancy at term.