Volume 345:3-8 |
|
July 5, 2001 |
|
Number 1 |
Risk of
Uterine Rupture during Labor among Women with a Prior Cesarean Delivery
Mona Lydon-Rochelle, Ph.D., Victoria L.
Holt, Ph.D., Thomas R. Easterling, M.D., and Diane P. Martin, Ph.D.
ABSTRACT
Background Each year in
the United States, approximately 60 percent of women with a prior
cesarean delivery who become pregnant again attempt labor. Concern
persists that a trial of labor may increase the risk of uterine
rupture, an uncommon but serious obstetrical complication.
Methods We conducted
a population-based, retrospective cohort analysis using data from
all primiparous women who gave birth to live singleton infants by
cesarean section in civilian hospitals in Washington State from 1987
through 1996 and who delivered a second singleton child during the
same period (a total of 20,095 women). We assessed the risk of
uterine rupture for deliveries with spontaneous onset of labor,
those with labor induced by prostaglandins, and those in which labor
was induced by other means; these three groups of deliveries were
compared with repeated cesarean delivery without labor.
Results Uterine
rupture occurred at a rate of 1.6 per 1000 among women with repeated
cesarean delivery without labor (11 women), 5.2 per 1000 among women
with spontaneous onset of labor (56 women), 7.7 per 1000 among women
whose labor was induced without prostaglandins (15 women), and 24.5
per 1000 among women with prostaglandin-induced labor (9 women). As
compared with the risk in women with repeated cesarean delivery
without labor, uterine rupture was more likely among women with
spontaneous onset of labor (relative risk, 3.3; 95 percent
confidence interval, 1.8 to 6.0), induction of labor without
prostaglandins (relative risk, 4.9; 95 percent confidence interval,
2.4 to 9.7), and induction with prostaglandins (relative risk, 15.6;
95 percent confidence interval, 8.1 to 30.0).
Conclusions For women
with one prior cesarean delivery, the risk of uterine rupture is
higher among those whose labor is induced than among those with
repeated cesarean delivery without labor. Labor induced with a
prostaglandin confers the highest risk.
Each year in the United States,
approximately 60 percent of women with a prior cesarean delivery
have a trial of labor in a subsequent pregnancy. Concern persists
that a trial of labor may increase the risk of maternal
complications as compared with elective cesarean delivery. Such
complications include uterine rupture, which is uncommon but serious
and may result in hysterectomy, urologic injury, a need for blood
transfusion, maternal death, and perinatal complications, including
neurologic impairment and death.1,2,3,4
Population-based studies of the relation between a trial of labor
and uterine rupture have had methodologic limitations and have
produced inconsistent findings. A study in Nova Scotia, Canada,
reported that a trial of labor was not significantly associated with
uterine rupture; however, in that study, too few women had uterine
rupture to provide meaningful results.2
In contrast, studies in Switzerland and California demonstrated a
significantly higher risk of uterine rupture among women undergoing a
trial of labor than among women with elective repeated cesarean delivery.1,5
However, these studies did not control for parity or the number of
prior cesarean deliveries. In addition, although the rates of
induction of labor among women with prior cesarean delivery have
been increasing, none of these studies distinguished the risk of
uterine rupture associated with a trial of labor with induction of
labor from that without induction.6 We
used statewide linked birth-certificate and hospital-discharge data
to examine the risk of uterine rupture associated with spontaneous onset
of labor, induction of labor not involving prostaglandins, induction
of labor with prostaglandins, and repeated cesarean delivery without
labor among women with one prior cesarean delivery.
Methods
Study Design
We conducted a population-based, retrospective cohort analysis
using data obtained from the Washington State Birth Events Record Database.
This data base links more than 95 percent of birth certificates in
Washington State with maternal and infant records from the Comprehensive
Hospital Discharge Reporting System for the hospitalization
associated with delivery. The present cohort included all
primiparous women who gave birth to live singleton infants by
cesarean section in civilian hospitals in Washington from January 1,
1987, through December 31, 1996, and who delivered a second
singleton child in Washington during the same period (a total of
20,525 women). Because a variable indicating that the women had a
second cesarean section without labor ("repeat cesarean no labor")
was not added to the birth certificates until 1989, we excluded the
430 women who had a second delivery before 1989. After the
exclusion, 20,095 subjects remained for analysis. Demographic
variables were derived from first and second birth certificates,
information on payers from maternal and infant
hospitalization-discharge data for the second delivery, and medical
information from maternal and infant hospitalization-discharge data
and birth certificates for both deliveries. The study was approved
by the Human Subjects Review Committee at the University of
Washington, Seattle, and the Human Research Review Board at the
Washington State Department of Health, Olympia.
Definitions
Hospitals were classified as level III (providing tertiary care,
with complete perinatal services), level II (with at least 500 births
per year, with board-certified obstetricians and pediatricians on
staff, and providing newborn intermediate care), or level I (having
a licensed obstetrical unit, with fewer than 500 births per year or
without one or more level II criteria).
A delivery was classified as a repeated cesarean delivery without
labor if "repeat cesarean no labor" was checked on the birth
certificate and if labor-related procedure or diagnosis codes of
the International Classification of
Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)7 were
not recorded on hospital-discharge forms. Labor was considered to
have been induced if "induction of labor" was checked on
the birth certificate or if any ICD-9-CM medical-induction procedure
or diagnosis codes were recorded in hospital-discharge data. Induced
labor was classified as induction of labor with prostaglandins if
ICD-9-CM procedure code 96.49 was recorded on the hospital-discharge
form. All other induced labor was classified as induction of labor
without prostaglandins. According to these criteria, there were 6980
women who had repeated cesarean delivery without labor (34.7 percent),
1960 women who had induction of labor without prostaglandins (9.8
percent), 366 women who had induction of labor with prostaglandins (1.8
percent), and 10,789 women who had spontaneous onset of labor (53.7
percent) available for analysis. Uterine rupture was considered to
have occurred if ICD-9-CM diagnosis code 665.0 or 665.1 was recorded
on the hospital-discharge form.
Statistical Analysis
To assess the risk of uterine rupture associated with spontaneous
onset of labor, induction of labor without prostaglandins, and induction
of labor with prostaglandins, as compared with repeated cesarean
delivery without labor, we used Mantel–Haenszel rate ratios to
estimate the relative risks and 95 percent confidence intervals.8
Interactions between the mother's labor status at the second
delivery and the type of uterine incision at the first delivery and
the year of the second delivery were assessed by the likelihood-ratio
test, with P values below 0.05 denoting statistical significance. No
significant interactions were found. The following variables,
reported at the time of the second delivery, were examined for
possible confounding effects in all analyses: maternal age; race or
ethnic background; marital status; smoking status during pregnancy;
presence or absence of preexisting diabetes mellitus, chronic
hypertension, moderate-to-severe preeclampsia, and genital herpes;
interval between the deliveries; payer; hospital level; infant birth
weight and estimated gestational age; presence or absence of breech
presentation; and presence or absence of placenta previa. Variables
were considered to be confounders if their inclusion changed the
model's relative risk for uterine rupture at the second delivery
associated with any labor-status category by 10 percent or more.
With this criterion, no variables were considered to confound the
models.
Because misoprostol was introduced into obstetrical practice for
induction of labor in Washington in 1996 and has been linked to a
higher risk of uterine rupture, we compared the risk of uterine
rupture associated with prostaglandin-induced labor with that
associated with repeated cesarean delivery without labor, with
stratification according to birth year (before 1996 or during 1996).
Because prostaglandin induction may be used differently according to
whether there are chronic or perinatal health conditions — which
may, in turn, be independently associated with uterine rupture — we
performed a secondary analysis limited to the 18,419 women without
diabetes mellitus, chronic hypertension, moderate-to-severe
preeclampsia, breech presentation, genital herpes, or placenta
previa (91.7 percent). Because a prior vertical uterine cesarean
incision may also affect a woman's risk of uterine rupture, we
conducted an analysis limited to the 19,822 women (98.6 percent)
without a vertical cesarean uterine incision at the first delivery.
Finally, since the severity of uterine rupture cannot be determined
from diagnostic codes, we examined the frequency of the diagnosis of
selected postpartum complications among women with and without
uterine rupture. Differences between the two groups were compared
with use of the Mantel–Haenszel chi-square test. Because the frequency
of postpartum complications was low, it was not possible to evaluate
the relation between labor status and specific complications of
uterine rupture.
Results
Demographic and perinatal characteristics at the time of the second
delivery were similar among women with spontaneous onset of labor
and women with no trial of labor (Table 1). Women
who underwent induction without prostaglandins were more likely
than women who did not have a trial of labor to deliver infants whose
estimated gestational age was more than 42 weeks. Women who
underwent prostaglandin induction were less likely to deliver within
two years of their first delivery and more likely to deliver at a
level II hospital than women who had no trial of labor. The
frequency of medical conditions and complications of pregnancy
varied substantially among the groups (Table 2). Women
who had spontaneous onset of labor were significantly less likely
than women with no trial of labor to have diabetes mellitus, chronic
hypertension, preeclampsia, a breech presentation, genital herpes,
or placenta previa. Women with induction without prostaglandins were
significantly less likely than women who did not undergo labor to
have breech presentation, genital herpes, or placenta previa.
Finally, women with prostaglandin induction were significantly less
likely to have breech presentation or genital herpes than women who
did not undergo labor.
|
|
In our study cohort, uterine rupture complicated 4.5 second singleton
deliveries per 1000 (91 women). Uterine rupture occurred at a rate
of 1.6 per 1000 among women with repeated cesarean delivery without
labor (11 women), 5.2 per 1000 among women with spontaneous onset of
labor (56 women), 7.7 per 1000 among women whose labor was induced
without prostaglandins (15 women), and 24.5 per 1000 among women
with prostaglandin-induced labor (9 women). Women with spontaneous
onset of labor were more likely than women who did not undergo labor
to have uterine rupture (relative risk, 3.3; 95 percent confidence
interval, 1.8 to 6.0) (Table 3). A
greater relative risk was observed among women with induced labor
without prostaglandins (relative risk, 4.9; 95 percent confidence
interval, 2.4 to 9.7), and particularly those with labor induced by
prostaglandins (relative risk, 15.6; 95 percent confidence interval,
8.1 to 30.0). For women giving birth before misoprostol became
generally available in 1996, the relative risk of uterine rupture
associated with prostaglandin-induced labor was 14.1 (95 percent
confidence interval, 6.1 to 33.0). The risk was similar among women
who gave birth in 1996 (relative risk, 12.2; 95 percent confidence
interval, 3.4 to 39.6). The risk estimates for uterine rupture
associated with spontaneous or induced labor were not materially
changed when we excluded women with diabetes mellitus, chronic
hypertension, preeclampsia, breech presentation, genital herpes, or
placenta previa (data not shown). There were no uterine ruptures
among the 272 women with previous vertical incisions, and the
results were unchanged when data on these women were excluded (data
not shown). Women with uterine rupture were significantly more
likely than women without uterine rupture to have postpartum
complications (Table
4).
|
|
Discussion
In this longitudinal cohort study of 20,095 women whose first
delivery of a singleton child was by cesarean delivery, a trial of
labor at the second delivery was associated with a significant increase
in the risk of uterine rupture. The risk of rupture was increased by
a factor of approximately three in women with spontaneous labor;
however, the rate of uterine rupture among these women was still
quite low. The risk of uterine rupture was highest among women whose
labor was induced, particularly when it was induced by
prostaglandins.
Our findings are consistent with the results of two
cross-sectional studies, which demonstrated increased risks of
uterine rupture with a trial of labor in women with prior cesarean
section.1,5
A countrywide study in Switzerland, which included 92 uterine ruptures,
observed a doubling of the risk among women with a trial of labor,
as compared with those who underwent elective repeated cesarean
delivery.1
Similar risks were found in a California cohort analysis that used
1995 statewide hospital-discharge data, in which 393 uterine
ruptures were reported.5 One
longitudinal cohort study in Nova Scotia estimated that the relative
risk of uterine rupture associated with a trial of labor, as
compared with elective repeated cesarean delivery, was 5.2, but only
11 women had uterine rupture, and this increase in risk was not
significant.2
None of those studies examined the effect of induction of labor on
the risk of uterine rupture associated with a trial of labor. Three
prior case series reported an increase in uterine rupture among
women with prior cesarean delivery in whose second deliveries labor
was induced by prostaglandins.9,10,11
However, these studies did not include a comparison group of women
whose second deliveries were by cesarean section without labor.
In prior studies, information about a trial of labor was based
on physician-survey or hospital-discharge data alone, and labor status
may have been misclassified. Information in the present study came
from linked birth-certificate and hospital-discharge data — an
approach that increases the accuracy and completeness of data on
obstetrical diagnoses and procedures.12
The use of longitudinally linked maternal data sets also allowed us
to use an entire state's cohort of women with one prior cesarean
delivery who had a second singleton delivery during a 10-year period;
this method ensured that the number of subjects was adequate for the
examination of a rare outcome. Since all levels of hospitals
statewide were included, our findings represent a wide range of
hospital settings.
Data derived from vital statistics and administrative records
may be limited in completeness and in the accuracy of coding of
obstetrical data. However, previous research has shown that 99.8
percent of cesarean deliveries are correctly classified when
Washington State linked birth-certificate and hospital-discharge files
are used.12
Inaccuracy in the measurement of exposure is possible, however; a
previous study found that only 72 percent of women with induction of
labor were correctly classified in the Washington State Birth Events
Record Database, and no studies have reported on the accuracy of
classification of the other exposure groups.12
Nonetheless, because the recording of labor status is unlikely to
depend on uterine-rupture status, any misclassification would be
random and would thus lead to underestimation of the risk associated
with labor. Although we could not document the accuracy of coding
for uterine rupture, the observed rate of uterine rupture of 4.5 per
1000 among women with prior cesarean delivery was consistent with
the results of other studies (range, 3.2 per 1000 to 6.4 per 1000).1,3,5
Furthermore, the increased frequency of adverse postpartum
complications among women with a diagnosis of uterine rupture
suggests that this diagnosis code was clinically meaningful.
We did not have information on specific types and dosages of prostaglandin
used, and therefore we could not evaluate the effects of different
preparations. Although the American College of Obstetricians and
Gynecologists currently advises against the use of misoprostol in
women with prior uterine surgery, because of the reported increased
frequency of uterine rupture, this prostaglandin analogue may have
been used during 1996, the last year of our study period.11,13
However, the observation that the risk of rupture associated with
prostaglandin-induced labor was increased in the years before
misoprostol was available indicates that this preparation alone
could not have been responsible for the increased risk seen with
prostaglandin use.
An increased risk of uterine rupture may be attributed to factors
other than labor status at a second delivery among women with prior
cesarean delivery, and these factors may also influence the decision
to undertake a trial of labor. We restricted our analysis to second
singleton births, eliminating the potentially confounding effects of
parity, more than one cesarean delivery, and multiple gestation, all
of which could have predisposed women both to a scheduled repeated
cesarean delivery and to uterine rupture.14,15
The results were similar when we excluded data from women with
preexisting medical conditions or complications of pregnancy that
might be expected to influence the mode of delivery and when we
excluded data from women with prior vertical incisions.
At present, the data suggest that induction of labor increases
the risk of uterine rupture among women with one prior cesarean delivery
and that labor induced with use of a prostaglandin confers a greater
relative risk. The overall effect of induction of labor with
prostaglandins on uterine rupture is still unclear and may vary
according to the preparation used, the regimen, and the degree of
cervical readiness for induction.
Supported in part by grants from the Agency for Healthcare
Research and Quality (5 T32 HS00034) and from the National
Institutes of Health–National Institute of Nursing Research (T32
NR07039).
We are indebted to Nancy Fugate-Woods, Ph.D., for assistance and
comments and to William O'Brien for data linkage and management support.
Source Information
From the Center for Women's Health Research, Department of Family
and Child Nursing, School of Nursing (M.L.-R.), the Department of Epidemiology,
School of Public Health and Community Medicine (V.L.H.), the Department of
Obstetrics and Gynecology, School of Medicine (T.R.E.), and the Department of
Health Services, School of Public Health and Community Medicine (M.L.-R.,
D.P.M.), University of Washington, Seattle.
Address reprint requests to Dr. Lydon-Rochelle at the Center for
Women's Health Research, Mailstop 357262, University of Washington, Seattle, WA
98195-7262, or at [log in to unmask].
References
Vaginal Delivery after
Cesarean Section — Is the Risk Acceptable?
Nearly 40 years ago,
Douglas et al. documented the risk of uterine rupture during a trial
of labor after cesarean section.1
Among more than 2000 women with scars from prior cesarean sections,
uterine rupture occurred during labor in just over 1 percent, and
more than a third of the fetuses in those cases died. The authors
concluded, "Probably the most vehement objections to the policy
of vaginal delivery after cesarean section allude to the occurrence
of catastrophic ruptures of the uterine scar." That experience,
coupled with the mantra "Once a cesarean, always a
cesarean," cast the practice of vaginal delivery after cesarean section
into widespread disrepute for decades.
Subsequently, the U.S. national cesarean-section rate rose
dramatically, peaking in the late 1980s at nearly 23 percent. This
increase attracted considerable attention, since cesarean sections
generally result in greater short-term and long-term maternal morbidity,
maternal mortality, and expense than vaginal deliveries. A third of
the cesarean sections were elective repeated procedures, with
absolute numbers rising as the rate of primary cesarean section
rose.
An obvious way to curb the rising cesarean-section rate was to
revive the practice of vaginal delivery after cesarean section. Enthusiastic
reports of success2,3 and
an optimistic meta-analysis that found no increase in perinatal
mortality with a trial of labor4
propelled the juggernaut. Insurers initiated programs to promote
vaginal delivery after cesarean section and to educate doctors and
patients regarding its benefits. Some insurers even felt justified
in refusing to pay for repeated cesarean sections that were not
preceded by attempts at vaginal delivery. Individual physicians, and
to some extent their professional organizations, were cajoled or
coerced into supporting these programs, lest they be dropped from
the roster of providers or portrayed as uncooperative and
nonprogressive. As experience again accumulated, however, so too did
reports of maternal and perinatal morbidity and mortality associated
with vaginal delivery after cesarean section; most of these adverse
effects were attributable to uterine rupture.5,6
Efforts to improve the safety of vaginal delivery after cesarean
section have focused on attempts to identify risk factors for
uterine rupture. Recent investigations into the effects of oxytocic
agents have yielded inconsistent results without a clear message.7,8,9
In this issue of the Journal,
Lydon-Rochelle et al. report on a study in which they used a large
statewide data base of more than 20,000 women in Washington State to
examine the risk of uterine rupture associated with various modes of
delivery and induction of labor.10
All the women had undergone one previous cesarean section and were
delivering a second singleton baby. The overall risk of uterine
rupture was 4.5 per 1000, which is consistent with rates in other
studies.7,8,9 The
rate of uterine rupture associated with spontaneous onset of labor
was 5.2 per 1000 and that associated with induction of labor without
prostaglandins was 7.7 per 1000. These rates were significantly higher
than the 1.6 per 1000 among women who had repeated cesarean sections
without labor, but not significantly different from one another.
Most striking was the rate of uterine rupture, 24.5 per 1000, when
labor was induced with a prostaglandin.
Previous reports of two small, uncontrolled case series11,12
suggested that a 5 to 10 percent risk of uterine rupture was associated
with use of misoprostol, a highly potent synthetic prostaglandin E1
analogue, to induce labor. The American College of Obstetricians and
Gynecologists subsequently warned against the use of misoprostol for
this purpose.13
Although Lydon-Rochelle et al. did not have information about the
prostaglandin preparations used to induce labor in the women in
their cohort, they performed an analysis with stratification
according to year to look specifically at deliveries before and
after 1996, when misoprostol began to be widely used. They found
that the risk associated with the use of a prostaglandin to induce
labor was similar during the two periods, suggesting that the risk
is not limited to misoprostol but extends to other prostaglandin
preparations as well.
The 91 women in the study by Lydon-Rochelle et al. who had uterine
ruptures had substantially greater rates of several postpartum complications,
suggesting that these ruptures were clinically important and not
merely instances of asymptomatic dehiscence. Notably, the incidence
of infant death was 10 times as high among the 91 women who had
uterine rupture as among the 20,004 who did not (5.5 percent vs. 0.5
percent). It is important to emphasize that this study, like all
others to date, was an observational study of the results of
clinical practice and not a randomized trial.
The relative risk of 3.3 in the present study for uterine rupture
in women with a spontaneous onset of labor, as compared with those
who underwent elective repeated cesarean section, is consistent with
the odds ratio of 2.1 for a similar comparison reported in a recent
meta-analysis of 11 studies involving a total of 39,000 subjects.14
This meta-analysis also found significant increases in the risks of
fetal death (odds ratio, 1.7) and of an Apgar score of less than 7
at five minutes (odds ratio, 2.2) associated with a trial of labor
as compared with elective repeated cesarean delivery. These risk
estimates reflect broad experience in a wide range of
clinical-practice settings. There is no reason to believe that
improvements in clinical care can substantially reduce the risks of
uterine rupture and perinatal mortality.
Given the potential risks, why might a woman choose a trial of
labor? Women who successfully deliver vaginally generally have less
postpartum discomfort, shorter hospital stays, and shorter periods
of disability than women who undergo repeated cesarean section. A
trial of labor may be associated with a lower risk of fever than
elective repeated cesarean section.14
Women who plan future pregnancies may prefer to avoid repeated cesarean
deliveries that further increase the risks of uterine rupture,
placenta accreta, and morbidity related to multiple abdominal
surgeries. Finally, there may be social and cultural reasons why some
women prefer vaginal delivery.
Slovic recognized that "experts" (such as health care
policy analysts, public health officials, and insurance company
executives) perceive risk differently from laypeople (patients).15
Experts judge risk according to technical estimates of numbers of
deaths. The rate of perinatal death, for example, is 5.8 per 1000
with trials of labor after cesarean section, as compared with 3.4
per 1000 with elective repeated cesarean section.14
The absolute difference between them is 2.4 per 1000 (or 1 per 417),
a relatively small number. Patients are more apt to judge risk
according to the degree to which they "dread" the unwanted
outcome. "Dread," in turn, is determined by the degree to
which the outcome is irreversible, potentially lethal, and
uncontrollable. By these criteria, the possibility of perinatal
death resulting from a trial of labor would probably be associated
with a high degree of dread.
The process of obtaining informed consent for medical care
requires that physicians provide patients with the information that
a reasonable person would want under the circumstances. Most
reasonable women considering a trial of labor after a prior cesarean
delivery would want to know that spontaneous labor is associated
with a tripling of the risk of uterine rupture and that induction
of labor with prostaglandins is associated with an increase in
that risk by a factor of 15. Should a rupture occur, the risk of
perinatal mortality increases by a factor of 10. Some reasonable
persons may conclude that these absolute risks are so small that
they are worth taking and are outweighed by the benefits of a
successful vaginal birth. Nonetheless, these issues must be
discussed with each patient, and she must make that decision for
herself. After a thorough discussion of the risks and benefits of
attempting a vaginal delivery after cesarean section, a patient
might ask, "But doctor, what is the safest thing for my
baby?" Given the findings of Lydon-Rochelle et al., my
unequivocal answer is: elective repeated cesarean section.
Michael F. Greene, M.D.
Massachusetts General Hospital
Boston, MA 02114
References
Edward E.
Rylander, M.D.
Diplomat American
Board of Family Practice.
Diplomat American
Board of Palliative Medicine.