Ananth CV, Berkowitz GS, Savitz DA, Lapinski RH. Placental Abruption and Adverse Perinatal Outcomes. JAMA. 1999;282(17):1646-1651. doi:10.1001/jama.282.17.1646
Author Affiliations: Division of Epidemiology and Biostatistics, Department of Obstetrics, Gynecology, and Reproductive Sciences, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, New Brunswick (Dr Ananth); Departments of Community and Preventive Medicine (Dr Berkowitz) and Obstetrics, Gynecology, and Reproductive Sciences (Dr Lapinski), Mount Sinai School of Medicine, New York City, New York; Department of Epidemiology, School of Public Health, University of North Carolina, Chapel Hill (Dr Savitz).
Context Pregnancies complicated by abruption result in increased frequency of
perinatal death and decreased fetal size and gestational duration, yet the
extent of placental separation and its effect on these adverse outcomes is
Objective To assess the contribution of placental abruption and extent of placental
separation to stillbirth, preterm delivery, and fetal growth restriction.
Design Hospital-based, retrospective cohort study.
Setting Mount Sinai Hospital, New York City, NY.
Participants A total of 53,371 pregnancies occurring in 40,789 women who were delivered
of singleton births between 1986 and 1996.
Main Outcome Measures Risks and relative risks for stillbirth (>20 weeks), preterm delivery
(<37 weeks), and fetal growth restriction (birth weight below 10th percentile
for gestational age) in relation to abruption.
Results The incidence of abruption was 1% (n = 530). Abruption was associated
with an 8.9-fold (95% confidence interval [CI], 6.0-13.0) adjusted relative
risk (aRR) of stillbirth. Preterm birth proportions among women with and without
abruption were 39.6% and 9.1%, respectively, yielding an aRR of 3.9 (95% CI,
3.5-4.4). In the abruption group, 14.3% of neonates were growth restricted,
compared with 8.1% among all other births (aRR, 2.0; 95% CI, 1.5-2.4). Extent
of placental separation had a profound effect on stillbirth (aRR for 75% separation,
31.5; 95% CI, 17.0-58.4), although evident only among those with at least
50% separation. However, the risk of preterm delivery was substantially increased
even for mild abruptions (aRR for 25% separation, 5.5; 95% CI, 4.2-7.3).
Conclusions In this cohort, placental abruption had a profound impact on stillbirth,
preterm delivery, and fetal growth restriction. The risk of stillbirth was
dramatically increased for severe placental separation, but preterm delivery
was common even among women with lesser degrees of placental separation.
Placental abruption results from a cascade of pathophysiologic processes
ultimately leading to the separation of the placenta prior to delivery. Pregnancies
complicated by abruption result in increased frequency of low birth weight,
preterm delivery, stillbirth, and perinatal death.1- 4
Attempts to understand this condition have focused on describing its etiology,
with little attention directed toward evaluating its sequlae on outcomes of
pregnancy. Even among studies that have reported associations between placental
abruption and adverse perinatal outcomes, many have been based on unadjusted
analysis without adequate control for the influences of potential confounding
Finally, the extent of placental detachment and its effect on adverse pregnancy
outcomes and the dose-response gradients between the extent of placental separation
and adverse pregnancy outcomes have not been examined.
The etiologies of low birth weight (<2500 g) are diverse and include
early delivery (prematurity), fetal growth restriction, fetal intrinsic disorders
(such as chromosomal disorders, genetic syndromes, or congenital infections),
or a combination of these. Therefore, low birth weight, by itself, as an indication
of fetal well-being is not helpful in trying to develop prevention strategies
unless the various causal pathways that ultimately lead to this outcome are
We designed this retrospective cohort study to address 2 objectives.
First, to delineate the components of low birth weight, preterm delivery,
and fetal growth restriction associated with abruption. The second objective
was to evaluate the association between extent of placental separation and
stillbirth and early delivery.
This study was approved by the institutional review board, and was based
on women who were delivered of a singleton birth at Mount Sinai Hospital,
New York City, NY, between January 1986 and December 1996. Approximately two
thirds of women who give birth at Mount Sinai Hospital are private patients,
while the remaining third seek care through the clinic service. A majority
of private patients are from greater New York City, while most of the clinic
patients come from neighboring Harlem and the South Bronx. Approximately 50%
of the women who give birth at Mount Sinai Hospital are white (non-Hispanic),
25% are Hispanic, 20% are non-Hispanic black, and the remainder are Asian
and other minorities. The Hispanic population is predominantly Puerto Rican.
Detailed descriptions of the patient population have been published elsewhere.6
Data for this study were derived from a detailed, computerized perinatal
database. The database includes extensive information on antepartum, intrapartum,
and postpartum course in the hospital, assembled from computer-adapted medical
record forms. The majority of these standardized forms are completed by the
attending physician, and, to a lesser extent, by nurses. All forms are routinely
reviewed for accuracy and completeness by a database coordinator. Routine
review of the data is performed every month to address inconsistent information
and missing data.
Information pertaining to gestational age includes date of last menstrual
period and an obstetric estimate of the expected date of delivery, with an
indication whether that estimate was based on menstrual date, sonography,
or clinical estimate. In addition, a pediatric estimate based on the Ballard
neonatal assessment was recorded.7 All these
estimates were routinely compared for any significant discrepancies. Assignment
of gestational age (in completed weeks) was based on the "best" obstetrical
estimate. If gestational age based on menstrual dates differed by less than
2 weeks when compared with a sonographic estimate before the third trimester,
then gestational age was assigned based on menstrual dates. If the patient
was unsure of the date of her last menstrual period, or if the discrepancy
was more than 2 weeks, then gestational age assignment was based on the sonographic
assessment. The final assignment of gestational age was based on last menstrual
period in 69% of births, ultrasound in 29%, and clinical estimate in 2% of
Placental abruption was defined as the partial or complete separation
of the placenta prior to delivery of the fetus. This condition is based on
a diagnosis by the attending or resident physician. In addition, the percentage
of placental separation was also available for about 82% of women diagnosed
as having abruption. This was based on gross clinical examination of the placenta
by the attending obstetrician at the time of delivery.
The study population comprised 53,675 singleton births. We excluded
292 pregnancies with a diagnosis of placenta previa and a further 12 pregnancies
with missing data on birth weight, leaving us with 53,371 births (occurring
to 40,789 women) for analyses.
Risk of stillbirth (occurring after 20 weeks) among pregnancies with
and without placental abruption was evaluated, with further stratification
based on the time of stillbirth (antepartum and intrapartum stillbirths).
Abruption-related pregnancy outcomes were examined among singleton livebirths,
including very low (<1500 g), moderately low (1500-2499 g), and all low
birth weight (<2500 g) infants. These categories were compared with infants
of normal birth weight (≥2500 g). Analyses relating to birth weight were
further stratified based on gestational duration (preterm vs term delivery).
Assessment of gestational duration in relation to placental abruption
included very preterm (<32 weeks), moderately preterm (32-36 weeks), and
all preterm deliveries (<37 weeks), and were compared with term (≥37
weeks) deliveries. Primary clinical manifestations leading to preterm delivery
were classified as (1) membrane rupture prior to onset of labor; (2) spontaneous
preterm labor; and (3) medically indicated preterm delivery, defined as early
delivery due to obstetric intervention. For analyses pertaining to clinical
manifestations of preterm delivery, we excluded women whose pregnancies were
treated with tocolytic agents and terminated after 37 weeks because such pregnancies
may have resulted in preterm delivery had tocolysis not been performed. Finally,
we examined the association between abruption and fetal growth restriction,
the latter defined as infants whose birth weight fell below the 10th percentile
for gestational age after adjustment for sex and race/ethnicity based on US
national standards.8 Since norms used for defining
growth restriction were available for pregnancies that ended between 25 and
42 weeks, women whose pregnancies terminated outside this range were excluded
from this particular analysis. Growth-restricted neonates were further divided
into preterm or term delivery.
The database contains extensive information on sociodemographic, lifestyle
risk factors, gynecologic history, and obstetric history, as well as medical
complications and procedures related to the current pregnancy. Sociodemographic
characteristics include maternal age, parity, marital status, self-reported
information on race/ethnicity (non-Hispanic white, Hispanic, non-Hispanic
black, and other), and type of insurance coverage or payment source (third
party, Medicaid, self-pay, and other). Lifestyle factors included cigarette
smoking and alcohol and drug (marijuana, methadone, heroin, and cocaine) use
Details on obstetric history included prior abortion, prior premature
birth, and family history of hypertensive disorders, bleeding disorders, and
diabetes. Details on obstetric complications of pregnancy included preexisting
or chronic hypertension, mild and severe preeclampsia, pregnancy-induced hypertension,
gestational and insulin dependent diabetes, oligohydramnios and polyhydramnios,
pyelonephritis, renal disease, and incompetent cervix.
We calculated risks of adverse pregnancy outcomes, including stillbirth,
low birth weight, preterm delivery, and fetal growth restriction in relation
to placental abruption. The effect of abruption on the outcome was evaluated
by computing both the unadjusted and adjusted relative risks (aRRs). Adjusted
odds ratios were derived from multivariable logistic regression analysis after
controlling for potential confounding variables. The unadjusted and adjusted
odds ratios were then transformed to aRRs based on the methods described by
Zhang and Yu.9
An important assumption of regression models is that observations are
independent of one another. However, analyses of multiple pregnancies to the
same woman violates the independence assumption and failure to account for
intracluster dependence will often result in biased variances of the effect
measure (ie, odds ratios) from the multivariable regression models. In order
to account for this dependence, we fit all regression models using generalized
estimating equations procedure.10 Adjusted
odds ratios and 95% confidence intervals (CIs) estimated from multivariable
logistic regression models based on generalized estimating equations carry
the same interpretation as those derived from a simple logistic regression
Risks of stillbirth and preterm delivery in relation to the degree of
placental separation was generated by allowing smooth terms for the extent
of placental separation. The smooth terms were based on the restricted cubic
spline smoothing procedure.11 Cubic splines
are nonparametric smoothing procedures that do not impose any restriction
on the shape of the distribution. Several regression models for stillbirth
and preterm delivery were fit by allowing different smoothing criteria based
on the number of knots; the most parsimonious model for these outcomes evaluated
based on the likelihood-ratio test11 was retained.
Knot locations for the model for stillbirth were assigned at 0%, 5%, and 70%
placental separation, while the locations for the model for preterm delivery
were assigned at 0%, 10%, 30%, and 50%.
Patient status (clinic vs private) and race/ethnicity were included
in all models for adjustment. Other potential confounding variables were considered
for adjustment if their presence in the regression model changed the odds
ratio for placental abruption by at least 10%. Parity was categorized as nulliparous,
1 through 3 pregnancies, and more than 3 pregnancies. Similarly, 3 indicator
variables for women of black, Hispanic, and other race/ethnicity were constructed,
and each was compared with white women.
During the study period, of a total of 53,371 pregnancies (occurring
to 40,789 women) resulted in a singleton birth, of which 530 (1%) were complicated
by placental abruption. The incidence of abruption increased with increasing
parity, although an association with maternal age was not apparent (Table 1). Black women, smokers, and drug
abusers were all at increased risk for abruption.
Frequency of stillbirth (occurring after 20 weeks' gestation) among
women with abruption was 5.3% compared with 0.5% among all other pregnancies,
resulting in an 8.9-fold (95% CI, 6.0-13.0) aRR (Table 2). The association was stronger for intrapartum than antepartum
Among livebirths, birth weight and gestational age were examined as
continuous variables as well as categorical outcomes. Infants born to mothers
with placental abruption weighed, on average, 494 g less than infants born
to women without this condition, after adjustment for potential confounding
variables (Table 3). Similarly,
pregnancies complicated by abruption ended approximately 2 weeks earlier than
Risks of giving birth to a low-birth weight infant among women with
and without placental abruption were 34.7% and 6.7%, respectively (Table 4). After adjustment for confounding
variables, women with abruption had an aRR of 4.6 (95% CI, 4.0-5.3) of giving
birth to a low-birth weight infant. When the analysis was further stratified
based on preterm and term delivery, the association was stronger for term
than preterm pregnancies.
Risks of preterm birth among women with and without placental abruption
were 39.6% and 9.1%, respectively (Table
5), yielding an aRR of 3.9 (95% CI, 3.5-4.4). The association was
much stronger for very preterm (gestational age <32 weeks) births than
for moderately preterm (32-36 weeks) births. The association between abruption
and clinical manifestations of preterm delivery revealed stronger aRRs among
those with spontaneous preterm labor as the precipitating cause. Preterm deliveries
after premature rupture of membranes and intervention for medical indications
had smaller but substantially elevated relative risks.
Risks of growth-restricted babies among women with and without placental
abruption were 14.3% and 8.1%, respectively (Table 6), conferring an aRR of 2.0 (95% CI, 1.5-2.4) in relation
to placental abruption.
Among women with abruption, the extent of placental separation was available
in 82% (435/530) of pregnancies. Of these 435 pregnancies, 54% had less than
25% placental separation; 16% had 25% through 49%; 13% had 50% through 74%;
and 17% had over 75%. Extent of placental separation had a profound effect
on stillbirth (Figure 1), especially
among those with severe placental separation (aRR for 75% placental separation,
31.5; 95% CI, 17.0-58.4). However, the risk of preterm delivery was increased
substantially even in pregnancies with mild placental separation (aRR for
25% separation, 5.5; 95% CI, 4.2-7.3) (Figure
Over the last decade, the incidence of prematurity has remained fairly
stable in the United States at 10%.12 Despite
recent advances in our understanding of the physiology of parturition, prevention
of prematurity and low birth weight continue to pose clinical and public health
concerns. Placental abruption has been associated with a 20% to 40% rate of
preterm delivery.13 Our results extend information
on the relationship of abruption and low birth weight and prematurity by demonstrating
that much of the contribution of abruption to low birth weight is mediated
through shortened gestations, and, to a lesser extent, through growth restriction.
We observed aRRs on the order of 2 through 11 for low birth weight and 2 through
16 for preterm delivery in the presence of abruption.
These results are in general agreement with what has been reported previously,1- 5
but neither the etiologies of low birth weight (prematurity vs fetal growth
restriction) nor the antecedent clinical presentations for preterm delivery
in relation to abruption has been explored previously. Our results indicate
that women with abruption were at substantially increased risk for preterm
labor when compared with those with premature rupture of membranes or medical
indication for preterm delivery. Similar results were reported in another
study based on a national database.14 An association
between peripheral placental separation and idiopathic preterm labor was demonstrated
by Harris and colleagues,15 who observed that
the incidence of fibrin deposition (both perivillous and marginal), decidual
necrosis, and marginal and placental floor hemorrhage were greater among placentas
of preterm compared with term pregnancies. The extravasation of blood at the
placental margin may lead to decidual necrosis, which, in turn, could initiate
the production of prostaglandin, thereby leading to preterm labor.
The second important finding of our study relates to the association
between extent of placental separation and stillbirth and preterm delivery.
Most notably, the risk of stillbirth increased dramatically for women with
over 50% separation of the placenta, implying that these abruptions were probably
the more severe and acute events. However, the risk for preterm delivery was
increased substantially even among women with milder abruptions. Hurd and
colleagues16 examined, through a descriptive
study, the extent of placental separation in their series of 54 cases of abruption
but did not present data on the association between degree of placental separation
and outcome of pregnancy.
The findings on placental abruption noted here have clinical and public
health implications. Established risk factors for placental abruption, such
as cigarette smoking, drug use, chronic hypertension, pregnancy-induced hypertension,
and preeclampsia, are potentially preventable. Some randomized trials and
observational studies have shown that low dosages of aspirin (60 and 80 mg/d)
and calcium supplementation (2 g/d) may prevent pregnancy-induced hypertension
and preeclampsia.17- 19
Similarly, prenatal patient counseling and education on the harmful effects
of smoking and drug abuse during pregnancy can help to reduce the incidence
of placental abruption and other adverse outcomes of pregnancy, including
preterm delivery and growth restriction. Preterm birth prevention programs
might, therefore, benefit if patients suspected to be at risk for developing
abruption are identified. A high level of clinical suspicion for placental
abruption in women presenting with preterm labor or premature rupture of membranes
is clearly warranted. Clinicians should suspect abruption in patients presenting
with spontaneous preterm labor or unexplained vaginal bleeding during the
latter half of pregnancy. Other common symptoms such as uterine tenderness,
excessive uterine contractions, and fetal distress are highly suggestive of
In our analysis pertaining to antecedent clinical presentations for
preterm delivery (Table 5), we
excluded term pregnancies in which the woman had undergone tocolysis. Although
the efficacy of tocolytic agents in prolonging pregnancy is debatable,20 our exclusion makes the study group more homogeneous.
We replicated the entire analysis presented in Table 5 with tocolyzed term pregnancies and the results (not presented)
were essentially unchanged.
The extent of placental separation was largely based on an assessment
by the attending obstetrician. This subjective assessment may have resulted
in some misclassification, especially for the milder forms of this condition.
Women with severe hemorrhage due to placental abruption are more likely to
be classified as having over 50% placental separation, whereas women with
abruption without external bleeding (concealed hemorrhage) are more likely
to be classified as having marginal separation of the placenta. The presence
of severe hemorrhage may serve as a reliable marker for the severity of placental
With regard to residual confounding, drug use, especially crack or cocaine,
is strongly associated with an increased risk for placental abruption, as
well as low birth weight and preterm delivery,21
and self-report is limited in its effectiveness as a marker of drug use.22 Cigarette smoking during pregnancy is associated
with a 2-fold risk for placental abruption.23
The low prevalence of cigarette smoking during pregnancy in this population
suggests some underreporting. Although these risk factors were adjusted in
all the analyses, some residual confounding may remain.
Our results indicate that neonates born to mothers with abruption have
poor rates of survival at birth and are delivered early compared with infants
born to women without abruption. Much of the risk in low birth weight due
to abruption is the consequence of shortened gestation, and, to a lesser extent,
fetal growth restriction. Furthermore, the risk of stillbirth is dramatically
increased for severe placental separation, but preterm delivery appears more
common even among women with lesser degrees of placental separation.