The association of placenta previa with history of cesarean

delivery and abortion: A metaanalysis

Cande V. Ananth, PhD, MPH, a John C. Smulian, MD, MPH," b and Anthony M. Vintzileos, MD b
New Brunswick, New Jersey

OBJECTIVE: Our purpose was to determine the incidence of placenta previa based on the available
epidemiologic evidence and to quantify the risk of placenta previa based on the presence and number of
cesarean deliveries and a history of spontaneous and induced abortion.
STUDY DESIGN: We reviewed studies on placenta previa published between 1950 and 1996 on the basis of
a comprehensive literature search with use of MEDLINE and by identifying studies cited in the references of
published reports. Studies were chosen for inclusion in the metaanalysis if the incidence of placenta previa
and its cross-classification with either prior cesarean delivery or abortions (both spontaneous and induced) or
both were available. We also extracted details about the study design (case-control or cohort study) and place
where they were conducted (United States or other countries). Published case reports dealing with placenta
previa and studies relating to abruptio placentae were excluded from this review. We also restricted the
search to studies published in English. No attempts were made to locate any unpublished studies. Data from
studies identified during the literature search were reviewed and abstracted by a single author. In case of
discrepancies or when the information presented in a study was unclear, abstraction by a (blinded) second
reviewer was sought to resolve the discrepancy.
RESULTS: Data on the incidence of placenta previa and its associations with previous cesarean delivery
and abortions were abstracted. Subgroup analyses were performed to identify potential sources of
heterogeneity by study design and place where they were conducted. Statistical methods used for the
metaanalysis included the fixed-effects logistic regression model, whereas potential sources of
heterogeneity among studies were evaluated by fitting random-effects models. The tabulation of 36
studies identified a total of 3.7 million pregnant women, of whom 13,992 patients were diagnosed with
placenta previa. The reported incidence of placenta previa ranged between 0.28% and 2.0%, or
approximately 1 in 200 deliveries. Women with at least one prior cesarean delivery were 2.6 (95%
confidence interval 2.3 to 3.0) times at greater risk for development of placenta previa in a subsequent
pregnancy. The results varied by study design, with case-control studies showing a stronger relative risk
(relative risk 3.8, 95% confidence interval 2.3 to 6.4) than cohort studies did (relative risk 2.4, 95%
confidence interval 2.1 to 2.8). Four studies, encompassing 170,640 pregnant women, provided data on
the number of previous cesarean deliveries. These studies showed a dose-response pattern for the risk of
previa on the basis of the number of prior cesarean deliveries. Relative risks were 4.5 (95% confidence
interval 3.6 to 5,5) for one, 7.4 (95% confidence interval 7.1 to 7.7) for two, 6.5 (95% confidence interval
3.6 to 11.6) for three, and 44.9 (95% confidence interval 13.5 to 149.5) for four or more prior cesarean
deliveries. Women with a history of spontaneous or induced abortion had a relative risk of placenta previa
of 1.6 (95% confidence interval 1.0 to 2.6) and 1.7 (95% confidence interval 1.0 to 2.9), respectively.
Substantial heterogeneity in the results of the metaanalysis was noted among studies.
CONCLUSION: There is a strong association between having a previous cesarean delivery, spontaneous
or induced abortion, and the subsequent development of placenta previa. The risk increases with number
of prior cesarean deliveries. Pregnant women with a history of cesarean delivery or abortion must be
regarded as high risk for placenta previa and must be monitored carefully. This study provides yet
another reason for reducing the rate of primary cesarean delivery and for advocating vaginal birth for
women with prior cesarean delivery. (Am J Obstet Gynecol 1997;177:1071-8.)

Key words: Cesarean delivery, induced abortion, metaanalysis, placenta previa, spontaneous
abortion

From the Centerfor PErinatal Health Initiatives~ and the Division of Presented at the Seventeenth Annual Meeting of the Society of Perinatal
Maternal-Fetal Medicine, b Department of Obstetrics, Gynecology, and Obstetricians, Anaheim, California, January 20-25, 199Z
Reproductive Sciences, University of Medicine and Dentistry of New Reprints not available from the authors.
Jersey, Robert Wood Johnson Medical School. Copyright © 0 by Mosby-Year Book, Inc.
The Centerfor Perinatal Health Initiatives is supported in part by grant 0002-9378/-1900 $5.00 + 0 6/6/84417
No. 029553from the Robert Wood Johnson Foundation, New Jersey.

1071
1072 Ananth, Smulian, and Vintzileos November 1997
AmJ Obstet Gynecol

Although placenta previa is relatively u n c o m m o n (in- published in English. No attempts were made to locate
cidence of 3 to 9 per 1000 pregnancies), it is regarded as any unpublished studies or studies in abstract form.
one of the leading causes of uterine bleeding during the Multiple articles resulting from the same data source
latter stages in gestation 1 and has been recognized as an (e.g., Collaborative Perinatal Project 2-5) were only in-
important determinant of maternal morbidity and ad- cluded once in the metaanalysis. However, if two studies
verse perinatal outcomes. Pregnancies complicated by came from the same data source but spanned nonover-
placenta previa have resulted in excessively high rates of lapping time periods of data accrual, they were both
preterm delivery, low birth weight, stillbirths, and neo- included in the metaanalysis.
natal and perinatal deaths. Risk factors associated with Data extraction. We identified a total of 41 published
placenta previa include advanced maternal age, multi- studies 2, 6-45on the basis of our inclusion criteria relating
parity, cigarette smoking and "crack" or cocaine use, to placenta previa. From these studies, information on
history of placenta previa, cesarean delivery, spontane- the incidence of placenta previa or its association with
ous and induced abortions, and prior gynecologic sur- history of cesarean delivery and abortions were available
geries. Nonetheless, the etiology of placenta previa from 36 studies. 2' 6-4o These studies were reviewed criti-
largely remains obscure and speculative. In spite of the cally by the first author (C.V.A.), and information on the
advent of ultrasonography to diagnose this disorder and total n u m b e r of pregnancies and the n u m b e r of preg-
the ability to assess fetal lung maturity to appropriately nancies complicated by placenta previa and data on
time delivery, efforts to improve perinatal outcomes in placenta previa cross-classified by prior cesarean delivery
cases of placenta previa continue to pose a challenge. and spontaneous and induced abortions were abstracted.
It appears that the rate of cesarean delivery has been Data on the type of study design (i.e., cohort or case-
increasing steadily over the past two decades. Some control studies) and the country where the study was
studies have observed an increased frequency of placenta carried out were also ascertained. Studies 2°' 4~-43that did
previa among women with a prior history of cesarean not provide sufficient information to carry out a meta-
delivery or abortions, suggesting an association with analysis or those that did not provide data for the
surgical procedures that disrupt the uterine cavity. None- comparison group (e.g., n u m b e r of pregnancies without
theless, the extent to which a history of cesarean delivery placenta previa among women with prior cesarean deliv-
or spontaneous and induced abortion predisposes ery) were excluded from our metaanalysis, although
women to the development of placenta previa is unclear these studies were still included in an analysis of inci-
from earlier studies. dence of previa, when data were available. In case of
We performed a systematic review and naetaanalysis of discrepancies or when the information presented in a
all published studies on placenta previa to determine its study was unclear, abstraction by a (blinded) second
incidence and to quantify the risk of placenta previa on reviewer @C.S.) was sought to resolve the discrepancy.
the basis of the presence and n u m b e r of cesarean Statistical methods for metaanalysis. The incidence
deliveries and a history of spontaneous and induced (risk) of placenta previa from cohort studies was ob-
abortions. In addition, the systematic review of all studies tained by dividing the n u m b e r of cases of placenta previa
enabled us to also identify sources of heterogeneity by the total n u m b e r of pregnancies. This information
among studies. was abstracted from case-control studies, if reported. We
calculated odds ratios and their SEs as the effect measure
Material and methods on the basis of the data abstracted from each study. Data
Literature review. We reviewed all studies published that were abstracted from each study were arranged in
between 1950 and 1996 on placenta previa. Studies 2 × 2 tables, and 0.5 was added to cells that contained no
chosen for the review were selected on the basis of a observations to improve the precision of the effect mea-
comprehensive literature search with use of MEDLINE sure. 46 Pooled estimates of odds ratios were obtained by
and by identifying studies cited in the "bibliography of weighting each study by the inverse variance of the effect
published reports. Key words that were used in the measure on a logarithmic scale. This approach to pool-
MEDLINE search included "placenta pr(a)evia," "pla- ing the results assumes that the study populations being
cental disorders," "antepartum h(a)emorrhage," and compared are similar and hence corresponds to a fixed-
"antepartum bleeding." In addition, the key words effects analysis. The validity of pooling the odds ratios was
"c(a)esarean delivery," "c (a) esarean section," "uterine tested (test for heterogeneity) on the basis of a X2 test. 47
surgery," "spontaneous abortion," "induced abortion," A violation of this test implies that the studies being
and "elective abortion" were also used in conjunction grouped differ frmn one another. In the presence of
with the search leading to studies on placenta previa. significant heterogeneity in the effect measure among
Published case reports dealing with placenta previa and studies being compared, we then performed a random-
studies relating to abruptio placentae were excluded effects analysis that was based on the method described
from this review. We also restricted the search to studies by DerSimonian and Laird. as The random-effects analysis
Volume 177, Number 5 Ananth, Srnulian, and Vintzileos 1073
Am J Obstet Gynecol

Incidence
(%)
2.0

1.8

1.8

1.4

1.2

1.0

0.8

0.6

0.zt

0.2

0.0
i . . . . i . . . . i ' , , r i . . . . i . . . . i . . . .

1950 1955 1960 1965 1970 1975 1980 1985 1990

Year of Data Collection

Fig. 1. Trends in incidence of placenta previa (smoothed line was generated based on locally weighted
scatterplot smoother procedure, a nonparametric regression smoothing procedure).

accounts for the interstudy variations. O u r goal of per- previa. T h e r e p o r t e d incidence of placenta previa ranged
f o r m i n g this metaanalysis was to identify sources of between 0.28% to 1.96%, or approximately- 1 in 200
h e t e r o g e n e i t y a m o n g studies. pregnancies. T h e incidence of placenta previa was the
T o e x a m i n e for the p r e s e n c e of any t r e n d s in the same for both c o h o r t and case-control studies. An exam-
i n c i d e n c e o f p l a c e n t a previa over time, we u s e d the ination for trends over time in the incidence of placenta
locally w e i g h t e d scatterplot s m o o t h e r p r o c e d u r e . 49 previa revealed that the incidence of this disorder was
This p r o c e d u r e is a n o n p a r a m e t r i c scatterplot almost similar until the mid-1980s (1966 to 1974: inci-
s m o o t h e r that down-weights observations that are dis- d e n c e was 0.36%; 1975 to 1984, 0.37%), but the inci-
tant f r o m its n e i g h b o r s and, conversely, assigns l a r g e r dence was 0.48% a m o n g studies c o n d u c t e d between
weights to observations that are closer to each other. 1985 and 1995 (Fig. 1).
For metaanalysis r e l a t i n g to the n u m b e r of p r i o r Association with prior cesarean delivery. Associations
c e s a r e a n deliveries, we c o m p u t e d an estimate of log between history of cesarean delivery and placenta previa
odds ratio with their SEs o n the basis of m e t h o d s were evaluated in 15 published studies 6 is, 4+45 (Fig. 2).
d e s c r i b e d by G r e e n l a n d and L o n g n e c k e r . 5° This Pooling of data from all these studies resulted in an odds
m e t h o d adjusts for the c o r r e l a t i o n that results f r o m ratio of 2.6 (95% confidence interval 2.3 to 3.0), al-
use o f a single r e f e r e n c e category (i.e., no p r i o r though the test for h o m o g e n e i t y of the p o o l e d odds
c e s a r e a n delivery) whiIe evaluating the risk of p l a c e n t a ratios was violated (X~ = 222.8, 14 degrees of freedom,
previa by n u m b e r of p r i o r cesarean deliveries. p < 0.0001) (Table I). H e n c e stratification of the studies
T o assess the public health implications of history of on the basis of their study designs resulted in a p o o l e d
cesarean delivery and abortions on placenta previa, we odds ratio of 2.4 (95% confidence interval 2.1 to 2.8) for
also c o m p u t e d the population attributable risk. 51 The cohort 6, 7, 10-12, 16, 17 and 3.8 (95% confidence interval 2.3
p o p u l a t i o n attributable risk can be i n t e r p r e t e d as the to 6.4) for case-control studies, s' 10, 13, 15, 18, 41, 42 Tests for
p r 6 p o r t i o n of the adverse o u t c o m e (i.e., placenta previa) h o m o g e n e i t y of odds ratios for b o t h comparisons were
that could be attributed to cesarean delivery or sponta- violated. Associations between prior cesarean delivery
neous and i n d u c e d abortions. All statistical analyses were and placenta previa were almost the same both for
p e r f o r m e d on the SAS system version 6.11 (SAS Institute, studies based in the U n i t e d States 6-9' 11-14, a7, 41 and else-
Cary, N.C.) operating on the U N I X system. where.V, 10, 11, 15, 16, 42
We further analyzed the association between placenta
Results previa in relation to the n u m b e r of prior cesarean
Incidence of placenta previa. Data abstracted from 36 deliveries. Data available from four studies 7' in, 14, 17 en-
studies 2, 6-4o resulted in a total of 3.7 million p r e g n a n t compassing a total of 170,640 p r e g n a n t w o m e n showed a
women, of w h o m 13,992 were identified with placenta dose-response pattern in the risk of previa with increas-
1074 Ananth, Smulian, and Vintzileos November 1997
AmJ Obstet GynecoI

G o r o d e s k i [41 ]
M a k s h e e d [13]
N i e l s e n [9]
S i n g h [6]
C l a r k [71
Rose [8]
ChaUopadhy [14]
Chelmow [1g I
Hershkowitz [17J
Williams [10]
Taylor [42) t--e~-q

Handler [151
Zhang [ 11]
Thomas [16 I
Parazzini [ 12] :i !:i d :i v .i d ~i ~ i
)

Fixed-Effects
Random-Effects
i
] I 'l
0.1 1.0 10.0 100.0

Odds Ratio (95% Confidence Interval)

Fig. 2. Association of placenta previa with history of cesarean delivery: O d d s ratios with 95% confidence
intervals (reference n u m b e r s in brackets).

Table I. A s s o c i a t i o n b e t w e e n p r i o r c e s a r e a n d e l i v e r y a n d p l a c e n t a previa: A m e t a a n a l y s i s b a s e d o n fixed- a n d

random-effects models

Fixed-effects )~2, Random-effects

Comparison No. of studies OR and 95% CI significance OR and 95% CI

Overalle-ls, 41, 42 15 2.9 (2.8-3.0) 222.8 (p < 0.0001) 2.6 (2.3-3.0)

U.S. studies 7' i0, 11, 15, 16, 18, 42 7 2.2 (1.9-2.5) 17.8 (p < 0.0001) 2.3 (1.7-2.8)
O t h e r countries 6' s, u, 12-15,17, 41 8 2.9 (2.8-3.0) 184.6 (p < 0.0001) 2.4 (2.0-2.9)
Cohort studies 6' 7, e, 11, 12, 16, 17 8 2.9 (2.8-3.0) 174.4 (p < 0.0001) 2.4 (2.1-2.8)
Case-control studies s' 10, 13, 15, 18, 41, 42 7 2.6 (2.2-3.1) 46.9 (p < 0.0001) 3.8 (2.3-6.4)

OR, Odds ratio; CI, confidence interval.

*Degrees of f r e e d o m for X2 test are n u m b e r of studies m i n u s one.

Table II. A s s o c i a t i o n b e t w e e n p r i o r s p o n t a n e o u s a n d i n d u c e d a b o r t i o n s a n d p l a c e n t a p r e v i a : A m e t a a n a l y s i s b a s e d
o n fixed- a n d r a n d o m - e f f e c t s m o d e l s

Fixed-effects X2 Random-effects
Comparison No. of studies OR and 95 % CI p value OR and 95 % C1

Spontaneous abortiont
Overalll5, i8, 28, 52, 53 5 1.7 (1.5-2.0) 3.4 (0.4933) 1.7 (1.0-2.9)
Cohort studies 28 1 3,0 (1.5-2.0) -- --
Case-control studies I5' 18, 52, 53 4 1,7 (1.5-2.0) 2.5 (0.4753) 1.7 (1.5-2.0)
I n d u c e d abortion t
Overall 9, 15, 28, 42, 52, 54 6 1.5 (1.3-1.7) 43.7 (p < 0.0001) 1.6 (1.0-2.6)
Cohort studies 28 1 6.7 (2.7-16.8) -- --
Case-control studies 9' ts, 42, 5~, 54 5 1.5 (1.3-1.7) 33.3 (p < 0.0001) 1.3 (0.7-2.3)

OR, O d d s ratio; CI, confidence interval.

*Degrees of D e e d o m f o r X2 test is n u m b e r of studies m i n u s o n e .
t a l l studies were based in the United States.

ing number o f p r i o r c e s a r e a n deliveries. O d d s ratios c o n f i d e n c e i n t e r v a l 3.6 to 11.6) f o r t h r e e p r i o r c e s a r e a n

w e r e 4.5 ( 9 5 % c o n f i d e n c e i n t e r v a l 3.6 to 5.5) f o r o n e a n d deliveries ( b a s e d o n two s t u d i e s 7, 17), a n d 44.9 (95%
7.4 ( 9 5 % c o n f i d e n c e i n t e r v a l 7.1 to 7.7) f o r two p r i o r c o n f i d e n c e i n t e r v a l 13.5 to 149.5) f o r f o u r o r m o r e p r i o r
c e s a r e a n d e l i v e r i e s ( b a s e d o n all f o u r s t u d i e s ) , 6.5 ( 9 5 % c e s a r e a n d e l i v e r i e s ( b a s e d o n o n e study7). T h e r e w a s a n
Volume 177, Number 5 Ananth, Smulian, and Vintzileos 1075
Am J Obstet Oynecol

Chelmow [18]
Barrett [28]
Handler [15]
g r a m e r [53]
Newton [52] t

iili~ ::i i. 'q:~ ::51:i ~;:vi'~.

Fixed-Effects
Random-Effects
i . . . . . . . . . . . . . . i

0.1 1.0 10.0

Odds Ratio (95% Confidence Intervals)

Fig. 3. Association of placenta previa with history Of spontaneous abortion: Odds ratios with 95%
confidence intervals (reference numbers in brackets).

Barrett [28} t

Handler [15]

Grimes [54] I

i
Taylor [4El

Newton [52]

Williams [10] i ;~.~i IL.' ~.:3.~.~.-:~;

Fixed -Effects

Random-Effects i P C::.'~i <"::,7i

i . . . . . . . . t . . . . . . . . i . . . . . . . . i

0.1 1.O I0.0 I00.0

Odds Ratio (95% Confidence Intervals)

Fig. 4. Association of placenta previa with history of induced abortion: Odds ratios with 95%
confidence intervals (reference numbers in brackets).

e x p o n e n t i a l increase in the risk of placenta previa with risk of subsequent d e v e l o p m e n t of placenta previa. Re-
n u m b e r of prior cesarean deliveries. stricting the analysis to the five case-control stud-
Association with p r i o r abortions. We identified five iesl0, 15, 49.52, 54 resulted in a decreased odds ratio (odds
studies15,1s, zs, 52, ss that evaluated the association be- ratio 1.3, 95% confidence interval 0.7 to 2.3) for placenta
tween prior spontaneous abortion and the subsequent previa, whereas the odds ratio f r o m the single c o h o r t
d e v e l o p m e n t of placenta previa (Fig. 3). The random- study 2s was 6.7 (95% confidence interval 2.7 to 16.8).
effects p o o l e d analysis indicated that the risk of placenta Additionally, with use of data f r o m the vital records,
previa was 1.7 (95% confidence interval 1.0 to 2.9) for Zhang and Savitz r e p o r t e d that the risk of placenta previa
w o m e n with at least one prior spontaneous abortion was 1.6 (95% confidence interval 1.3 to 1.8) for w o m e n
(Table II). All five studies were c o h o r t studies. T h e test with one abortion (either spontaneous or induced), 2.3
for h o m o g e n e i t y of odds ratios was n o t violated (p = (95% confidence interval 1.8 to 3.0) for those with two,
0.4933). T h e association between prior spontaneous and 3.7 (95% confidence interval 2.7 to 5.2) for those
abortion and the subsequent d e v e l o p m e n t of placenta with three or m o r e abortions.
previa was the same when stratified on study design.
Six studies ~°' 15, 2s, 42, 52, 54 r e p o r t e d the association be- Comment
tween placenta previa and history of i n d u c e d abortions Placenta previa has b e e n r e p o r t e d to occur in approx-
(Fig. 4), all of w h o m were based in the U n i t e d States. Five imately 0.3% to 0.8% of p r e g n a n c i e s J A variation in this
of the six studies were case-control studies. A l t h o u g h the incidence has b e e n attributed to m e t h o d s of diagnosis,
test for h o m o g e n e i t y of the p o o l e d odds ratio was definitions used, and diverse nature of patient popula-
violated (X2 = 43.7, 5 degrees of freedom, p < 0.0001), a tions b e i n g studied. A l t h o u g h the overall incidence of
history of spontaneous abortion was associated with a placenta previa has b e e n remarkably stable for almost
70% (95% c o n f i d e n c e interval 1.0 to 2.9) increase in the three decades, the incidence of this disorder was almost
1076 Ananth, Smulian, and Vintzileos November I997
AmJ Obstet Gynecol

similar until the mid-1980s (1966 to 1974: incidence was delivery or abortions as a secondary analysis. Neverthe-
0.36%; 1975 to 1984, 0.37%), but the incidence was less, to account for this bias on our findings, we report
0.48% among studies conducted between 1985 and 1995 the results from random-effects regression models, which
(see Fig. 1). The increased incidence of placenta previa assume that studies included in our metaanalysis is a
in the last decade may be the result of increasing (random) sample from a larger population of similar
cesarean delivery rates during this period or the more studies. In addition, we generated a "funnel-plot" (by
widespread use of ultrasonography for detecting pla- plotting the log odds ratio against their corresponding
centa previa. SEs, graph not shown) to examine for indications of
Several studies, based on ultrasonography findings, publication bias for history of cesarean delivery and
have shown that the incidence of placenta previa is about abortions. The plots did not indicate the presence of any
3% to 5% in a normal obstetric population during publication bias.
midtrimesterY However, this frequency falls dramati- Another shortcoming of this metaanalysis is that we
cally to almost 0.3% to 0.7% among term pregnancies as were unable to evaluate the risk of placenta previa in
a result of the so-called placental "migration." Almost relation to uterine rupture, complications occurring with
four decades ago Bender 56 first observed an i n c r e a s e d an abortion, curettage for postpartum hemorrhage or
frequency of placenta previa among women with uterine retained products of conception, postpartum complica-
scarring (because of cesarean delivery or abortions) in tions such as endometritis, multiple gestations, and other
prior pregnancies. Recently, few studies have explored potential risk factors for this disorder. These data were
this association and have unequivocally observed in-
not universally available from each of the individual
creased risks of placenta previa among women with a
studies. Future prospective studies should evaluate these
history of cesarean delivery. An association between
factors to better understand the etiology of placenta
placenta previa and prior cesarean delivery is biologically
previa.
plausible. Damage to the endometrial and myometrial
Public health implications. The rates of primary cesar-
uterine lining (during cesarean delivery) can predispose
ean delivery have been steadily increasing in the past
to a low implantation of the placenta in the uterus. This
decade. Although this increase has probably improved
metaanalysis quantifies the risk on the basis of the
fetal and neonatal morbidities and other adverse repro-
number of previous cesarean deliveries, implying a dose-
ductive outcomes as well, the public health implications
response effect of multiple uterine procedures. Likewise,
for the rise in cesarean delivery rates have been poorly
curettage of the uterus during a spontaneous or induced
addressed.
abortion may significantly damage the endometrium and
Given that the rate of cesarean delivery is 20% in the
uterine cavity so as to increase the risk for placenta
general population and if one is interested in reducing
previa. Unfortunately, we were unable to evaluate the
this rate by 50% (i.e., from 20% to 10%), the population-
association between curettage and subsequent develop-
ment of placenta previa because of insufficient informa- attributable risk5I for prior cesarean delivery on the
tion from published studies. subsequent risk of placenta previa is 14%. This implies
The strength of this metaanalysis is the sheer size of that by reducing the primary and repeat cesarean deliv-
the study. Although some of the associations observed in ery rates by half the risk for placenta previa could be
this study violated the homogeneity assumptions because reduced by 14%. Similarly, assuming that the rates of
of pooling of several studies, interstudy heterogeneity spontaneous and induced abortions are 90% and 5%,
was adequately addressed through the fit of models on respectively, and if these rates are reduced by 50%, then
the basis of random-effects analysis. Nonetheless, a few 6.5% and 1.5% of placenta previa cases could potentially
methodologic limitations in this study must also be be averted. This suggests that a reduction in uterine
noted. First, the risk of placenta previa increases both instrumentation rates for the management of both spon-
because of aging effects of the uterus and repeated taneous and induced abortions could further reduce the
pregnancies, s9 Hence an association between prior uter- risk of placenta previa. The public health ramifications of
ine scarring from cesarean delivery or abortion and these findings needs careful assessment in future pro-
placenta previa may have been confounded, in part, spective studies.
because of repeated pregnancies. Unfortunately, insuffi- In conclusion, out (meta)analysis clearly demonstrates
cient data from published studies precluded us from an elevated risk for placenta previa among women with
adequately controlling for parity effect. Another poten- prior cesarean delivery or abortions. Moreover, this risk
tial limitation of the metaanalysis is publication bias. We increases dramatically with increasing number of prior
may have missed identifying published studies during cesarean deliveries. Pregnant women with a history of
our literature search. In spite of our best efforts to cesarean delivery or abortion must be regarded as being at
identify studies, we may have missed some that may have increased risk for the subsequent development of placenta
reported data on placenta previa and prior cesarean previa. This study provides yet another reason for reducing
Volume 177, Number 5 Ananth, Smulian, and Vintzileos 1077
Am J Obstet Oynecol

the primary cesarean delivery rate and for advocating 23. Chervenak F, Lee Y, Handier MA, Monoson RF, gerkowitz
vaginal birth for w o m e n with prior cesarean delivery. RL. Role of attempted vaginal delivery in the management
of placenta previa. Obstet Gynecol 1984;64:798-801.
24. MacGillivray I, Davey D, Isaac S. Placenta previa and sex
REFERENCES
ratio at birth. BMJ 1986;292:371-2.
25. Iyasu S, Saftlas AK, Rowley DL, Koonin LM, Lawson HW,
1. Cnnningham GF, MacDonald PC, Gant NF, Leveno KJ, Atrash HK. The epidemiology of placenta previa in the
Gilstrap LC, editors. Williams' obstetrics. 19th ed. Norwalk United States, 1979 through 1987. Am J Obstet Gynecol
(CT): Appleton & Lange; 1993. p. 819-51. 1993;168:142-9.
2. Niswander KR, Friedman EA, Hoover DB, Pietrowski H, 26. Cotton DB, ReidJA, Paul RH, Culligan EJ. The conservative
Westphal M. Fetal morbidity following potentially anoxi- aggressive management of placenta previa. Am J Obstet
genic obstetric conditions. II. Placenta previa. Am J Obstet Gynecol 1980;137:687-95.
Gynecol 1966;95:846-52. 27. Brenner WE, Edelman DA, Hendricks CH. Characteristics
3. Naeye RL. Effects of maternal smoking on the fetus and of patients with placenta previa and results of "expectant
placenta. BrJ Obstet Gynaecol 1978;85:732-7. management." Am J Obstet Gynecol 1978;132:180-91.
4. Naeye RL. Abruptio placentae and placenta previa: fre- 28. BarrettJM, Boehm FH, Killam AP. Induced abortion: a risk
quency, perinatal mortality, and cigarette smoking. Obstet factor for placenta previa. Am J Obstet Gynecol t981;141:
Gynecol 1980;55:701-4. 769-72.
5. Naeye ILL. Placenta previa: predisposing factors and effects 29. Neri A, Gorodeski I, Bahar:t C, Ovadia Y. Impact of placenta
on the fetus and surviving infants. Obstet Gynecol 1978;52: previa on intrauterine fetal growth. IsrJ Med Sci 1980;16:
521-5. 429-32.
6. Singh DM, Rodriguez C, Gupta AM. Placenta previa and 30. Crenshaw C Jr, Jones DED, Parker RT, Placenta previa: a
previous cesarean section. Acta Obstet Gynecol Scand 1981; survey of twenty years experience with improved perinatal
60:367-8. survival by expectant therapy and cesarean delivery. Obstet
7. Clark SL, Koonings PP, Phelan JP. Placenta previa/accreta Gynecol Surv 173;28:416-7.
and prior caesarean section. Obstet Gynecol 1985;66:89-92. 31, Wexler P, Gottesfeld KR. Early diagnosis of placental previa.
8. Rose GL, Chapman MG. Aetiological factors in placenta Obstet Gynecol 1979;54:231-4.
previa--a case-controlled study. BrJ Obstet Gynaecol 1986; 32. Parsons MT, Wineger A, Siefert L, Spellacy WN. Preg-
93:586-8. nancy outcomes in short 'women. J Reprod Med 1989;34:
9. Nielsen TF, Hagberg H, Ljungbald U. Placenta previa and 357-61.
antepartum hemorrhage after previous caesarean section. 33. Jakobovits M, Zubek L. Sex ratio and placenta previa. Acta
Gynecol Obstet Invest 1989;27:88-90. Obstet Gynecol Scand 1989;68:503-5.
10. Williams MA, Mittendorf R, Leiberman E, Monson RR,
34. Khashoggi T. Maternal and neonatal outcome in major
Schoenbaum SC, Genest DR. Cigarette smoking during
placenta previa. Ann Saudi Med 1995;15:313-6.
pregnancy in relation to placenta previa. Am J Obstet
35. Wolf EJ, Mallozzi A, Rodis JF, Egan JFX, Vintzileos AM,
Gynecol 1991;165:28-32.
Campbell WA. Placenta previa is not an independent risk
11. Zhang J, Savitz DA. Maternal age and placenta previa: a
factor for small for gestational age infant. Obstet GDaecol
population-based case-control study. Am J Obstet Gynecol
1991;77:707-9.
1993;166:641-5.
12. Parazzini F, Dindelli M, Luchini L, La Rosa M, Potenza MT, 36. Leiberman JR, Frasier D, Kasis A, Mazor M. Reduced
Frigerio L, et al. Risk factors for placenta previa. Placenta frequency of hypertensive disorders in placenta previa.
1994;15:321-6. Obstet Gynecol 1991;77:83-6.
13. Maksheed M, EI-Tomi N, Moussa M. A retrospective analysis 37. Ananth CV, Wilcox AJ, Savitz DA, Bowes WA Jr, Luther
of pathological placental implantation--site and penetra- ER. Influence of maternal age and parity on the risk of
tion. IntJ Obstet Gynecol 1994;47:127-34. uteroplacental bleeding disorders. Obstet Gynecol 1996;
14. Chattopadhyay SK, Kharif H, Sherbeeni MM. Placenta pre- 88:511-6.
via and accreta after previous cesarean section. EurJ Obstet 38. Nicolaides KH, Faratian B, Symonds EM. Effective low
Gynecol Reprod Biol 1993;52:151-6. implantation of the placenta on maternal blood pressure
15. Handler AS, Mason ED, Roseberg DL, Davis FG. The and placental function. Br J Obstet Gynaecol 1982;89:806-
relationship between exposure during pregnancy to ciga- 10.
rette smoking and cocaine use and placenta previa. Am J 39. Meyer MB, Jonas BS, Tonascia JA. Perinatal events associ-
Obstet Gynecol 1994;170i884-9. ated with maternal smoking during pregnancy. AmJ Epide-
16. Thomas AG, Alvari M, Friedman S Jr, Brodman ML, KimJ, miol 1976;103:464-76.
Lockwood C. The effect of placenta previa on blood loss in 40. Bianco A, Stone J, Lynch L, Lapinski R, Berkowitz G,
second trimester pregnancy termination. Obstet Gynecol Berkowitz RL. Pregnancy outcome at 40 years. Obstet
1994;84:58-60. Gynecol 1996;87:917-22.
17. Hershkowitz R, Fraser D, Mazor M, Leiberman JR. One or 41. Gorodeski IG, Bahari CM, Schachter A, Neri A. Recurrent
multiple previous cesarean sections are associated with placenta previa. EurJ Obstet Gynecol Reprod Biol 1981;12:
similar increased frequency of placenta previa. EurJ Obstet 7-11.
Gynecol Reprod Biol 1995;62:185-8. 42. Taylor VW, I~amer MD, Vaughan TL, Peacock S. Placenta
18. Chelmow D, Andrew E, Baker ER. Maternal cigarette smok- previa in prior cesarean delivery: how strong is the associa-
ing and placenta previa. Obstet Gynecol 1996;87:703-6. tion? Obstet Gynecol 1994;84:55-7.
19. McShane PM, Hey1 PS, Epstein MP. Maternal and perinatal 43. Ballas S, Gitstein S, Jaffa AJ, Peyser MR. Midtrimester
morbidity resulting from placenta previa. Obstet Gynecol placenta previa: normal or pathologic finding. Obstet Gy-
1985;65:176-82. necol 1979;54:12-4.
20. Silver R, Depp R, Sabbagha RE, Dooley SL, Socol ML, 44. D'Angelo LJ, Erwin LF. Conservative management of pla-
Tamura RK. Placenta previa: aggressive expectant manage- centa previa: a cost-benefit analysis. Am J Obstet Gynecol
ment. Am J Obstet Gynecol 1984;150:15-22. 1984;149:320-6.
21. Lilja GMC. Placenta previa, maternal smoking and recur- 45. Morgan J. Placenta previa: report on a series of 538 cases
rence risk. Acta Obstet Gynecol Scand 1995;74:341-5. (1938-1962). j Obstet Gynaecol Br Commonw 1965;72:700-5.
22. Hemminki E, Glebatis DM, Therriault GD, Janerich DT. 46. Walter SD, Cook R. A comparison of several point estima-
Incidence of placenta previa and abruptio placentae in New tors of the odds ratio in a single 2X2 contingency table.
York State. N Y State J Med 1987;87:594-8. Biometrics 1991;47:795-811.
1078 Ananth, Smulian, and Vintzileos November1997
Am J Obstet Oynecol

47. FleissJL. The statistical basis ofmeta-analysis. Star Meth Med clinical history of asymptomatic midtrimester placenta pre-
Res 1993;2:121-45. xda. A m J Obstet Gynecol 1984;148:743-8.
48. DerSimonian R, Laird N. Meta-analysis in clinical trials. 53. Kramer MD, Taylor V, Hickok DE, DalingJR, Vaughan TL,
Control Clin Trials 1986;7:177-88. Hollenbach KA. Maternal smoking and placenta previa.
49. Hastie TJ, Tribshirani R. Generalized additive models. New Epidemiology 1991;2:221-3.
York: Chapman and Hall; 1995. 54. Grimes DA, Techman T. Legal abortion and placenta
50. Greenland S, Longnecker MP. Methods for trend estima- previa. Am J Obstet Gynecol 1984;149:501-4.
tion from summarized dose-response data, with applications 55. Varma TR. The implication of low implantation of the
to meta-analysis. A m J Epidemiol 1993;135:1301-9. placenta detected by ultrasonography in early pregnancy.
51. Walter SD. Estimation and interpretation of attributable risk Acta Obstet Gynecol Scand 1981;60:265-8.
on health sciences. Biometrics 1978;32:829-49. 56. Bender S. Placenta previa and prmdous lower segment
52. Newton ER, Barss V, Cetrulo CL. The epidemiology and cesarean section. Surg Gynecol Obstet 1954;98:625-7.