Research www.AJOG .

org

IMAGING
Accuracy of ultrasound for the prediction of placenta accreta
Zachary S. Bowman, MD, PhD; Alexandra G. Eller, MD; Anne M. Kennedy, MB BCh, BAO; Douglas S. Richards, MD;
Thomas C. Winter III, MD, MA; Paula J. Woodward, MD; Robert M. Silver, MD

OBJECTIVE: Ultrasound has been reported to be greater than 90%
sensitive for the diagnosis of accreta. Prior studies may be subject to
bias because of single expert observers, suspicion for accreta, and
knowledge of risk factors. We aimed to assess the accuracy of ul-
trasound for the prediction of accreta.
STUDY DESIGN: Patients with accreta at a single academic center were
matched to patients with placenta previa, but no accreta, by year of
delivery. Ultrasound studies with views of the placenta were collected,
deidentified, blinded to clinical history, and placed in random sequence.
Six investigators prospectively interpreted each study for the presence
of accreta and findings reported to be associated with its diagnosis.
Sensitivity, specificity, positive predictive, negative predictive value, and
accuracy were calculated. Characteristics of accurate findings were
compared using univariate and multivariate analyses.
RESULTS: Six investigators examined 229 ultrasound studies from
55 patients with accreta and 56 controls for 1374 independent
observations. 1205/1374 (87.7% overall, 90% controls, 84.9% cases)
studies were given a diagnosis. There were 371 (27.0%) true posi-
tives; 81 (5.9%) false positives; 533 (38.8%) true negatives, 220
(16.0%) false negatives, and 169 (12.3%) with uncertain diagnosis.
Sensitivity, specificity, positive predictive value, negative predictive
value, and accuracy were 53.5%, 88.0%, 82.1%, 64.8%, and 64.8%,
respectively. In multivariate analysis, true positives were more likely to
have placental lacunae (odds ratio [OR], 1.5; 95% confidence interval
[CI], 1.4e1.6), loss of retroplacental clear space (OR, 2.4; 95% CI,
1.1e4.9), or abnormalities on color Doppler (OR, 2.1; 95% CI,
1.8e2.4).
CONCLUSION: Ultrasound for the prediction of placenta accreta may
not be as sensitive as previously described.
Key words: placenta accreta, prenatal diagnosis, sensitivity and
specificity, ultrasound

Cite this article as: Bowman ZS, Eller AG, Kennedy AM, et al. Accuracy of ultrasound for the prediction of placenta accreta. Am J Obstet Gyneol 2014;211:177.e1-7.

P lacenta accreta is dened as an

abnormal adherence of placental
villi to underlying myometrium with an infection, venous thromboembolism or
incidence of cesarean delivery,1,4,8,9 pa-
tients and providers are more frequently
confronted with difcult decisions such
absence of decidua basalis. Failure to prolonged hospitalization.4-6 Accurate as whether to plan for a scheduled hys-
anticipate placenta accreta and prepare antenatal diagnosis of placenta accreta terectomy or to transfer care to a tertiary
for its appropriate management can can allow arrangements to be made for a care center. With its implications for
lead to emergency hysterectomy with planned delivery at a tertiary care center surgical morbidity and future fertility,
profuse, life-threatening hemorrhage,1,2 utilizing a multidisciplinary approach, this is not a decision taken lightly.
disseminated coagulopathy, renal fail- which has been shown to signicantly Placenta previa and history of prior
ure, acute respiratory distress, or even reduce maternal morbidity.5-7 cesarean delivery remain the most im-
death.3 Additional surgical complica- As the incidence of placenta accreta portant predictors of placenta accreta.4
tions include cystotomy, ureteral injury, increases concurrently with an increased In addition to clinical risk factors, ul-
trasound is often used antenatally as
an adjunct to clinical history to modify
From the Departments of Obstetrics and Gynecology (Drs Bowman, Eller, Richards, and Silver) and
risk estimation for placenta accreta. The
Radiology (Drs Kennedy, Winter, and Woodward), University of Utah Health Sciences Center, and the
Department of Obstetrics and Gynecology, Intermountain Healthcare (Drs Bowman, Eller, Richards, accuracy of ultrasound for the prediction
and Silver), Salt Lake City, UT. of placenta accreta is generally reported
Received Dec. 17, 2013; revised Jan. 30, 2014; accepted March 10, 2014. to be good with sensitivities ranging
The authors report no conict of interest. from 77e97%.10-15 However, prior
University of Utah Center for Clinical and Translational Sciences grant support (CTSA
studies on the accuracy of ultrasound
5UL1RR025764-02) enabled the use of Research Electronic Data Capture (REDCap) for this project. for the prediction of accreta may be
Presented, in part, in poster format at the 34th annual meeting of the Society for Maternal-Fetal subject to bias because of single expert
Medicine, New Orleans, LA, Feb. 3-8, 2014. observers, suspicion for accreta, and
Reprints not available from the authors. knowledge of risk factors. Our objective
0002-9378/free
2014 Mosby, Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ajog.2014.03.029 was to assess the accuracy of ultrasound
for the prediction of placenta accreta
See related editorial, page 87 using multiple observers blinded to
clinical status.

AUGUST 2014 American Journal of Obstetrics & Gynecology 177.e1

Research Imaging www.AJOG.org

Six investigators consisting of 3 expe-

TABLE 1 rienced obstetric radiologists (A.M.K.,
Patient characteristics T.C.W., and P.J.W.) and 3 maternal-fetal
Characteristic Accreta (Cases) Controls P value medicine physicians (A.J.E., D.S.R., and
Maternal age, 32.4  5.2 31.3  6.9 .36a R.M.S.) prospectively reviewed and in-
y (mean  SD) terpreted each ultrasound study. All
Body mass index, kg/m2 29.0  9.4 31.9  6.9 .15a 3 radiologists are fellowship trained in
(mean  SD) abdominal imaging and have more than
Gravidity (median, range) 5 (2e18) 4 (1e14) .0047b 10 years experience in obstetric ultra-
sound, including evaluation of accreta.
Parity (median, range) 3 (1e7) 2 (0e10) .0090b Similarly, all 3 maternal-fetal medicine
Cesarean delivery (n, %) < .001c specialists are fellowship trained and
0 1 (1.8) 16 (28.6) have a minimum of 8 years of experience
in diagnosing placenta accreta (>20
1 17 (30.9) 25 (44.6)
years for 2 of the 3 physicians). In-
2 16 (29.1) 9 (16.1) vestigators were asked to score each
3 21 (38.2) 6 (10.7) imaging study for the presence of accreta
Delivery gestational age, 33.8  3.0 34.9  4.0 .12a (yes, no, or unable to determine)
wk (mean  SD) and indicate the presence or absence
Gestational age at 29.2 (25.1e32.4) 29.3 (23.9e33.6) .11b of specic ndings that have been re-
study, wk (median, ported to be associated with its diag-
interquartile range) nosis. These ndings included the
a
Students t test; b Wilcoxon-Mann-Whitney test; c Fisher exact test. following: number of lacunae,17-20 loss
Bowman. Accuracy of ultrasound for the prediction of placenta accreta. Am J Obstet Gynecol 2014. of retroplacental clear space,13,17,18,20,21
loss of visualization of the myome-
trium,20 and bladder wall irregularity.20
M ATERIALS AND M ETHODS of abnormal adherence of the placenta or
If color Doppler was used, investigators
Patients who delivered at the University evidence of gross placental invasion at the
further identied the presence or
of Utah between 2000 and 2012 with time of surgery. Patients were included
absence of the following: subplacental
documentation of a clinical or histopath- if they had ultrasound images of the
vascularity,10,11,22 vessels bridging from
ologic diagnosis of placenta accreta5,7,16 placenta available at a gestational age of
the placenta to the uterine margin,10,13,20
were identied and matched to patients greater than or equal to 16 weeks. For
gaps in myometrial blood ow,10,20 ves-
with placenta previa but no accreta by each patient, every image of the placenta
sels crossing interface disruption sites,13
year of delivery. Histopathalogic diagno- was collected, de-identied, and blinded
or turbulent lacunae.10,23 If a diagnosis
sis of accreta was conrmed by docu- to clinical history. Images from each study
was made, investigator condence for
mentation of placental invasion into the were then placed in random sequence
each imaging study was scored on a scale
myometrium, and clinical diagnosis of using the Microsoft Excel random num-
of 0 (none) to 10 (certain), and image
accreta was conrmed by documentation ber generator.
quality was scored on a scale from 1 (very
poor) to 10 (best).
Study data were collected and
TABLE 2 managed using REDCap electronic data
Diagnostic performance characteristics of ultrasound for the diagnosis capture tools hosted at the University of
of placenta accreta Utah.24 REDCap (Research Electronic
Excluding Missing/uncertain Data Capture) is a secure, web-based
missing/uncertain diagnoses assigned application designed to support data
Characteristic diagnoses as no accreta capture for research studies, providing:
Sensitivity 62.8 (58.7e66.7) 53.3 (49.5e57.1) (1) an intuitive interface for validated
Specificity 86.8 (83.9e89.4) 88.1 (85.4e90.4) data entry; (2) audit trails for tracking
data manipulation and export pro-
PPV 82.1 (78.7e85.0) 82.1 (78.8e85.4)
cedures; (3) automated export proce-
NPV 70.8 (68.5e73.0) 64.8 (62.9e66.7) dures for seamless data downloads to
Accuracy 75.0 (72.5e77.4) 65.8 (63.2e68.3) common statistical packages; and (4)
Data are % (95% confidence interval). procedures for importing data from ex-
NPV, negative predictive value; PPV, positive predictive value. ternal sources.
Bowman. Accuracy of ultrasound for the prediction of placenta accreta. Am J Obstet Gynecol 2014. Continuous variables were analyzed
with the Student t test. Categorical

177.e2 American Journal of Obstetrics & Gynecology AUGUST 2014

www.AJOG.org Imaging Research
variables were analyzed by the
Wilcoxon-Mann-Whitney, c2 or Fisher FIGURE 1
exact test, where appropriate. Sensitivity, Receiver operating characteristic curve for ultrasound to accurately
specicity, positive predictive value, diagnose placenta accreta
negative predictive value, and accuracy
were calculated according to standard
denitions. For multivariate analyses, all
variables of interest were included in a
logistic regression model. Covariates
were then removed in a stepwise fashion
until all covariates in the nal model
for a particular outcome had a P value
of < .2. Receiver operating characteristic
curves were generated and the area un-
der the curve computed to determine the
discriminative ability of ultrasound to
correctly identify placenta accreta.25 A
P value of < .05 was considered statisti-
cally signicant. All statistical analyses
were performed using Stata 12.1 (Stata-
Corp, College Station, TX). The institu-
tional review board of the University of
Utah approved this study.

R ESULTS
After exclusions, we identied 55 women
with placenta accreta with available im-
aging studies at the University of Utah
between 2000 and 2012. Fifty-six women
with placenta previa but no accreta
and appropriate imaging studies and
matched to cases by year of delivery were
chosen as controls. Clinical data are
summarized in Table 1. Compared with
patients with placenta previa only (no
accreta), those with placenta accreta had
signicantly higher parity, more prior
cesarean deliveries and an earlier ges-
tational age at delivery. Maternal age and
body mass index at the time of delivery
were similar.
The 55 women with placenta accreta
had a total of 116 ultrasound studies
and the 56 women with placenta
previa but no accreta had 113 studies.
Thus, a total of 229 ultrasound studies
were available for review. All ultra-
sound images were collected, deiden-
tied, and placed in a random order.
Each of the 6 investigators reviewed all
studies for a total of 1374 independent
observations.
A specic diagnosis regarding the
presence or absence of placenta accreta
was reported for 1205/1374 (87.7%)
studies, including 614/678 (90.6%) of
Areas under the curve are presented corresponding to inclusion or exclusion of uncertain or missing
diagnoses.
Bowman. Accuracy of ultrasound for the prediction of placenta accreta. Am J Obstet Gynecol 2014.
controls and 591/696 (84.9%) of cases
(P .001). Of studies receiving a diag-
nosis, diagnostic performance charac-
teristics were as follows: 371 true
positives (30.8%), 81 false positives
(6.7%), 533 true negatives (44.2%), and
220 false negatives (18.3%). 165/1374
studies (12.0%) were designated unable
to determine and 4/165 studies (2.4%)
were not given a diagnosis. Results for
sensitivity, specicity, positive predictive
value, negative predictive value and
overall accuracy are shown in Table 2.
Two analyses are presented; one ex-
cluding studies that were interpreted as
uncertain and one that assigned un-
certain diagnoses as no accreta. Re-
ceiver operator characteristic curves,
which accounted for ultrasound as a
binary test (absence or presence of
accreta), are shown in Figure 1. Overall,
the diagnostic performance characteris-
tics were improved when uncertain di-
agnoses were excluded.
Given that individual patients may
have had more than one ultrasound
study performed, we examined accuracy
of diagnoses by patient. Diagnoses for
the 56 women with previa only (ie, con-
trols) were correct signicantly more
often than for the 55 cases (75% vs
60.4%, respectively, P < .0068). Simi-
larly, concordant with higher overall
specicity, incorrect results were lower
for controls compared with cases (13.8%
vs 27.0%, respectively, P .0093). The
ratio of unknown or missing diagnoses
was not signicantly different between
cases and controls (13.3% vs 10.2%,
respectively, P .21).

AUGUST 2014 American Journal of Obstetrics & Gynecology 177.e3

Research Imaging
FIGURE 2
Receiver operating characteristic curve for multivariable regression
model to predict placenta accreta based on specific ultrasound findings
Bowman. Accuracy of ultrasound for the prediction of placenta accreta. Am J Obstet Gynecol 2014.
To determine which ndings on ul-
trasound were associated with the
presence of placenta accreta, an initial
model was created with accreta as the
outcome and included the following
specic ndings as covariates: number
of lacunae, loss of the retroplacental
clear space, loss of visualization of the
myometrium, irregular bladder wall, or
any color Doppler abnormalities. After
stepwise removal of covariates from
the model until remaining covariates
had a P value of < .2, only loss of
visualization of the myometrium was
excluded in the nal model. Placenta
accreta was associated with more
placental lacunae (odds ratio [OR], 1.4;
95% CI, 1.3e1.6), loss of the retro-
placental clear space (OR, 2.2; 95% CI,
1.6e3.0), an irregular bladder wall
(OR, 1.3; 95% CI, 1.0e1.6) and color
Doppler abnormalities (OR, 1.3; 95%
CI, 1.1e1.4). The receiver operating
characteristic curve for this model is
shown in Figure 2.
Next, we examined whether specic
ultrasound ndings and/or patient
177.e4 American Journal of Obstetrics & Gynecology AUGUST 2014
demographic factors were associated
with accurate and inaccurate diagnoses
(ie, true positives, true negatives, false
positives, false negatives, and uncertain
diagnosis). The results of univariate an-
alyses are shown in Table 3. True posi-
tives were signicantly associated with
lower maternal BMI (26.9 vs 29.1,
P .0034), better image quality (6.1 vs
5.7, P .0026), more lacunae (5 vs 1.2,
P < .001), loss of the retroplacental clear
space (91.9% vs 19.0%, P < .001), an
irregular bladder wall (39.4% vs 4.3%,
P < .001) and color Doppler abnor-
malities (97.3% vs 65.7%, P < .001).
False positives were associated with
similar ndings, but color Doppler ab-
normalities were not signicantly dif-
ferent from other diagnoses (P .45).
In contrast to the positive results
(accreta thought to be present), true
negatives had fewer lacunae (0.7 vs 3.2,
P < .001), and fewer instances of loss
of retroplacental clear space (5.3% vs
59.9%, P < .001), loss of visualization
of the myometrium (5.3% vs 59.2%,
P < .001), irregular bladder wall (1.9%
www.AJOG.org
vs 21.3%, P < .001) or abnormal color
Doppler ndings (60.4% vs 83.0%,
P < .001). False negatives were similarly
associated with these ndings.
Studies given an uncertain diagnosis
were associated with relatively poorer
image quality (4.0 vs 6.0, P < .001). Also,
the magnitude of the differences be-
tween cases and controls with regard
to placental lacunae, loss of the retro-
placental clear space, loss of visualization
of the myometrium, an irregular bladder
wall, and color Doppler abnormalities
was less pronounced than for studies
given a diagnosis.
For each diagnosis, an initial multi-
variate model was created that included
specic ultrasound ndings, perceived
image quality, gestational age at which
the study was performed, maternal age at
delivery, and maternal BMI. Covariates
were then removed in a stepwise fashion
until all covariates in the nal models
had a P value of < .2. True positives were
more likely to have placental lacunae
(OR, 1.5; 95% CI, 1.4e1.6), loss of the
retroplacental clear space (OR, 2.4; 95%
CI, 1.1e4.9), or 1 or more abnormalities
on color Doppler (OR, 2.1; 95% CI,
1.8e2.4), earlier study year (OR, 0.93;
95% CI, 0.88e0.99), and absence of an
irregular bladder wall (OR, 0.66; 95%
CI, 0.50e0.89). True negatives were
more likely to have fewer lacunae (OR,
0.64; 95% CI, 0.57e0.73), better image
quality (OR, 1.2; 95% CI, 1.12e1.29),
and more advanced gestational age at the
time of the study (OR, 1.05; 95% CI,
1.02e1.08). As might be expected, true
negatives were less likely to have loss of
the retroplacental clear space (OR, 0.34;
95% CI, 0.15e0.75) or 1 or more color
Doppler abnormalities (OR, 0.50; 95%
CI, 0.39e0.65).
False positives were associated with
loss of the retroplacental clear space
(OR, 2.7; 95% CI, 1.9e3.8) or 1 or more
color Doppler abnormalities (OR, 1.2;
95% CI, 1.1e1.4) and absence of an
irregular bladder wall (OR, 0.47; 95%
CI, 0.30e0.73). False negatives were
associated with an earlier gestational
age (OR, 0.96; 95% CI, 0.94e0.99),
placental lacunae (OR, 1.2; 95% CI,
1.1e1.3), and inversely associated with
loss of visualization of the myometrium
 www.AJOG.org
TABLE 3
Specific findings associated with accurate and inaccurate diagnoses
True positive True negative False positive False negative Diagnosis given
P P P P P
Variable Yes No value Yes No value Yes No value Yes No value Yes No value
Gestational age 28.9  4.9 28.3  5.5 .043 28.7  5.6 28.3  5.2 .28 28.0  5.3 28.5  5.4 .46 27.5  5.2 28.7  5.4 .0034 28.5  5.3 28.3  5.6 .59
at study, wk
Maternal age 33.0  5.0 31.8  6.7 .0015 32.4  5.5 31.6  7.5 .021 31.4  5.8 32.1  6.4 .28 31.3  5.8 32.2  6.4 .042 32.0  6.4 33.1  5.6 .029
at study, y
Maternal BMI 26.9  9.9 29.1  10.3 .0034 28.5  10.1 28.5  10.4 .94 23.6  11.6 28.8  10.1 < .001 31.8  9.5 27.7  10.3 < .001 28.4  10.2 29.4  10.6 .37
at delivery
Number of 5.0  2.7 1.2  2.7 < .001 0.74  1.0 3.2  2.8 < .001 2.9  2.4 2.2  2.6 .014 1.4  1.4 2.4  2.7 < .001 2.3  2.6 1.7  1.9 .009
placental
lacunae
Image qualitya 6.1  1.9 5.7  2.2 .0026 6.2  2.0 5.6  2.1 < .001 5.6  2.1 5.8  2.1 .32 5.7  2.0 5.9  2.1 .30 6.0  2.0 4.0  2.0 < .001
Confidence of 63.2  22.3 69.5  22.4 < .001 76.4  19.5 61.6  22.6 < .001 48.0  25.8 69.0  21.7 < .001 68.9  21.5 67.4  22.8 .39 69.0  22.7 49.0  9.4 < .001
interpretationb
Loss of 341 (91.9) 191 (19.0) < .001 28 (5.3) 504 (59.9) < .001 70 (86.4) 462 (35.7) < .001 25 (11.4) 507 (43.9) < .001 464 (38.5) 68 (40.2) < .001
retroplacental
clear space
AUGUST 2014 American Journal of Obstetrics & Gynecology

Loss of 338 (91.1) 188 (18.7) < .001 28 (5.3) 498 (59.2) < .001 67 (82.7) 459 (35.5) < .001 25 (11.4) 501 (43.4) < .001 458 (38.5) 68 (40.2) < .001
visualization
of the
myometrium
Irregular 146 (39.4) 43 (4.3) < .001 10 (1.9) 179 (21.3) < .001 18 (22.2) 171 (13.2) .014 12 (5.45) 177 (15.3) .001 186 (15.4) 3 (1.8) < .001
bladder wall
Any abnormal 361 (97.3) 659 (65.7) < .001 322 (60.4) 698 (83.0) < .001 66 (81.5) 954 (73.8) .45 149 (67.7) 871 (75.5) .022 898 (74.5) 122 (72.2) < .001
Doppler finding
Data are mean  SD or n (%); P value determined by Student t test or c2 test where appropriate.
a
Image quality was scored on a scale from 1 to 10 with 10 representing the best image quality; b Investigator confidence was scored on a scale from 0 to 100% with 100% representing the highest confidence.

Imaging
Bowman. Accuracy of ultrasound for the prediction of placenta accreta. Am J Obstet Gynecol 2014.

Research
177.e5
Research Imaging
(OR, 0.30; 95% CI, 0.19e0.46) and
1 or more color Doppler abnormalities
(OR, 0.54; 95% CI, 0.43e0.68). Image
studies for which a diagnosis was unable
to be determined were inversely associ-
ated with image quality (OR, 0.68; 95%
CI, 0.62e0.76) and 1 or more color
Doppler abnormalities (OR, 0.62; 95%
CI, 0.49e0.79), whereas loss of visuali-
zation of the myometrium (OR, 3.7;
95% CI, 2.7e5.1) and an irregular
bladder wall (OR, 1.7; 95% CI, 1.3e2.1)
showed positive associations.
C OMMENT
Placenta previa and history of prior ce-
sarean delivery are known risk factors for
placenta accreta. The risk for placenta
accreta with a previa and 1, 2, 3, or 4
prior cesarean deliveries is 11, 40, 61, and
67 percent, respectively.4 Thus, with a
previa and 2 prior cesarean deliveries,
the pretest probability for placenta
accreta is approximately 40%. Ultra-
sound is the mainstay for prenatal as-
sessment of accreta, and the ability of
ultrasound to accurately predict the
presence of placenta accreta is reported
to be excellent. One recent systematic
review and metaanalysis examined the
diagnostic value of ultrasound and noted
a pooled sensitivity of 83% and a robust
area under the summary receiver oper-
ating characteristic curve of 0.9485.26
However, the 13 studies included were
limited by a low number of cases per
study, a priori knowledge of risk factors,
and single expert (or unspecied) ob-
servers. Our results show that when im-
ages are reviewed by a diverse group
of providers, blinded to any clinical his-
tory, ultrasound for the prediction of
placenta accreta may not be as sensitive or
accurate as previously described.
Chalubinski et al,27 Esakoff et al,15
Warshak et al,11 and Chou et al10 re-
ported the largest studies (80 patients)
of ultrasound for the prediction of
accreta with 232, 108, 453, and 80 pa-
tients included, respectively. The total
number of patients with conrmed
accreta in these studies (ie, cases) was 35,
19, 46, and 16 with sensitivities of 91,
89, 77, and 82 percent, respectively. The
corresponding accuracies (number of
true positives and true negatives divided
177.e6 American Journal of Obstetrics & Gynecology AUGUST 2014
by total number of patients in the study)
were 95, 91, 95, and 94 percent. For each
study, patient history and risk for
accreta were known. When only studies
that were given a diagnosis were con-
sidered in our study, the sensitivity of
62.8% was lower than previously re-
ported. If images with an uncertain
diagnosis are treated as a negative result,
our sensitivity of 53.5% and accuracy
of 65.8% are even worse with over 1/3
of cases misidentied.
Chalubinski et al,27 Esakoff et al,15
Warshak et al,11 and Chou et al10 re-
ported specicities of 96, 91, 96, and 97
percent respectively. Our specicity was
similar regardless of how uncertain di-
agnoses were treated, 86.8% if excluded
and 88.1% if included. Although this
suggests that a negative ultrasound is
able to accurately identify those patients
without accreta, the lower negative pre-
dictive value (64.8%) suggests that an
important number of those with a neg-
ative ultrasound will be misdiagnosed
(ie, have an accreta).
When considering a diagnostic test,
ultimately one needs to decide whether
the results of that test will change
the management of a given patient.
Accordingly, one should consider likeli-
hood ratios, the percentage of affected
people with a positive test result divided
by the percentage of unaffected in-
dividuals with a negative test result,
along with pre- and posttest probabili-
ties. 28 Using the pooled sensitivity and
specicity from the previously men-
tioned metaanalysis (83% and 95%,
respectively), the likelihood ratio posi-
tive is calculated to be 16.6 with ratios
greater than 10 generally thought to
yield a large increase in posttest proba-
bility. Thus, in a patient with a previa
and 1 or 2 prior cesarean deliveries,
the pretest probabilities would be 11
and 40%, respectively, and the posttest
probabilities would be approximately
65 and 90% for a positive ultrasound
result. Conversely, the likelihood ratio
negative would be 0.18 with posttest
probabilities of approximately 2 and
10% respectively. Even with a relatively
good test, we cannot completely elimi-
nate the possibility of an accreta. In the
case of a previa and 2 prior cesarean
www.AJOG.org
deliveries, the posttest probability is
reduced from 40% to 10%, but for
some providers this might not be a suf-
cient reduction in risk to avoid a
planned cesarean hysterectomy. Con-
sider then that these likelihood ratios
and diagnostic performance character-
istics have already taken into account
pre-test probabilities and may be overly
optimistic. If the blinded diagnostic
performance characteristics are consid-
ered (sensitivity of 53.5% and specicity
of 88%), the likelihood ratio positive is
only 4.5 (associated with only a moder-
ate increase in the post-test probability)
and the likelihood ratio negative only
0.53. Thus for a patient with a previa and
2 prior cesarean deliveries, the posttest
probability of a positive and negative
ultrasound result would be approxi-
mately 65% and 20%, respectively. Re-
gardless of the ultrasound result, these
posttest probabilities are sufciently
imprecise so as to be useful as an
adjunctive test but not as a denitive
gold standard.
Previous studies have shown that the
number of placental lacunae,17-20 loss of
retroplacental clear space,13,17,18,20,21
loss of visualization of the myome-
trium,20 and bladder wall irregularity,
and color Doppler abnormal-
ities12,13,15,21,23,24 are associated with
placenta accreta. We found that each
incremental increase in placental
lacunae on a scale from 0 (none) to 10
(many) was associated with a 1.4-fold
increased risk for accreta as well as loss
of the retroplacental clear space, an
irregular bladder wall, and abnormalities
using color Doppler, consistent with the
observations of other studies. Further-
more, when the receiver operator char-
acteristic curve was computed from this
logistic regression model, the area under
the curve was 0.82, suggesting that
these factors may improve the diagnostic
accuracy of ultrasound (Figure 2).
Although the presence of these factors
were associated with positive ultrasound
result and the absence was associated
with negative results, they do not appear
to be useful for distinguishing accurate
from inaccurate results (eg, true posi-
tives from false positives or true nega-
tives from false negatives).
www.AJOG.org
Our study has several strengths. We
were able to use a relatively large num-
ber of patients with accreta (55) that
represents the largest number in the
literature for which ultrasound was
evaluated. Each patient had the potential
for multiple studies (116 total) that
were then evaluated independently by
6 investigators for 696 unique observa-
tions. Second, the images were evaluated
by 3 radiologists and 3 maternal-fetal
medicine physicians, which allowed for
a diversity of evaluation that has not
previously been reported and may be
more applicable to general practice.
Perhaps most importantly, the inves-
tigators evaluating ultrasound images
were blinded to all clinical history
and knowledge of risk factors. In this
manner, the actual accuracy of ultra-
sound could be evaluated absent pretest
bias based on history of prior cesarean
delivery. Finally, all images were evalu-
ated prospectively allowing for collect-
ion of the data points of interest.
We acknowledge some weaknesses to
our study as well. In practice, practi-
tioners are aware of patient history and
use this information to shape a diagnosis
by performing the scan in real time and
understanding certain practical mea-
sures, such as how much transducer
pressure was used or sonographer
experience. By blinding investigators,
the environment is clearly an articial
one and in that respect may not be
directly relevant to actual practice. In
addition, although the images were
prospectively evaluated by the inves-
tigators, the images themselves were
collected retrospectively, which pre-
cluded the ability to use cine clips or
real-time evaluation.
In conclusion, the published accu-
racy of ultrasound for the diagnosis
of placenta accreta is likely biased
by knowledge of patient risk factors,
which can signicantly increase pretest
probability. When evaluated by a
diverse group of providers who blinded
from patient history, the diagnostic
performance characteristics of ultra-
sound may not be as high as previously
reported. -
ACKNOWLEDGMENTS
University of Utah Center for Clinical and
Translational Sciences grant support (CTSA
5UL1RR025764-02) enabled the use of
REDCap (Research Electronic Data Capture) for
this project.
REFERENCES
1. Miller DA, Chollet JA, Goodwin TM. Clinical
risk factors for placenta previa-placenta accreta.
Am J Obstet Gynecol 1997;177:210-4.
2. Belfort MA. Publications Committee SfM-FM.
Placenta accreta. Am J Obstet Gynecol
2010;203:430-9.
3. Bauer ST, Bonanno C. Abnormal placenta-
tion. Semin Perinatol 2009;33:88-96.
4. Silver RM, Landon MB, Rouse DJ, et al.
Maternal morbidity associated with multiple
repeat cesarean deliveries. Obstet Gynecol
2006;107:1226-32.
5. Eller AG, Porter TF, Soisson P, Silver RM.
Optimal management strategies for placenta
accreta. BJOG 2009;116:648-54.
6. Warshak CR, Ramos GA, Eskander R, et al.
Effect of predelivery diagnosis in 99 consecutive
cases of placenta accreta. Obstet Gynecol
2010;115:65-9.
7. Eller AG, Bennett MA, Sharshiner M, et al.
Maternal morbidity in cases of placenta accreta
managed by a multidisciplinary care team
compared with standard obstetric care. Obstet
Gynecol 2011;117(Pt 1):331-7.
8. Clark SL, Koonings PP, Phelan JP. Placenta
previa/accreta and prior cesarean section.
Obstet Gynecol 1985;66:89-92.
9. MacDorman M, Declercq E, Menacker F.
Recent trends and patterns in cesarean and
vaginal birth after cesarean (VBAC) deliveries in
the United States. Clin Perinatol 2011;38:179-92.
10. Chou MM, Ho ES, Lee YH. Prenatal diag-
nosis of placenta previa accreta by trans-
abdominal color Doppler ultrasound. Ultrasound
Obstet Gynecol 2000;15:28-35.
11. Warshak CR, Eskander R, Hull AD, et al.
Accuracy of ultrasonography and magnetic
resonance imaging in the diagnosis of pla-
centa accreta. Obstet Gynecol 2006;108(Pt 1):
573-81.
12. Dwyer BK, Belogolovkin V, Tran L, et al.
Prenatal diagnosis of placenta accreta: so-
nography or magnetic resonance imaging?
J Ultrasound Med 2008;27:1275-81.
13. Wong HS, Cheung YK, Zuccollo J, Tait J,
Pringle KC. Evaluation of sonographic
AUGUST 2014 American Journal of Obstetrics & Gynecology
Imaging Research
diagnostic criteria for placenta accreta. J Clin
Ultrasound 2008;36:551-9.
14. Shih JC, Palacios Jaraquemada JM, Su YN,
et al. Role of three-dimensional power Doppler
in the antenatal diagnosis of placenta accreta:
comparison with gray-scale and color Doppler
techniques. Ultrasound Obstet Gynecol
2009;33:193-203.
15. Esakoff TF, Sparks TN, Kaimal AJ, et al.
Diagnosis and morbidity of placenta accreta.
Ultrasound Obstet Gynecol 2011;37:324-7.
16. Wu S, Kocherginsky M, Hibbard JU.
Abnormal placentation: twenty-year analysis.
Am J Obstet Gynecol 2005;192:1458-61.
17. Comstock CH, Love JJ Jr, Bronsteen RA,
et al. Sonographic detection of placenta accreta
in the second and third trimesters of pregnancy.
Am J Obstet Gynecol 2004;190:1135-40.
18. Finberg HJ, Williams JW. Placenta accreta:
prospective sonographic diagnosis in patients
with placenta previa and prior cesarean section.
J Ultrasound Med 1992;11:333-43.
19. Twickler DM, Lucas MJ, Balis AB, et al.
Color ow mapping for myometrial invasion in
women with a prior cesarean delivery. J Matern
Fetal Med 2000;9:330-5.
20. Isaacs DL, Rouse GA, Lange Md,
Goeser CD, Lyko E. The sonographic appear-
ance of placenta accreta. J Diagn Med Sonog
1998;14:15-20.
21. Hudon L, Belfort MA, Broome DR. Diag-
nosis and management of placenta percreta: a
review. Obstet Gynecol Surv 1998;53:509-17.
22. Comstock CH. Antenatal diagnosis of pla-
centa accreta: a review. Ultrasound Obstet
Gynecol 2005;26:89-96.
23. Moodley J, Ngambu NF, Corr P. Imaging
techniques to identify morbidly adherent
placenta praevia: a prospective study. J Obstet
Gynaecol 2004;24:742-4.
24. Harris PA, Taylor R, Thielke R, Payne J,
Gonzalez N, Conde JG. Research electronic
data capture (REDCap)a metadata-driven
methodology and workow process for
providing translational research informatics
support. J Biomed Inform 2009;42:377-81.
25. Cantor SB, Kattan MW. Determining the area
under the ROC curve for a binary diagnostic test.
Med Decis Making 2000;20:468-70.
26. Meng X, Xie L, Song W. Comparing the
diagnostic value of ultrasound and magnetic
resonance imaging for placenta accreta: a sys-
tematic review and meta-analysis. Ultrasound
Med Biol 2013;39:1958-65.
27. Chalubinski KM, Pils S, Klein K, et al. Pre-
natal sonography can predict degree of
placental invasion. Ultrasound Obstet Gynecol
2013;42:518-24.
28. Grimes DA, Schulz KF. Rening clinical di-
agnosis with likelihood ratios. Lancet 2005;365:
1500-5.
177.e7