Screening Ultrasonography in Pregnancy
Recommendation
Routine third-trimester ultrasound examination of the fetus is not recommended.
There is insufficient evidence to recommend for or against routine ultrasound examination
in the second trimester in low-risk pregnant women (see Clinical Intervention).
Burden of Suffering
Ultrasonography is widely used in pregnancy in the U.S.
According to 1992 U.S. natality data, 58% of
mothers who had live births received ultrasonography
in pregnancy, compared to 48% in 1989.1 The
highest rates occurred in white women and those ages 25_39 years.
For asymptomatic low-risk women, a single scan in the second trimester may be
used to estimate gestational age in women with unreliable dates of last menses
and to detect multiple gestation and fetal malformations. A third-trimester
scan may be used to screen for intrauterine growth retardation (IUGR) and fetal
malpresentation as well as previously undetected
multiple gestations and malformations.2
These conditions may be
associated with increased maternal or perinatal
morbidity and mortality. Inaccurate estimation of gestational age may lead to
repeated testing of fetal well-being and induction of labor in pregnancies
erroneously thought to be postterm.3,4 About 25-45% of
women are unable to provide an accurate menstrual history;3,5,6 the
estimated date of confinement derived from the last menstrual period differs by
more than 2 weeks from the actual date of birth in nearly one quarter of pregnancies.5
Multiple gestation is associated with increased perinatal
mortality, preterm delivery, and other obstetric complications,7 and it
is more likely to result in cesarean delivery (56%
compared to a baseline rate of 23%).8 The
ratio of multiple gestation births to all births (currently 24/1,000 live births) has risen steadily since 1972 and is
the highest reported in the past 50 years.1
Congenital anomalies are the leading cause of death before 1 year
of age in the U.S., with a mortality rate of 1.7/1,000 live births, and are also important contributors to childhood morbidity
and shortened life expectancy.10,11 Fetal
growth retardation has been associated with poor pregnancy outcomes, including
fetal and neonatal death, reduced intelligence, seizures, and cerebral palsy,
although most term growth-retarded infants develop normally.12 Breech
and other malpresentations may be associated with
poor outcome and result in cesarean delivery in 84% of
cases.8 Malpresentation occurs in 38/1,000 live births, with the risk increasing with increasing age of the mother.1
Accuracy of the Screening Test
Real-time ultrasound consists of high-frequency sound waves that allow
two-dimensional imaging of both structural and functional characteristics of
the fetus, as well as the location and morphology of the placenta.2 (This
chapter will not address the topic of umbilical Doppler ultrasound13).
Ultrasound is the recommended test for determination of gestational age in
women with uncertain menstrual dates because measurement of the biparietal diameter, when performed early in the second
trimester, has been shown to be accurate in determining gestational age.5,6 Ninety percent of patients
deliver within 2 weeks of the due date when gestational age is determined by early second-trimester
ultrasound.5
Ultrasound can also detect
multiple gestations, which are missed by clinical examination in nearly one
third of cases.14 One center that provided the only maternity services in its community
reported that 98% of all twins were detected antenatally when
routine ultrasound screening was performed.4,6 The average gestational age at
detection fell from 27 weeks to 20 weeks. Randomized controlled trials of routine ultrasound before 20 weeks
found higher rates of early detection of multiple gestation with screening (83-100%)
compared to unscreened controls (60- 76%).15-19
False-positive ultrasound diagnoses also occur, however, primarily in the first
trimester; over 20% of multiple fetuses identified in the first trimester are either artifacts
or die early in pregnancy.20
Many fetal structural
malformations, including cardiac, gastrointestinal, renal, limb, and neural
tube defects, can also be detected by current ultrasound techniques (for
detailed discussion of ultrasound screening to detect chromosomal abnormalities
and neural tube defects, see Chapters 41 and 42,
respectively). Detection rates depend on the quality of the equipment and the
expertise of the ultrasonographer. In a trial in
low-risk pregnant women, routine serial ultrasonography
at 15-22 and 31-35 weeks of gestation had a sensitivity of 35% for
detecting fetuses with at least one major anomaly before delivery but only 17% for
detection before the typical gestational-age limit for legal abortion (<24 weeks).21 Only 4% of
missed cases occurred in women who did not comply with scheduled screening
ultrasounds. The sensitivity of selective ultrasound, performed only for
obstetric or medical indications, was significantly lower than routine
ultrasound: 11% before delivery and 5% before 24 weeks.
In this study, the sensitivity of routine midtrimester
ultrasound was significantly higher at tertiary compared to other scanning
facilities (35% vs. 13%). False-positive diagnoses were reported for 7 cases,
or 0.9/1,000 pregnant women scanned before 24 weeks, with most reported from
other than tertiary facilities. In another trial, the rates of detection of
major malformations by screening before 20 weeks (confirmed at abortion
or delivery) were 36% and 77% at two hospitals.19 Ten of the thirty cases with
suspected major malformations were judged normal at follow-up ultrasound
examinations at 20- 36 weeks and an 11th was found to have only a minor anomaly at delivery; 2.7/1,000 pregnant women received a false diagnosis of a major fetal malformation.
Large case series evaluating routine ultrasound in low-risk women have reported
sensitivities ranging from 21% to 74% for
detecting major fetal abnormalities prior to 22-24 weeks
among women who were scanned in the second trimester.22-24 False-positive
rates of 0.2-1.0/1,000 women scnned were reported; in one study, 6 of 8
initially false-positive diagnoses were corrected on follow-up evaluation.
Direct comparisons of the trials and series results are hampered by varying
definitions of "fetal malformation."
The ultrasound examination is
the most accurate means of detecting IUGR, although the lack of consensus on
standards for the definition or diagnosis of IUGR12 makes
evaluating screening tests for this condition difficult. Measurements of the
fetal abdomen and head, and indices that compare the relative sizes of these
structures, are accurate in assessing fetal growth.25-32 A
small abdominal circumference, for example, the most commonly affected anatomic
measurement,3 has a sensitivity of 80-96% and a
specificity of 80-90% in detecting growth-retarded fetuses in the third trimester.3,26,33,34 The
product of the crown-rump length and the trunk area has a sensitivity of 94% and a
specificity of 90%.35 Because of the relatively low risk of IUGR in the general population,
however, the likelihood that an abnormal test indicates IUGR is relatively low.
For example, an abnormal abdominal circumference at 34-36 weeks'
gestation indicates IUGR in only 21-50% of
cases.26,33,36 The generalizability
of these studies has also been questioned; many had small samples, used only
expert ultrasonographers, and/or suffered from methodologic limitations.26,37 In
addition, the definitions commonly used in these studies may cause normal but
constitutionally small fetuses to be labeled as IUGR.
Effectiveness of Early
Detection
For routine ultrasonographic screening to be proven
beneficial, evidence is needed that interventions in
response to examination results lead to improved clinical outcome. Twelve
randomized controlled trials have examined the effectiveness of routine
ultrasound screening in improving maternal or neonatal outcomes. Four of these
evaluated a single ultrasound before 20 weeks,15-17,19 three
trials assessed serial ultrasound at 18-20 weeks
and 31-35 weeks,18,21,38-40 three trials evaluated one or two ultrasounds between 32 and 37 weeks
when all subjects received one ultrasound before 24 weeks,35,41,42 and
two tested multiple scans (plus Doppler flow studies in one trial) every 3-4 weeks
beginning at 24-28 weeks, with all subjects receiving a single midtrimester
scan.43,44 In a 13th trial, all subjects received three ultrasounds, but the results of
placental grading at 34-36 weeks were reported only for the experimental group.45 In
addition, four meta-analyses have been published, none of which included the
U.S. RADIUS trial, the most recent and largest to date.46-49 In
most of the trials, large proportions of the controls also received ultrasound
results, although not with the same timing or frequency as in the screened
groups.
The most important potential
benefit of ultrasound screening is reduced perinatal
mortality. Among the seven trials that evaluated an ultrasound before 20 weeks
(with or without additional late ultrasound), only the Helsinki trial19 and a
meta-analysis heavily influenced by that trial's results47 were
able to demonstrate a statistically significant benefit in lowering perinatal mortality. Two trials17,40 showed nonsignificant reductions in mortality
while the remaining four trials and another meta-analysis48 showed
no mortality benefit. In the Helsinki trial, the overall perinatal death rate was 4.6/1,000 deliveries (n = 18) in screened women versus 9.0/1,000 deliveries (n = 34) in unscreened women. In the experimental group, 11 induced
abortions were performed because of ultrasound findings and two babies died
with major anomalies, compared to no abortions and 10 deaths
with anomalies in the control group. There was no difference in perinatal mortality when the induced abortions resulting
from ultrasound detection of congenital anomalies were included as deaths in
the analysis. The meta-analysis47 that reported a significant
mortality reduction included the four then-published trials16-19 that compared routine to selective ultrasound scanning and that reported
number of pregnancies, deliveries, and perinatal
deaths. It also evaluated the live birth rate, which takes
into account induced abortions for malformations, and found it to be identical
in the screened and control groups. The largest trial to date, the RADIUS trial,38
randomized 15,151 low-risk pregnant women to routine ultrasound scans at 15-22 and 31-35 weeks
of gestation or to usual care, which included ultrasounds performed for
indications that developed after randomization. The risk of fetal or neonatal
death was the same in the screened (0.6%, n = 52) and
control (0.5%, n = 41) groups. Including induced abortions for fetal anomalies (9 vs. 5 in the
routinely and selectively screened groups, respectively) did not affect these
estimates.
Effects on neonatal and
maternal morbidity from a single second-trimester scan have also been
evaluated. Most of the trials and meta-analyses showed no statistically
significant benefit of prenatal ultrasound on neonatal morbidity (including low
birth weight, admission to special care nursery, neonatal seizures, mechanical
ventilation, and Apgar scores), or on maternal
outcomes such as antenatal hospitalization.15,17-19,46,47 In one
randomized controlled trial of early second-trimester ultrasound,16 babies
born to screened women had a significantly greater mean birth weight (3,521 g vs. 3,479 g) than did those born to controls, with most of the benefit accruing to
smokers. The Cochrane Database meta-analysis reported significantly fewer low
birth weight singleton births and reduced risk of admission to special care
nurseries with routine early ultrasound, but no effect on Apgar
scores.48 The RADIUS trial reported a slightly lower rate of tocolysis
in screened women (3.4% vs. 4.2%), but no other differences in maternal outcomes (e.g., amniocentesis,
external version, cesarean delivery, or days of hospitalization)39 or in
overall or individual indicators of perinatal
morbidity.38
Accurate dates determined by
second-trimester ultrasound might help prevent routine tests of fetal
well-being and the induction of labor for fetuses thought to be postterm on the basis of erroneous dating.3,4,26 Rates
of induced labor for postterm pregnancy were
significantly reduced in three trials16,39,40 but
were unaffected in two others;17,18 meta-analysis demonstrated
significantly decreased inductions for postterm
pregnancy.48 These trials may have underestimated such effects by including women with
reliable dates, who are less likely to benefit from ultrasound dating. Trials
and meta-analyses have not established whether overall rates of induced labor
are reduced by a second-trimester ultrasound.15-18,39,40,47,48 In the RADIUS trial, the significant decreases in induced labor for postterm pregnancy were completely offset by significant
increases in inductions for IUGR.39 Two meta-analyses46,47
reported significant heterogeneity among the trials, suggesting that other
factors, such as differences in obstetric management between countries or over
time, may also influence this outcome. In one community, the incidence of postterm inductions fell from 8% to 2.6% after
ultrasound screening was instituted,4 but it
was not proved that this trend was due specifically to improved accuracy of
estimating gestational age. Two trials of second-trimester ultrasound reported
other outcomes potentially related to inaccurate dates. The RADIUS trial found
no significant effect of ultrasound screening on adverse perinatal
outcomes among postdate pregnancies38 or on the number of tests
performed to assess fetal well-being.39 Another trial reported
significantly fewer days of inpatient neonatal care after treatment for
"overdue pregnancy" among screened cases.40 Other
potential benefits of prenatal ultrasound, including the early detection of multiple
gestations and congenital anomalies, are often cited in support of screening.
The early detection of multiple gestation, a risk factor for intrapartum and neonatal complications,3 might allow improved antenatal surveillance and management, but direct
evidence of clinical benefits from early detection, such as improved maternal
or neonatal outcome, is lacking. No significant improvements in fetal,
neonatal, or maternal outcomes in multiple gestations were reported in any of
the screening trials, except for a small reduction in use of tocolytics in the RADIUS trial.15-19,38,39 Numbers of multiple gestations were small in all trials, however, and
power to detect improved outcomes from screening was generally inadequate.
There is also no clear evidence that early intervention for identified multiple
gestation, including routine hospital admission for bed rest, cervical cerclage, or prophylactic oral tocolysis,
results in improved perinatal outcome.50
While ultrasound before 20 weeks
allows earlier detection of fetal structural malformations, it is not clear
that this results in improved outcome. In the Helsinki trial, early detection
led to an increased rate of elective abortions (2.7/1,000 screened women vs. 0/1,000 control women) and therefore to reduced perinatal
deaths (see above).19 On the other hand, in the RADIUS trial,38
screening had no statistically significant effect on the rate of induced
abortion (n = 9 or 1.2/1,000 screened women compared to n = 4 or 0.5/1,000 controls). Although early detection might theoretically improve survival
for infants with fetal anomalies if they could be delivered at tertiary care
centers capable of immediate medical and surgical intervention, no significant
effects of early detection on overall perinatal
mortality, or on survival rates among infants born with acute life-threatening
anomalies or with any major anomalies, were seen in the RADIUS trial.21,38 Other
trials of routine ultrasound before 20 weeks have detected too few
(i.e., 0-2) malformations to allow meaningful comparisons of outcomes.15-18,40 None
of the trials has evaluated whether routine screening improves outcomes in
newborns with nonlethal anomalies.
Eight randomized controlled
trials and one meta-analysis have evaluated the effectiveness of routine
third-trimester ultrasound focused on fetal anthropometry and morphology in
improving outcomes.18,35,38-44,49 Six trials involved low-risk patients or patients selected from the
general population,18,35,38-40,42,43 while two were restricted to women with suspected IUGR or at increased
risk for IUGR or other complications (with results of the scan either released
or withheld based on randomization).41,44
Several of these trials had methodologic problems
such as inadequate reporting of results,40 use of hospital number for
randomization,35 and the revealing of test results for nearly one third of cases in the
control group because of obstetrician requests.41 These
studies reported no significant reductions in low Apgar
scores, admission to or length of stay in special care nursery, low birth
weight or preterm delivery, perinatal morbidity, or perinatal mortality (excluding lethal malformations). There
were also no consistent beneficial effects on antenatal hospitalization or
induction of labor. The meta-analysis49 reported that third-trimester
ultrasound was associated with a significantly increased risk of antenatal
hospital admission. One additional randomized controlled trial in unselected
women, all of whom received ultrasounds at midtrimester
and twice in the third trimester, evaluated whether reporting the result of
placental grading by third-trimester ultrasound to the clinician responsible
for care improved neonatal outcome.45 Reporting the placental
grading was associated with significant reductions in meconium
staining in labor, low Apgar scores at 5
minutes, and perinatal mortality in normally formed
babies. One previously cited trial43 of serial third-trimester
ultrasounds also assessed placental morphology and reported no beneficial
effects of ultrasound on perinatal mortality or
morbidity, but the method of assessing placental morphology was not described.
Additional trials of third-trimester placentalgrading
are needed to assess its effectiveness.
There is no clear evidence of
important adverse effects related to screening ultrasonography
reported from the published randomized controlled trials, although such effects
might be difficult to detect given the small number of ultrasounds (usually one
or two per patient) and the fact that the controls in many trials were also
scanned, with results concealed. One randomized controlled trial compared
routine multiple ultrasound scans plus Doppler flow studies to selective
ultrasound for indications, with four or more scans being done in 91% of
screened vs. 8% of control women.43 The screened group had a
significantly higher percentage of infants with birth weight below the 3rd and 10th
percentiles. Although this was not a primary endpoint of the study, it suggests
a possible adverse effect of frequent ultrasound examinations with Doppler
studies on fetal growth, which is supported by several studies in mice and
monkeys.51-53 Long-term follow-up of singleton live births to age 8-9 years
from the two Norwegian trials (in which only 19% of
controls received ultrasound) was performed to evaluate possible adverse
effects of ultrasound on neurologic development.54,55 These
two studies, with 83-89% response rates, found no differences between the two groups in school
performance; deficits in attention, motor control, or perception (by parent
questionnaire); development in infancy; or prevalence of dyslexia. Although
false-positive diagnoses of major fetal malformations occurred in both the
Helsinki trial (2.7/1,000 women in the screened group) and in the RADIUS trial (0.9/1,000), none of these pregnancies was electively aborted as a result.19,21 Case reports have
suggested adverse psychological effects of early and false-positive diagnoses
of fetal abnormalities,56-58 but no
controlled studies that evaluated adverse effects of ultrasound diagnosis of
fetal anomalies were found.
Recommendations of Other Groups
A National Institutes of Health consensus development conference recommended
that ultrasound imaging during pregnancy be performed only for a specific
medical indication and not for routine screening.59 This
is also the position of the American College of Obstetricians and Gynecologists.2 The
Canadian Task Force on the Periodic Health Examination found fair evidence to
recommend a single second-trimester ultrasound examination in women with normal
pregnancies, but concluded that there was insufficient evidence to recommend
the inclusion or exclusion of routine serial ultrasound screening for IUGR in
normal pregnancies.60
Discussion
Neither early, late, nor serial ultrasound in normal
pregnancy has been proven to improve perinatal
morbidity or mortality. Clinical trials show that a single midtrimester
ultrasound examination detects multiple gestations and congenital malformations
earlier in pregnancy, but there is currently insufficient evidence that early
detection results in improved outcomes. In the U.S., it is
not clear whether early detection of fetal anomalies by routine ultrasound
leads to increased rates of induced abortion. In addition, many of the major
fetal anomalies discoverable by routine ultrasound might be detected anyway
during screening for Down syndrome (see Chapter 41) or
neural tube defects (see Chapter 42). Routine second-trimester
ultrasound can lower the rate of induction for presumed postterm
pregnancy, a benefit likely to accrue primarily to women with unreliable dates,
among whom ultrasound is more accurate than dates for predicting actual date of
delivery. Early ultrasound has not been proven to reduce overall rates of
induction, however, due to increases in inductions for other indications. It is
also unclear whether the likeliest potential benefits of routine
second-trimester ultrasound (reduced induction of labor for postterm
pregnancy and increased induced abortions for fetal anomalies) would justify
the significant economic implications of widespread testing. No benefits of
routine ultrasound examination of the fetus in the third trimester have been
demonstrated despite multiple randomized controlled trials. Additional trials
of third-trimester placental grading are needed to adequately evaluate the
potential benefits of screening for placental appearance. Further research to
evaluate possible adverse effects of ultrasound and the cost-effectiveness of
routine screening is also needed.
CLINICAL
INTERVENTION
Routine ultrasound examination of the fetus in the third trimester is not recommended, based on multiple trials and meta-analyses
showing no benefit for either the pregnant woman or her fetus ("D"
recommendation). There is currently insufficient evidence to recommend for or
against a single routine midtrimester ultrasound in
low-risk pregnant women ("C" recommendation). These recommendations
apply to routine screening ultrasonography and not to
diagnostic ultrasonography for specific clinical
indications (e.g., follow-up evaluation of elevated maternal serum a-fetoprotein).