Keywords diagnostic effectiveness - fetal Doppler - neonatal mortality - stillbirth
Introduction
India has a declining perinatal mortality rate that is still very high relative to
the global rates.[1 ]
[2 ]
[3 ] The poor perinatal health in India is exacerbated by the high prevalence of pre-eclampsia
(PE), fetal growth restriction (FGR), and preterm births in India.[3 ]
[4 ] An estimated 8 to 10% of pregnant women in India develop PE during pregnancy and
an estimated 3.5 million children are born preterm every year.[5 ]
[6 ]
Fetal Doppler studies help to ascertain hemodynamic redistribution suggestive of fetal
adaptation to undernutrition/hypoxia, placental disease, higher risk of PE, and distinguish
FGR from small for gestational age (SGA) babies.[7 ] Conventionally, the umbilical artery (UA) has been used to identify FGR with an
abnormal UA Doppler study predicting poorer outcomes among small fetuses.[7 ]
[8 ]
[9 ] However, UA Doppler indices have limitations as they may not pick up a mild placental
disease.[10 ] Previous studies have reported that a significant proportion of fetuses with normal
UA pulsatility indices (PI) have worse outcomes than fetuses with normal growth.[10 ]
[11 ]
[12 ]
[13 ] The cerebroplacental ratio (CPR) is more sensitive to hypoxia than UA or middle
cerebral artery (MCA) PI and has a better correlation with adverse perinatal outcomes.[7 ]
[14 ]
[15 ] The uterine artery (UtA) PI can be abnormal even in the presence of a normal UA
PI and predict poorer perinatal outcomes.[7 ] This recent information led to the integration of UtA PI, CPR (includes both UA
and MCA PI) and estimated fetal weight (EFW) < third percentile as core components
of the diagnosis of FGR.[7 ] However, there is a lack of information on the effectiveness of these components
to identify FGR in India.
Samrakshan is an ongoing national program of the Indian Radiological and Imaging Association
(IRIA) started in July 2019 that integrates trimester-specific fetal Doppler studies
with routine antenatal ultrasound studies to determine a customized risk assessment
of preterm PE and FGR for each pregnant woman.[16 ] In this manuscript, we present the diagnostic effectiveness of individual abnormal
fetal Doppler parameters with late- and-term stillbirths and neonatal deaths in a
cohort of pregnant women screened in the third trimester of pregnancy using the Samrakshan
protocols.
Materials and Methods
Samrakshan utilizes an opportunistic screening approach focused on pregnant women
seeking imaging services at departments of clinical radiology at hospitals, clinics,
and diagnostic centers. A unique identification number was assigned to all participants
in the screening program. Demographic details, clinical details including previous
obstetric and imaging history, co-morbidity including prior history of PE, development
of PE in the current pregnancy, gestational age at diagnosis of PE in the current
pregnancy, and personal risk behaviors were collected from each participant. All women
underwent routine third trimester-specific antenatal ultrasound exams for fetal biometric
parameters. This analysis includes consecutive pregnant women in the third trimester
screened in the Samrakshan program from September 2019 (after the first training continuous
medical education program) to February 2022 and for whom Samrakshan radiologists could
access details of childbirth. Women screened in the third trimester of pregnancy without
childbirth outcome details or yet to deliver were not considered for this analysis.
Fetal Doppler studies were integrated with the antenatal scans for all women. Doppler
studies included assessments of the mean UtA PI, UA PI, MCA PI, and CPR. All Doppler
studies were performed by experienced FMF-certified fetal radiologists who had attended
the Samrakshan continuous medical education (CME) workshops on trimester-specific
ultrasound assessments as trainers and faculty for the Samrakshan program.
A transabdominal approach was used to assess the Doppler measures of interest. The
UtA PI was assessed at the apparent crossover of the right and left uterine arteries
at the external iliac arteries with the pulsed wave Doppler sampling gate set at nearly
2 mm.[17 ] A peak systolic velocity > 60 cm/s confirmed that the uterine artery was being assessed.[17 ] The PI was estimated when three similar consecutive waveforms were obtained and
a PI > 95th percentile was considered abnormal ([Fig. 1 ]).[17 ] The UA Doppler indices were measured at a free loop and a PI > 95th percentile was
considered abnormal ([Fig. 2 ]).[18 ] The MCA was assessed by placing the pulsed wave Doppler gate at the proximal third
of MCA close to its origin in the internal carotid artery and keeping the angle between
the direction of blood flow and ultrasound beam as close to 0 as possible.[18 ] A minimum of 3 but fewer than 20 waveforms were recorded with the highest point
of the waveform (the peak systolic volume) measured using auto trace or manual calipers.[18 ] An MCA PI < fifth percentile was considered abnormal ([Fig. 3 ]).[18 ] The CPR was estimated by dividing the MCA Doppler with the UA Doppler indices and
a CPR < fifth percentile was considered abnormal.[7 ]
[16 ] The percentile values were determined using the Barcelona calculator available online
and as an App.
Fig. 1 Abnormal uterine artery waveform.
Fig. 2 Reversal of diastole in umbilical artery in early FGR stage III. FGR, fetal growth
restriction.
Fig. 3 Middle cerebral artery waveform in early FGR. FGR, fetal growth restriction.
Doppler measures were integrated with the fetal biometry and EFW assessments to clinically
stage FGR for further management of the fetus.[7 ] The clinical staging was used to recommend the frequency and interval of repeat
Doppler assessments and were communicated to the obstetrician to plan interactively
for childbirth.
Data from the ultrasound assessments were anonymized and transcribed to an online
Google form integrated with a Google spreadsheet located in a dedicated password-protected
Google Drive folder for live updating of entries. The data were subsequently exported
to STATA v14.0 (College Station, Texas, USA) for statistical analysis. The fetal Doppler
measures closest to childbirth were considered for analysis for women who had multiple
Doppler studies. An abnormal Doppler study was defined as the presence of any one
or more of an abnormal mean UtA PI, UA PI, MCA PI, and CPR PI. The proportion of abnormal
Doppler studies was expressed as a proportion and the 95% confidence intervals around
the point estimates. The loss of a fetus from 28 weeks of pregnancy until childbirth
was defined as SB in the third trimester. SB was further categorized as late SB (from
28 weeks to 36 weeks) and term SB from 37 weeks till delivery. Neonatal deaths were
defined as the loss of a liveborn baby within 28 days of childbirth. The data were
initially explored for missing data, errors in data entry, outliers, the distribution
of data (normality of distribution, the width of the interquartile range [IQR] and
the IQR/median ratio), and floor and ceiling effects of the data, and pooling around
certain ranges or values. We considered a <5% cut-off for errors as reasonable for
data inclusion. Additionally, potential data errors that were identified were verified
with the medical records of the participating investigators. A bivariate logistic
regression model was used to determine the magnitude and direction of the association
of abnormal Doppler studies with late stillbirth, term stillbirth, and neonatal deaths.
The diagnostic effectiveness of abnormal Doppler studies and late stillbirths, term
stillbirths and neonatal deaths was expressed as the sensitivity, specificity, positive
and negative predictive values (PPV and NPV) and likelihood ratios (LR+ and LR − ),
and area under receiver operator characteristic (AUROC) curves. The point estimates
and 95% confidence intervals (CI) of the point estimates were estimated.
Results
Third-trimester Samrakshan protocol-specific assessment and childbirth outcomes details
from September 2019 to February 2022 were available for 1,326 women and these were
included in the analysis. There were 308 (23.23%, 95% CI: 20.95, 25.50) abnormal Doppler
studies, 11 (0.83%, 95% CI: 0.47, 1.48) SB including 5 late SB and 6 term SB, and
11 (0.84%, 95% CI: 0.47, 1.49) neonatal deaths in this cohort ([Table 1 ]). [Tables 2 ] to [4 ] present the diagnostic effectiveness of Doppler studies for late-and-term stillbirths
and neonatal deaths in this population. An abnormal Doppler study was significantly
associated with late stillbirths (OR 37.2, 95% CI: 2.05, 674) but not with term SB
(OR: 3.38, 95% CI: 0.76, 15) or neonatal deaths (OR 1.39, 95% CI: 0.40, 4.87). Mean
UtA PI, Umbilical artery PI, MCA PI and CPR were significantly associated with late
SB but not with term SB. The AUROC of Doppler measures was excellent for late SB (all
AUROC > 0.85) but did not show discriminatory ability for term SB or neonatal deaths
([Tables 2 ] to [4 ]). Of the 308 subjects with an abnormal Doppler test, 295 (95.78%) had normal liquor,
13 (4.22%) had oligohydramnios, and none had polyhydramnios. The liquor was normal
in all the cases with stillbirths or neonatal mortality in this cohort.
Table 1
Prevalence of abnormal Doppler studies, stillbirths, and neonatal deaths in the 1,326
pregnant women screened in Samrakshan in the third trimester of pregnancy
n , %, (95% CI)
Abnormal Doppler Study
308, 23.23% (20.95, 25.50)
Mean Uterine Artery PI >95th percentile
156, 11.56% (10.03,13.50)
Umbilical Artery PI >95th percentile
89, 6.71% (5.36, 8.06)
Middle Cerebral Artery PI <5th percentile
123, 9.28% (7.71, 10.84)
Cerebroplacental Ratio <5th percentile
154, 11.61% (9.89, 13.34)
Late stillbirth (28 to 36 weeks)
5, 0.38% (0.16,0.88)
Term stillbirth (37 weeks until childbirth)
6, 0.45% (0.21, 0.99)
Neonatal death (<28 days of childbirth)
11, 0.84% (0.47, 1.49)
Table 2
Diagnostic effectiveness of abnormal Doppler studies for late stillbirths in the population
screened using Samrakshan protocol in the third trimester
Mean UtA PI > 95th percentile
(95% CI)
UA PI >95th percentile
(95% CI)
MCA PI <5th percentile
(95% CI)
CPR <5th percentile
(95% CI)
Abnormal Doppler study
(95% CI)
Sensitivity
100% (47.8,100)
100% (47.8,100)
80%
(28.4,99.5)
100% (47.8,100)
100% (47.8,100)
Specificity
88.6%
(86.8, 90.3)
93.6%
(92.2, 94.9)
91%
(89.4, 92.5)
88.8%
(87.0, 90.5)
77.2%
(74.8, 79.4)
PPV
3.23%
(1.06,7.37)
5.62%
(1.85,12.6)
3.28%
(0.9, 8.18)
3.29%
(1.08, 7.51)
1.64%
(0.53, 3.78)
NPV
100%
(99.7, 100)
100%
(99.7, 100)
99.9%
(99.5, 100)
100%
(99.7, 100)
100%
(99.6, 100)
LR+
8.02
(6.03, 10.70)
14.3
(10.4,19.6)
8.33
(5.09, 13.6)
8.18
(6.15,10.9)
4.01
(3.09, 5.21)
LR-
0.09
(0.01, 1.34)
0.09
(0.01,1.27)
0.28
(0.07,1.1)
0.09
(0.01, 1.33)
0.11
(0.01, 1.53)
Odds Ratio
85.2
(4.69, 1548)
160
(8.79, 2924)
30.3
(4.73, 194)
87.1
(4.79, 1584)
37.2
(2.05, 674)
AUROC
0.94
(0.93, 0.95)
0.97
(0.96, 0.98)
0.86
(0.66, 1)
0.94
(0.93, 0.95)
0.89 (0.87,0.90)
Abbreviations: AUROC, area under receiver operator characteristic curve; CI, confidence
interval; CPR, cerebroplacental ratio; LR, negative likelihood ratio; LR + , positive
likelihood ratio; MCA, middle cerebral artery; NPV, negative predictive value; PI,
pulsatility index; PPV, positive predictive value; UA, umbilical artery; UtA, uterine
artery.
Table 3
Diagnostic effectiveness of abnormal Doppler studies for term stillbirths in the population
screened using Samrakshan protocol in the third trimester
Mean UtA PI > 95th percentile
UA PI >95th percentile
MCA PI <5th percentile
CPR <5th percentile
Abnormal Doppler study
Sensitivity
16.7%
(0.42, 64.1)
0%
(0, 45.9)
16.7%
(0.42, 64.1)
33.3%
(4,33, 77.7)
50%
(11.8, 88.2)
Specificity
88.6%
(86.8, 90.3)
93.6%
(92.2, 94.9)
91%
(89.4, 92.5)
88.8%
(87, 90.5)
77.2%
(74.8, 79.4)
PPV
0.67%
(0.02, 3.63)
0%
(0, 4.3)
0.84%
(0.02, 4.59)
1.34%
(0.16, 4.76)
0.99%
(0.21, 2.87)
NPV
99.6%
(99, 99.9)
99.5%
(98.9, 99.8)
99.6%
(99, 99.9)
99.7%
(99.1,99.9)
99.7%
(99.1, 99.9)
LR+
1.87
(0.45, 7.8)
1.11
(0.08, 16.2)
2.38
(0.57, 9.93)
3.19
(1.17, 8.71)
2.19
(1.04, 4.62)
LR-
0.89
(0.61, 1.31)
0.99
(0.81, 1.22)
0.86
(0.59, 1.27)
0.72
(0.42, 1.26)
0.65
(0.31,1.36)
Odds ratio
2.11
(0.35, 12.9)
1.12
(0.07, 20.1)
2.76
(0.45, 16.9)
4.4
(0.93, 20.9)
3.38
(0.76, 15)
AUROC
0.52
(0.36, 0.69)
0.47
(0.46, 0.48)
0.54
(0.38, 0.71)
0.61
(0.40, 0.82)
0.64
(0.42, 0.86)
Abbreviations: AUROC, area under receiver operator characteristic curve; CI, confidence
interval; CPR, cerebroplacental ratio; LR, negative likelihood ratio; LR + , positive
likelihood ratio; MCA, middle cerebral artery; NPV, negative predictive value; PI,
pulsatility index; PPV, positive predictive value; UA, umbilical artery; UtA, uterine
artery.
Table 4
Diagnostic effectiveness of abnormal Doppler studies for neonatal deaths in the population
screened using Samrakshan protocol in the third trimester
Mean UtA PI > 95th percentile
UA PI >95th percentile
MCA PI <5th percentile
CPR <5th percentile
Abnormal Doppler study
Sensitivity
0%
(0, 28.5)
27.3%
(6.02, 61)
18.2%
(2.28, 51.8)
18.2%
(2.28, 51.8)
27.3%
(6.02, 61)
Specificity
88.5%
(86.6, 90.2)
93.8%
(92.3, 95)
91.1%
(89.4, 92.6)
88.9%
(87, 90.5)
77.2%
(74.8, 79.5)
PPV
0%
(0, 2.43)
3.57%
(0.74, 10.1)
1.69%
(0.21, 5.99)
1.36%
(0.17, 4.83)
1%
(0.21, 2.89)
NPV
99.1%
(98.3, 99.5)
99.4%
(98.7, 99.7)
99.2%
(98.6, 99.7)
99.2%
(98.5, 99.6)
99.2%
(98.5, 99.7)
LR+
0.36
(0.02, 5.47)
4.67
(1.89, 11.6)
2.33
(0.77, 7.13)
1.87
(0.62, 5.69)
1.28
(0.53, 3.11)
LR-
1.08
(0.96, 1.22)
0.76
(0.5, 1.09)
0.87
(0.65, 1.16)
0.89
(0.67, 1.19)
0.92
(0.64, 1.32)
Odds ratio
0.34
(0.02, 5.69)
6.18
(1.74, 21.9)
2.68
(0.66, 11)
2.1
(0.52, 8.54)
1.39
(0.40, 4.87)
AUROC
0.44
(0.43, 0.45)
0.61
(0.47, 0.75)
0.55
(0.43, 0.67)
0.54
(0.42, 0.66)
0.52
(0.38, 0.67)
Abbreviations: AUROC, area under receiver operator characteristic curve; CI, confidence
interval; CPR, cerebroplacental ratio; LR, negative likelihood ratio; LR + , positive
likelihood ratio; MCA, middle cerebral artery; NPV, negative predictive value; PI,
pulsatility index; PPV, positive predictive value; UA, umbilical artery; UtA, uterine
artery.
Discussion
The results from Samrakshan show a significantly strong association of the individual
Doppler tests with late stillbirths. Abnormalities of the individual Doppler tests
showed AUROC curves suggestive of an excellent discriminatory ability to differentiate
between late stillbirths and normal childbirth. Clinically, we can use the sensitivity
of a test to rule out the condition of interest. A normal result in a test with a
very high sensitivity suggests a low probability for the condition of interest in
that person. Individual Doppler tests in this cohort had a very high sensitivity suggesting
that the presence of a normal Doppler test can be used to clinically rule out a high
risk for late stillbirth in this population. The upper limits of the 95% confidence
interval of the LR+ for the individual Doppler tests were >10 suggesting that fetal
Doppler tests have a clinical utility in the identification of pregnant women at high
risk for late stillbirths. A low positive predictive value of fetal Doppler was anticipated
as the PPV is dependent on the prevalence of the condition and the prevalence of late-and-term
stillbirths and neonatal deaths is low in this cohort.
Fetal Doppler tests did not have good diagnostic effectiveness to identify pregnant
women at risk for term stillbirths or neonatal mortality. The tests had a poor AUROC
curve suggesting poor discrimination between pregnant women at high risk for term
stillbirths and neonatal mortality and those with normal childbirth and were not significantly
associated with term stillbirths or neonatal mortality. An LR+ of 10 or more suggests
the test has diagnostic utility in clinical practice and indicates an increased probability
of the conditions of interest. None of the fetal Doppler tests had an LR+ >5 for term
stillbirth and neonatal deaths. Clinically, we can use the specificity of a test to
rule in the condition of interest. An abnormal result in a test with a very high specificity
suggests a high probability of the abnormality of interest in that person. The fetal
Doppler tests had a very high specificity (>85%) for term stillbirths and neonatal
mortality in this cohort suggesting that an abnormal Doppler test could be used clinically
to stratify pregnant women that must be monitored more intensely than pregnant women
with normal fetuses.
The lack of statistical significance for fetal Doppler tests in the third trimester
to identify pregnant women at risk for term stillbirths or neonatal mortality may
be a true lack of significance or may be driven by the low prevalence of term stillbirths
and neonatal mortality in this cohort. The lack of significance must also be considered
within the context of the absence of a natural slow progression for late FGR.[7 ]
[10 ] Late FGR usually has a mild placental disease and may not show abnormal UA PI measures.[10 ] These fetuses may have an abnormal MCA PI suggestive of chronic hypoxia and cerebral
vasodilation and an abnormal CPR.[10 ]
[19 ] Late FGR fetuses may show rapid acute fetal deterioration without identifiable signs
before labor and may have intrapartum fetal distress and neonatal acidosis leading
to mortality.[7 ]
[19 ]
[20 ] The lack of a significant association of third-trimester Doppler studies with term
stillbirths and neonatal mortality in this cohort may reflect an acute deterioration
in late-stage FGR and diminished placental reserve.[7 ]
[19 ]
[20 ] Kady et al had reported that a stillborn fetus had a high chance of FGR before demise,
with an OR of 5.3 if EFW was <10th customized centile, and OR of 11.2 for EFW <2.5
centiles.[19 ] Figueras et al had previously reported that perinatal outcome in small-for-gestational-age
fetuses with normal umbilical artery Doppler is suboptimal and questioned the role
of umbilical artery Doppler to discriminate between normal-SGA and growth-restricted
fetuses.[20 ] Exploring the associations in a larger cohort of pregnant women will help to determine
the magnitude and direction of associations with term stillbirths and neonatal deaths
more accurately.
An EFW < third percentile is a core component of the diagnosis of FGR[7 ] but has its limitations in clinical practice in India. We must consider that the
determination of EFW percentiles in USG machines in India is not based on normative
data from an Indian population and therefore introduces limitations in interpretation.
These limitations extend to fetal biometry assessments as well as they are based on
normative data from other populations. Serial longitudinal assessments using the same
formula to estimate biometry and fetal weights help to minimize this limitation. However,
antenatal care service uptake is suboptimal in India and with limited longitudinal
follow-ups with the same practitioner. Additionally, migration of pregnant women to
their parental house for childbirth is a common cultural occurrence and might result
in antenatal assessments by multiple practitioners at different locations limiting
the possibility of serial assessments. We must also consider that changes in fetal
biometry and fetal weights is a slow process and rarely shows significant acute changes.
It is not pragmatic to expect a significant discernible shift in fetal weights or
biometry within a short period closer to term, which limits the utility of biometry
and EFW to stratify risk in pregnant women with late FGR.
Doppler studies provide objective measures that can help early identification of fetal
deterioration. Previous studies have reported that absent or reversed end-diastolic
velocities, is found nearly 1 week before the acute deterioration of the fetus and
in nearly 40% of fetuses with acidosis.[21 ] A large systematic review has previously reported that ductus venosus (DV) Doppler
has predictive capacity for perinatal mortality.[22 ] Previous studies have reported that an abnormal DV is observed before the loss of
short-term variability (STV) in computerized cardiotocography (cCTG) in 50% of cases
and is abnormal 48 to 72 hours before the biophysical profile (BPP) in ∼90% of cases.[23 ]
[24 ] Data on the DV Doppler assessment is not presented in this manuscript as the test
was performed only in cases with an abnormal UA Doppler PI or abnormal CPR or suspicion
of advanced FGR based on the discretion of the radiologist and was not routinely done
for all pregnant women in the Samrakshan protocol.
Conclusion
The integration of fetal Doppler studies provides an objective assessment of fetal
hemodynamic circulation and can be used clinically to identify pregnant women at risk
for stillbirths and neonatal mortality in India. Early identification of fetuses at
risk for fetal deterioration will help the clinical management of childbirth and reduce
perinatal mortality in India.
