Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
To cite this article: Francesca Crovetto, Nicola Cesano, Federica Rossi, Stefano Acerboni,
Stefano D. E. Marinis, Annachiara Basso, Raigam Jafet Martinez Portilla, Barbara Acaia, Luigi
Fedele, Enrico Ferrazzi & Nicola Persico (2019): Intrapartum prediction of emergency delivery
due to non-reassuring fetal status at 40 weeks’ gestation in low-risk pregnancies: contribution
of Doppler parameters, maternal history, and intrapartum clinical characteristics, The Journal of
Maternal-Fetal & Neonatal Medicine, DOI: 10.1080/14767058.2019.1671338
ORIGINAL ARTICLE
CONTACT Francesca Crovetto fcrovetto@clinic.cat Barcelona Center for Maternal-Fetal and Neonatal Medicine, Hospital Clınic and Hospital Sant
Joan de Deu, Natal, IDIBAPS, University of Barcelona, and Centre for Biomedical Research on Rare Diseases (CIBER-ER), Barcelona, Spain
ß 2019 Informa UK Limited, trading as Taylor & Francis Group
2 F. CROVETTO ET AL.
term. Over the last decade, several studies have sug- tachycardia and the assessment of variability as
gested that Doppler assessment could be valuable in reported elsewhere [14]. In cases with two or more cri-
the detection of placental insufficiency in adequate- teria of indeterminate tracing and one or more criteria
for-gestational-age fetuses (AGA). The rationale behind of abnormality, indication for emergency delivery
this hypothesis is that Doppler parameters could cap- was considered for non-reassuring fetal status.
ture a subgroup of fetuses with placental insufficiency Obstetricians and midwives who carried out labor
that has not reached an estimated fetal weight (EFW) were blinded to the results of Doppler evaluation.
below the 10th percentile. In these cases, the only
sign of placental insufficiency would be a redistribu-
Ultrasound evaluation
tion of the fetal circulation shown by a reduction in
the middle cerebral artery (MCA) pulsatility index (PI) Pregnancies were dated by first-trimester crown-rump
and cerebroplacental ratio (CPR) [7–9], which has been length measurements [15]. The EFW was calculated
associated with an increased risk for emergency cesar- using the Hadlock formula [16].
ean section and neonatal acidosis [10–12]. The object- According to the hospital protocol for low-risk
ive of this study is to assess the added value of pregnancies, all patients were evaluated at 40 weeks
Doppler parameters combined with maternal history of gestation for the assessment of fetal well-being, by
and intrapartum clinical characteristics for the intrapar- the biophysical profile modified based on fetal move-
tum prediction of emergency delivery due to non- ments and the measurement of the amniotic fluid
reassuring fetal status in low-risk pregnancies. index (AFI). Ultrasound Doppler examinations were
performed by one of three experienced operators
(N.C., S.A., or SD), using a General Electric Voluson E6
Materials and methods (GE Medical Systems, Zipf, Austria) ultrasound machine
Study population equipped with a 6–2 MHz linear curved-array trans-
ducer. Doppler recordings were performed in the
This is a blinded prospective cohort of low-risk preg-
absence of fetal movements, voluntarily suspended
nancies conducted in a third-level reference hospital
maternal breathing, and maternal contractions.
in Italy between October 2014 and December 2015.
Spectral Doppler parameters were performed automat-
Criteria for inclusion were (1) low-risk pregnant
ically from three or more consecutive waveforms, with
woman (where low-risk was considered in the absence
the angle of insonation as close to 0 as possible.
of any major maternal morbidity or pregnancy compli-
Umbilical artery PI was calculated from a free-floating
cation), (2) admitted for hospital evaluation at
cord loop, and it was considered abnormal if >95th
40 weeks’ gestation, (3) with history of an adequate-
centile [17]. Middle cerebral artery PI was measured in
for-gestational age fetus (EFW 10th percentile) at
a transversal view of the fetal head, at the level of its
third trimester ultrasound screening (30–34 weeks’
origin from the circle of Willis; it was considered
gestation) according to local standards [13], (4) deliv-
abnormal if below the fifth centile [17,18]. Finally, CPR
ering within the next 7 days from evaluation. Exclusion
was calculated as a ratio of the MCA PI to the UA PI,
criteria were multiple pregnancies, pregnancies com-
and its value was considered abnormal if below the
plicated with neonatal chromosomal or structural
fifth centile [18]. Three images were taken from each
abnormalities, intrauterine infection, and those fetuses
vessel, and the mean of them was used for subse-
without a cephalic presentation. The hospital ethics
quent analysis.
committee approved the study protocol, and written
consent was obtained for the study from all recruited
Management
patients (protocol # 354_2014).
No clinical decisions were based on Doppler results
since Doppler examinations were blinded to the
Main outcome
attending obstetrician. Indications for hospital admis-
The main outcome was the indication of emergency sion were active labor, premature rupture of the mem-
cesarean or vaginal operative delivery for non-reassur- branes (PROM), oligohydramnios, vaginal bleeding,
ing fetal status was based on abnormal fetal heart and 41 þ 0 weeks’ gestation if no signs of labor
rate tracing [14]. In brief, fetal heart monitoring was appeared according to the hospital protocol.
carried out, and tracings were classified as normal, Active labor was defined as cervical ripening and
suspicious, or abnormal, according to the presence, dilatation 3 cm in the presence of at least three
type, and length of decelerations; bradycardia; uterine contractions in 10 min recorded at
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE 3
cardiotocography [19]. Diagnosis of PROM was based combination of predictors for the main outcome
on history and confirmed by the presence of pooled [24,25]. Data were analyzed using Stata for Mac, v.14.1
amniotic fluid on a sterile speculum or positive result (College Station, TX) and R v15.1 (the R Foundation
from a prom-test. Oligohydramnios was defined if AFI for Statistical Computing) [package “pROC” and
was below 5 cm. Induction of labor was carried out by “Party”] [26]. A p values .05 was considered statistic-
cervical ripening with a release vaginal prostaglandin ally significant.
E2 or oral prostaglandin E1. If the onset of labor did
not occur within 18 h, oxytocin induction was started.
Results
Participants and baseline characteristics
Data collection and outcome measures
A total of 428 pregnancies fulfilled the selection crite-
Data on maternal characteristics including age, ethni-
ria, from these, six patients refused to participate in
city, body mass index, parity, smoking status, known
the study and 18 were excluded because of delivery
chronic disease as hypertension, diabetes mellitus,
renal disease, autoimmune disease, and previous after 7 days from initial evaluation, leaving 403 partici-
maternal history were recorded in the hospital data- pants for analysis. Baseline maternal characteristics of
base at inclusion. In addition, data regarding preg- the study population are shown in Table 1. The mean
nancy follow-up, complications developed during gestational age at admission was 40.4 weeks (SD
pregnancy, ultrasound evaluation and perinatal data 5 days) and the mean gestational age at delivery was
were prospectively collected. 40.5 weeks (SD 5 days).
Neonatal metabolic acidosis was defined as the
presence of UA pH less than or equal to 7.15 and Labor results and perinatal outcome
base excess greater than 12 mEq/L at birth [20]. SGA
was defined as birth weight was less than 10th centile, Induction of labor was carried out in 194 women
while large-for-gestational-age (LGA) was defined as (48.1%); the remaining 209 patients (51.9%) had the
birthweight was above the 90th centile, according to spontaneous onset of labor. During labor, 75 fetuses
local standards [13]. (18.6%) had an emergency delivery due to non-reas-
suring fetal status that required either an emergency
cesarean section (n ¼ 57, 76%) or an operative vaginal
Statistical analysis
Quantitative variables were assessed for normality Table 1. Maternal characteristics of the study popula-
tion (n ¼ 403).
using Shapiro–Wilk’s test: normally distributed
Characteristics Mean (SD) or n (%)
variables were compared using t-test and expressed as
Age (years) 33.9 (5)
mean and standard deviation (SD). Non-normally Age <35 years 237 (58.8)
distributed variables were compared using Age 35–39 years 126 (31.3)
Age 40 years 40 (9.9)
Mann–Whitney U-test and expressed as median and Ethnicity
interquartile range (IQR). Qualitative variables were Caucasian 347 (86.1)
Black 5 (1.2)
compared using v2 or Fisher’s exact test. The associ- Asian 21 (5.2)
ation between Doppler parameters (UA, MCA, CPR) Latin–American 20 (5)
Others 10 (2.5)
and emergency delivery was assessed by logistic Nulliparous 275 (68.2)
regression, where the basal comparison was a model Obstetrical history
Previous miscarriage 87 (21.6)
comprising maternal history and intrapartum, and the Previous delivery 118 (29.3)
full model included the addition of Doppler parame- Previous cesarean section 11 (2.7)
Maternal disease 55 (13.6)
ters using a nested logistic regression [21]. Models Thyroid disorders 28 (6.9)
were compared by assessing the improvement in their Autoimmune diseases 10 (2.5)
Infections 7 (1.7)
Naegelkerke R2 as a measure of goodness of fit (the Others 10 (2.5)
proportion of uncertainty explained by the model) ART pregnancy 20 (5)
Smoking during pregnancy 43 (10.7)
using a Wald v2 test [22]. The predictive performance Maternal BMI before pregnancy (kg/m2) 22.1 (3.6)
for the models was determined by receiver operating Weight gained during pregnancy (kg) 13 (4.6)
characteristic (ROC) curve analysis and the area under Pregnancy complicationsa 23 (5.7)
ART: assisted reproductive technologies; BMI: body mass index.
the curve (AUC) [23]. A Conditional Decision Tree a
Refers to those cases with chronic hypertensions (n ¼ 4), gestational dia-
Analysis was also performed to present the best betes mellitus (n ¼ 9) or other minor conditions (n ¼ 10).
4 F. CROVETTO ET AL.
Table 2. Delivery outcome of the study population (n ¼ 403). Table 3. Neonatal outcome of the study popula-
Characteristics Mean (SD) or n (%) tion (n ¼ 403).
Gestational age at delivery (weeks) 40.6 (0.4) Characteristics Mean (SD) or n (%)
Hospital admission GA at delivery (weeks) 40.4 (3)
Labor 163 (40.4) Birth weight (g) 3450 (356)
PROM 46 (11.4) Birth weight centile 46.2 (28.8)
Oligohydramnios 39 (9.7) Small-for-gestational-age 48 (11.9)
Maternal disease 16 (4) Large-for-gestational-age 44 (10.9)
Labor induction 194 (48.1) Male 200 (49.6)
Prolonged pregnancy (>41 weeks) 109 (56.2)a Apgar 1 minute 8.8 (0.8)
Oligohydramnios 41 (21.1)a Apgar 5 minutes 9.9 (0.5)
PROM 10 (5.2)a Umbilical cord evaluation
Maternal disease 14 (7.2)a pH artery 7.22 (0.1)
Fetal disease 7 (3.6)a BE artery 4.36 (3.4)
Others 13 (6.7)a Lactates artery 4.34 (2.2)
Meconium stained amniotic fluid 69 (17.1) Asphyxias’ parameters
Oxytocin during labor 218 (54.1) Apgar 1 minute <7 11 (2.7)
Maternal anesthesia 294 (73) Apgar 5 minutes <7 2 (0.5)
Type of delivery pHA <7.15 55 (13.6)
Vaginal delivery 264 (65.5) pHA <7.10 24 (6)
Operative delivery (vacuum) 46 (11.4) pHA <7.00 3 (0.7)
Cesarean section 93 (23.1) Neonatal problems 30 (7.4)
Emergent delivery 99 (24.6) Respiratorya 9 (30)
Non-reassuring fetal statusb 75 (75.8) Infectiousa 12 (40)
Mechanical dystociac 15 (15.2) Othersa 9 (30)
Othersd 29 (7.2) Neonatal hospitalization
PROM: premature rupture of the membranes. NICU admission 16 (4)
a
Refers to women underwent labor induction. NICU daysb 5.6 (7.5)
b
Refers to women underwent emergent delivery due to non-reassuring BE: base excess deficit; GA: gestational age; NICU: neonatal intensive
fetal status. care unit.
c
Refers to meconium stained amniotic fluid, non-progressive labor, mater- a
Refers to newborns with problems (n ¼ 30).
nal election, persistent bradycardia. b
Refers to newborns admitted to NICU.
d
Fetal chord prolapse, abruptio placentae.
emergency deliveries in a multiparous woman with study, we found that women requiring emergency
normal fetal Doppler was only 4%. delivery due to non-reassuring fetal status had fetuses
with lower MCA and CPR and that the addition of
Doppler MCA to a base model comprising maternal
Discussion history and intrapartum clinical characteristics, signifi-
cantly improved the prediction of an emer-
Main findings
gency delivery.
There is controversy on whether Doppler parameters During periods of either acute or chronic hypoxia,
are useful in low-risk pregnancies at term. In this the fetus redistributes its cardiac output to ensure
6 F. CROVETTO ET AL.
0.731 (p ¼ .008)
0.66 (p ¼ .027)
0.621
p-Value
<.001
<.001
<.001
.029
<.001
.02
3.74
0.69
8.7
8.3
9.4
3.7
95% CI
1.12
0.35
1.9
2.1
1.1
2
4.17
0.49
4.4
1.9
OR
4
2
Figure 1. Doppler comparisons between normal fetuses and
those requiring emergent delivery due to non-reassuring
fetal status.
MCA Z-score
Nulliparity
Nulliparity
Nulliparity
interaction of alterations in blood flow to the brain,
as a result of cerebrovascular dilatation revealed by
the MCA and increased placental resistance, resulting
in a decreased diastolic flow of the UA. The CPR
becomes abnormal earlier, showing a greater sensitiv-
ity compared to its individual components, but with
Maternal characteristics þ intrapartum findings
evaluated CPR within 2 weeks of delivery as a pre- utility of CPR for cesarean section. Similarly, our study
dictor of intrapartum fetal compromise and composite suggests that CPR and MCA Doppler are lower
adverse neonatal outcome, finding a fair predictive in women undergoing emergency delivery for non-
reassuring fetal status but most important, the add-
ition of MCA in a clinical setting would significantly
improve the prediction of an emergency delivery.
et al. [30] conducted a retrospect-
Morales-Rossello
ive cohort of 11 576 AGA term fetuses finding a rela-
tionship between a lower CPR and lower birth weight,
suggesting an increasing prevalence of fetal hypox-
emia in AGA fetuses. They proposed the CPR as a tool
to identify AGA fetuses that are failing to reach their
growth potential. These findings are in accordance
with other studies showing the potential role CPR has
to identify term AGA fetuses at risk of neonatal acide-
mia [9,31], and NICU admission [32,33].
With the rationale to show the potential role of
Doppler parameters as indicators of fetal hypoxia,
rather than low estimated fetal weight, Bakalis et al.
[11] and Akolekar et al. [12] evaluated the perform-
ance of a low CPR at 33 and 36 weeks’ gestation, find-
ing a lower CPR in fetuses with adverse outcome.
The optimal gap between Doppler assessment and
Figure 2. Performance of the predictive models for emergency labor has not been determined [34]. Prior et al. [7]
delivery due to non-reassuring fetal status: model 2 based on demonstrated that an abnormal CPR measured
nulliparity and meconium-stained amniotic fluid, and model 3 within 72 h of labor among AGA neonates, was signifi-
comprising the previous model plus Doppler parameters. cantly associated with emergency delivery due to
Figure 3. Decision tree analysis on emergency delivery due to non-reassuring fetal status: black bars represent the percentage of
cases requiring emergency delivery among each subgroup.
8 F. CROVETTO ET AL.
non-reassuring fetal status. Recently Dell’Asta et al. Any part of the work was appropriately investigated and
[10], reported that a reduced CPR in low-risk term approved by the authors.
woman undergoing labor was associated with a two-
fold higher risk of fetal compromise and three-fold Disclosure statement
higher incidence of composite adverse perinatal out-
No potential conflict of interest was reported by the authors.
come. Our data suggest that Doppler evaluation may
be useful to predict those fetuses at risk of intrapar-
tum fetal compromise in a context of clinical decision References
analysis, and regardless of the type of labor, which [1] Kushtagi P, Deepika KS. Amniotic fluid index at admis-
emphasizes the benefit of induction in comparison sion in labour as predictor of intrapartum fetal status.
with expectant management at the end of preg- J Obstet Gynaecol. 2011;31(5):393–395.
nancy [33,34]. [2] Alfirevic Z, Gyte GM, Cuthbert A, et al. Continuous
The strengths of our study are first, the prospective cardiotocography (CTG) as a form of electronic fetal
monitoring (EFM) for fetal assessment during labour.
design, whereas a substantial proportion of the data Cochrane Database Syst Rev. 2017;2:2006–2008.
on this topic comes from retrospective studies. [3] Martis R, Emilia O, Nurdiati DS, et al. Intermittent aus-
Secondly, we recruited a cohort of low-risk pregnan- cultation (IA) of fetal heart rate in labour for fetal
cies that underwent Doppler assessment within 7 days well-being. Cochrane Database Syst Rev. 2017;2:
of delivery, potentially minimizing Doppler changes CD008680.
[4] Graham EM, Ruis KA, Hartman AL, et al. A systematic
between the first assessment and the intrapartum
review of the role of intrapartum hypoxia-ischemia in
management. And finally, this is one of the few stud- the causation of neonatal encephalopathy. Am J
ies in which attending obstetricians and midwives Obstet Gynecol. 2008;199(6):587–595.
were blinded to Doppler results, reducing potential [5] Gunn AJ, Bennet L. Timing of injury in the fetus and
bias the management of these patients. The main limi- neonate. Curr Opin Obstet Gynecol. 2008;20(2):
175–181.
tation of this study is the relatively small sample size.
[6] Low JA, Pickersgill H, Killen H, et al. The prediction
Therefore, larger studies, ideally clinical trials, would and prevention of intrapartum fetal asphyxia in term
be needed to add quality information on the useful- pregnancies. Am J Obstet Gynecol. 2001;184(4):
ness of Doppler assessment in low-risk pregnancies. 724–730.
In conclusion, in low-risk pregnancies at term, the [7] Prior T, Mullins E, Bennett P, et al. Prediction of intra-
partum fetal compromise using the cerebroumbilical
addition of Doppler parameters such as MCA Z-score
ratio: a prospective observational study. Am J Obstet
to a base model comprising maternal history and Gynecol. 2013;208(2):124.e1–124.e6.
intrapartum clinical characteristics, significantly [8] Bligh LN, Al Solai A, Greer RM, et al. Diagnostic per-
improves the intrapartum prediction of emergency formance of cerebroplacental ratio thresholds at term
delivery due to non-reassuring fetal status. This study for prediction of low birthweight and adverse intra-
adds evidence on the possible usefulness of Doppler partum and neonatal outcomes in a term, low-risk
population. Fetal Diagn Ther. 2018;43(3):191–198.
evaluation in low-risk pregnancies. Meanwhile, there is [9] Morales-Rosello J, Khalil A, Morlando M, et al. Poor
still not enough quality information to support the neonatal acid–base status in term fetuses with low
use of Doppler examination for clinical decision mak- cerebroplacental ratio. Ultrasound Obstet Gynecol.
ing in this type of population outside the context of a 2015;45(2):156–161.
clinical trial. [10] Dell’Asta A, Ghi T, Rizzo G, et al. Early labor cerebro-
placental ratio assessment in uncomplicated term
pregnancies and prediction of adverse perinatal out-
Author contributions comes: a prospective, multicentre study. Ultrasound
Obstet Gynecol. 2018;10:19113.
All the authors had a substantial contribution with specific [11] Bakalis S, Akolekar R, Gallo DM, et al. Umbilical and
responsibilities. fetal middle cerebral artery Doppler at 30–34 weeks’
Francesca Crovetto was responsible for conception and gestation in the prediction of adverse perinatal
design of the work, analysis and interpretation of data, critic- outcome. Ultrasound Obstet Gynecol. 2015;45(4):
ally revision and final approval; Nicola Cesano, Federica 409–420.
Rossi, Stefano Acerboni, Stefano De Marinis had a substantial [12] Akolekar R, Syngelaki A, Gallo DM, et al. Umbilical
role in the acquisition of data; Annachiara Basso, Raigam and fetal middle cerebral artery Doppler at 35–37
Jafet Martinez Portilla, Rosalia Pascal Capdevila had a sub- weeks’ gestation in the prediction of adverse peri-
stantial contribution for writing and drafting of the work natal outcome. Ultrasound Obstet Gynecol. 2015;
and analysis of data; Barbara Acaia, Luigi Fedele, Enrico 46(1):82–92.
Ferrazzi, Nicola Persico were responsible for the final revision [13] Bertino E, Coscia A, Boni L, et al. Weight growth vel-
and approval of the version to be published. ocity of very low birth weight infants: role of gender,
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE 9
gestational age and major morbidities. Early Hum [26] Hothorn T, Hornik K, Zeileis A. Unbiased recursive par-
Dev. 2009;85(6):339–347. titioning: a conditional inference framework.
[14] American College of Obstetricians and Gynecologists. J Comput Graph Stat. 2006;15(3):651–674.
ACOG Practice Bulletin No. 106: Intrapartum fetal [27] Cruz-Martınez R, Figueras F, Hernandez-Andrade E,
heart rate monitoring: nomenclature, interpretation, et al. Fetal brain Doppler to predict cesarean delivery
and general management principles. Obstet Gynecol. for nonreassuring fetal status in term small-for-gesta-
2009;114(1):192–202. tional-age fetuses. Obstet Gynecol. 2011;117(3):
[15] Robinson HP, Fleming JEE. A critical evaluation of 618–626.
sonar “crown-rump length” measurements. BJOG. [28] Figueras F, Savchev S, Triunfo S, et al. An integrated
1975;82(9):702–710. model with classification criteria to predict small-for-
[16] Hadlock FP, Harrist RB, Sharman RS, et al. Estimation gestational-age fetuses at risk of adverse perinatal
of fetal weight with the use of head, body, and femur outcome. Ultrasound Obstet Gynecol. 2015;45(3):
measurements – A prospective study. Am J Obstet 279–285.
Gynecol. 1985;151(3):333–337. [29] Bligh LN, Alsolai AA, Greer RM, et al. Cerebroplacental
[17] Arduini D, Rizzo G. Normal values of Pulsatility Index ratio thresholds measured within 2 weeks before birth
from fetal vessels: a cross-sectional study on 1556 and risk of Cesarean section for intrapartum fetal
compromise and adverse neonatal outcome.
healthy fetuses. J Perinat Med. 1990;18(3):165.
Ultrasound Obstet Gynecol. 2018;52(3):340–346.
[18] Baschat AA, Gembruch U. The cerebroplacental
[30] Morales-Rosello ~ J, Khalil A, Morlando M, et al.
Doppler ratio revisited. Ultrasound Obstet Gynecol.
Changes in fetal Doppler indices as a marker of fail-
2003;21(2):124–127.
ure to reach growth potential at term. Ultrasound
[19] Cunningham FG, Leveno KJ, Bloom SL, et al. Williams
Obstet Gynecol. 2014;43(3):303–310.
obstetrics. 24th ed. New York: McGraw-Hill Education;
[31] Parra-Saavedra M, Simeone S, Triunfo S, et al.
2014. Correlation between histological signs of placental
[20] Gregg AR, Weiner CP. Normal umbilical arterial and underperfusion and perinatal morbidity in late-onset
venous acid-base and blood gas values. Clin Obstet small-for-gestational-age fetuses. Ultrasound Obstet
Gynecol. 1993;36(1):24–32. Gynecol. 2015;45(2):149–155.
[21] Lindsey C, Sheather S. Best subsets variable selection [32] Khalil AA, Morales-Rosello J, Elsaddig M, et al. The
in nonnormal regression models. The Stata J. 2015; association between fetal Doppler and admission to
15(4):1046–1059. neonatal unit at term. Am J Obstet Gynecol. 2015;
[22] Manjo n M, Martınez O. The chi-squared goodness-of- 213(1):57.e1–57.e7.
fit test for count-data models. The Stata J. 2014;14(4): [33] Nicholson JM, Kellar LC, Henning GF, et al. The associ-
798–816. ation between the regular use of preventive labour
[23] De Long ER, De Long DM, Clarke-Pearson DL. induction and improved term birth outcomes: find-
Comparing the areas under two or more correlated ings of a systematic review and meta-analysis. BJOG:
receiver operating characteristic curves: a nonpara- Int J Obstet Gy. 2015;122(6):773–784.
metric approach. Biometrics. 1988;44(3):45. [34] Morales-Rosello J, Khalil A, Fornes-Ferrer V, et al.
[24] Zhang H. Classification trees for multiple binary Accuracy of the fetal cerebroplacental ratio for the
responses. J Am Stat Assoc. 1998;93(441):180–193. detection of intrapartum compromise in nonsmall
[25] Shih Y-S. Families of splitting criteria for classification fetuses. J Matern Fetal Neonatal Med. 2019;32(17):
trees. Stat Comput. 1999;9(4):309–315. 2842–2852.