Prediction of perinatal and neonatal outcomes after preterm hypertensive
complications of pregnancy: a systematic review of prediction models
J. Kooiman, MD-PhD
Laymen’s summary
A small proportion of women develop complications during the preterm period of pregnancy (the period before 37 weeks of gestation). An increase in blood
pressure (hypertension) in combination with organ failure is one of these complications, also known as HELLP syndrome or preeclampsia. In some pregnancies, this increase in blood pressure goes hand in hand with a growth restriction of the fetus due to placental insufficiency. A high blood pressure (with or without accompanying organ failure) can be dangerous for mother whereas placental insufficiency can result in chronic brain damage of the fetus or even stillbirth. Consequently, health of both mother and child is sometimes best served by medically-induced preterm labor.
To inform couples about the risks for the newborn of medically-induced preterm labor (such as serious breathing problems or dying due to prematurity) versus the chance of stillbirth when continuing pregnancy, risk estimates of these undesirable pregnancy outcomes are required. Such estimates can be made using prediction models, mathematical models containing information about the mother, fetus and current course of pregnancy. Using these risk estimates, couples together with their healthcare professionals can come to a decision on medical management. The aim of this research project was to summarize all available prediction models on adverse outcomes of the fetus or newborn for women dealing with complications of an high blood pressure in the preterm period of pregnancy.
To this end, a systematic search of medical literature was performed. After model identification, the quality of the models and their robustness when using the models in external patient populations (for instance, patients with the same medical problem in another hospital or country) was studied using standardized tools and checklists by two independent researchers. Information on the
performance of the prediction models was studied, which can be expressed using statistical formulas.
This literature search identified 8 studies describing 16 prediction models. None of the models were tested outside of the original study population. This is
cumbersome as models tend to predict best in the population in which they were developed. Most models included ultrasonography measurements to calculate the risks of stillbirth or complications after birth, such as serious illness or death of the newborn. The methods used to create the prediction models were not in line with current standards for this type of research. Furthermore, most studies lacked important information on how the prediction models were created and how well they can predict the outcome of interest.
This research project has four main conclusions. First, 8 studies reporting on 16 prediction models were identified and none of these models were tested outside the original study population. Second, the methods used to develop the models are not in line with current standards. Third, important information about the predictive performance of the models was lacking in most publications.
Consequently and fourth, none of the prediction models described in this research project are suitable for clinical use.
Abstract
Introduction Prediction models could aid counseling on obstetric and neonatal
management in preterm hypertensive disorders of pregnancy. The aim of this systematic review was to identify all risk scores on perinatal and neonatal
outcomes in this patient population. Second, we aimed to appraise the quality of the models, their performance in external patient populations and to pool their predictive performance measures in a meta-analysis.
Methods A systematic literature search was performed in Pubmed, Web of
Science and Embase. Studies describing prediction models on perinatal or
neonatal outcomes after preterm hypertensive complications of pregnancy were included. Data extraction was performed using the CHAMS-checklist. Risk of bias was assessed using PROBAST. Literature selection and data extraction were performed by two researchers independently.
Results Our literature search yielded 15,956 unique publications, of which 8 were
included after full text readings. These 8 studies derived 16 prediction models, 2 in the setting of preeclampsia and the remaining in fetal growth restriction. None of the models were externally validated and internal validation was performed by one study. Final models contained mostly ultrasound (Doppler) markers as
predictors. Discriminative properties were reported for 4/16 models (c-statistics between 0.7-0.9). Only one study presented a calibration plot. The risk of bias was assessed as high for all included models, mainly due to inappropriate statistical methods.
Conclusion This systematic review identified 16 prediction models for fetal and
neonatal outcomes in pregnancies complicated by preterm hypertensive disorders. At this time, none of these models are ready for clinical use, mainly due to a lack of external validation and issues in analytical conduct.
Introduction
Hypertensive disorders of pregnancy comprise a spectrum of obstetric
complications including gestational hypertension, preeclampsia, HELLP syndrome and fetal growth restriction (FGR)[1]. As expected, neonatal outcomes are
negatively affected by hypertensive complications, especially when they occur preterm. For instance, neonatal mortality was reported 4-10% after preterm delivery in mothers with preeclampsia and as high as 52% in a large cohort of pregnancies complicated by early FGR[2-4]. This cumbersome neonatal prognosis is partly driven by the need for iatrogenic preterm birth to prevent further
maternal or neonatal morbidity and mortality, and partly by the disease itself, especially in case of FGR[5].
Prediction of perinatal and neonatal outcomes of pregnancies complicated by preterm hypertensive disorders is attractive from a clinical point of view[6]. First of all, counseling about willingness for fetal surveillance and active neonatal support requires patient-tailored estimates of fetal and neonatal prognosis.
Second, in case of active fetal surveillance and neonatal management, risk
assessment of neonatal morbidity and mortality could indicate whether there is a need for intrauterine (maternal) transfer to an hospital with a neonatal intensive care unit. Third, prediction of stillbirth could guide decisions on timing of delivery.
Hence, prediction models containing antenatal available predictors for fetal and neonatal outcomes could considerably facilitate clinical decision making in pregnancies complicated by preterm hypertensive disorders.
From development of a prediction model to bedside use, certain steps need to be undertaken[7]. After derivation and internal validation studies, external validation of a prediction model should determine whether the performance remains
adequate in external patient populations. A third and final step including an impact study analyzing the effect of the use of the prediction model on fetal and
neonatal outcomes should be undertaken before the prediction model is implemented in daily practice.
The aim of this systematic review was to identify all risk scores on perinatal and neonatal outcomes in the setting of preterm hypertensive complications of
pregnancy. These risk scores should solely contain predictors that are known prior to delivery, to be able to facilitate clinical decision making during pregnancy.
Second, we aimed to appraise the quality of the models, their performance in external patient populations and lastly to pool their predictive performance measures in a meta-analysis.
Methods
Data sources and searches
A literature search was performed on April 10th 2020 in Embase, Web of Science and Pubmed. The search strings contained terms on hypertensive disorders of pregnancy, prediction models and adverse fetal or neonatal outcomes (for full search string, see Appendix). No language or publication date restrictions were applied. The protocol of this systematic review was published at Prospero (https://www.crd.york.ac.uk/prospero/), international prospective register of systematic reviews, registration number CRD42020178976.
Study selection
Studies developing or validating a prediction model on perinatal and neonatal outcomes in the setting of preterm hypertensive disorders (gestational
hypertension, preeclampsia, FGR or HELLP syndrome) were eligible for inclusion.
Preterm was defined as gestational age at birth < 37 weeks. Fetal and neonatal outcomes were defined as any of the following: perinatal and neonatal mortality, cystic leukomalacia, grade three or four intracranial hemorrhage, retinopathy of prematurity requiring laser coagulation, necrotizing enterocolitis or infant
respiratory distress syndrome. End of follow-up on neonatal outcomes was defined as until 28 days of age, hospital discharge to home or death, whichever occurred first.
To be eligible for inclusion, studies should report original data and be published in the form of a research letter, meeting abstract or full-text paper. Studies on term pregnancies, fetuses with congenital malformations or restricted to multiple gestations were excluded. We also excluded studies on prediction models containing neonatal outcome variables (such as Apgar score) as predictors in their model, as these models do not fit with the timing of the research question.
Study selection for inclusion was performed by two independent researchers after duplicates were removed. Discrepancies were solved by consulting a third
independent researcher. Literature selection was conducted in two stages. The first stage of title and abstract screening was followed by a second phase of full text readings of remaining articles. Corresponding authors were contacted in case of missing information relevant for inclusion, data extraction or quality
assessment.
Data extraction and Quality assessment
Data extraction of included prediction models was performed using CHARMS (CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies) and PROBAST (Prediction model Risk of Bias Assessment tool) by means of a standardized scoring form by two independent researchers[7, 8]. Discrepancies were resolved by consulting a third independent researcher. CHARMS was used to extract information on study design, population characteristics, predictors and outcomes assessed by the prediction models and the statistical analysis including handling of missing data, model building
strategies and model performance measurements[8]. Relevant performance measures included amongst others: c-statistic (or area under the receiver-
operating characteristic curve), calibration measures (curve, slope, calibration-in- the-large), sensitivity, specificity, positive predictive value and negative
predictive value, with corresponding 95% confidence intervals.
PROBAST was used to assess the risk of bias of included studies[7, 9]. This tool is especially developed to assess the risk of bias and applicability issues in
prediction model studies using four domains: participants, predictors, outcome and analysis. Thereafter, an overall conclusion on the risk of bias and applicability can be formed.
Data synthesis
Meta-analysis will be performed on model performance estimates in case at least two independent external validation studies provide information on the same measures of the same model. Pooled summary estimates will be calculated using a random effects model with restricted maximum likelihood (REML) estimations for calculation of 95% prediction intervals. For the c-statistic, a logit
transformation will be applied to result in a normal distribution. For the
observed:expected ratio, log (ln) transformation will be applied for the same reason.
Subgroup analysis will be performed if at least two studies are available per subgroup. Subgroups of interest are: gestational hypertension (i.e. hypertension starting after 20 weeks of gestation), preeclampsia, HELLP syndrome and FGR.
The amount of between-studies heterogeneity will be calculated using the I² (%).
Statistical analysis will be conducted using R-studio software.
Results
The literature search yielded 15,956 unique results. After title and abstract screening, 156 full-text articles were gaged for inclusion. After full-text readings, 8 studies were included (Figure 1), evaluating 16 individual prediction models[6, 10-16]. All studies were derivation studies, meaning that none of the included studies validated a pre-existing prediction model. Consequently, meta-analysis on prediction model performance measures was not possible. Characteristics of included studies are reported in Table 1. One study was performed in Brazil and 1 in South Africa[10, 12]. The remaining studies were from European or North American origin. A summary of candidate predictors and predictors included in the final models is presented in Figure 2.
Prediction models within the setting of preeclampsia (Table 2)
Two studies were performed in women with preeclampsia, describing two models.
The first study was performed by Geerst, et al, and included 113 prospectively enrolled patients diagnosed with preeclampsia between 24 and 34 weeks of gestation[12]. The authors developed a model to predict a composite endpoint of neonatal morbidity starting with 20 candidate predictors, which were reduced to 7 predictors in the final model. A sensitivity of 79% with a specificity of 88.2%
was reported, without information on the c-statistic or calibration properties of the model. No internal validation steps were reported.
The second study, published by Gomez-Arriaga, et al, was a prospective monocenter study[13]. They aimed to derive a model to predict a composite outcome of perinatal morbidity and mortality in pregnancies complicated by preeclampsia before 34 weeks of gestational age. The final model contained three predictors, i.e. gestational age at time of diagnosis, mean uterine artery pulsatility index and the soluble fms-like tyrosine kinase-1 (sFlt-1)/ placental
growth factor (PlGF) ratio. The model yielded a c-statistic of 0.89 (0.79-0.99). No calibration or internal validation was performed.
For both models, the final models including intercepts and predictor weights were presented in the publication.
Prediction models within the setting of fetal growth restriction (Table 3)
Six studies derived a total of 14 prediction models for fetal or neonatal outcomes in pregnancies complicated by FGR.
Baiao, et al, developed a model for neonatal morbidity and a second model for overall mortality (i.e. stillbirth combined with neonatal mortality before
discharge)[10]. One hundred and fifteen neonates developed the combined endpoint of neonatal morbidity for which the authors evaluated 11 candidate predictors. The final model included 2 predictors, absent end-diastolic flow of the umbilical artery and reversed flow of the same vessel. No performance measures of the final model were reported. For the outcome of overall mortality, which occurred in 45/265 fetuses or neonates, the same 11 candidate predictors were analyzed. The final model included the z-score of the estimated fetal weight and an elevated index of the ductus venosus. Again, no performance measures of the model were reported.
Baschat, et al, aimed to assess which combination of fetal Doppler profiles best predicts stillbirth, perinatal and neonatal mortality, using 224 pregnancies
complicated by FGR[11]. All women underwent ultrasonography 48 hours prior to delivery, and these values were used for modelling. The same set of candidate and final predictors were used for all three outcomes. For each outcome, two final models were derived with the first model including abnormal umbilical vein (pulsatile instead of constant flow) and ductus venosus (absent flow during the A- wave) flow patterns as predictors in the final model (models A, C, E respectively,
in Table 3). The second model for each outcome included the combination of absent or reversed end-diastolic umbilical artery flow and abnormal umbilical vein flow as final predictors (models B, D and F, Table 3). Test efficacy of the models ranged between 91-94%. Discrimination and calibration estimates were not reported.
Hernandez-Andrade, et al, developed a prediction model for the combined endpoint of perinatal and neonatal mortality[14]. They mainly included fetal ultrasound markers and started with 7 candidate predictors which were reduced to 3 dichotomized predictors in the final model (gestational age < 28 weeks, ductus venosus absent or reversed A-wave, myocardial performance index > 95th percentile). Their final model was simplified to a fetal cardiovascular risk score showing the highest mortality rates in those with the highest score. No
discrimination or calibration measures were reported.
The paper by Odibo, et al, described the development of a prediction model for a combined endpoint of stillbirth, neonatal morbidity and mortality in 66
pregnancies resulting in 17 outcome events[15]. They combined features of the biophysical fetal profile with a multi-vessel Doppler assessment, starting with a model with 6 candidate predictors, of which 4 ended up in the final model
(pulsatility index middle cerebral artery < 5th percentile, peak systolic velocity of middle cerebral artery, ductus venosus pulsatility index, cerebroplacental ratio).
The model yielded a c-statistic of 0.73 (95% CI 0.59-0.87). No calibration measures or internal validation steps were reported.
Sharp, et al, developed prediction models within the STRIDER UK population, including clinical features, ultrasonography measurements and angiogenic biomarkers[6]. They restricted their analysis to 78% of the cohort (105 patients) with complete information on the angiogenic markers sFlt-1 and PlGF. Six models were developed of which 3 were included in this systematic review based on the
predicted outcome. The included models predict overall survival, neonatal
morbidity and stillbirth (for definitions see Table 3), and start with the same list of 17 candidate predictors. All final models were drastically reduced using the Akaike information criterion in the forward stepwise selection process. The final models included estimated fetal weight for all 3 outcomes and the sFlt-1:PlGF- ratio in models predicting overall survival and stillbirth. All other candidate
predictors were removed. The models yielded varying discriminative values, with c-statistics ranging from 0.70 to 0.90. Internal validation was performed by bootstrapping, resulting in similar c-statistics for all three models compared to the original values.
The last study and model included in this systematic review was published by Vergani, et al [16]. They included 39 patients of which 16 developed the outcome of interest, i.e. perinatal mortality and severe neonatal morbidity, using 13
candidate predictors. Their final model included the cerebroplacental ratio and estimated fetal weight at last ultrasound before birth as final predictors. Besides a Nagelkerke R2 of 0.72, no other model performance measures were described.
Four of the above described 14 prediction models in FGR reported intercepts and predictor weights of the final models[6, 16].
Risk of bias evaluation
Using PROBAST, applicability and the risk of bias were assessed on a model level.
For the domains of ‘Participants’, ‘Predictor’ and ‘Outcome assessment’ risk of bias or concerns regarding applicability were generally low (Figure 3). However, all models were regarded high risk of bias on the domain of ‘Analysis’, mainly due to too low event per predictor values, performance of a complete case analysis or unclear handling of missing data. Also, most models did not report on criteria for selection of candidate predictors into the final model during the model building
process and lacked interval validation steps in order to control for overfitting and optimism. As a result, all studies were regarded high risk of bias in the overall assessment. Thirteen out of 16 models scored low concern of applicability issues, the remaining models were scored unclear.
Discussion
This is the first systematic review on prediction models of fetal and neonatal outcomes in pregnancies complicated by preterm hypertensive disorders. Based on the results, the following conclusions can be made: Firstly, 16 models have been derived for this patient population of which none were externally validated.
Second, judgement of their predictive performance is cumbersome due to incomplete model evaluation, in which discrimination measures are
underreported and calibration has been ignored by the vast majority of studies.
Third, methods used to derive the models, notwithstanding the models derived by Sharp, et al as a notable exception, are questionable in light of current
standards. Consequently, the risk of overfitting of the models to the source data and consequently overoptimistic performance estimates is likely. This is also reflected in the high risk of bias for all included models, as determined by use of PROBAST. As such, at this time, none of the prediction models described in this review are suitable for clinical use.
On the whole, the main conclusions of this systematic review are in line with conclusions and recommendations made by other systematic reviews on
prediction modelling studies across different medical specialties[17-20]. It is this repetition of the same problems in prediction modelling studies that strongly emphasizes the need for better understanding of conductance and reporting of this type of studies by researchers in general. Tools and initiatives such as
TRIPOD (transparent reporting of a multivariable prediction models for individual prognosis or diagnosis, https://www.tripod-statement.org), PROBAST and
CHARMS, which are publicly available, can provide research groups with key information on these important matters[7, 8, 21].
Timing of the prediction models included in this systematic review is a concern that needs to be raised. The intended clinical use of the models likely involves counseling about obstetric and neonatal management at time of diagnosis or worsening of preterm preeclampsia or FGR. Unfortunately, almost all models were build using ultrasound measurements taken at the time point closest to delivery, not at time of diagnosis or counseling. This might overestimate the true performance estimates of the models when using them at earlier stages in
pregnancy[9]. Also, ideally, prediction models in the setting of FGR or
preeclampsia would be dynamic, instead of static, providing updated estimates on fetal and neonatal outcomes when predictor values (such as Doppler
assessment) change during pregnancy. On top of that, when thinking about clinical applicability, another problem seems to arise. Models predicting stillbirth generally include the same predictors as models predicting neonatal morbidity or mortality. If the risk of stillbirth is low and the risk of neonatal complications is estimated as high, continuation of pregnancy with close fetal surveillance is likely regarded the preferred option in terms of neonatal outcomes. However, given the fact that stillbirth and neonatal morbidity and mortality share common predictors, it is more likely that risks of all of these outcomes would be calculated as high by prediction models. Thereby, these models do not seem appropriate for clinical decision making on timing of delivery, balancing the risk of stillbirth with neonatal complications due to iatrogenic prematurity.
Another aspect worthy of discussion is the variance in definitions. Different qualification terms for FGR have been used by included studies. Not only do varying definitions result in clinical generalizability issues, they also disturb head- to-head comparisons of prediction models created for the same patient
populations and outcomes. Using definitions constructed by Delphi procedures or based on guidelines such as the International Society for the Study of
Hypertension in Pregnancy (ISSHP) would strongly improve future research[22].
As regards to predictor definitions, although most models share common predictors in their final models, continuous variables like gestational age or estimated fetal weight were dichotomized at varying cut-offs, based on the best fit to the studied population. This is unfortunate as dichotomization in general leads to a loss in power[23]. Moreover, the cut-offs used were not externally validated. As a result, prediction models containing dichotomized predictors in general have reduced predictive ability[24]. Finally, different outcome definitions were used, mainly for the combined outcome of neonatal morbidity. Future
research should use endpoints based on core outcome sets such as those derived by COMET to increase generalizability and head-to-head comparisons[25, 26].
Incomplete reporting on the derivation process, and in particular on the statistical analysis phase, is another important issue for most studies included in this
systematic review. For instance, handling of missing outcome or predictor variables and selection criteria for inclusion in the final model during
multivariable analysis were not mentioned by the majority of studies. In addition, important information on predictive properties of the models such as calibration and discrimination measures were underreported[27]. Moreover, the final
regression model equation was only described for 10 out of 16 models. This equation is vital for external model validation and model updating. Lastly, reporting on definitions of individual components of the composite endpoint of neonatal morbidity such as necrotizing enterocolitis or sepsis, and how these outcomes were assessed, was limited for some models.
Future research should focus on externally validating existing instead of deriving new prediction models. The main focus should be on improving their predictive performance. Deriving new models will not results in new scientific knowledge as the existing models already have a broad overlap in the lists of predictors
included in their final models. Good candidate models for external validation studies include the models published by Gomez, et al, and Sharp, et al, although their models include the angiogenic markers sFlt-1 and PlGF, which are not readily available in most healthcare settings and countries[6, 13].
Strengths of this systematic review are underlined by the systematic and comprehensive search of the literature and the thorough data extraction using PROBAST and CHARMS by two independent researchers. The results provide and up-to-date summary of the available prediction models and their predictive value in patients with preterm hypertensive complications during pregnancy. Moreover, important analytical issues of included studies have been identified and
discussed by this study, and this knowledge could prevent similar methodological problems in future research. This systematic review was limited by the fact that external validation studies were lacking, and consequently a meta-analysis of performance measures of the models in external populations was not possible.
Second, we restricted our literature search to prediction models derived or validated in patients in the preterm phase of pregnancy. This is logical in the sense that prediction of fetal and neonatal outcomes is especially interesting in the preterm (and thus high risk) period when facing hypertensive complications.
Also, the current literature search already yielded 15,956 records. However, we cannot exclude that there are prediction models derived or validated in the term setting of hypertensive complications that could also have predictive value in the preterm setting. These models were not identified by our literature search.
To conclude, this systematic review identified 16 prediction models for fetal and neonatal outcomes in pregnancies complicated by preterm hypertensive
disorders. At this time, none of these models are ready for clinical use, mainly due to a lack of external validation and issues in analytical design and conduct leading to a high risk of bias and therefore overestimation of the true
performance of the models in clinical practice. Future research should focus on external validation of these prediction models using up-to-date and well-
recognized validation strategies using standardized predictor and outcome definitions.
Figure 1: PRISMA flow diagram.
Figure legend: PE = preeclampsia, FGR = fetal growth restriction
Figure 2: Summary of predictors per category
Figure legend: Main categories of predictors considered (blue bars) or included (green bars) in prediction models.
apH= arterial umbilical cord pH, BW= birth weight, CPR= cerebroplacental ratio, DV=
ductus venosus, EDF= end-diastolic flow, EFW= estimated fetal weight, GA= gestational age, MCA= middle cerebral artery, PE= preeclampsia, PI= pulsatility index, PIH=
pregnancy induced hypertension, PlGF= placental growth factor, sEng= soluble endoglin, sFlt-1= soluble fms-like tyrosine kinase-1, UA= uterine artery
Figure 3: Risk of bias assessment
Legend: Risk of bias and applicability assessment according to the PROBAST tool.
Tables
Table 1: Characteristics of included studies Author
(year) Countr
y Hypertensive complication (definition)
Study
design Inclusio
n period Type of prediction model study
N mod els
N patie nts
Mean BW
in grams Mean GA at birth in weeks
Baião (2019) [10]
Brazil FGR
(UA PI >> 2 SD)
Retrospectiv e,
multicenter
2002- 2016
Derivation 2 265 913
(range 315- 1995)
intact survivors median 30.1, (range 25-33) non-intact survivors median 27.5 (range 24-33) Baschat
(2003) [11]
USA and German y
FGR
(BW < 10th
percentile AND UA PI > 2 SD
Cohort, multicenter
1994- 2001
Derivation 6 224 1090
(range 360- 2010)
31+4
Geerts (2007) [12]
South- Africa
PE
(severe PE according to the 1988 ISHHP criteria)
Prospective, monocenter
Not reported
Derivation 1 113 1412.3
(SD 412.6)
31+5 (SD 2 weeks)
Gómez- Arriaga (2014) [13]
Spain PE
(ISSHP criteria 2001)
Prospective,
monocenter 2007-
2012 Derivation 1 51 1302 (600) 30.3 (SD 3.9)
Hernand ez- Andrade (2009) [14]
Spain and UK
FGR
(EFW < p10 AND UA PI > p95)
Prospective, multicenter
2006- 2008
Derivation 1 97 956
(IQR 694- 1170)
30 (range 28-32)
Odibo (2014) [15]
USA FGR
(EFW < p10 AND UA PI > p95)
Prospective, monocenter
5-year period not specified
Derivation 1 66 1778.3
(SD 678.7)
34.1 (SD 3.5)
Sharp (2019) [6]
UK FGR
(AC or EFW < p 10 AND A/REDF of UA)
RCT,
multicenter 2014-
2016 Derivation 3 105 590
(IQR 480- 769)
28.3 (IQR 26.9-29.7)
Vergani
(2005) Italy FGR
(AC < p10 with Retrospectiv
e, 1995-
2004 Derivation 1 39 838
(SD 29.3 (SD unknown)
[16] absent end-diastolic flow of UA)
monocenter) unknown)
Abbreviations: AC= abdominal circumference, A/REDF= absent or reversed end-diastolic flow, BW= birth weight, EFW= estimated fetal weight, FGR= fetal growth restriction, GA= gestational age, ISHHP= International Society for the Study of Hypertension in Pregnancy, IQR
= interquartile range, PE= preeclampsia, PI= pulsatility index, RCT = randomized controlled trial, SD= standard deviation, UA= umbilical artery
Table 2. Prediction models in pregnancies complicated by preeclampsia Auth
or (yea r)
Predicted outcome (definition)
Candidate
predictors Final predict ors
Time of mod el use
Modelli ng metho d
EPV Handl ing of missi ng data
Selection criteria in
multivari able analysis
Discrimina tion (C- statistic or AUC)
Calibrati
on Other
performa nce measure s
Geert s (2007 ) [12]
Neonatal morbidity (respiratory failure requiring ventilation, neonatal sepsis, pulmonary or gastrointestina l
haemorrhage, pneumothorax , NEC,
anaemia requiring transfusion, metabolic acidosis, periventricular leukomalacia, grade 3 or more IVH, convulsions and
hydrocephaly)
GA at
recruitment; GA at birth; Days gained; initial EWF; BW; Initial PCR; Final PCR;
recruitment <
31w; initial EFW
< 1284; BW
<10th perc;
Initial UA PI;
final UA PI;
Initial AREDF;
final AREDF;
Initial PCR >1;
Final PCR >1;
Initial DV >
p95; final DV >
p95; initial asymmetry;
final asymmetry
initial UA PI; Initial AREDF;
final growth asymme try;
initial GA;
initial EFW >
1284 g;
Initial PCR;
final DV PI
Final ultra- sound befor e delive ry
Logistic regressi on
3 Not
reporte d
Not
reported Not reported Not
reported Sens 79%, spec 88.2%, PPV 89.1%. NPV 77.6%
Góm ez- Arria ga
Combined endpoint of morbidity and mortality (5
maternal age, BMI, race, parity, smoking, aspirin use,
GA, mean UtA-PI, sFlt-
Uncle
ar Logistic regressi on
0 Comple
te case analysi s
Not
reported 0.89 (0.79-
0.99) Not
reported Sens 95%, spec 41%, PPV 37%, NPV 96%
(2014 ) [13]
min AS <7, apH < 7.0, fetal or neonatal death, IVH grade ≥3, PVL, hypoxic ischemic encephalopath y, NEC, ARF, cardiac failure requiring inotropic agents)
chronic hypertension, renal disease;
pregestational diabetes;
thrombophilia;
GA at delivery, reason for delivery, time- to-delivery interval, BW, BW percentile, SGA, cesarean section, UtA-PI;
UtA-PI MoM, Mean UtA- PI >
Q3; sFlt-1/PlGF ratio; sFlt- 1/PlGF ratio>Q3
1/PlGF ratio
Abbreviations: A/REDF= absent or reversed end-diastolic flow, apH = umbilical cord artery pH, ARF = acute renal failure, AS = apgar score, BW= birth weight, DV = ductus venosus, EFW= estimated fetal weight, EPV= events per candidate predictor variable, GA= gestational age, IVH = intraventricular haemorrhage, MoM= multiple of mean, NEC = necrotising enterocolitis, NPV = negative predictive value, PCR = placental-cerebral ratio, PI= pulsatility index, PlGF = placental growth factor, PPV = positive predictive value, sFlt-1= soluble fms-like tyrosine kinase-1, UA= umbilical artery, UtA = uterine artery, sens = sensitivity, SGA = small for gestational age, spec = specificity.
Table 3. Prediction models in pregnancies complicated by fetal growth restriction Author
(year)
Predicted outcome (definition)
Candidate predictors
Timing of model use
Modelli ng metho d
EP V
Handli ng missin g data
Selection criteria in
multivari able analysis
Discrimin ation (C-
statistic or AUC with 95% CI)
Calibra tion
Other
performance measures
Baião (2019) [10]
a Neonatal morbidity (BPD, IVH gr 3-4, NEC and ROP)
b Mortality
(stillbirth or death before discharge)
a, b: UA EDV;
DV doppler;
DV A-wave;
EFW; EFW z- score; FGR;
oligohydramni os; GA at birth; Birth weight < 600;
BW < 800; AS
<7
a, b 24 hours prior to delivery
a, b
Logistic regressi on
a10
b 4
a, b NR a, b Unclear a, b NR a, b NR a, b NR
Baschat (2003) [11]
a, b Perinatal mortality (stillbirth and neonatal mortality combined)
c,d Neonatal mortality
e,f Stillbirth
A,b,c,d,e,f UA AREDV; DV A- wave;
pulsatile UV blood flow; GA at delivery;
birth weight percentiles;
arterial blood gas
A,b,c,d,e,f 48 hours prior to delivery
A,b,c,d,e,f
Logistic regressi on
A,b 6
c,d 3
e,f 3
A,b,c,d,e,f
NR
A,b,c,d,e,f
Unclear
A,b,c,d,e,f NR A,b,c,d,e,f
NR
A Sens 55%, spec 98%, PPV 82%, NPV 93%
B Sens 52%, Spec 98%, PPV 81%, NPV 92%
C Sens 44%, spec 98%, PPV 64%, NPV 95%
D Sens 38%, Spec 98%, PPV 60%, NPV 95%
E Sens 65%, spec 95%, PPV 50%, NPV 97%
F Sens 52%, spec 97%, PPV 65%,
NPV 95%
Hernand ez- Andrade (2009) [14]
Combined endpoint of perinatal and neonatal mortality (stillbirth or neonatal mortality within 28 days after birth)
GA at time of ultrasound, UA-PI, MCA-PI, DV-PI, IFI, MPI, DV A-wave
72 hours prior to delivery or fetal demise
Logistic regressi on
3 NR Unclear NR NR Nagelkerke R2
50%, Hosmer- Lemeshow x2 2.135, p 0.55
Odibo (2014) [15]
Combined endpoint of stillbirth, a-min AS
<3, apH < 7.2, seizures, NEC, IVH grade 3-4, PVL, neonatal death
Biophysical profile,
oligohydramni os, MCA-PSV, MCA-PI, DV-PI, CPR
Last ultrasoun d
assessme nt prior to delivery
Logistic regressi on
3 NR Unclear 0.73
(0.59-0.87) NR Sens 35.1%, spec 91.8%, LR+ 4.3 (3.3-5.3)
Sharp (2019) [6]
a Overall survival (hospital
discharge of a live child)
b Neonatal morbidity (liveborn fetus surviving until discharge experiencing at least one of the following events:
NEC< ROP, BDP, patent ductus arteriorus
requiring medical or surgical
treatment, need for vasopressor therapy, neonatal infection, IVH or a confirmed SAE as
a,b,c Gestational age,
gestational hypertension, preeclampsia, EFW, SBP, DBP, MAP, parity, UA EDF, DV a- wave, MCA-PI, UtA notching, PlGF, sFlt-1, sEng, sFlt- 1:PlGF ratio
a,b,c At time of trial enrollme nt
a,b,c Generali zed linear modellin g
a,b 3
c 2
a,b,c Comple te case analysi s
a,b,c Akaike information criterion
a 0.88 (0.83- 0.94)
b 0.70 (0.58- 0.81)
c 0.90 (0.85- 0.96)
Slope in appendix
a,b,c NR
defined by the STRIDER protocol)
c Live birth Vergani
(2005) [16]
Composite of perinatal mortality and severe
neonatal morbidity (PVL, IVH grade 3 or more, ROP; NEC, BPD)
UA-RI; GA at AEDF; First UA PI centile; first MCA-PI; first CPR; GA at last doppler;
days from AEDF to delivery; last EFW centile;
last EFW (g);
Last UA-PI; last MCA PI; last CPR; fetal gender
Last doppler measure ment before birth
Logistic regressi on
1 NR Nagelkerke
R2
NR NR Nagelkerke R2 0.72
Abbreviations: A/REDF= absent or reversed end-diastolic flow, BPD= bronchopulmonary dysplasia, BW= birth weight, CPR
=cerebroplacental ratio, DV = ductus venosus, EFW= estimated fetal weight, EPV= events per candidate predictor variable, GA=
gestational age, IFI = aortic isthmus flow index, IVH = intraventricular haemorrhage, LR = likelihood ratio, MCA = middle cerebral artery, MPI = myocardial performance index, NEC = necrotising enterocolitis, NPV = negative predictive value,NR= not reported, PI= pulsatility index, PlGF = placental growth factor, PPV = positive predictive value, PSV = peak systolic velocity, PVL=periventricular leukomalacia, RI
= resistance index, ROP= retinopathy of prematurity, sFlt-1= soluble fms-like tyrosine kinase-1, SAE = serious adverse event, sEng = soluble endoglin, sens = sensitivity, spec = specificity, UA= umbilical artery, UtA = uterine artery.
Appendix 1: Literature search EMBASE
(‘Maternal hypertension’/exp OR ’eclampsia and preeclampsia’/exp OR ‘eclamp*’:ti,ab,kw OR ‘preeclamp*’:ti,ab,kw OR ‘pre-eclamp*’:ti,ab,kw OR ‘pre eclamp*’:ti,ab,kw OR
(‘Maternal Hypertension’/exp AND (‘proteinuria’/exp OR ‘multiple organ failure’/exp)) OR (‘Maternal Hypertension’/exp AND (‘proteinuria*’:ti,ab,kw OR ‘multiple organ
failure*’:ti,ab,kw)) OR 'hellp syndrome'/exp OR ‘hellp syndrome*’:ti,ab,kw OR 'hemolysis, elevated liver enzymes, lowered platelets':ti,ab,kw OR ('hemolysis'/exp AND 'liver
enzymology'/exp AND 'thrombocytopenia'/exp) OR ('hemolys*':ti,ab,kw AND 'elevated liver enzymes':ti,ab,kw AND ('thrombocytopenia*':ti,ab,kw OR 'thrombopenia*':ti,ab,kw)) OR 'intrauterine growth retardation'/exp OR 'fetal growth retard*':ti,ab,kw OR ‘fetal growth restrict*’:ti,ab,kw OR 'foetal growth retard*':ti,ab,kw OR ‘foetal growth
restrict*’:ti,ab,kw OR ‘intra uterine growth restrict*’:ti,ab,kw OR ‘intra uterine growth retard*’:ti,ab,kw OR ‘intrauterine growth restrict*’:ti,ab,kw OR ‘intrauterine growth retard*’:ti,ab,kw OR 'placenta insufficiency'/exp OR 'placenta insufficien*':ti,ab,kw OR 'placental insufficien*':ti,ab,kw OR 'uteroplacental insufficien*':ti,ab,kw OR
'uteroplacental dysfunction*':ti,ab,kw OR 'placental dysfunction*':ti,ab,kw OR ‘Premature labor’/exp OR ‘prematurity’/exp OR ‘extremely premature infant’/exp OR ‘premature labor’:ti,ab,kw OR ‘preterm labor’:ti,ab,kw OR ‘prematurity’:ti,ab,kw OR ‘premature deliver*’:ti,ab,kw OR ‘preterm deliver*’:ti,ab,kw OR ‘obstetric labor, premature’:ti,ab,kw OR ‘obstetric labor, preterm’:ti,ab,kw OR ‘premature labor’:ti,ab,kw OR ‘preterm
labor’:ti,ab,kw OR 'extremely premature infant*':ti,ab,kw OR 'extremely preterm infant*':ti,ab,kw OR ‘infant, premature’:ti,ab,kw OR ‘infant, preterm’:ti,ab,kw OR 'premature fetus membrane rupture'/exp OR 'premature rupture of fetal
membrane*':ti,ab,kw OR 'preterm rupture of fetal membrane*':ti,ab,kw OR
‘PROM’:ti,ab,kw OR 'low birth weight'/exp OR 'low birth weight':ti,ab,kw) AND
('validat*':ti,ab,kw OR 'predict*':ti,ab,kw OR 'rule*':ti,ab,kw OR ('predict*':ti,ab,kw AND ('outcome*':ti,ab,kw OR 'risk*':ti,ab,kw OR 'model*':ti,ab,kw)) OR (('history':ti,ab,kw OR 'variable*':ti,ab,kw OR 'criteria':ti,ab,kw OR 'scor*':ti,ab,kw OR 'characteristic*':ti,ab,kw OR 'finding*':ti,ab,kw OR 'factor*':ti,ab,kw) AND ('predict*':ti,ab,kw OR 'model*':ti,ab,kw OR 'decision*':ti,ab,kw OR 'identif*':ti,ab,kw OR 'prognos*':ti,ab,kw)) OR
('decision*':ti,ab,kw AND ('model*':ti,ab,kw OR 'clinical*':ti,ab,kw OR 'logistic
models':ti,ab,kw)) OR ('prognostic':ti,ab,kw AND ('history':ti,ab,kw OR 'variable*':ti,ab,kw OR 'criteria':ti,ab,kw OR 'scor*':ti,ab,kw OR 'characteristic*':ti,ab,kw OR 'finding*':ti,ab,kw OR 'factor*':ti,ab,kw OR 'model*':ti,ab,kw))) AND ('perinatal mortality'/exp OR 'perinatal mortalit*':ti,ab,kw OR 'perinatal death'/exp OR 'perinatal death*':ti,ab,kw OR 'neonatal death*':ti,ab,kw OR 'neonatal mortalit*':ti,ab,kw OR 'fetus mortality'/exp OR 'fetus mortalit*':ti,ab,kw OR ‘fetal mortalit*':ti,ab,kw OR 'fetal death*':ti,ab,kw OR 'fetal dismise*':ti,ab,kw OR 'IUFD':ti,ab,kw OR 'stillbirth'/exp OR 'stillbirth*':ti,ab,kw OR 'intra uterine fetal death':ti,ab,kw OR ‘intrauterine fetal death':ti,ab,kw OR 'intra uterine fetal demise':ti,ab,kw OR ‘intrauterine fetal demise':ti,ab,kw OR 'periventricular
leukomalacia'/exp OR 'periventricular leukomalacia*':ti,ab,kw OR 'periventricular encephalomalacia*':ti,ab,kw OR 'periventricular leucomalacia*':ti,ab,kw OR 'neonatal cerebral leukomalacia*':ti,ab,kw OR 'cystic periventricular leukomalacia*':ti,ab,kw OR 'cystic periventricular leucomalacia*':ti,ab,kw OR 'cystic leukomalacia*':ti,ab,kw OR 'cystic leucomalacia*':ti,ab,kw OR 'cystic encephalomalacia*':ti,ab,kw OR (((‘grade
three’:ti,ab,kw OR ‘grade 3’:ti,ab,kw) AND (‘brain hemorrhage’/exp OR ‘intracranial hemorrhage*’:ti,ab,kw OR ‘ICH’:ti,ab,kw OR ‘Intraventricular hemorrhage*’:ti,ab,kw OR
‘IVH’:ti,ab,kw OR ‘brain hemorrhage*’:ti,ab,kw OR ‘cranial hemorrhage*’:ti,ab,kw)) OR (‘venous’:ti,ab,kw AND (‘brain infarction’/exp OR ‘brain infarct*’:ti,ab,kw OR ‘cerebral infact*’:ti,ab,kw))) OR ('laser coagulation'/exp OR 'low level laser therapy'/exp) AND ('retrolental fibroplasia'/exp OR ‘retinopath* of prematurity':ti,ab,kw OR ‘ROP’:ti,ab,kw OR
‘prematurity retinopath*':ti,ab,kw OR ‘retrolental fibroplasia*’:ti,ab,kw) OR ('laser coagulation':ti,ab,kw OR 'low level laser therapy':ti,ab,kw) AND ('retrolental
fibroplasia'/exp OR ‘retinopath* of prematurity':ti,ab,kw OR ‘ROP’:ti,ab,kw OR ‘prematurity retinopath*':ti,ab,kw OR ‘retrolental fibroplasia*’:ti,ab,kw) OR 'necrotizing
enterocolitis'/exp OR 'necrotizing enterocolitis':ti,ab,kw OR 'NEC':ti,ab,kw OR 'neonatal respiratory distress syndrome'/exp OR 'neonatal respiratory distress syndrome':ti,ab,kw
OR 'newborn respiratory distress syndrome':ti,ab,kw OR 'infant* respiratory distress syndrome':ti,ab,kw OR 'IRDS':ti,ab,kw OR 'NRDS':ti,ab,kw OR 'perinat* outcome*':ti,ab,kw OR 'neonat* outcome*':ti,ab,kw ) AND [embase]/lim NOT ([embase]/lim AND
[medline]/lim) AND ('article'/it OR 'conference abstract'/it OR 'article in press'/it)
WEB OF SCIENCE
(ts=(pre-eclamp*) OR ts=preeclamp* OR ts=(pre eclamp*) OR ts=eclamp* OR ts=
(hypertens* AND pregnan* AND proteinuria*) OR ts=(hypertens* AND pregnan* AND multi* organ failure) OR ts= (pregnancy induced hypertens* AND proteinuria*) OR ts=(gestational hypertens* AND proteinuria*) OR ts= (pregnancy induced hypertens*
AND multi* organ failure) OR ts= (gestational hypertens* AND multi* organ failure) OR ts=(HELLP syndrom*) OR ts=(Hemolysis, elevated liver enzyme*, Lower* platelets) OR ts=(hemolys* AND liver/enzymology AND trombocytop*) OR ts=(hemolys* AND elevated liver enzyme* AND trombocytop$nia*) OR ts= (hemolys* AND elevated liver enzyme*
AND thrombopenia*) OR ts=(fetal growth retardation) OR ts= (foetal growth retardation) OR ts= (fetal growth restricted) OR ts= (foetal growth restricted) OR ts=(fetal growth restriction) OR ts=(foetal growth restriction) OR ts=(intra uterine growth restricted) OR ts= (intra uterine growth restriction) OR ts= (intrauterine growth restricted) OR ts=
(intrauterine growth restriction) OR ts=(intrauterine growth restrictions) OR ts=(intrauterine growth retarded) OR ts=(intrauterine growth retardation) OR
ts=(intrauterine growth retardations) OR ts=(intra uterine growth retarded) OR ts=(intra uterine growth retardation) OR ts=(intra uterine growth retardations) OR ts=(Placental Insufficiency) OR ts=( placental insufficienc*) OR ts=(placental insufficien*) OR ts=
(placenta insufficiency) OR ts=(placenta insufficien*) OR ts=(uteroplacental insufficien*) OR ts= (uteroplacental dysfunction*) OR ts= (placental dysfunction*) OR ts=(Premature Birth*) OR ts= (Preterm Birth*) OR ts=(Obstetric Labor, Premature) OR ts= (Obstetric Labor Premature) OR ts= (Obstetric Labor, Preterm) OR ts=(Labor, Premature) OR ts=(Labor, Preterm) OR ts= (Preterm deliver$ ) OR ts=(Infant, Extremely premature) OR ts= (Extremely Premature Infant$) OR ts= ( Extremely Preterm Infant$) OR ts=(Infant, Premature) OR ts=(Infant$, Premature) OR ts=(Preterm Infants) OR ts=(Preterm Infant ) OR ts=(Premature Infants) OR ts=(Neonatal Prematurity) OR ts=(Fetal Membranes, Premature Rupture) OR ts= (Premature Rupture of Fetal Membrane*) OR ts=(Preterm Rupture of Fetal Membrane* ) OR ts= (Preterm Premature Rupture of Fetal Membrane* ) OR ts=(PROM) OR ts= (Preterm PROM) OR ts=(Infant, Low Birth Weight) OR ts=(Infant, Low Birth weight*) OR ts=(low birth weight)) AND TS=((Validat$ OR Predict$.ti. OR Rule$) OR (Predict$ AND (Outcome$ OR Risk$ OR Model$)) OR ((History OR Variable$ OR Criteria OR Scor$ OR Characteristic$ OR Finding$ OR Factor$) AND (Predict$ OR Model$ OR Decision$ OR Identif$ OR Prognos$)) OR (Decision$ AND (Model$ OR Clinical$ OR Logistic Models/)) OR (Prognostic AND (History OR Variable$ OR Criteria OR Scor$ OR
Characteristic$ OR Finding$ OR Factor$ OR Model$))) AND (ts=(perinatal mortality) OR ts=(perinatal mortalit*) OR ts=(perinatal death) OR ts=(perinatal death*) OR
ts=(Neonatal death*) OR ts=(Neonatal mortalit*) OR ts=(Fetal mortality) OR ts=(Fetal mortalit*) OR ts=(Fetal death*) OR ts=(Fetal demise*) OR ts=(Stillbirth) OR
ts=(Stillbirth*) OR ts=(IUFD) OR ts=(Intra uterine fetal death) OR ts=(Intrauterine fetal death) OR ts=(Intra uterine fetal demise) OR ts=(Intrauterine fetal demise) OR ts=(Fetal death) OR ts=(Fetal mortality) OR ts=(Periventricular leukomalacia) OR
ts=(Periventricular leukomalacia*) OR ts=(Periventricular Encephalomalacia*) OR ts=(Periventricular leucomalacia*) OR ts=(Neonatal cerebral leukomalacia*) OR
ts=(Cystic periventricular leukomalacia*) OR ts=(Cystic periventricular leucomalacia*) OR ts=(Cystic leukomalacia*) OR ts=(Cystic leucomalacia*) OR ts=(Cystic
encephalomalacia*) OR ts=((Intracranial Hemorrhages) AND (Grade three OR Grade 3)) OR ts=((Intracranial Hemorrhage*) AND (Grade three OR Grade 3)) OR ts=((ICH) AND (Grade three OR Grade 3)) OR ts=((Intraventricular Hemorrhage*) AND (Grade three OR Grade 3)) OR ts=((IVH) AND (Grade three OR Grade 3)) OR ts=((Brain Hemorrhage*) AND (Grade three OR Grade 3)) OR ts=((Cranial Hemorrhage*) AND (Grade three OR Grade 3)) OR ts=(Brain infarction AND Venous) OR ts=(Brain infarct* AND Venous) OR ts=(Cerebral infarct* AND Venous) OR ts=(Venous brain infarct*) OR ts=(Venous cerebral infarct*) OR ts=((Retinopathy of Prematurity) AND (Laser coagulation OR Laser therapy)) OR
ts=((Retinopath* of Prematurity) AND (Laser coagulation OR Laser therapy)) OR
ts=((ROP) AND (Laser coagulation OR Laser therapy)) OR ts=((Prematurity Retinopath*)
AND (Laser coagulation OR Laser therapy)) OR ts=((Retrolental Fibroplasia*) AND (Laser coagulation OR Laser therapy)) OR ts=Necrotizing Enterocolitis OR ts=Necrotizing Enterocolitis OR ts=NEC OR ts=(Newborn Respiratory Distress Syndrome) OR ts=(Neonatal Respiratory Distress Syndrome) OR ts=(Newborn Respiratory Distress Syndrome) OR ts=(Infant Respiratory Distress Syndrome) OR ts=(Infantile Respiratory Distress Syndrome) OR ts=IRDS OR ts=NRDS OR ts=(perinat* outcome*) OR ts=(neonat*
outcome*))
Review & meeting abstract exclude: 9644 hits
PUBMED
(((((((Pregnancy-induced hypertension [MeSH Terms] OR pre-eclamp* [Title/Abstract] OR preeclamp* [Title/Abstract] OR pre eclamp* [Title/Abstract] OR eclamp* [Title/Abstract]
OR (Hypertension [MeSH Terms] AND Pregnancy [MeSH Terms] AND (Proteinuria [MeSH Terms] OR multiple organ failure [MeSH Terms])) OR (Hypertens*[Title/Abstract] AND Pregnan* [Title/Abstract] AND (Proteinuria* [Title/Abstract] OR Multiple Organ Failure [Title/Abstract])) OR (Gestational Hypertens* [Title/Abstract] AND (Proteinuria*
[Title/Abstract] OR Multiple Organ Failure [Title/Abstract])))) OR ((HELLP Syndrome*
[Title/Abstract] OR Hemolysis Elevated Liver Enzymes Lowered Platelets [Title/Abstract]
OR (Hemolysis [MeSH Terms] AND Liver/enzymology [MeSH Terms] AND
Thrombocytopenia [MeSH Terms]) OR (Hemolys* [Title/Abstract] AND Elevated Liver Enzyme*[Title/Abstract] AND Thrombocytopenia* [Title/Abstract]) OR (Hemolys*
[Title/Abstract] AND Elevated Liver Enzyme* [Title/Abstract] AND Thrombopenia*
[Title/Abstract])))) OR ((Fetal growth retardation [MeSH Terms] OR Fetal growth retard*
[Title/Abstract] OR fetal growth restrict* [Title/Abstract] OR foetal growth retard*
[Title/Abstract] OR foetal growth restrict* [Title/Abstract] OR intra uterine growth restrict*
[Title/Abstract] OR intrauterine growth restrict* [Title/Abstract] OR intrauterine growth retard* [Title/Abstract] OR intra uterine growth retard* [Title/Abstract]))) OR ((Placental Insufficiency [MeSH Terms] OR placenta insufficien*[Title/Abstract] OR placental
insufficien*[Title/Abstract] OR uteroplacental insufficien*[Title/Abstract] OR uteroplacental dysfunction*[Title/Abstract] OR placental dysfunction*[Title/Abstract]))) OR ((Infant, Low Birth Weight [MeSH Terms] OR low birth weight [Title/Abstract]))) AND ((Validat*
[Title/Abstract] OR Predict* [Title/Abstract] OR Rule* [Title/Abstract] OR (Predict*
[Title/Abstract] AND (Outcome* [Title/Abstract] OR Risk* [Title/Abstract] OR
Model*[Title/Abstract])) OR ((History [Title/Abstract] OR Variable* [Title/Abstract] OR Criteria [Title/Abstract] OR Scor* [Title/Abstract] OR Characteristic* [Title/Abstract] OR Finding* [Title/Abstract] OR Factor*[Title/Abstract]) AND (Predict* [Title/Abstract] OR Model* [Title/Abstract] OR Decision* [Title/Abstract] OR Identif* [Title/Abstract] OR Prognos*[Title/Abstract])) OR (Decision* [Title/Abstract] AND (Model* [Title/Abstract] OR Clinical* [Title/Abstract] OR Logistic Models[Title/Abstract])) OR (Prognostic
[Title/Abstract] AND (History [Title/Abstract] OR Variable* [Title/Abstract] OR Criteria [Title/Abstract] OR Scor* [Title/Abstract] OR Characteristic* [Title/Abstract] OR Finding*
[Title/Abstract] OR Factor* [Title/Abstract] OR Model* [Title/Abstract]))))) AND
((((((((Perinatal mortality [MeSH Terms] OR Perinatal mortalit* [Title/Abstract] OR Perinatal death [MeSH Terms] OR Perinatal death* [Title/Abstract] OR Neonatal death*
[Title/Abstract] OR Neonatal mortalit* [Title/Abstract] OR Fetal mortality [MeSH Terms] OR Fetal mortalit* [Title/Abstract] OR Fetal death* [Title/Abstract] OR Fetal demise*
[Title/Abstract] OR Stillbirth [MeSH Terms] OR Stillbirth* [Title/Abstract] OR IUFD [Title/Abstract] OR Intra uterine fetal death [Title/Abstract] OR Intrauterine fetal death [Title/Abstract] OR Intra uterine fetal demise [Title/Abstract] OR Intrauterine fetal demise [Title/Abstract] OR perinatal demise[Title/Abstract] OR neonatal demise [Title/Abstract])) OR ((Periventricular leukomalacia [MeSH Terms] OR Periventricular leukomalacia*
[Title/Abstract] OR Periventricular Encephalomalacia* [Title/Abstract] OR Periventricular leucomalacia* [Title/Abstract] OR Neonatal cerebral leukomalacia* [Title/Abstract] OR Cystic periventricular leukomalacia* [Title/Abstract] OR Cystic periventricular
leucomalacia* [Title/Abstract] OR Cystic leukomalacia* [Title/Abstract] OR Cystic leucomalacia* [Title/Abstract] OR Cystic encephalomalacia* [Title/Abstract]))) OR ((((Grade three [Title/Abstract] OR Grade 3 [Title/Abstract]) AND (Intracranial Hemorrhages [MeSH Terms] OR Intracranial Hemorrhage* [Title/Abstract] OR ICH [Title/Abstract] OR Intraventricular Hemorrhage* [Title/Abstract] OR IVH [Title/Abstract]
OR Brain Hemorrhage* [Title/Abstract] OR Cranial Hemorrhage* [Title/Abstract])) OR (Venous [Title/Abstract] AND (Brain infarction [MeSH Terms] OR Brain infarct*
[Title/Abstract] OR Cerebral infarction [MeSH Terms]))))) OR ((((Laser coagulation [MeSH Terms] OR Laser therapy [MeSH Terms]) AND (Retinopathy of Prematurity [MeSH Terms]
OR Retinopath* of Prematurity [Title/Abstract] OR ROP [Title/Abstract] OR Prematurity
Retinopath* [Title/Abstract] OR Retrolental Fibroplasia* [Title/Abstract])) OR ((Laser coagulation [Title/Abstract] OR Laser therapy [Title/Abstract]) AND (Retinopathy of Prematurity [MeSH Terms] OR Retinopath* of Prematurity [Title/Abstract] OR ROP [Title/Abstract] OR Prematurity Retinopath* [Title/Abstract] OR Retrolental Fibroplasia*
[Title/Abstract]))))) OR ((Necrotizing Enterocolitis [MeSH Terms] OR Necrotizing Enterocolitis [Title/Abstract] OR NEC [Title/Abstract]))) OR ((Respiratory Distress Syndrome, Newborn [MeSH Terms] OR Neonatal Respiratory Distress Syndrome [Title/Abstract] OR Newborn Respiratory Distress Syndrome [Title/Abstract] OR Infant*
Respiratory Distress Syndrome [Title/Abstract] OR IRDS [Title/Abstract] OR NRDS [Title/Abstract]))) OR ((perinat* outcome* [Title/Abstract] OR neonat* outcome*
[Title/Abstract])))
References
1. Brown MA, Magee LA, Kenny LC, Karumanchi SA, McCarthy FP, Saito S, et al. Hypertensive Disorders of Pregnancy: ISSHP Classification, Diagnosis, and Management Recommendations for International Practice. Hypertension (Dallas, Tex : 1979). 2018;72(1):24-43. Epub 2018/06/15. doi:
10.1161/hypertensionaha.117.10803. PubMed PMID: 29899139.
2. Friedman SA, Schiff E, Kao L, Sibai BM. Neonatal outcome after preterm delivery for preeclampsia. American journal of obstetrics and gynecology.
1995;172(6):1785-8; discussion 8-92. Epub 1995/06/01. doi: 10.1016/0002- 9378(95)91412-9. PubMed PMID: 7778633.
3. Bossung V, Fortmann MI, Fusch C, Rausch T, Herting E, Swoboda I, et al.
Neonatal Outcome After Preeclampsia and HELLP Syndrome: A Population-Based Cohort Study in Germany. Frontiers in pediatrics. 2020;8:579293. Epub
2020/11/07. doi: 10.3389/fped.2020.579293. PubMed PMID: 33154958; PubMed Central PMCID: PMCPMC7586782.
4. Lawin-O'Brien AR, Dall'Asta A, Knight C, Sankaran S, Scala C, Khalil A, et al.
Short-term outcome of periviable small-for-gestational-age babies: is our
counseling up to date? Ultrasound in obstetrics & gynecology : the official journal of the International Society of Ultrasound in Obstetrics and Gynecology.
2016;48(5):636-41. Epub 2016/11/18. doi: 10.1002/uog.15973. PubMed PMID:
27854384.
5. Garite TJ, Combs CA, Maurel K, Das A, Huls K, Porreco R, et al. A multicenter prospective study of neonatal outcomes at less than 32 weeks
associated with indications for maternal admission and delivery. American journal of obstetrics and gynecology. 2017;217(1):72.e1-.e9. Epub 2017/03/08. doi:
10.1016/j.ajog.2017.02.043. PubMed PMID: 28267444.
6. Sharp A, Jackson R, Cornforth C, Harrold J, Turner MA, Kenny L, et al. A prediction model for short-term neonatal outcomes in severe early-onset fetal growth restriction. European journal of obstetrics, gynecology, and reproductive biology. 2019;241:109-18. Epub 2019/09/10. doi: 10.1016/j.ejogrb.2019.08.007.
PubMed PMID: 31499415.
7. Wolff RF, Moons KGM, Riley RD, Whiting PF, Westwood M, Collins GS, et al.
PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies. Ann Intern Med. 2019;170(1):51-8. Epub 2019/01/01. doi: 10.7326/m18- 1376. PubMed PMID: 30596875.
8. Moons KG, de Groot JA, Bouwmeester W, Vergouwe Y, Mallett S, Altman DG, et al. Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist. PLoS medicine. 2014;11(10):e1001744.
Epub 2014/10/15. doi: 10.1371/journal.pmed.1001744. PubMed PMID: 25314315;
PubMed Central PMCID: PMCPMC4196729.
9. Moons KGM, Wolff RF, Riley RD, Whiting PF, Westwood M, Collins GS, et al.
PROBAST: A Tool to Assess Risk of Bias and Applicability of Prediction Model Studies: Explanation and Elaboration. Ann Intern Med. 2019;170(1):W1-w33.
Epub 2019/01/01. doi: 10.7326/m18-1377. PubMed PMID: 30596876.
10. Baiao AER, de Carvalho PRN, Moreira MEL, de Sa RAM, Gome SC. Predictors of perinatal outcome in early-onset fetal growth restriction: A study from an emerging economy country. Prenatal diagnosis. 2020;40(3):373-9. doi:
10.1002/pd.5596. PubMed PMID: WOS:000505305400001.
11. Baschat AA, Gembruch U, Weiner CP, Harman CR. Qualitative venous Doppler waveform analysis improves prediction of critical perinatal outcomes in premature growth-restricted fetuses. Ultrasound in obstetrics & gynecology : the official journal of the International Society of Ultrasound in Obstetrics and
Gynecology. 2003;22(3):240-5. Epub 2003/08/28. doi: 10.1002/uog.149. PubMed PMID: 12942494.
12. Geerts L, Odendaal HJ. Severe early onset pre-eclampsia: prognostic value of ultrasound and Doppler assessment. J Perinatol. 2007;27(6):335-42. doi:
10.1038/sj.jp.7211747. PubMed PMID: WOS:000246905800003.
13. Gómez-Arriaga PI, Herraiz I, López-Jiménez EA, Escribano D, Denk B, Galindo A. Uterine artery Doppler and sFlt-1/PlGF ratio: prognostic value in early- onset pre-eclampsia. Ultrasound in obstetrics & gynecology : the official journal of the International Society of Ultrasound in Obstetrics and Gynecology.
2014;43(5):525-32. Epub 2013/11/05. doi: 10.1002/uog.13224. PubMed PMID:
24185845.
14. Hernandez-Andrade E, Crispi F, Benavides-Serralde JA, Plasencia W, Diesel HF, Eixarch E, et al. Contribution of the myocardial performance index and aortic isthmus blood flow index to predicting mortality in preterm growth-restricted fetuses. Ultrasound in obstetrics & gynecology : the official journal of the
International Society of Ultrasound in Obstetrics and Gynecology. 2009;34(4):430- 6. Epub 2009/10/01. doi: 10.1002/uog.7347. PubMed PMID: 19790100.
15. Odibo AO, Goetzinger KR, Cahill AG, Odibo L, Macones GA. Combined sonographic testing index and prediction of adverse outcome in preterm fetal growth restriction. American journal of perinatology. 2014;31(2):139-44. Epub 2013/04/03. doi: 10.1055/s-0033-1341574. PubMed PMID: 23546845; PubMed Central PMCID: PMCPMC3932308.
16. Vergani P, Roncaglia N, Locatelli A, Andreotti C, Crippa I, Pezzullo JC, et al.
Antenatal predictors of neonatal outcome in fetal growth restriction with absent end-diastolic flow in the umbilical artery. American journal of obstetrics and gynecology. 2005;193(3 Pt 2):1213-8. Epub 2005/09/15. doi:
10.1016/j.ajog.2005.07.032. PubMed PMID: 16157140.
17. Wynants L, Van Calster B, Collins GS, Riley RD, Heinze G, Schuit E, et al.
Prediction models for diagnosis and prognosis of covid-19 infection: systematic review and critical appraisal. BMJ (Clinical research ed). 2020;369:m1328. Epub 2020/04/09. doi: 10.1136/bmj.m1328. PubMed PMID: 32265220; PubMed Central PMCID: PMCPMC7222643 www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no competing interests with regards to the submitted work; LW discloses support from Research
Foundation–Flanders (FWO); RDR reports personal fees as a statistics editor for The BMJ (since 2009), consultancy fees for Roche for giving meta-analysis teaching and advice in October 2018, and personal fees for delivering in-house training courses at Barts and The London School of Medicine and Dentistry, and also the Universities of Aberdeen, Exeter, and Leeds, all outside the submitted work.
18. Damen JA, Hooft L, Schuit E, Debray TP, Collins GS, Tzoulaki I, et al.
Prediction models for cardiovascular disease risk in the general population:
systematic review. BMJ (Clinical research ed). 2016;353:i2416. Epub 2016/05/18.
doi: 10.1136/bmj.i2416. PubMed PMID: 27184143; PubMed Central PMCID:
PMCPMC4868251 www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have
influenced the submitted work.
19. Kleinrouweler CE, Cheong-See FM, Collins GS, Kwee A, Thangaratinam S, Khan KS, et al. Prognostic models in obstetrics: available, but far from applicable.
American journal of obstetrics and gynecology. 2016;214(1):79-90.e36. Epub 2015/06/14. doi: 10.1016/j.ajog.2015.06.013. PubMed PMID: 26070707.
20. Bellou V, Belbasis L, Konstantinidis AK, Tzoulaki I, Evangelou E. Prognostic models for outcome prediction in patients with chronic obstructive pulmonary disease: systematic review and critical appraisal. BMJ (Clinical research ed).
2019;367:l5358. Epub 2019/10/06. doi: 10.1136/bmj.l5358. PubMed PMID:
31585960; PubMed Central PMCID: PMCPMC6776831
www.icmje.org/coi_disclosure.pdf (available on request from the corresponding
author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
21. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med.
2015;162(1):W1-73. Epub 2015/01/07. doi: 10.7326/m14-0698. PubMed PMID:
25560730.
22. Gordijn SJ, Beune IM, Thilaganathan B, Papageorghiou A, Baschat AA, Baker PN, et al. Consensus definition of fetal growth restriction: a Delphi procedure.
Ultrasound in obstetrics & gynecology : the official journal of the International Society of Ultrasound in Obstetrics and Gynecology. 2016;48(3):333-9. Epub 2016/02/26. doi: 10.1002/uog.15884. PubMed PMID: 26909664.
23. Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous predictors in multiple regression: a bad idea. Stat Med. 2006;25(1):127-41. Epub 2005/10/12.
doi: 10.1002/sim.2331. PubMed PMID: 16217841.
24. Collins GS, Ogundimu EO, Cook JA, Manach YL, Altman DG. Quantifying the impact of different approaches for handling continuous predictors on the
performance of a prognostic model. Stat Med. 2016;35(23):4124-35. Epub
2016/05/20. doi: 10.1002/sim.6986. PubMed PMID: 27193918; PubMed Central PMCID: PMCPMC5026162.
25. Healy P, Gordijn SJ, Ganzevoort W, Beune IM, Baschat A, Khalil A, et al. A Core Outcome Set for the prevention and treatment of fetal GROwth restriction:
deVeloping Endpoints: the COSGROVE study. American journal of obstetrics and gynecology. 2019;221(4):339.e1-.e10. Epub 2019/06/04. doi:
10.1016/j.ajog.2019.05.039. PubMed PMID: 31152710.
26. Myatt L, Redman CW, Staff AC, Hansson S, Wilson ML, Laivuori H, et al.
Strategy for standardization of preeclampsia research study design. Hypertension (Dallas, Tex : 1979). 2014;63(6):1293-301. Epub 2014/04/02. doi:
10.1161/hypertensionaha.113.02664. PubMed PMID: 24688121.
27. Steyerberg EW, Vickers AJ, Cook NR, Gerds T, Gonen M, Obuchowski N, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128-38. Epub 2009/12/17. doi:
10.1097/EDE.0b013e3181c30fb2. PubMed PMID: 20010215; PubMed Central PMCID: PMCPMC3575184.
