FIGO (International Federation of Gynecology and Obstetrics) initiative on fetal growth: Best practice advice for screening, diagnosis, and management of fetal growth restriction

Nir Melamed1 | Ahmet Baschat2 | Yoav Yinon3 | Apostolos Athanasiadis4 | Federico Mecacci5 | Francesc Figueras6 | Vincenzo Berghella7 | Amala Nazareth8 | Muna Tahlak9 | H. David McIntyre10 | Fabrício Da Silva Costa11 | Anne B. Kihara12 | Eran Hadar13,14 | Fionnuala McAuliffe15 | Mark Hanson16,17 | Ronald C. Ma18,19 | Rachel Gooden20 | Eyal Sheiner21 | Anil Kapur22 | Hema Divakar23 | Diogo Ayres-de-Campos24 | Liran Hiersch25 | Liona C. Poon26 | John Kingdom27 | Roberto Romero28 | Moshe Hod13,14*


| E XECUTIVE SUMMARY
Fetal growth restriction (FGR) is defined as the failure of the fetus to meet its growth potential due to a pathological factor, most commonly placental dysfunction. Worldwide, FGR is a leading cause of stillbirth, neonatal mortality, and short-and long-term morbidity.
Ongoing advances in clinical care, especially in definitions, diagnosis, and management of FGR, require efforts to effectively translate these changes to the wide range of obstetric care providers.
This article highlights agreements based on current research in the diagnosis and management of FGR, and the areas that need more research to provide further clarification of recommendations.
The purpose of this article is to provide a comprehensive summary of available evidence along with practical recommendations concerning the care of pregnancies at risk of or complicated by FGR, with the overall goal to decrease the risk of stillbirth and neonatal mortality and morbidity associated with this condition. To achieve these goals, FIGO (the International Federation of Gynecology and Obstetrics) brought together international experts to review and summarize current knowledge of FGR.
This summary is directed at multiple stakeholders, including healthcare providers, healthcare delivery organizations and providers, FIGO member societies, and professional organizations.
Recognizing the variation in the resources and expertise available for the management of FGR in different countries or regions, this article attempts to take into consideration the unique aspects of antenatal care in low-resource settings (labelled "LRS" in the recommendations). This was achieved by collaboration with authors and FIGO member societies from low-resource settings such as India, Sub-Saharan Africa, the Middle East, and Latin America.
Aspects of FGR addressed in this article include prediction, diagnosis, investigation, management, and postpartum counselling. The main recommendations are given below and are summarized in Table 1  There is no evidence to support the routine use of biochemical markers for the prediction of FGR. However, when such information is available as part of the prenatal genetic screening for trisomy 21, it may be reasonable to use this information for the purpose of risk stratification for FGR (and other placenta-mediated complications).
Postpartum assessment and counselling for future pregnancies in women with a history of fetal growth restriction (FGR)

Quality of evidence
Strength of recommendation 1. Growth restricted infants are at an increased risk of short-and long-term morbidity and should be followed postnatally more closely than normally grown infants.

| TARG E T AUD IEN CE
This article is directed at multiple stakeholders with the intention of bringing attention to the assessment of fetal growth, with a particular focus on the screening, diagnosis, and management of FGR, which is a leading cause of stillbirth and neonatal mortality and morbidity. This article proposes to standardize and provide guidance for the screening, prevention, diagnosis, and management of FGR.
The intended target audience includes: Healthcare providers: all those qualified to care for pregnant women (obstetricians, maternal-fetal medicine specialists, general practitioners, midwives, nurses, advance practice clinicians, radiologists, sonographers, pediatricians, and neonatologists).
Healthcare delivery organizations and providers: governments, federal and state legislators, healthcare management organizations, health insurance organizations, international development agencies, and nongovernmental organizations.
Professional organizations: international, regional, and national professional organizations of obstetricians and gynecologists, obstetric ultrasound, family practitioners, pediatricians, neonatologists, and worldwide national organizations dedicated to the care of pregnant women and their offspring.

| A SS E SS MENT OF QUALIT Y OF E VIDEN CE AND G R ADING OF S TRENG TH OF RECOMMENDATIONS
In assessing the quality of evidence and grading of strength of recommendations, the article follows the terminology proposed by the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) Working Group. 5 This system uses consistent language and graphical descriptions for the strength and quality of the recommendations and the evidence on which they are based.
Recommendations are classified as strong or conditional (weak) ( Table S1). 6 The strength of recommendation is dependent not only on the quality of evidence, but also on factors such as risk-benefit, cost, resource allocation, values, and preferences. Thus, some recommendations may be based on low-quality evidence but still represent a benefit that outweighs the risks and burdens, and therefore may be strongly recommended.
The overall quality of evidence was assessed for each of the recommendations and expressed using four levels of quality: very low, low, moderate, and high (Table S2). 7 Considerations for quality of evidence include primarily the study design and methodology. As such, evidence based on randomized controlled trials is considered high-quality evidence, observational studies provide moderate or low quality of evidence, and all others are very low. However, other parameters must be considered while assessing the level of evidence: risk of bias, study limitations, consistency of results, precision, publication bias, indirectness of evidence, and scarcity of evidence. For the quality of evidence, cross-filled circles are used: ⊕ ○○○ denotes very low-quality evidence; ⊕⊕ ○○ low quality; ⊕⊕⊕ ○ moderate quality; and ⊕⊕⊕⊕ high-quality evidence.
The purpose of this article is to provide a summary of the available evidence and provide recommendations regarding the early prediction and prevention, diagnosis, investigation, monitoring, and timing of delivery of pregnancies complicated by FGR, with the overall goal to decrease the risk of stillbirth and neonatal mortality and morbidity associated with this pregnancy complication.
Given the variation in resources and expertise available for the assessment and monitoring of pregnancies complicated by FGR in different countries or regions, we have included, in addition to the standard of care or "best" recommendations, specific recommendations for low-resource settings, which are marked as LRS in the recommendation tables. Management algorithms for women in high-resource and low-resource settings are summarized in Figure 1a and 1b, respectively.

| Terminology and definitions
FGR is defined as the failure of the fetus to meet its growth potential due to a pathological factor, most commonly placental dysfunction.
Clinically, this is reflected by a drop in fetal size percentiles over the course of gestation. However, fetal growth potential is difficult to determine, and serial assessments of fetal size to detect a drop in fetal weight percentile are usually not available. Instead, care providers most commonly have only a "snapshot" of fetal weight estimation at a given point in time. Therefore, in clinical practice, small for gestational age (SGA), defined as estimated fetal weight (EFW) or abdominal circumference below a certain threshold such as the 10th or 3rd percentile, is most commonly used to suspect FGR.
The use of SGA as a proxy for FGR has several limitations that need to be recognized. First, most SGA fetuses are constitutionally healthy small fetuses, whose smallness is merely the result of their predetermined growth potential (i.e. false-positive diagnosis of FGR). Second, some growth-restricted fetuses, depending on their original growth potential and timing of insult, may remain above the percentile threshold described above and may thus not be SGA (i.e. false-negative diagnosis of FGR). Third, the use of SGA as a proxy for FGR is limited by the accuracy of sonographic fetal weight estimation, which has an estimation error of up to ±15%-20%. Finally, the diagnosis of SGA is highly dependent on the growth chart being used, which can therefore have a considerable effect on the proportion of fetuses or infants flagged as SGA in a given population.
It should be noted that there is inconsistency in the literature regarding the terminology described above, where some use the   term FGR to describe a fetus with an estimated weight below the   10th percentile for gestational age and the term SGA to describe   an infant with birth weight below the 10th percentile for

| Early-versus late-onset FGR
It has been suggested that FGR should be broadly classified, based on gestational age at the time of diagnosis, into early-onset FGR (<32 weeks) and late-onset FGR (≥32 weeks). The rationale underlying this classification is based on differences between these two phenotypes of FGR in severity, natural history, Doppler findings, association with hypertensive complications, placental findings, and management. [16][17][18] Early-onset FGR has a prevalence of 0.5%-1%, is usually more severe, and is more likely to be associated with abnormal umbilical artery Doppler than late-onset FGR. The underlying placental pathology is frequently similar to that observed in cases of early-onset pre-eclampsia (maternal vascular malperfusion), which explains the strong association of early-onset FGR with pre-eclampsia. Therefore, early-onset FGR is usually easier to detect, and the natural history tends to follow a predictable sequence of Doppler changes in the umbilical artery and ductus venosus. The main challenge in cases of early-onset FGR is management (i.e. timing of delivery), by attempting to determine the optimal balance between the opposing risks of stillbirth and prematurity. 19 Late-onset FGR is more common than early-onset FGR with a prevalence of 5%-10%. In contrast to early-onset FGR, it is usually milder, is less likely to be associated with pre-eclampsia, and is usually associated with normal umbilical artery Doppler.
Therefore, the main challenge with regard to late-onset FGR is diagnosis, while management (i.e. delivery) is relatively simple given that the diagnosis is commonly made during the late-preterm or term periods, where the risks associated with delivery are relatively small. The diagnosis of late-onset FGR mainly relies on adaptive changes in the cerebral circulation ("redistribution" or "brain-sparing effect"), which is reflected by low resistance to flow in the middle cerebral artery thereby generating a low cerebroplacental ratio, as described in section 8.

| Etiology of fetal growth restriction
FGR is often the result of one or more maternal, placental, or fetal disorders that interfere with the normal mechanisms regulating fetal growth. 20,21 The most common etiologies of FGR are listed in Box 2. It is important to note that there is often confusion in the literature between "etiologies" (or pathogenetic pathways) and "risk factors" for FGR. For example, although maternal conditions such as chronic hypertension, kidney disease, systemic lupus erythematosus, and long-standing diabetes are often listed as "maternal etiologies" for FGR, these conditions should probably be viewed instead as maternal risk factors for abnormal placentation that may result in placenta-mediated FGR.
Given that maternal nutrition and fetal growth are closely related, 22,23 maternal undernutrition is an important cause of FGR worldwide. [24][25][26] The impact of maternal undernutrition on fetal growth depends on its timing and severity. 20 To date, maternal interventions in dietary advice and modifications have lacked significant success in preventing FGR. While the mechanisms by which maternal anemia contributes to FGR are unclear, both impaired nutrient transport to the fetus 27 and abnormal placental adaptation to low maternal hemoglobin 28 have been suggested as potential mechanisms.
Abnormal placentation is a common cause of FGR, 29 which is often diagnosed by ultrasound Doppler studies 30 and typical histopathological placental findings. [31][32][33] Chromosomal abnormalities have been suggested to contribute to up to 5% of FGR cases; triploidy and trisomy 13 and 18 are important considerations in early-onset FGR and the risk of many aneuploidies is higher in the presence of structural fetal anomalies. [34][35][36] In 1%-6% of cases of FGR with normal karyotype, submicroscopic (micro) duplications/deletions can be found using chromosomal microarray analysis, 35 even when FGR is an apparently isolated finding. 37 FGR is also more prevalent in fetuses with structural malformations, and the risk increases when multiple anomalies are present. 38 FGR is related to intrauterine infection in up to 5% of cases. 20,39 Viral agents such as rubella, cytomegalovirus, HIV, and Zika are common causes of infection-related FGR. [40][41][42][43][44] Protozoan infections like toxoplasmosis and malaria are another important cause, especially in endemic areas. 45,46 The main mechanism involved in the pathogenesis of FGR in these cases is a decline in cell population. 20 Finally, maternal exposure to teratogens such as radiation, 47 illicit drugs, 48,49 and alcohol 50 is another important etiology of FGR.

| Risks associated with fetal growth restriction
The main short-and long-term risks associated with FGR are listed in Box 3. It is associated with both fetal and obstetric complications. The most devastating complication is stillbirth, [51][52][53] and there is a well-established inverse relationship between weight percentile and the risk of stillbirth, [54][55][56][57] which is more pronounced in the early preterm period than at term. 58 FGR is an important cause of iatrogenic preterm birth, 59 as early delivery remains the main and perhaps only strategy for the prevention of stillbirth in cases of severe FGR. 16,60 FGR is also an independent risk factor for spontaneous preterm birth. 61 Other obstetric complications associated with FGR include pre-eclampsia and placental abruption, as the pathophysiology of these conditions is often closely related. 29,30,[62][63][64][65][66] Despite ongoing improvements in neonatal care, FGR is associated with increased neonatal mortality and short-term morbidity.
The risk of perinatal mortality in term FGR is reported to be fiveto 10-fold higher than in appropriately grown neonates. 57,61,67 The severity of FGR, Doppler abnormalities, and associated prematurity are independent predictors of neonatal complications. 68 Among preterm infants, the co-presence of FGR further increases the risk of certain prematurity-related complications such as respiratory morbidity, intraventricular hemorrhage, necrotizing enterocolitis, and metabolic disorders. 57 Among term infants, FGR increases the risks of low cord artery pH, 69 low Apgar score, 69 and neonatal complications such as hypoglycemia, hypothermia, and jaundice. [70][71][72] Growth-restricted infants are also at risk of long-term complications including neurodevelopmental impairment 11,73-78 and noncommunicable diseases. 15,[79][80][81][82] This is discussed in greater detail in section 9.1 (Infant follow-up).

| Recommendations
FIGO recommends the following for the definition of fetal growth restriction (FGR) prediction of FGR is the lack of a gold standard for the antenatal or postnatal diagnosis of FGR. As such, there is wide variation among studies regarding the outcomes being predicted, including either SGA (birth weight below the 10th or 3rd percentile) or adverse perinatal outcomes that are associated with (but are not specific to) FGR.

Quality of evidence
As many SGA infants are constitutionally small and healthy, differentiating between healthy small fetuses and those that are small due to FGR is critically important. As a rule, the prediction of early-onset severe FGR is better than of late-onset FGR. and texture (calcifications, echogenic cystic lesions). gonadotropin (hCG) levels greater than 2.5 MoM in the second trimester, alone or combined with high alpha-fetoprotein levels, are also associated with an increased risk of SGA. 99 Angiogenic factors play a key role in the regulation of placental vascular development. 100 Placental growth factor (PlGF) is a proangiogenic factor highly expressed in the syncytiotrophoblast and the maternal endothelium. Impaired placentation is associated with reduced placental production of this protein. Low first-trimester PlGF levels have been shown to be associated with adverse pregnancy outcome including pre-eclampsia and SGA. [101][102][103][104] In a casecontrol study of 296 pregnancies with SGA and 609 controls, the detection rate of low PlGF for SGA at a false-positive rate of 5% and 10% was 15% and 21%, respectively. The combined use of PlGF and PAPP-A increased the detection rate to 19% and 27%, respectively. 103 A multicenter screening study found that the detection rate of a combined screening by maternal factors, fetal biometry, and serum PlGF and alpha-fetoprotein at 19-24 weeks for the delivery of SGA infants below the 5th percentile at less than 32, 32-36, and greater than or equal to 37 weeks of gestation was 100%, 76%, and 38%, respectively, at a false-positive rate of 10%. 96 Findings are less consistent for soluble fms-like tyrosine kinase-1 (sFlt-1), an antiangiogenic factor released from the placenta that results in maternal endothelial dysfunction characteristic of pre-eclampsia. 105 Although maternal serum sFlt-1 levels are known to be elevated in pre-eclamptic pregnancies, a large case-control study demonstrated that high levels of sFlt-1 at 10-14 weeks were actually associated with a slightly reduced risk of SGA (OR 0.92; 95% CI, 0.88-0.96). 101 Therefore, the sFlt-1:PlGF ratio test used to diagnose pre-eclampsia should not be used in the first trimester as a screening test for FGR. 106

| Ultrasound markers
Several ultrasound-based markers have been shown to be predictive of FGR, including uterine artery Doppler, placental morphology, and placental volumes. However, given their modest predictive accuracy, they cannot be recommended for universal screening for FGR.
Increased uterine artery resistance largely reflects a failure of extravillous cytotrophoblast invasion and transformation of the spiral arteries and is associated with the development of pre-eclampsia and FGR due to maternal vascular malperfusion of the placenta. 107 First-and second-trimester abnormal uterine artery Doppler waveforms, defined as mean pulsatility index above the 95th percentile, have been shown to be associated with FGR. [108][109][110] In a large prospective cohort study of 4610 nulliparous women, uterine artery pulsatility index at 11 +0 to 13 +6 weeks predicted 60% of preterm and 17% of term SGA infants at a false-positive rate of 10%. 111 Although uterine artery Doppler shows promise, especially for the prediction of early-onset FGR, current evidence does not support routine screening with uterine artery Doppler for FGR in low-or high-risk pregnancies. 112 Sonographic evaluation of the placenta is a routine part of the obstetric ultrasound examination. A method for systematic two-dimensional (2D) placental ultrasound examination has been described, often in combination with other parameters 30,113,114 Abnormal placental morphology is defined by placental dimensions, shape, texture, and cord insertion. Placental shape is considered abnormal when the placental thickness is above 4 cm or greater than 50% of placental length. Placental texture is defined as normal when it is homogenous, and abnormal when the placenta is heterogeneous and contains multiple echogenic cystic lesions or has a jelly-like appearance with turbulent uteroplacental flow. 115 Another parameter is the placental quotient, defined as the ratio of the placental volume to the fetal crown-rump length. The placental quotient was reported to have a high negative predictive value for perinatal complications but was not very useful when used for screening of SGA in a low-risk population, with a sensitivity of 27.1%. 120 The discriminatory ability of placental volume alone for SGA appears to be modest, but may be integrated into a multivariable screening model. However, the use of 3D placental volume as a routine screening tool for FGR is limited by the need for proper equipment and training required to obtain these measurements in a reproducible manner.  Substance use, including smoking, alcohol, and illicit drugs, is associated with low birth weight and increased perinatal morbidity and mortality. 90 Interventions to promote smoking cessation during pregnancy have been shown to result in a reduction in low birth weight (RR 0.81) and an increase in mean birth weight (+33 g). 127 Women should be advised that smoking cessation at any point in gestation is of benefit, and that the greatest benefit is associated with cessation before 15 weeks of pregnancy. 128 The risk of SGA with alcohol intake is increased with as little as one drink per day. 129

| Medical interventions
Most studies on early prevention of placental complications have focused on pre-eclampsia, with the results often being extrapolated to FGR due to the common pathophysiology. However, to date, other than lifestyle modifications, no medical interventions to prevent FGR have been clearly established.
Aspirin is recommended for women at increased risk of pre-eclampsia, but there is some evidence that it may also reduce the risk of FGR 130,131 In a recent meta-analysis of 45 trials that included 20 909 women at high risk of pre-eclampsia, the administration of aspirin starting at less than or equal to 16 131 One randomized trial found that evening but not morning administration of aspirin is associated with reduction in the rate of pre-eclampsia and FGR. 133 However, it should be emphasized that most of the available data on aspirin come from studies that focused on the prevention of pre-eclampsia as the primary outcome in women at high risk of pre-eclampsia, with the prevention of FGR considered only as a secondary outcome. Furthermore, in the largest trial to date on the use of aspirin for the prevention of pre-eclampsia (ASPRE trial), aspirin was not associated with a reduction in the risk of SGA below the 10th, 5th, or 3rd percentile. 130 However, we believe that given the safety of aspirin and the overlap in the risk factors and pathogenesis of pre-eclampsia and FGR, it is reasonable to recommend aspirin to women at high risk of FGR, using the same regimen of aspirin used for women at high risk of pre-eclampsia. Most international guidelines recommend 100-150 mg aspirin to prevent FGR in women at high risk. 134 The adjunct role of heparin in combination with aspirin to prevent placenta-mediated complications in high-risk situations was originally attributed to its anticoagulant properties and the speculative prevention of placental thrombosis. However, in vitro and in vivo data suggest heparins may have other biological properties including anti-inflammatory, complement inhibition, and proangiogenic activities. [135][136][137][138] A study-level meta-analysis of six trials including 848 women showed that low-molecular-weight heparin (LMWH) was associated with a reduction in the composite outcome of pre-eclampsia, birth weight below the 10th percentile, placental abruption, or pregnancy loss after 20 weeks (RR 0.52; 95% CI, 0.32-0.86) with similar risk reduction for SGA below the 10th and 5th percentiles. However, the higher-quality trials suggest no treatment effect, 139 and a subsequent individual patient data meta-analysis looking at the same composite outcome found no beneficial effect of LMWH treatment (RR 0.64; 95% CI, 0.36-1.11). 140 Likewise, the enoxaparin for pre-eclampsia and intrauterine growth restriction (EPPI) trial included women at high risk for placenta-mediated complications (with a high proportion of women with prior FGR) and showed no difference in the rate of the composite outcome (pre-eclampsia or SGA <5th percentile) between treated and nontreated women. 141 Therefore, based on the most up-to-date evidence, LMWH cannot be recommended for the prevention of FGR in women at high risk of placenta-mediated complications. Its use for the prevention of FGR should therefore be limited to research settings, for example in women already on aspirin who are found to have abnormal levels of angiogenic markers prior to fetal viability. 142

Strength of recommendation
1. Women should undergo risk stratification for FGR (and other placenta-mediated complications) at the time of the first-trimester antenatal visit using history-based (medical and obstetric) risk factors.
There is no evidence to support the routine use of biochemical markers for the prediction of FGR. However, when such information is available as part of prenatal genetic screening for trisomy 21, it may be reasonable to use this information for the purpose of risk stratification for FGR (and other placentamediated complications).

⊕⊕⊕ ○
Strong 3. Ultrasound-based markers and multiparameter algorithms have only a moderate predictive accuracy for FGR, and therefore currently cannot be recommended for universal screening. There are insufficient data to recommend routine treatment with aspirin in all women at high risk of FGR. Treatment with aspirin at a dose of 100-150 mg starting at 12-16 weeks may be considered in selected cases such as women who are at high risk of pre-eclampsia or those with a history of placenta-mediated FGR.

⊕⊕ ○○
Weak 8. Low-molecular-weight heparin is not recommended for the prevention of FGR in women at high risk of FGR and its use should be limited to research settings.

| DE TEC TI ON OF FE TAL G ROW TH RE S TRI C TI ON
Detection of FGR is based on the identification of a fetus that is smaller than expected for gestational age, through either physical examination (symphysis-fundal height, SFH) or ultrasound.

| Symphysis-fundal height
Measurement of SFH using a tape is a simple, inexpensive, and widely used strategy to screen for FGR. [143][144][145][146] SFH is measured with the woman in a supine position using a nonelastic metric tape after she has emptied her bladder. To decrease the interobserver variability, a standardized technique for measuring SFH should be followed. 144,145 SFH is defined as the distance from the upper border of the symphysis pubis bone to the top of the uterine fundus. 145 SFH measured in centimeters between 24 and 38 weeks of gestation approximates the gestational age. 147 Numerous local charts are currently used worldwide, [148][149][150][151][152][153][154][155][156] with the recent addition of an international standard for SFH based on serial measurements. 145 However, the accuracy of SFH measurement in predicting SGA (EFW <10th percentile) is limited, and there are no randomized controlled trials that compare SFH measurement with serial ultrasound evaluation of fetal biometry. 157 In a meta-analysis of 34 observational studies, SFH was reported to have a sensitivity of 58% and a specificity of 87% for predicting birth weight below the 10th percentile. There was marked heterogeneity between studies, mainly due to the use of different SFH charts. 158 A single SFH measurement at 32-34 weeks of pregnancy has been reported to be approximately 65%-85% sensitive and 96% specific for detecting FGR. 143 It is important to acknowledge that factors such as maternal obesity, uterine leiomyomas, and polyhydramnios may further limit the accuracy of SFH as a screening tool. 144,159

| Sonographic fetal weight estimation
Sonographic fetal biometry is the cornerstone for detection of fetal growth disorders. Standard fetal biometry includes assessment of head circumference (HC), biparietal diameter, abdominal circumference (AC), and femur length (FL). Measurement of these biometric indices should be obtained by an experienced individual and in a standardized manner, as has been previously described. 160 Fetal weight is estimated based on various combinations of the four biometric indices described above, using one of many published equations. [161][162][163][164][165] The accuracy of most equations falls within the range of ±10%, and the error has been shown to be greater at the extremes of fetal weight, and to be affected by factors such as fetal sex, presentation, and plurality (greater in twin gestations). 162 an equation that has been validated within their local population and within the gestational age range in which it will be used. However, if such information is not available-a very frequent scenario-it seems reasonable to use the Hadlock equation as described above.

| Is there a role for routine third-trimester ultrasound to assess fetal growth?
In many countries, measurement of SFH is the primary screening tool for FGR in low-risk pregnancies and ultrasound measurement of fetal biometry is performed only when indicated on the basis of risk factors or abnormal SFH. 134,143,[172][173][174] However, this approach fails to identify the majority of FGR infants, 146 a concerning finding given that undetected FGR is associated with increased risk of adverse perinatal outcome and stillbirth. 53,175 An alternative approach is to perform a routine third-trimester ultrasound for fetal weight estimation. However, a strategy for routine third-trimester ultrasound in low-risk pregnancies is not supported by available data and cannot be recommended. [176][177][178] A meta-analysis of 13 trials assessed the effect of routine sonographic weight estimation at more than 24 weeks of gestation on pregnancy outcomes in both unselected and low-risk pregnancies. 178 The authors found no association between routine sonographic EFW and adverse pregnancy outcomes including perinatal mortality, preterm birth, induction of labor, or cesarean section. In a recent randomized controlled trial of women with uncomplicated pregnancies, the use of serial (every 4 weeks) third-trimester ultrasound was superior to routine care in the detection of a composite outcome of fetal growth or amniotic fluid abnormalities (RR 3.43; 95% CI, 1.64-7.17). 179 However, it is important to note that the incidence of maternal or fetal morbidity was not significantly different between the groups. Similar results were reported by others. 180 In contrast, the Pregnancy Outcome Prediction (POP) study prospectively assessed 3977 women and compared the detection of SGA (birth weight <10th percentile) by routine ultrasound versus clinically indicated ultrasound in the third trimester. 181 The detection rate of SGA was nearly tripled in the routine ultrasound group (57% vs 20%). The risk of neonatal morbidity was increased only in the subset of SGA fetuses with fetal abdominal circumference growth velocity in the lowest decile (RR 3.9; 95% CI, 1.9-8.1), emphasizing the importance of combined analysis of fetal biometry and fetal growth velocity for better detection of fetuses at risk. 182 Furthermore, it has been suggested that the prediction of FGR based on routine third-trimester ultrasound can be improved by integrating EFW with additional biomarkers. A combined screening model that included maternal characteristics, third-trimester EFW and placental Doppler, and biochemical markers (PlGF and estriol) achieved better performance than EFW alone in the detection of FGR (77% vs 64%) at a 10% false-positive rate. 183 There are many conceptual explanations to support thirdtrimester ultrasound as it can assist in the diagnosis of clinically significant findings other than FGR, including fetal malpresentation, 184 disorders of amniotic fluid, and fetal anomalies, 185,186 especially when combined with Doppler measurements and biochemical markers. 95,[187][188][189] However, there is no evidence that this information improves outcomes when performed routinely in lowrisk pregnancies.

| Which growth chart should be used to determine fetal weight percentile?
The interpretation of sonographic EFW depends on gestational age and is commonly classified as appropriate for gestational age, SGA, or large for gestational age, based on the calculation of EFW percentile using one of the many available growth charts. The choice of growth chart has been shown to have a considerable impact on the proportion of fetuses classified as either SGA or large for gestational age. 190,191 Over the past several years there has been an ongoing debate regarding the optimal growth chart that should be used, and numerous studies have compared the performance of a wide variety of charts in different populations with conflicting results. Prior to further discussion of specific charts, it is important to clarify the terminology and the types of charts that are currently available.

| Growth references versus growth standards
Growth references are descriptive charts that provide information on the distribution of weight of all newborns in a given population, and as such they include both normal and complicated pregnancies.
Although growth references are useful as they provide information on the overall distribution of birth weight in the population, their use for the purpose of antenatal detection of FGR may be challenging as they are affected by the rate of pathologies in the population. For example, in populations with a high rate of large fetuses (e.g. due to a high rate of obesity and diabetes), the reference would be shifted upward. Similarly, in populations with a high rate of FGR (e.g. due to a high rate of malnutrition), the reference would be shifted downward.
For that reason, it may be reasonable to prefer growth standards over growth references for the antenatal detection of FGR. Growth standards are prescriptive charts that are based only on low-risk or uncomplicated pregnancies, and as such provide information on what is the optimal fetal growth. There is variation between different growth standards with regard to the definition of "low-risk" pregnancies; while some standards excluded women with pre-existing medical conditions and pregnancy complications, others also excluded women below or above certain height or weight, women with suboptimal nutrition, low socioeconomic status, exposure to air pollution, high altitude etc. Since growth standards include only low-risk uncomplicated pregnancies, their distribution is usually narrower (i.e. the 10th and 90th percentiles are closer to the mean) compared with growth references.
One important and practical aspect regarding the use of reference versus standard charts relates to the weight percentile threshold that should be used to trigger further evaluation for FGR. When using a growth reference, it is reasonable to use the 10th percentile for that purpose, as a considerable proportion of infants below the10th percentile will be affected by pathology. In the case of growth standard, however, using the same threshold of the 10th percentile would, per definition, identify 10% of the low-risk pregnancies as suspected for FGR, which is not practical. Therefore, when using a growth standard, a lower threshold-such as the 5th or 3rd percentile-should be used to indicate further evaluation for FGR. are more likely to be affected by placental dysfunction and to be growth restricted. Therefore, these charts are likely to underestimate the optimal weight of fetuses during the preterm period, which in turn may lead to an underdiagnosis of FGR before 37 weeks. This is illustrated in Figure 2, where the birth weightbased chart of Alexander (USA) 192 is compared with several ultrasound-based charts.
Therefore, it seems reasonable to prefer growth charts that are based on sonographic EFW over those that are based on birth weight.
Ultrasound-based growth charts are more difficult and expensive to develop, as they are usually based on data from prospective longitudinal studies where women undergo several sonographic weight estimations during pregnancy. However, these charts do not share the limitation of birth weight-based charts, described above, and are thus more likely to reflect the optimal fetal growth throughout pregnancy ( Figure 2). Another reason why ultrasound-based charts should be preferred is that the measure used during pregnancy to assess fetal growth is sonographic EFW; it is therefore more appropriate to compare it to charts based on the same measure (i.e. sonographic EFW) rather than to charts based on birth weight.
Some of the commonly used ultrasound based charts are presented in Figure 2. 193  Others, however, believe that the variation in fetal growth between countries and races is not solely the result of environmental factors. Instead, it is suggested that genetic variation in growth potential contributes to the observed differences in fetal growth between race groups, and that race-specific charts should therefore be preferred over universal charts. Examples of such  199 According to the third approach, growth charts should be adjusted not only for maternal race but also for other physiologic factors that are thought to determine fetal growth potential, such as maternal height, weight, parity, and fetal sex. One such example is the Gestation Related Optimal Weight (GROW) software for customized growth percentiles. 200,201 At the other end of the spectrum is the individualized growth assessment (IGA) approach, which is based on estimation of the growth potential of the individual fetus, calculated from the secondtrimester growth velocity of that fetus. These estimates are used to generate individualized trajectories that are used to interpret fetal growth during the third trimester (https://igap.resea rch.bcm. edu). [202][203][204] While compelling, this approach requires earlier ultrasound exams during pregnancy, as well as appropriate software, and is therefore challenging at present for the purpose of FGR screening in the general population, and especially in low-resource settings.

| Description of commonly available charts
The 10th percentile curves of some of the charts described above are compared in Figure 2. The Hadlock chart (1991), one of the most commonly used growth charts, is an ultrasound-based standard. It is based on a cohort of 392 low-risk, primarily white women from Texas. The Alexander chart (1996) is based on over 3 million singleton live births in the USA and is included as an example of a birth weight-based reference to illustrate their limitation, which is the underestimation of optimal fetal growth during the preterm period.
The goal of the Intergrowth-21st project (2014) was to develop a universal ultrasound-based prescriptive growth chart. This was a prospective longitudinal study of 4321 low-risk women from eight centers located in eight high-and middle-income countries. 194 The study had strict inclusion and exclusion criteria to ensure that participants were not exposed to environmental factors known to affect fetal growth, and it therefore aimed to reflect optimal fetal growth.
Based on predetermined criteria, the authors concluded that the differences between participants from different countries in measures of skeletal growth (crown-rump length and head circumference) were similar enough to justify pooling the data, and they therefore generated a single universal chart. No information was provided on the differences between countries with respect to measures such as fetal weight estimation and abdominal circumference, which are used in clinical practice to detect FGR and are known to be associated with adverse perinatal outcomes. Interestingly, there were considerable differences in birth weight between infants from different countries, even in this highly selected group of women free from the negative influence of environmental factors known to affect fetal growth. For example, the mean birth weight at term in India was 2.9 kg, which was approximately 600 g lower than the mean birth weight in the UK (3.5 kg). 195 205,206 As demonstrated in Figure 2, the 10th percentile of the Intergrowth-21st chart is significantly lower throughout gestation than most other ultrasound-based standards.
At around the same time, the results of the NICHD growth study (2015) were published. 198 The overall design of this study was similar to that of the Intergrowth-21st study. It was a prospective longitudinal study of 2334 low-risk women from 12 centers in the USA. The authors found substantial differences in fetal weight between different race groups, and therefore developed separate race-specific growth charts for white, black, Hispanic, and Asian women. The 10th percentile of the NICHD growth chart for white women is included as an example in Figure 2.
The WHO fetal growth charts were published in 2017. 197 Similar to the Intergrowth-21st project, this study aimed to develop a prescriptive universal chart to extend the previously published WHO child growth standard 207 to the fetal period. The design of this study was also similar to that of Intergrowth-21st-a prospective longitudinal study of 1387 low-risk women from 10 centers in 10 high-and middle-income countries. Despite this, the results of the WHO study differed from those of Intergrowth-21st in two aspects.
First, the 10th percentile of the WHO chart is considerably higher than the Intergrowth-21st chart, and is in fact almost identical to the 10th percentile of the Hadlock standard ( Figure 2). Second, unlike Intergrowth-21st, the investigators of the WHO study found substantial differences in fetal growth between the various countries, and concluded that "…populations, even under optimal nutritional conditions and environment, vary and that fetal growth varies and should be considered when the WHO fetal growth charts or any growth references are applied". 208 They expressed concern that use of a universal chart carries a risk of misclassification of FGR, [209][210][211] and recommended that their chart should be adjusted in each country to the local population.
The benefit of customized charts remains a matter of debate.
The GROW software incorporates certain factors that are believed to determine fetal growth potential (maternal race, height, weight, parity, and fetal sex) to calculate the predicted optimal (customized) weight at 40 weeks for each individual fetus. 200,201 The customized fetal growth curve is then determined retrospectively, based on a proportionality growth function derived from the ultrasound-based Hadlock standard. 193 The use of customized charts is appealing, especially in the setting of ethnically mixed populations where their use has been shown to decrease over-and underestimation of FGR rates in certain race groups. 212 A large number of studies investigated the association of customized charts with adverse pregnancy outcomes compared with other birth weight-and ultrasound-based charts, with conflicting results. Several studies found that customized charts performed better at predicting stillbirth and adverse neonatal outcomes, 66,201,211,[213][214][215][216] while others found no benefit and concluded that the benefit reported by others is merely because they are based on an ultrasound-based chart and are thus more likely to reflect optimal fetal growth, while the act of customization has a minimal contribution to the stronger association with adverse outcome. [217][218][219] Another criticism is that the GROW approach assumes that all fetuses follow the same growth trajectory (which is derived from the Hadlock chart)-an assumption that may not be true. Finally, it has been suggested that the required adjustment for multiple factors may be too complex for low-resource countries and, in that setting, a simple adjustment to only one factor-mean birth weight at 40 weeks in the local population-is as predictive for adverse perinatal outcomes as the fully customized GROW charts. 220 It may thus be reasonable for care providers to compare the performance of customized growth charts in their population with that of noncustomized charts (as discussed below), especially in regions or countries with a mixed population where the benefit of customization is expected to be greatest.

| How to choose the best chart
The conflicting results and conclusions regarding the growth charts described above have led to an ongoing debate about the best approach (i.e. universal versus customized charts), as well as to considerable confusion among care providers over which chart they should be using in their local population.
The FIGO Safe Motherhood and Newborn Health Committee recently published a position paper on the choice of reference charts for fetal growth and size at birth. 3 In that paper, the committee reviewed in detail the commonly available charts and the available data on their predictive accuracy. The main conclusions were as follows: (1) local or regional charts are likely to be best to identify the 10th percentile of infants at highest risk, given that universal charts such as Intergrowth-21st are likely to under detect SGA fetuses in high-resource countries and, at the same time, over detect SGA in low-and middle-income countries; (2) as an alternative, universal standards such as Intergorwth-21st and WHO may be used with locally adjusted thresholds (e.g. 3rd or 5th percentile in lowor middle-income countries versus 15th or 20th percentile in highincome countries) to avoid under or overdetection of SGA; and (3) when assessing fetal size antenatally by ultrasound, fetal (i.e. ultrasound-based) charts should be used rather than birth weight-based charts. We fully endorse and support these recommendations.
Furthermore, we as well as others, 160,208 believe that the decision on which chart to use can be further based on a comparison of performance of the various charts in the population of interest, using a local data set. This can be achieved by the following approaches: (1) statistical validation: finding the chart that matches best the distribution of fetal weight in low-risk pregnancies in the local population. That is, identifying the chart that when applied to the local population yields weight percentiles that follow a normal distribution centered at approximately the 50th percentile, and identifies approximately 10% of the low-risk population as being below the 10th percentile and above the 90th percentile, and approximately 5% of the population as being below the 5th percentile and above the 95th percentile. An example of this approach is provided in Figure 3; (2) outcome-based validation: finding the chart for which the diagnosis of SGA has the best predictive value for adverse outcomes related to FGR.. 211,221 While this approach seems compelling, interpretation of the predictive value of the different charts for adverse outcomes may be challenging, as there is a trade-off between detection rate and false positive rate for adverse outcomes. 221 Thus, charts that are shifted upward (e.g. Hadlock, WHO) would have a higher detection rate but also a high false-positive rate, while charts that are shifted downward (e.g. Intergrowth-21st) would have a lower false-positive rate but would also have a lower detection rate for SGA fetuses at risk of adverse outcomes (Figure 4). Finding the chart that provides the best balance between these two measures requires careful consideration and should be based on a clear definition of the goals of screening.

| How to assess fetal growth in twin gestations
Twin fetuses grow more slowly than singletons, starting from 28-32 weeks of gestation onward. 222 238 or twin-specific charts. 239 As a result, singleton-based standards are used by default in most centers to assess the growth of twins. However, recent data provide support to the hypothesis that the relative smallness of twins is a benign adaptive mechanism and, therefore, for the use of twin-specific charts. For example, several studies suggest that the slower growth of twins is the result of differences in programming that is determined as early as the first trimester.. 229,[240][241][242][243] In addition, it was found that the use of twin-specific (versus singleton-based) charts was associated with a marked decrease in the rate of twins classified as SGA, without affecting the detection rate of stillbirth, suggesting that twin-specific charts can be used safely. 227,244,245 Similar findings were reported in studies that investigated the association between the type of chart used (twin versus singleton charts) and other outcomes such as perinatal complications and long-term morbidity. 246,247 Studies that investigated placental pathology findings reported that SGA twins (based on singleton charts) are less likely to have placental histopathological evidence of placental insufficiency when compared with SGA singletons. 248,249 In another recent study on the association between SGA and pre-eclampsia, it was found that in contrast to singletons, the diagnosis of SGA in twins based on singleton charts was not associated with a greater risk of pre-eclampsia, while the association of SGA in twins diagnosed using twin-specific charts had the same magnitude of association with pre-eclampsia to that observed between SGA and pre-eclampsia in singletons. 63 Overall, these findings provide support to the hypothesis that the relative smallness of twins is less likely to be the result of placental insufficiency and, thus, less likely to reflect true growth restriction. Based on that, we believe that it seems reasonable to use twin-specific charts for the assessment of fetal growth in twin gestations, as this has the potential to avoid overdiagnosis of There is no evidence to support routine third-trimester ultrasound for the detection of FGR, as this practice has not been shown to be associated with improved perinatal outcomes. We support the recommendation of the FIGO Safe Motherhood and Newborn Health Committee that local or regional growth charts should be preferred over universal charts. 3 Alternatively, universal standards may be used with locally adjusted thresholds to avoid under-or overdetection of FGR.

⊕⊕⊕⊕
⊕⊕⊕ ○ Strong 6. The decision regarding which growth chart to use may be further guided by comparing the performance of the various charts in the population of interest, using local data sets.

⊕⊕ ○○
Weak 7. Based on the available evidence it seems reasonable to use twin-specific charts for the assessment of fetal growth in twin gestations as this has the potential to avoid overdiagnosis of FGR in this population.

⊕⊕ ○○
Weak 8. In twin gestations, the diagnosis of FGR should also take into consideration intertwin size discordance, especially in the case of monochorionic placentation.

| WHAT K IND OF INVE S TI G ATI ON S S HOULD B E PERFORMED WHEN FE TAL G ROW TH RE S TRIC TI ON IS SUS PEC TED?
Once FGR is suspected, a systematic investigation should be per-

| Detailed anatomy scan
Detailed anatomy scan should be routinely performed when FGR is suspected, especially in cases of early-onset severe FGR. The It should be noted that umbilical artery Doppler findings may be normal in the early stages of placental FGR. Therefore, normal umbilical artery Doppler studies do not rule out placental dysfunction, and therefore serial monitoring is recommended in all cases of suspected FGR. 275 deletion, and 7q11.23 deletion. The incremental yield of microarray over karyotyping was even higher at 10% (95% CI, 6%-14%) in the presence of associated fetal malformations. 35 Based on these data, it seems reasonable to offer amniocentesis with karyotype and microarray analysis (when available) to women with FGR, with the decision based on factors such as ultrasound findings, gestational age, lack of evidence of placental dysfunction, and whether the results of the amniocentesis would affect management. These data also suggest that the temptation to substitute amniocentesis by noninvasive prenatal testing (NIPT) using cell-free fetal DNA analysis in this context should be strongly resisted.

Quality of evidence
Strength of recommendation 1. Women with suspected FGR should undergo systematic assessment that includes the following: (1) detailed history; (2) detailed sonographic assessment for structural anomalies, soft markers, and sonographic signs related to fetal infection; (3) Doppler studies that include at least the umbilical artery and, when available, also the uterine and middle cerebral arteries; and (4) maternal screening for relevant congenital infections, which should be focused on cytomegalovirus and toxoplasmosis, but may also include rubella, herpes, syphilis, malaria, and Zika virus in cases at high risk.

LRS The extent of investigation may be limited by available resources. Assessment should include
screening for infections such as malaria and Zika virus in endemic areas.
⊕⊕⊕⊕ Strong 2. Confirmation of gestational age should be the first step when FGR is suspected. With the exception of pregnancies achieved by assisted reproductive technology, first-trimester crown-rump length is the most accurate method to date pregnancy when in the range of 7-60 mm. If more than one scan is performed in the first trimester, the earliest scan with a crown-rump length of at least 10 mm should be used.
LRS In low-resource settings, dating may need to be based on menstrual history or symphysis-fundal height.
⊕⊕⊕⊕ Strong 3. Amniocentesis for karyotype (as well as microarray and polymerase chain reaction for infectious agents when available) should be offered to women with suspected FGR, especially in cases with early-onset severe (estimated fetal weight <3rd percentile) FGR, in the presence of sonographic findings associated with genetic or infectious etiologies, no obvious signs of placental dysfunction, and when the findings are likely to affect management.

LRS
The availability of genetic testing may be limited by available resources. There is no effective antenatal treatment for placental dysfunction and therefore once FGR has been identified, the principal management steps are institution of fetal surveillance and determination of appropriate thresholds for delivery. Perinatal outcome in early-onset FGR is improved when pregnancies are managed in a high-level fetal medicine and neonatology unit utilizing a uniform management protocol. 301 Likewise, the optimal management setting for late-onset FGR is a unit that has access and experience in interpretation of surveillance tests, together with an appropriate level neonatal unit. The recommendations for monitoring, timing and mode of delivery, and potential treatments for placentamediated FGR are described below and summarized in Table 1 and

| Monitoring
The primary goal of fetal monitoring is prevention of stillbirth by

| Fetal movement counting
Fetal activity is established from the first trimester onward and as gestational age advances becomes organized into coordinated behavioral states. Progressive fetal hypoxemia is accompanied by a reduction of fetal activity that can be perceived most accurately by the mother when she is lying down and paying focused attention to fetal movements. 305 Decreased fetal movement is often defined as less than 10 movements in 2 hours during focused maternal counting. 306 Although reports on whether quality improvement tools to promote awareness and management of reduced fetal movements can effectively decrease the risk of stillbirth have been conflicting, most of these interventions were focused on unselected populations rather than in pregnancies with suspected FGR. [307][308][309] Given that fetal movement counting is a simple and inexpensive tool that may provide a safety net between scheduled outpatient monitoring visits, it seems reasonable to use movement counting as an adjunct to monitoring in FGR. In addition to reactivity, FHR patterns display "variability"-the average oscillations in the FHR signal, evaluated in bpm in 1-minute windows. Reduced variability appears later than absent reactivity in the process of progressive fetal hypoxia. It reflects reduced sympathetic-parasympathetic activity, secondary to diminished brainstem oxygenation.
The FHR pattern reflects fetal oxygenation and acid-base status at the time of evaluation but does not predict deterioration TA B L E 1 Recommendations for monitoring, timing, and mode of delivery in cases with suspected fetal growth restriction.  In unselected pregnancies, the rate of stillbirth in the week following a normal modified or five-component BPP is 0.8/1000 (negative predictive rate >99.9%). FGR fetuses show a sequential loss of heart rate reactivity, breathing movements, gross body movement, and tone with decrease in pH. 287,288,301,302 In FGR pregnancies, an abnormal BPP (score of 4 or less) is associated with an umbilical artery pH of less than 7.20, with sensitivity increasing to 100% at a score of 0/10. 301,302,319 The BPP is a more accurate predictor of fetal acid-base status at the time of testing than CTG/NST, with a similar accuracy as cCTG.
Therefore, a five-component BPP can be used to clarify fetal acidbase status when a nonreactive CTG/NST is obtained. The frequency of BPP testing is guided by the same principles as timing of fetal heart rate testing.

| Umbilical artery Doppler
Umbilical artery Doppler is universally recommended for monitoring of FGR because it assesses the hemodynamic aspect of placental dysfunction. 134,143,308,310 It is estimated that approximately one-third of the villous circulation needs to be damaged before a decrease in umbilical artery end-diastolic velocity occurs. Absent or reversed umbilical artery end-diastolic velocity corresponds to malperfusion of 50%-70% of the villous vascular tree. 328 Because elevated villous blood flow resistance is predominantly associated with the placental pathology found in early-onset FGR, umbilical artery Doppler does not reliably predict outcome in late-onset FGR. [329][330][331] The umbilical artery Doppler waveform can be quantified using the pulsatility index, or by visual classification of end-diastolic velocity as absent (AEDV) or reversed (REDV). With increasing degrees of placental blood flow resistance, an abnormal umbilical artery waveform is defined as either having an elevated pulsatility index, AEDV, or REDV. The degree of placental blood flow resistance elevation is the primary factor determining the rate of clinical progression and the associated risk for fetal deterioration and stillbirth in early-onset-FGR. 286,289,291,292 When the umbilical artery pulsatility index is elevated but end-diastolic forward flow is still present, the median time interval to additional surveillance abnormalities is 2 weeks. Once AEDV occurs, cardiovascular deterioration advances after a median of 5 days and the weighted odds ratio for stillbirth is 3.6 (2.3-5.6). 286,291,332 When REDV occurs, the median interval for further fetal deterioration is 2 days and the weighted odds ratio for stillbirth is 7.3 (4.6-11.4). 291,331 In patients with normal umbilical artery Doppler, the recommended frequency to repeat Doppler monitoring ranges from weekly to every other week. However, when AEDV develops, Doppler surveillance is recommended at minimum twice weekly, and for REDV at least three times weekly unless delivery criteria have been met.

| Cerebral artery Doppler
The majority of professional societies now recommend middle cerebral artery Doppler for monitoring in late-onset FGR. Concurrent measurement of the umbilical artery and middle cerebral artery pulsatility index allows calculation of the cerebroplacental Doppler ratio. Both the cerebroplacental ratio and middle cerebral artery pulsatility index decrease as a hemodynamic response to fetal hypoxemia and therefore reflect placental dysfunction, even in those pregnancies where the villous blood flow resistance is not elevated enough to produce an abnormal umbilical artery pulsatility index. Approximately 20% of term SGA fetuses with normal umbilical artery Doppler have a decreased middle cerebral artery pulsatility index, which is associated with a higher rate of cesarean section for intrapartum distress, poor neonatal transition, and adverse developmental outcome. [333][334][335] The cerebroplacental Doppler ratio is more closely related to fetal hypoxia than its individual components, 336 but has a similar predictive accuracy for perinatal death, fetal distress, or poor neonatal transition as the umbilical artery pulsatility index. 337 Cardiovascular deterioration in late-onset FGR is characterized by abnormal cerebral artery Doppler. Therefore, an important role of middle cerebral artery Doppler is to provide an estimate of perinatal risk in patients with normal umbilical artery Doppler. 292,331 Because of the higher risk for adverse outcome within 1 week of a decrease in middle cerebral artery pulsatility index, it is recommended to utilize at least twice weekly surveillance in this setting.

| Ductus venosus Doppler
The few professional societies that recommend ductus venosus Doppler evaluation specify that it should be performed in specialized centers that have expertise in the comprehensive perinatal management of early-onset FGR. 312 The relative forward flow in atrial systole in the ductus venosus decreases with worsening placental function or reduced fetal cardiac function, leading to an increase in the pulsatility index for veins, absent, or reversal of the a-wave. 286,288,291,292,338 Abnormal ductus venosus Doppler is primarily observed in early-onset FGR and can provide an estimate of fetal acid-base balance and the risk of stillbirth. The odds ratio of absent or reversed atrial systolic velocity for an umbilical artery pH less than 7.20 at birth is 4.4 (1.2-17.2). 339,340 The weighted odds ratio of absent or reversed ductus venosus atrial systolic velocity for fetal death is 11.6 (6.3-19.7). 331 Abnormal ductus venosus Doppler also predicts fetal decompensation to an abnormal BPP, reduced variability on cCTG, or stillbirth. In fetuses with elevated ductus venosus pulsatility index for veins but forward flow during atrial systole, the median interval to progressive venous Doppler deterioration can be as short as 2 days. 291 In patients that do not yet meet delivery criteria, ductus venosus Doppler is recommended at minimum twice weekly in patients with AEDV and three times weekly when REDV is observed. 286,291,292,341 When ductus venosus Doppler indices increase as a new finding, the frequency of monitoring needs to be increased further.

| Surveillance strategy
Monitoring in FGR pregnancies is intended to prevent fetal compromise or stillbirth, and the choice of tests and their timing is heavily influenced by gestational age. A robust plan is essential, since expectant management with ongoing monitoring, particularly in the setting of early-onset FGR, can result in a three to five-fold increased stillbirth rate when compared with immediate delivery, depending on the degree of cardiovascular compromise that is tolerated before triggering delivery. 294,342,343 The optimal monitoring frequency in FGR has not been determined due to the varying circumstances of gestational age and severity of FGR. A combination of surveillance modalities is needed to accurately determine fetal acidbase status at the time of testing, as well as allowing anticipation of future deterioration. [289][290][291][292]298,344 The accurate prediction of fetal acid-base status is required to prevent unnecessary intervention and  (Table 1).
When umbilical artery pulsatility index is elevated, weekly Doppler is suggested, and when there is AEDV or REDV, more frequent assessment is recommended (Table 1). In early-onset FGR with AEDV or REDV, the risk of stillbirth increases when the ductus venosus Doppler or the CTG/NST patterns become abnormal. 292,319,340 However, there is currently no evidence that adjusting the timing of monitoring based on ductus venosus Doppler improves outcome. In patients with AEDV the stillbirth rate is 0%-1% when at least once daily CTG/NST, cCTG, or BPP is performed with predefined delivery criteria. 341,342 When monitoring is continued to allow for an increase in the ductus venosus pulsatility index for veins, the stillbirth rate is 2%, and 11% of deliveries occur for abnormal STV, 19% for an abnormal BPP, and 22% for FHR decelerations. 304,342 When monitoring is continued in anticipation of reversal of the ductus venosus a-wave velocity, the stillbirth rate is 4%, and 20% of deliveries occur for abnormal STV, 29% for an abnormal BPP, and 31% for FHR decelerations (Table 1). 342,345 This indicates that with ongoing monitoring the risk of FHR abnormalities or an abnormal BPP requiring delivery cannot be predicted by the ductus venosus Doppler. 319,346 Based on the regional pattern of practice, this indicates that in patients who are admitted for AEDV, the minimum frequency of CTG/NST or BPP should be daily and more frequent with REDV (Table 1). 319

| Timing of delivery
The timing of delivery in FGR is determined by gestational age, severity of FGR, findings of fetal monitoring tests, and maternal factors such as pre-eclampsia (Table 1 and Figure 5). Delivery indications can be considered as absolute if they are independent of gestational age, and relative if the threshold to deliver based on the surveillance findings varies across gestational age. Absolute delivery criteria are findings associated with important health risks to the mother or fetus, and therefore require delivery without consideration of gestational age ( Figure 5). Because patients with ductus venosus Doppler monitoring also had FHR monitoring, safety net delivery criteria based on cCTG were also applied in these groups. These included STV below 2.6 ms irrespective of gestational age and below 3.0 ms from 29 weeks onward. In addition, umbilical artery Doppler findings were utilized as relative delivery criteria from 30 weeks onward for REDV and 32 weeks onward for AEDV. The choice of these thresholds is supported by a recent meta-analysis that found that in undelivered FGR pregnancies, umbilical artery REDV has a 19% stillbirth rate, which exceeds mortality for neonates delivered from 30 weeks onward, while AEDV carries a 6.8% stillbirth risk, which favors delivery due to lower neonatal mortality from 32 weeks onward (Table 1 and Figure 1). 332 The TRUFFLE study demonstrated that a predefined management strategy produces better outcomes than expected in all FGR pregnancies. 343 The primary endpoint was less frequently observed in patients randomized to deliver for late ductus venosus abnormal-  291 in which case the likelihood of the fetus tolerating labor is low and the rate of cesarean section has been reported to be greater than 80%. 351 In addition, labor induction in general is less likely to be successful during the preterm period. 352,353 For these reasons, primary cesarean section is usually the preferred option when delivery is indicated in cases of severe early-onset FGR. 354 In contrast, late-onset FGR is usually less severe and fetal hypoxia or acidosis is less likely to be present at the time when delivery is It is recommended that the placenta is sent for histopathological evaluation after delivery. Ideally this should be done in accordance with the Amsterdam workshop consensus statement. 358 Highquality evaluation of the placental pathology is not only likely to increase the precision of the diagnosis but also provides information on the risks of recurrence. 18,359,360

| Antenatal corticosteroids
The efficacy of antenatal corticosteroids in cases of FGR has been questioned, based on reports of elevated endogenous cortisol levels in this population when compared with normally grown fetuses. [361][362][363][364] In addition, the unique cardiovascular, hormonal, and metabolic  376,377 Despite this, recent data support the efficacy and safety of antenatal corticosteroids in the subgroup of SGA fetuses, 378,379 which should be administered when delivery is anticipated, ideally within 1-7 days before birth. 380 When administered in cases of severe FGR with late Doppler changes, an inpatient setting is advised where the fetus can be closely monitored.
Finally, it is important to recognize that the "improvement" in umbilical artery Doppler that is often seen following administration of antenatal corticosteroids is transient, and is thought to be the result of vasodilation of the fetoplacental arterial tree and increased fetal cardiac output rather than a true decrease in placental resistance. 381 Therefore, these transient changes should not be interpreted as an improvement in fetal status and should not affect the management plan. Of note, the absence of any change in end-diastolic flow in response to antenatal corticosteroids is a concern and predicts subsequent fetal deterioration. 372

| Magnesium sulfate for neuroprotection
Administration of magnesium sulfate to women at risk of preterm birth has been shown to have a neuroprotective role, with a decrease in the risk of perinatal mortality, cerebral palsy, and gross motor dysfunction. 382,383 Possible mechanisms thought to be involved in the beneficial effects of magnesium sulfate include reducing intracellular calcium levels, stabilizing blood pressure, normalizing cerebral blood flow, blocking the effects of excitatory neurotransmitters such as glutamate, and antioxidant and anti-inflammatory effects. 384,385 However, the optimal protocol for the administration of magnesium sulfate for the purpose of neuroprotection remains unclear and available protocols vary with regard to the timing of administration, upper gestational age limit, dose, duration, and need for repeat doses. [386][387][388][389] The observation that term FGR infants have higher cord blood magnesium levels compared with normally grown infants raises the theoretical concern that maternal administration of magnesium sulfate in cases of FGR might result in toxic magnesium levels in the fetus. 390,391 However, there are currently no data on the efficacy and safety of magnesium sulfate in FGR fetuses that can support or refute these theoretical concerns. Therefore, there is currently no evidence in favor or against recommending administration of magnesium sulfate for neuroprotection in women at risk of preterm birth with suspected FGR. 379 We believe that, at the current time, it is reasonable to extrapolate the efficacy of magnesium sulfate to specific subgroups of pregnancies, including those complicated by FGR, especially given that FGR is an independent risk factor for cerebral palsy.

| Treatments under investigation
Several novel therapies aiming to improve poor placentation and uterine blood flow are being explored, some of which are described below. However, there are currently no proven treatments for FGR, and any of the therapies currently under investigation should be evaluated only in an appropriately regulated research setting. 392 Phosphodiesterase type-5 inhibitors, such as sildenafil citrate, potentiate nitric oxide availability, lead to vasodilatation, 393,394 and can improve umbilical artery and middle cerebral artery Doppler. 395 However, in the recently published STRIDER trial, which randomized 135 women with early-onset FGR to 25 mg sildenafil three times daily or placebo, sildenafil did not prolong pregnancy or improve pregnancy outcomes. 396 More recently, a similar randomized trial was halted prematurely due to lack of benefit along with concerns that sildenafil may cause neonatal pulmonary hypertension. 397 Another approach is to target the uteroplacental circulation with maternal vascular endothelial growth factor gene therapy, thereby improving local vasodilatation and angiogenesis. 392 Clinically, vector delivery into the uterine arteries can be achieved with intervention radiology. This approach is currently being investigated in the ongoing EVERREST trial. 398 Protein pump inhibitors have been shown in vitro to decrease sFlt-1 and soluble endoglin and improve markers of endothelial dysfunction. However, in a recent randomized trial involving 120 women with preterm pre-eclampsia, esomeprazole did not improve pregnancy outcomes. 399 Pravastatin has been shown to have anti-inflammatory, antioxidant, and proangiogenic properties. 400,401 However, in a recently published randomized trial of 94 women with early-onset pre-eclampsia, the administration of 40 mg pravastatin daily did not lower maternal sFlt-1 levels or prolong pregnancy when compared with placebo. 402 Other novel potential therapies include nanoparticles and microRNAs that deliver drugs locally to the uterine arterial endothelium or trophoblasts, to improve uterine blood flow and placental function.

| Recommendations
FIGO recommends the following for the management of fetal growth restriction (FGR)

| Infant follow-up
Growth-restricted infants are at increased risk of both immediate and long-term complications, and therefore require closer follow-up than normally grown infants in the first years of life.
Growth-restricted infants have lower survival rates compared with those appropriate for gestational age. 403 Although this may be attributed in part to prematurity that is often associated with FGR, birth weight has been shown to be an independent prognostic factor for neonatal mortality, irrespective of gestational age. 404  FGR infants are also at increased risk of adverse long-term neurodevelopmental outcomes. A systematic review of this topic found that FGR infants are at higher risk of poor neurodevelopmental outcomes measured up to 3 years of age; however, high levels of heterogeneity in primary outcomes were reported in the studies included in the review. 12 Of note, adverse neurodevelopmental outcomes may at least partly be related to coexisting increased prematurity rates. 408 In line with the developmental origins of health and disease hypothesis, FGR has been associated, in both animal and human studies, with an increased risk of future noncommunicable diseases including obesity, diabetes, hypertension, and cardiovascular disease. [409][410][411] The risk is especially high in those infants who experience rapid catch-up growth in the first few years of life. 412,413 The mechanisms underlying these associations are not entirely clear.
However, fetal programming by means of epigenetic changes as well as direct organ damage are thought to play a role. 414 Ongoing studies are investigating the optimal follow-up and prevention strategies to decrease the risk of these complications. 415

| Risk of recurrence based on severity and onset
Most of the data on the risk of recurrence of placenta-mediated complications come from studies evaluating hypertensive complications of pregnancy. In a recent individual patient data meta-analysis of 22 studies, the overall risk of recurrence of hypertensive complications was 21%, and was higher in women who experienced early-onset hypertensive complications. 418 Data on the recurrence of FGR are limited. [419][420][421][422] In a population-based study, the overall recurrence rate of FGR in women who gave birth to an infant with a birth weight below the 10th percentile was 24%, compared with 6% in women without a history of FGR (OR 3.9; 95% CI, 3.7-4.0). The risk of recurrence was related to the severity of FGR, and was nearly six-fold when the infant birth weight was below the 5th percentile (OR 5.7; 95% CI, 5.4-6.0). 423 Thus, couples with FGR in the first pregnancy can be reassured that the overall chance of recurrence in subsequent pregnancies is less than 25%.
However, interpretation of the data is limited by the lack of distinction between constitutionally SGA infants and infants who were truly growth restricted, as much of the association described in that study may be driven by the recurrence of constitutional SGA.
Therefore, counselling regarding the risk of recurrence should be further refined based on the risk factors of the individual patient, severity of FGR as reflected by timing of onset and Doppler findings, the co-presence of pre-eclampsia, and placental histopathological findings.  Table 2. 424-426

| Role of thrombophilia screening
Whether women who experienced placenta-mediated pregnancy complications should be screened for antiphospholipid syndrome is a matter of debate. Although the consensus criteria for antiphospholipid syndrome include premature birth before 34 weeks for severe pre-eclampsia or features consistent with placental insufficiency including birth weight below the 10th percentile, 427 the association of antiphospholipid (aPL) antibodies with these conditions is relatively weak and conflicting, especially for FGR. [428][429][430] In addition, although some care providers recommend treatment with LMWH during pregnancy to women with aPL syndrome and previous preterm birth for placenta-mediated complications, this practice is mostly extrapolated from women with aPL syndrome and recurrent pregnancy loss, where there is some evidence in favor of LMWH. [431][432][433] However, the only trial on LMWH in women with aPL syndrome and prior placenta-related complications (FRUIT trial) found no evidence that LMWH improves outcomes in these cases. 434 Given the above, there is insufficient evidence to justify routine screening for aPL antibodies in women with prior FGR. 435 However, screening for aPL antibodies is recommended in women with a history of thromboembolism or recurrent pregnancy loss (or ≥1 late fetal loss), and may be considered in selected cases of women with a history of severe FGR associated with severe earlyonset pre-eclampsia, when placental examination shows features of severe maternal vascular malperfusion, especially central or multiple areas of villous infarction that are due to multiple spiral artery thromboses.
Management of women already diagnosed with antiphospholipid syndrome based on a history of placenta-mediated complications is also under debate. Based on the evidence from the FRUIT trial described above, some only recommend treatment with aspirin in this setting, 436 while others recommend either surveillance or LMWH during the antepartum and postpartum periods. 437 Based on available evidence we only recommend treatment with aspirin, and suggest that LMWH be considered only in selected cases, such as for women who have experienced recurrent complications despite aspirin treatment (aspirin failure).
The findings are clearer for inherited thrombophilias. Most prospective studies found no significant association between inherited thrombophilia and placenta-mediated complications. [438][439][440][441][442][443] Furthermore, the TIPPS and FRUIT trials found no benefit of LMWH in women with thrombophilia and a history of placenta-mediated pregnancy complications. 444,445 These findings were confirmed by a recent individual patient data meta-analysis that found no benefit of LMWH in decreasing the risk of recurrence of placenta-mediated complications, including in women with thrombophilia. 140 Therefore, there is no indication for routine screening for inherited thrombophilia in women with prior FGR. 446,447 9.3.4 | Preconception counselling and management of future pregnancies Given the considerable risk of recurrence of FGR, efforts should be made to decrease this risk in future pregnancies. 448 Modifiable risk factors for FGR such as smoking or poor nutritional status should be identified as early as possible and managed accordingly, as discussed in section 5.5.1.
There is some evidence that administration of aspirin can reduce the risk of FGR. However, as described in section 5.5.2 most available data focused on the prevention of pre-eclampsia as the primary outcome in women at high risk of pre-eclampsia, with the prevention of FGR being considered a secondary outcome. Data on the prevention of recurrence of FGR in women with a history of FGR are limited. 449 Therefore, some recommend that aspirin should be considered in women with past FGR only if they have risk factors for pre-eclampsia at the time of the next pregnancy. 450 However, we believe that given the safety of aspirin and the overlap in pathogenesis of pre-eclampsia and FGR, it is reasonable to recommend aspirin to women with a history of placenta-mediated FGR in the previous pregnancy, using the same regimen of aspirin used for the prevention of pre-eclampsia. This recommendation is shared by most professional societies. 134 Data on the role of LMWH to prevent recurrence of placenta-  Anecdotal reports of treatment with aspirin, heparin, and IVIG 426,463,470 Abbreviations: CNS, central nervous system; FGR, fetal growth restriction; IVIG, intravenous immunoglobulins.

| Recommendations
FIGO recommends the following for postpartum assessment and counselling for future pregnancies in women with a history of fetal growth restriction (FGR)

Strength of recommendation
1. Growth-restricted infants are at an increased risk of short-and long-term morbidity and should be followed postnatally more closely than normally grown infants. ⊕⊕ ○○ Strong 7. Low-molecular-weight heparin is not recommended for the prevention of FGR in women with a history of placenta-mediated FGR.

⊕⊕ ○○
⊕⊕⊕ ○ Strong 8. In women with antiphospholipid syndrome and a history of placenta-mediated FGR, low-molecularweight heparin may be considered in selected cases, such as in women who have experienced recurrent complications despite aspirin treatment (aspirin failure).

⊕⊕ ○○
Weak 9. Women with a history of FGR should undergo close surveillance of fetal growth starting at 24-28 weeks.

| SUMMARY AND FUTURE RE S E ARCH DIREC TIONS
FGR is an important cause of stillbirth, neonatal mortality, and shortand long-term neonatal morbidity. Early prediction and preventive strategies, timely diagnosis, and management using a standardized protocol to determine the proper monitoring and timing of delivery can decrease the risk of stillbirth and improve perinatal outcomes in pregnancies complicated by FGR.
Future research should focus on the development of new fetal assessment tools that may improve the accuracy of the prediction of fetal deterioration and thus further optimize timing of delivery of FGR fetuses, as well as on novel treatments that may improve placental function in cases of placenta-mediated FGR and thereby deferring delivery in cases of early-onset FGR.