Diagnose and manage fetal growth restriction, including Doppler, timing and route of delivery

In one line

Fetal growth restriction (FGR) is the failure of a fetus to reach its genetically determined growth potential, almost always from placental insufficiency — and the entire management problem is a single trade-off: balancing the rising risk of intrauterine demise against the iatrogenic harms of prematurity, arbitrated by a Doppler-driven surveillance ladder and a gestation-specific delivery threshold.

This chapter assumes the Intermediate groundwork on growth-restriction pathophysiology and SFH screening and the macrosomia counterpart; it does not re-teach what SGA is or how a basic umbilical-artery Doppler waveform is obtained. The Final-level work is to phenotype the small fetus, read the deterioration sequence, and time delivery against a moving target — so the words below are spent on the subtype distinctions, the staging systems, the appraisal of the timing trials, and the consultant judgement calls.

Assessment

FGR is not the same as small-for-gestational-age (SGA), and the most consequential cognitive error at consultant level is treating "small" as a single disease. There are at least four populations under the <10th-centile umbrella, each with a different driver and a different management consequence:

• Constitutionally small (true "SGA"). Normal placenta, normal Dopplers, appropriate interval growth, no perinatal-risk signal. The error is over-delivering this fetus — iatrogenic prematurity for a healthy baby.
• Early-onset placental FGR (<32 weeks). The "respiratory failure of the placenta" phenotype: severe placental disease, steeply escalating UA→DV Doppler abnormality, a strong association with pre-eclampsia, and high mortality and morbidity. Prevalence ~1%. This is the phenotype that follows the classic deterioration cascade and forces a viability-versus-prematurity decision.
• Late-onset placental FGR (≥32 weeks). Mild placental disease, a normal or near-normal umbilical artery (only an extensive lesion lifts the UA-PI), and the abnormality declared instead by the middle cerebral artery / cerebroplacental ratio — "brain-sparing" without UA derangement. Far commoner (~60% of FGR), lower mortality, but it accounts for a large share of unexplained term stillbirth precisely because the UA looks reassuring and the clinician is falsely reassured.
• Non-placental smallness. Aneuploidy, early CMV/malaria, structural anomaly, or simply wrong dates. The signature is a symmetrically small fetus with normal Doppler and often normal liquor — the placenta is innocent and the work-up is genetic/infective, not a delivery-timing problem.

The mechanism→consequence link that drives surveillance: in early FGR the UA is the index test because gross placental loss is what drives it, so the UA leads the deterioration sequence; in late FGR the placental reserve loss is too subtle to move the UA, so the CPR (more sensitive to hypoxia than either vessel alone) and the MCA become the discriminators, and a normal UA must never be read as "reassured."

The diagnostic definition — and why it is built the way it is

• Diagnosis (Delphi consensus, Gordijn 2016). Early FGR (<32 weeks): a solitary criterion — AC or EFW <3rd centile, or absent end-diastolic flow (AEDF) in the umbilical artery — or AC/EFW <10th centile combined with UA-PI or uterine-artery PI >95th centile. Late FGR (≥32 weeks): AC/EFW <3rd centile alone, or AC/EFW <10th centile with either CPR <5th centile or UA-PI >95th centile, or AC/EFW crossing >2 quartiles on serial charts. The architecture is deliberate: a <3rd-centile fetus is defined as restricted (the fixed low centile carries an adverse-outcome signal on its own — an isolated EFW <3rd centile predicts adverse perinatal outcome irrespective of Doppler), whereas a 4th–10th-centile fetus is "small enough to be suspicious but not enough to be certain" and so requires Doppler or interval-growth corroboration before it earns the FGR label.
• The customised-versus-population chart controversy. Population charts call constitutionally small-but-healthy fetuses "abnormal" and miss the large baby who has dropped below his own potential; customised/GROW charts (adjusting for maternal height, weight, parity, ethnicity) reclassify a meaningful fraction and improve the correlation with stillbirth, but the GAP/DESiGN cluster-randomised evidence did not show that mandating customised charts reduces stillbirth at a programme level, so this remains a tool that sharpens individual judgement rather than a proven population intervention. State the limitation; do not present customised charts as a settled win.
• Confirm dating first. A wrong LMP manufactures false FGR; first-trimester CRL trumps a late scan. An undated fetus presenting late and small is a diagnostic trap — you cannot distinguish "small" from "younger than you think" without a growth interval.

History and the SA risk landscape

• History/risk-stratify. Prior FGR/stillbirth/pre-eclampsia, chronic hypertension, renal/SLE/APS disease, smoking, multiple pregnancy, antepartum haemorrhage. In SA, add HIV (and the PVT cascade), TB and under-nutrition as population drivers — and remember the antiretroviral nuance, that older protease-inhibitor regimens were associated with growth restriction whereas the current dolutegravir-based first-line is not, so the modern SA HIV cohort's FGR is more often the placenta than the drug.

Investigations and what each one is actually telling you

• EFW + AC on customised or population charts; liquor volume (oligohydramnios = redistribution of cardiac output away from the kidneys, i.e. a decompensation marker, not a primary diagnostic criterion — its inclusion in management protocols is, on the meta-analytic evidence, of doubtful independent value); umbilical artery Doppler (the index test that changes outcomes); MCA-PI and CPR for the late/term phenotype; ductus venosus (DV) for early severe disease; uterine artery PI to confirm the placental aetiology. Screen for the cause: anomaly scan, and where indicated karyotype/microarray and CMV/malaria serology — a structurally abnormal or symmetrically tiny fetus with normal Doppler is more likely aneuploid or infective than placental.
• Read the venous Dopplers as a hierarchy, not a checklist. The deterioration sequence in early FGR is stereotyped: raised UA-PI → absent end-diastolic flow → reversed end-diastolic flow (arterial/placental side) running in parallel with CPR fall and MCA brain-sparing, then aortic isthmus reversal (the bridge between arterial and venous compromise — it precedes DV abnormality by roughly a week but does not predict mortality better than the DV, so it is physiologically informative but not the trigger), and finally the venous signs of cardiac decompensation: DV-PI >95th centile → absent then reversed a-wave, with UV pulsations and a falling cCTG short-term variation as the terminal events. The DV is the single strongest predictor of short-term risk of fetal death; it precedes the loss of cCTG short-term variability in ~50% of cases and precedes a biophysical-profile deterioration by 48–72 h in ~90%.

Management

Structure management as immediate → ongoing → long-term.

Immediate (at diagnosis). Establish phenotype (early vs late) and aetiology. Refer SGA-with-abnormal-Doppler from district to a regional/tertiary unit with neonatal capacity — the single most consequential SA decision, because a 28-week REDF fetus delivered at a district hospital without ventilation will die. If delivery before 34 weeks is foreseeable, give a course of betamethasone 12 mg IM × 2, 24 h apart (dexamethasone 6 mg IM × 4 12-hourly is the SAMF/NDoH default), and magnesium sulphate for neuroprotection if delivery <32 weeks is imminent. A practical sequencing point: an abnormal DV is considered sufficient to recommend delivery at any gestation after the steroid course is complete — so in a stage III–IV fetus you start steroids the moment the DV turns, because the 48-hour maturation window is the rate-limiting step, not the decision to deliver.

Ongoing (surveillance ladder). Doppler dictates frequency and triggers delivery:

Below ~32 weeks the ductus venosus is the discriminator: abnormal DV (PI >95th centile or absent/reversed a-wave) signals myocardial decompensation and is the TRUFFLE-validated trigger, with a computerised-CTG short-term-variation floor as the safety net. Above 32 weeks the umbilical artery has usually already declared itself and DV adds little.

The stage-based protocol (Figueras–Gratacós / "Barcelona")

The surveillance table above maps onto a formal staging system that collapses the bewildering number of Doppler permutations into five action-linked groups and supplies the language of "stage." It assumes each waveform can already be obtained and interpreted — staging governs the decision, not the measurement.

Two consequences flow from the staging that change practice. First, mode of delivery escalates with stage: at stage II the risk of intrapartum non-reassuring fetal status forcing an emergency caesarean already exceeds ~50%, so an elective caesarean becomes the reasonable default; at stages III–IV elective caesarean is effectively mandatory because these fetuses will not tolerate the contraction-driven placental compression of labour. Second, the viability floor is explicit: below ~26–28 weeks intact survival is under 50%, so a stage IV fetus at 25 weeks is a multidisciplinary counselling conversation (neonatology + parents) about whether any intervention is in the baby's interest — not an automatic caesarean. This is the formalisation of "do not deliver early FGR on the umbilical artery alone."

Route of delivery. AEDF/REDF tolerate labour poorly — chronic hypoxaemia plus contraction-driven placental compression precipitates pathological CTGs and emergency caesarean. Reversed EDF, or any FGR with abnormal DV, is effectively a caesarean indication. With merely elevated UA-PI and a favourable cervix, induction with continuous monitoring is reasonable but carries a high crash-section conversion rate, which you must consent for explicitly. A late-FGR fetus with brain-sparing (low CPR) but positive EDF can usually be induced, but the brain-sparing itself predicts intolerance of labour and a higher emergency-caesarean rate, so set up continuous monitoring and a low threshold from the start.

Intrapartum nuance. Continuous CTG is mandatory; the growth-restricted fetus has minimal reserve and decompensates fast. Avoid prostaglandin hyperstimulation. The combination of meconium, oligohydramnios and a chronically hypoxic fetus makes intrapartum acidosis and meconium aspiration more likely, so anticipate a neonatal team at delivery.

Long-term. Document a recurrence-risk plan: aspirin 150 mg nocte from <16 weeks in the next pregnancy, early-pregnancy uterine-artery Doppler, smoking cessation, and inter-pregnancy optimisation of hypertension/renal/autoimmune disease. There is no effective in-utero treatment that improves growth — bed rest, oxygen, nutrient supplementation and sildenafil have all failed or harmed.

Differential diagnosis — the mimics that change management

The differential is not academic; each mimic redirects the work-up away from a delivery-timing problem:

• Constitutional SGA — the commonest "mimic": normal Doppler, appropriate interval growth, normal liquor → surveillance, deliver at term, do not induce prematurity.
• Wrong dates — an undated late presenter; resolve with a growth interval before labelling FGR.
• Aneuploidy (especially triploidy and trisomy 18) — early-onset, often symmetric, frequently with structural markers and normal or only mildly abnormal Dopplers; offer karyotype/microarray before attributing to the placenta.
• Congenital infection — CMV (and, in the SA setting, malaria, syphilis and TB) cause symmetric smallness, sometimes with intracranial calcification, ventriculomegaly or hepatosplenomegaly; serology/PCR, not delivery timing, is the answer.
• Pre-eclampsia–FGR overlap — early FGR and early pre-eclampsia are the same placental disease wearing two faces; finding one mandates screening for the other (BP, proteinuria, bloods), because the maternal syndrome can force delivery before the fetal indication does. Cross-link: early-onset pre-eclampsia with severe features.

The evidence & the controversy

The defining FGR controversy is timing of delivery in early disease, and two trials bracket it. GRIT showed that when the obstetrician was genuinely uncertain, delaying delivery did not change death-or-disability at 2 years — so reflexive early delivery buys nothing and costs prematurity. TRUFFLE then asked how to time the inevitable delivery and found that waiting for late DV changes (rather than reacting to reduced cCTG variation) yielded more 2-year survivors free of neuroimpairment (95% vs 85%), at the cost of a non-significant rise in mortality. The synthesis: before 32 weeks, monitor with DV + cCTG and do not deliver on umbilical-artery Doppler alone — but hold a cCTG safety net, because the gain came with a mortality signal that cannot be ignored.

The appraisal point: TRUFFLE did not randomise "deliver now versus wait" — it randomised the trigger. Women were allocated to one of three indications for delivery: reduced cCTG short-term variation, early DV change (PI >95th centile), or late DV change (a-wave at or below baseline), each with the cCTG as a hard safety floor. The cCTG-STV floor thresholds were operationalised as <3.5 ms at 26+0–28+6 weeks and <4.0 ms at 29+0–31+6 weeks. The "late-DV" arm won on intact survival, but the trial only worked because every arm had the cCTG backstop — so the lesson is not "wait for the a-wave to reverse and ignore everything else," it is "let the DV buy you time underneath a computerised-CTG safety net."

At term the question inverts. DIGITAT randomised suspected FGR ≥36 weeks to induction vs expectant management and found equivalence in adverse neonatal outcome and caesarean rate — so neither is wrong, and you individualise. The pragmatic reading is to induce by ~37–38 weeks once growth restriction is confirmed at term, because expectant management leaves a small but real stillbirth tail with no offsetting benefit. The long-term DIGITAT follow-up is reassuring on the point that worries parents — induced and expectantly managed children showed equivalent neurodevelopment and behaviour at 2 years — so you can counsel induction without implying it harms the brain.

The therapeutic graveyard is instructive. Maternal sildenafil was a biologically plausible vasodilator; the UK STRIDER trial showed no benefit, and the Dutch STRIDER trial was halted when sildenafil-exposed neonates developed persistent pulmonary hypertension (27% vs 5%) with 11 neonatal deaths. This is the cautionary tale that plausible-but-unproven antenatal therapy can kill the baby it aims to save, and it governs the appraisal of any novel FGR treatment. Aspirin prevents placental dysfunction prospectively but does nothing once FGR is established; do not start it as treatment. The ASPRE trial defined the prophylactic case: 150 mg of aspirin from 11–14 weeks in screen-positive women cut preterm pre-eclampsia by ~62% with a number-needed-to-treat of 39 (95% CI 23–100) — and because early FGR and early pre-eclampsia share the placental lesion, the same prophylaxis is what you offer the next pregnancy after a placental FGR. The dose, timing and threshold matter: ≥150 mg, started before 16 weeks, taken at night (the chronotherapy signal), and stopped near term.

How the guidelines actually disagree

There is no single universal delivery threshold. The major bodies converge on the direction and diverge on the exact week:

• For AEDF, the SMFM/ISUOG-aligned reading is delivery around 32 weeks (some protocols 32–34); for REDF, around 30 weeks (some 30–32). The Barcelona stage III threshold sits at 30 weeks for either REDF or DV-PI >95th — slightly more conservative on REDF than the "deliver at AEDF=34" simplification.
• ISUOG and SMFM both anchor early-FGR timing on the DV + cCTG combination (the TRUFFLE position); the divergence is mostly in how aggressively they act on isolated AEDF before 32 weeks.
• The pragmatic SA reconciliation: the number matters less than the system — what kills the SA FGR fetus is not a two-week difference in the threshold but being in the wrong hospital when the threshold is reached. Get the woman to neonatal-capable care while the DV still has an a-wave.

The South African evidence — and why it is world-leading

The SA-specific evidence is genuinely world-leading. Symphysis-fundal height misses most growth-restricted fetuses (antenatal detection of the at-risk SGA fetus falls short by up to ~75% in some series, and SFH alone detects only a minority), and the South African-developed Umbiflow continuous-wave umbilical-artery Doppler, deployed by primary-care nurses at 28–34 weeks, cut stillbirths by ~43% (10.1 vs 17.8/1000) in a 6536-woman programme — a low-cost, task-shifted screen tailored exactly to the district-level reality where most of SA's ~16 000 annual stillbirths occur. The mechanism explains the gain: it converts an unselected low-risk population (where most stillbirths actually happen, not in the high-risk clinic) into a triaged one, pulling the abnormal-UA fetus into referral before demise — task-shifting the index test, not the expert. It is the practical answer to reducing FGR stillbirths in a resource-limited setting.

Landmark trials & key evidence

Worked viva — how to structure the answer

A stem might read: "a 27-year-old, 29 weeks, EFW <3rd centile, umbilical-artery reversed end-diastolic flow, ductus venosus PI just above the 95th centile; she is at a district hospital." A high-scoring answer runs:

1. Phenotype it — "This is early-onset (placental) FGR with advanced arterial deterioration: REDF places her at Barcelona stage III, and the rising DV-PI signals early venous/cardiac decompensation."
2. Make the system-level call first — "The single most important step is in-utero transfer to a neonatal-ICU–capable unit now; a 29-week stage III fetus delivered at a district hospital without ventilation will die. I would not deliver locally."
3. Buy maturity safely — "Start a betamethasone course immediately, because the 48-hour window is the rate-limiting step, and give magnesium sulphate for neuroprotection as delivery before 32 weeks is likely. Admit for inpatient daily DV + computerised-CTG surveillance with a cCTG short-term-variation floor."
4. Set the delivery trigger and route — "I would deliver from ~30 weeks for stage III, or sooner if the DV a-wave reverses, cCTG STV falls below the gestation-specific floor, or spontaneous decelerations appear — i.e. progression to stage IV. Mode is elective caesarean; this fetus will not tolerate labour."
5. Justify from evidence — "TRUFFLE supports monitoring with DV plus a cCTG safety net rather than delivering on the umbilical artery alone; GRIT cautions against reflexive early delivery; there is no in-utero treatment — sildenafil is harmful (Dutch STRIDER)."
6. Screen the mother and close the loop — "Screen for pre-eclampsia (same placental disease); counsel ~recurrence risk; aspirin 150 mg before 16 weeks next pregnancy with early uterine-artery Doppler; document the neonatal-outcome conversation honestly given the gestation."

Exam traps & red flags

• Calling SGA "FGR." A <10th-centile fetus with normal UA Doppler, normal liquor and appropriate interval growth is usually constitutionally small — over-delivering it causes iatrogenic prematurity. Conversely a "5th-centile" fetus with REDF is an emergency, not a number.
• Being reassured by a normal umbilical artery in late FGR. After 32 weeks the UA may stay normal until the placenta is grossly compromised; the late-FGR fetus declares itself through the CPR/MCA (brain-sparing), and a normal UA is exactly how the unexplained-term-stillbirth FGR fetus slips through. Measure the CPR.
• Delivering early FGR on umbilical artery Doppler alone. TRUFFLE's lesson: below 32 weeks the ductus venosus (with a cCTG floor) times delivery — AEDF alone before viability-plus is not an automatic delivery order.
• Forgetting that an abnormal DV warrants delivery after steroids, and starting steroids too late. The 48-hour maturation window is the rate-limiting step; start betamethasone the moment the DV turns, not when you finally decide to deliver.
• Allowing AEDF/REDF a trial of labour. These fetuses decompensate with contractions; REDF (and any abnormal DV) is a caesarean.
• Missing the non-placental causes. A symmetrically tiny early fetus with normal Doppler — think aneuploidy, CMV, malaria, or wrong dates — and karyotype/serology before you blame the placenta.
• Forgetting the floor of viability and the steroid/MgSO₄ window. A 25-week REDF fetus at a district hospital needs in-utero transfer now, plus antenatal steroids and magnesium — but a 25-week stage IV fetus is also a multidisciplinary counselling conversation about whether intervention serves the baby.
• Offering "treatment." Bed rest, oxygen, nutritional supplements and sildenafil do not work; sildenafil is harmful. The only interventions are surveillance, timely delivery, and (prospectively, next pregnancy) aspirin.
• Treating oligohydramnios as a primary diagnostic criterion. It is a decompensation marker (cardiac-output redistribution); acting on amniotic fluid index alone increases inductions and caesareans without a proven outcome benefit.

Evidence anchors

• ISUOG Practice Guidelines: SGA and FGR (Lees, 2020) — diagnosis, Doppler surveillance, delivery thresholds.
• SMFM Consult Series #52 / ISUOG–SMFM comparison (Abuhamad, 2021) — delivery timing by UA-Doppler category (AEDF 32–34 wk; REDF 30–32 wk).
• Delphi consensus definition of FGR (Gordijn, 2016)
• Figueras & Gratacós — an integrated, stage-based approach to FGR (Best Pract Res Clin Obstet Gynaecol 2017)
• TRUFFLE (Lees, 2015) · GRIT 2-yr (Thornton, 2004) · DIGITAT (Boers, 2010)
• Cochrane fetal & umbilical Doppler in high-risk pregnancy (Alfirevic, 2017)
• UK STRIDER sildenafil trial (Sharp, 2018)
• ASPRE — aspirin for preterm pre-eclampsia prevention (Rolnik, NEJM 2017)
• Umbiflow CW-Doppler stillbirth-prevention programme, South Africa (Hlongwane, 2022)