Evaluate a patient at risk of Rhesus iso-immunisation and red-cell alloimmunisation, including anti-D shortages

In one line

Red-cell alloimmunisation is a prevention disease first and a fetal-medicine disease second: stop anti-D sensitisation with correctly dosed, correctly timed immunoglobulin, and once a clinically significant antibody exists, drive surveillance off MCA peak systolic velocity (>1.5 MoM) rather than amniocentesis — escalating to intrauterine transfusion in a fetal-medicine centre.

This chapter assumes the Intermediate groundwork — Rh antigen biology, the mechanics of sensitisation, ABO/Rh inheritance and routine prophylaxis — from Rh iso-immunisation basics; it covers which antibody behaves how and why, the tests that mislead, the named transfusion technique, and the judgement calls (including how to ration anti-D in a shortage and genotype a fetus in an African population).

Why this matters in South Africa

The epidemiology here is not the European epidemiology. The D-negative phenotype runs at roughly 3–7% in Black African populations (versus ~15% in Caucasians), so the number of at-risk pregnancies is lower — but two SA realities make the disease disproportionately dangerous. First, anti-D immunoglobulin is chronically short and expensive in the public sector, so the prophylaxis programme that all but eliminated Rh disease in high-income countries is delivered imperfectly here; established severe disease still presents. Second, the molecular basis of D-negativity differs: in Africans the D-negative phenotype is usually caused not by deletion of the RHD gene (the Caucasian mechanism) but by an intact-but-silenced RHD pseudogene (RHDψ) carrying a 37-bp insert and stop codons. This has a direct clinical consequence developed below — non-invasive fetal RHD genotyping that screens only one exon will misclassify these women, so the assay and its interpretation must be African-aware. The combination — a population with non-deletional D-negativity, an unreliable anti-D supply, and send-away fetal genotyping — is exactly why this is a judgement topic in SA, not a protocol topic.

Pathophysiology — why the antibody, not the titre, decides

Assume the two-hit story from Intermediate (a sensitising bleed primes the maternal immune system; on re-exposure, IgG anti-D crosses the placenta and haemolyses antigen-positive fetal cells). The antibodies classify by mechanism, because each subtype produces a different disease and a different management trigger.

• Anti-D — the classic IgG-mediated extravascular haemolysis. Maternal IgG coats D-positive fetal red cells; the fetal reticuloendothelial system (splenic macrophages) destroys them. Fetal anaemia drives compensatory extramedullary haemopoiesis (hepatosplenomegaly), then high-output cardiac failure and, at the extreme, hydrops fetalis (ascites, effusions, skin oedema, placentomegaly) once the haematocrit falls below roughly a third of normal. Here titre tracks severity reasonably well and the disease worsens with each pregnancy (the anamnestic response), so a prior affected pregnancy is one of the strongest predictors.
• Anti-c and anti-E (Rh system, non-D). Anti-c can cause disease as severe as anti-D and is now a leading cause of severe non-D haemolytic disease; anti-E alone is usually mild but the combination anti-c + anti-E is synergistic and more severe than either alone. These matter because RhD prophylaxis does nothing to prevent them — a woman can be diligently anti-D-protected and still alloimmunise to c or E.
• Anti-Kell (anti-K) — a different disease. Kell glycoprotein is expressed on erythroid progenitor cells (BFU-E/CFU-E), not just mature red cells. Anti-K therefore suppresses erythropoiesis at the progenitor level in addition to haemolysing — it is a marrow-failure picture as much as a haemolytic one. The mechanistic consequences: (1) anaemia is more severe than the maternal titre, the antibody history, or ΔOD450 predicts, because there is little haemolysis to generate bilirubin or reticulocytosis to read; (2) because there is less haemolysis there is less amniotic bilirubin, so the old ΔOD450 method systematically underestimates Kell disease; and (3) fetal anaemia can develop early and fast, and rare case reports describe profound anaemia that even weekly MCA-PSV missed. So in anti-Kell, refer and survey on antibody presence and paternal/fetal Kell status, not on titre.
• The non-haemolytic mimics that present as "anaemia + antibody". A positive antibody screen plus fetal hydrops is not always alloimmune. Non-immune hydrops (parvovirus B19 aplasia, alpha-thalassaemia major / Bart's hydrops, fetal arrhythmia, structural cardiac disease, twin–twin transfusion) produces the same MCA-PSV picture and the same ultrasound, but the antibody is a red herring — see the differential below.

The unifying point: the antibody's specificity, not its concentration, sets the rules. A "low titre" is reassuring in anti-D and meaningless in anti-Kell.

Assessment

Assume the booking-bloods routine from Intermediate. Two questions sharpen at this level: is this antibody capable of causing fetal anaemia, and how severe is it likely to be this pregnancy?

• Screen every pregnancy, not just RhD-negative women. The booking indirect antiglobulin (Coombs) test must look for all clinically significant antibodies — anti-D, anti-c, anti-E, anti-C, anti-e and anti-Kell — because non-D antibodies now cause a large share of severe haemolytic disease.
• Know which antibodies matter. Anti-D, anti-c and anti-Kell cause severe disease; anti-E and most others are usually milder. Anti-Kell behaves differently: it suppresses erythropoiesis as well as haemolysing, so anaemia is more severe than the titre or the ΔOD450 predicts, and titre correlates poorly with severity.
• Titre drives the first decision. A critical titre of 1:16 or higher (or any prior affected pregnancy) mandates fetal-medicine referral. Below that, repeat titres 2–4 weekly in the same laboratory — inter-laboratory drift is a real source of error.
• Determine the fetal antigen. If the antibody is present, establish whether the fetus even carries the target antigen. Cell-free fetal DNA (cffDNA) genotyping from maternal plasma is non-invasive and highly accurate; an antigen-negative fetus needs no surveillance at all. In SA this is send-away, costly and not reimbursed by medical aids — so paternal genotyping (with the caveat of non-paternity/new partner) or specialist referral is the pragmatic fallback.
• Surveillance once at risk = MCA-PSV. From 16 weeks, serial middle cerebral artery peak systolic velocity, plotted in multiples of the median (MoM) for gestation. A value >1.5 MoM predicts moderate-to-severe fetal anaemia and triggers fetal blood sampling ± transfusion. Look also for early hydrops (ascites, skin oedema, effusions) — a late, ominous sign.

The advanced judgement: prior history beats every number

The single most useful prognostic variable is the outcome of the last affected pregnancy, and it reshapes the whole surveillance plan. Alloimmune disease tends to present 2–3 weeks earlier and more severely with each successive pregnancy (the anamnestic response strengthens). So a woman whose previous fetus needed transfusion at 28 weeks should start MCA-PSV surveillance well before that gestation in the index pregnancy — typically from 16–18 weeks — rather than waiting for a critical titre, because in a sensitised multip the titre is no longer a useful trigger. As a rule: titre triages the first sensitised pregnancy; history triages every one after it.

Why cffDNA genotyping needs an African-aware assay

For an unsensitised D-negative woman, fetal RHD genotyping can spare anti-D entirely if the fetus is D-negative; for a sensitised woman it can abolish surveillance if the fetus lacks the target antigen. But the SA molecular reality from the opening section bites here. Because most D-negative Africans carry the silent RHD pseudogene rather than a deleted gene, an assay that detects only one RHD exon will read "RHD present" and call the fetus D-positive when it is in fact D-negative — a false positive that wastes anti-D and prompts needless surveillance. Robust non-invasive RHD genotyping in an African population therefore amplifies multiple exons (commonly exons 5, 7 and 10) so that the pseudogene (exon-7-negative) is distinguished from a truly D-positive fetus. Do not order or interpret cffDNA RHD typing in this population without knowing the assay is multi-exon and pseudogene-aware — a single-exon result is uninterpretable here.

The protective effect you can use to read the history

When counselling on the likelihood of sensitisation, note that ABO incompatibility between mother and fetus is partially protective against Rh sensitisation: incompatible fetal cells leaking into a group-O mother are rapidly cleared by maternal anti-A/anti-B before the D antigen can be processed. Quantitatively, the spontaneous sensitisation risk of a D-positive ABO-compatible fetus (no prophylaxis) is roughly 16%, falling to about 1.5–2% when the pair is ABO-incompatible. This explains the otherwise-puzzling D-negative multip who never sensitised, and it is why the residual sensitisation risk in the era of anti-D is not zero — most third-trimester sensitisation is from silent feto-maternal haemorrhage in the weeks before a missed or under-dosed prophylaxis.

Investigations at depth — and the tests that mislead

• MCA-PSV is excellent but not perfect. Its physiological basis is sound — anaemic blood is less viscous and fetal cardiac output rises, so peak systolic velocity climbs — and >1.5 MoM has near-complete sensitivity for moderate-to-severe anaemia. But the ~12% false-positive rate (Mari) means a single elevated value is not a mandate to needle a fetus; repeat it, correlate with the trend, the gestation and the antibody, and remember the velocity is operator- and angle-dependent (insonate close to a zero angle, sample just distal to the origin of the MCA from the internal carotid, avoid fetal breathing or activity which falsely elevate it). After ~35 weeks the MoM thresholds lose specificity, which is one reason surveillance converts to delivery planning near term rather than continued transfusion.
• MCA-PSV after the first transfusion is a different test. Once a fetus has been transfused with adult donor cells, the relationship between velocity and haematocrit changes (donor blood has different rheology and the fetus's own marrow is suppressed). The threshold for a second IUT is conventionally raised (many centres use ~1.69 MoM) and timing increasingly relies on the predicted rate of haematocrit decline rather than Doppler alone — the substance of the Dodd trial below.
• ΔOD450 and the Liley/Queenan curves — know them, don't use them. Historically, amniotic-fluid bilirubin (the change in optical density at 450 nm) was plotted on Liley's three-zone chart (valid only from ~27 weeks) and later Queenan's four-zone chart (extended to ~14 weeks) to grade severity and time delivery. Doppler retired this for routine surveillance (DIAMOND, below). It still matters for two reasons: it is how disease was monitored before Doppler, and it explains why Kell disease was so lethal in that era — with little haemolysis there is little amniotic bilirubin, so ΔOD450 reads falsely reassuring while the marrow is failing.
• Titre technique matters more than the number. Report titres from the same accredited laboratory using the same (saline/AHG) method, because a one-tube difference between labs is within assay noise and can falsely look like a rising trend. The critical titre (1:16) is a property of the method, not a universal constant.
• At fetal blood sampling, measure, don't estimate. Cordocentesis gives the true fetal haematocrit/haemoglobin and a direct Coombs; it is both the confirmatory test and the access for transfusion in the same sitting, which is why it is only done where you are prepared to transfuse immediately.

Differential diagnosis — the mimics that change management

A raised MCA-PSV or fetal hydrops with a positive antibody screen is presumed alloimmune, but the consultant explicitly excludes the mimics, because their treatment diverges completely:

• Parvovirus B19 — causes fetal anaemia by direct red-cell aplasia (and myocarditis), often producing hydrops that is self-limiting. Maternal serology/PCR; the management is the same transfusion technique but the natural history and recurrence counselling are entirely different (no recurrence risk, no anti-D implication).
• Alpha-thalassaemia major (Hb Bart's hydrops) — in at-risk ancestry, severe fetal anaemia and hydrops from the second trimester; the antibody screen is negative. Recognising it changes counselling toward a genetic, not an alloimmune, frame.
• Non-immune hydrops generally — fetal tachy/bradyarrhythmia, structural cardiac disease, twin–twin transfusion, fetal infection (CMV, toxoplasma, syphilis), chromosomal/syndromic causes. The hint is hydrops with a normal MCA-PSV (the heart, not the blood, is the problem) or a negative/clinically-insignificant antibody.
• A "positive antibody" that is not the cause — naturally occurring or passively acquired antibodies (including the residual passive anti-D after a recent prophylactic dose, which the lab will report and which must not be mistaken for active immunisation). Always ask whether anti-D was recently given before calling a low-titre anti-D "sensitisation".

Getting this right matters because IUT into a non-anaemic fetus is a needless hazard, and missing a treatable alloimmune anaemia while pursuing a wrong mimic is lethal.

Management

Frame the answer immediate (prevention) → ongoing (surveillance) → definitive (fetal therapy & delivery).

Immediate — prevent sensitisation (the unsensitised RhD-negative woman). Anti-D works only before an immune response exists; it is useless once anti-D antibodies are detectable. SA practice (SASOG/NDoH, using Rhesugam 100 µg = 500 IU IMI; note 1 µg = 5 IU):

Sensitising events requiring anti-D: any invasive procedure (amniocentesis, CVS, cordocentesis), antepartum haemorrhage, ECV, abdominal trauma, fetal death, ectopic, molar pregnancy, termination, and miscarriage/evacuation. Route is IMI into deltoid (buttock absorption is unreliable); Rhesugam must never be given IV (anaphylaxis risk).

The dose arithmetic

The standard ≥500 IU dose neutralises roughly 4 mL of fetal red cells (~8 mL whole fetal blood) — i.e. the SA convention is ~125 IU per mL of fetal red cells (500 IU ÷ 4 mL). So when a Kleihauer (or flow cytometry, which is more precise and quantifies FMH down to ~0.1%) estimates a larger bleed, the dose scales: a 12 mL fetal-red-cell bleed needs roughly 12 ÷ 4 × 500 = 1500 IU, given as the standard dose plus the calculated top-up, with a repeat Kleihauer at ~48–72 h to confirm clearance of fetal cells (residual fetal cells mean the dose was insufficient — give more). Large concealed FMH is the classic cause of "correctly given anti-D that still failed": the event that mandates the Kleihauer is precisely the one where one ampoule is not enough (abruption, manual removal of placenta, traumatic delivery, intrauterine death, external cephalic version).

Why the trimester of the event changes the dose

The <20-week / ≥20-week split is mechanistic, not arbitrary: fetal red-cell mass is tiny in the first half of pregnancy, so even a complete feto-maternal bleed is small and ≥250 IU suffices and a Kleihauer is not informative (too few fetal cells to quantify reliably); from 20 weeks the fetal blood volume is large enough that a significant FMH is possible, so the dose rises to ≥500 IU and a Kleihauer becomes worth doing to detect the large bleed that needs scaling. This is why a 1:1 substitution of doses across gestation is wrong.

Ongoing — surveillance of the sensitised pregnancy. Refer to fetal medicine. Serial MCA-PSV from 16 weeks; the velocity rises because anaemic fetal blood is less viscous and cardiac output climbs. Anti-Kell pregnancies are surveilled regardless of titre unless the father (hence fetus) is confirmed Kell-negative.

Definitive — intrauterine transfusion (IUT) and delivery. MCA-PSV >1.5 MoM → fetal blood sampling; if haematocrit is low, transfuse leucodepleted, irradiated, CMV-negative, RhD-negative, antigen-negative, cross-matched packed red cells into the umbilical vein. IUT is a tertiary, fetal-medicine procedure. Delivery timing: mild disease to 38 weeks; disease requiring IUT, deliver at 34–36 weeks (steroids for lung maturity), as the procedural risk eventually exceeds the prematurity risk. Alert neonatology in advance — these babies need cord group/Coombs/FBC/bilirubin and may need phototherapy, exchange transfusion or top-up transfusion for late hyporegenerative anaemia for weeks.

Intrauterine transfusion in depth — the named technique, the numbers, the trade-offs

The procedure must be described in full, not just named.

• Blood product specification. Group O, RhD-negative, and negative for the implicated antigen(s) (e.g. K-negative for anti-Kell); cross-matched against maternal serum; leucodepleted and irradiated (to prevent transfusion-associated graft-versus-host disease in the immunoincompetent fetus); CMV-negative; fresh (<5 days); and packed to a high haematocrit (~0.75–0.85) so a small volume delivers a large red-cell mass into a fetus that cannot tolerate volume overload.
• Route. The intravascular route — needle into the umbilical vein at the cord insertion (or the intrahepatic portion of the umbilical vein) under continuous ultrasound — is standard; the older intraperitoneal route is reserved for very early gestation when the vessel is inaccessible (absorption is slow and unreliable). Fetal paralysis (e.g. a neuromuscular blocker into the fetus) reduces movement during the needle is in the vessel.
• Target and volume. The aim is to raise the fetal haematocrit to roughly 0.40–0.45 without over-transfusing (a hyperviscous post-transfusion haematocrit risks fetal demise). A workable estimate is a transfusion coefficient of ~0.02 × estimated fetal weight (g) of packed cells to raise the haematocrit by ~10 percentage points, scaled to the measured starting haematocrit; the actual volume is confirmed on the post-transfusion sample. In a hydropic fetus, deliberately under-correct on the first procedure (a partial correction, then a second IUT a few days later) because rapid full correction of a chronically anaemic, failing fetal circulation precipitates death.
• The interval lengthens as the marrow shuts down. A practical schedule is the second IUT ~10–14 days after the first, the third ~2 weeks later, then ~3–4 weekly. Each transfusion suppresses fetal erythropoiesis, fetal reticulocytes all but disappear after about two transfusions, and the circulating red-cell mass becomes overwhelmingly transfused (antigen-negative, non-haemolysed) donor blood — so the haematocrit falls more slowly and predictably, which is exactly why post-IUT timing shifts from Doppler toward the predicted decline.
• Outcomes. In experienced fetal-medicine hands, overall fetal survival after IUT exceeds ~90%; survival is high (>90%) in the non-hydropic fetus but falls to roughly 75% with hydrops, and lower still (around half) with severe hydrops — so the priority is getting to the fetus before hydrops develops: surveillance is the life-saving step and IUT is the rescue. Procedure-related loss is in the low single-digit-percent-per-procedure range and accumulates with the number of transfusions, which is why delivery is planned once the prematurity risk drops below the cumulative procedural risk (commonly the last IUT by ~34–35 weeks, delivery at ~34–37 weeks after steroids).

Pharmacological attenuation — the frontier, and where it sits in SA

For the worst recurrent cases — a woman who has already lost fetuses or needed very early IUT — there is now a pharmacological lane, but it is referral-centre, not routine. High-dose IVIG ± plasma exchange before the gestation at which IUT becomes technically feasible (roughly <18–20 weeks, when the umbilical vessel is too small to needle safely) can delay the onset of severe anaemia by reducing maternal antibody and blocking placental IgG transfer. The newer agent is the neonatal Fc receptor (FcRn) blocker nipocalimab, which reduces transplacental IgG transfer (see UNITY/AZALEA below). In the SA public sector none of this is standard — the deliverable is early referral to a fetal-medicine unit so these options are considered before hydrops, not the drugs themselves.

Long-term and neonatal. These infants are not "fixed" by delivery. Because their own marrow was suppressed by transfusion and they carry donor cells, they characteristically develop late hyporegenerative anaemia for weeks to months, needing top-up transfusions and sometimes erythropoietin, on top of the immediate risks of severe hyperbilirubinaemia, kernicterus, phototherapy and exchange transfusion. Mandatory neonatal handover with cord group/Coombs/FBC/bilirubin and a planned haematology follow-up is part of the obstetric plan, not an afterthought.

The evidence & the controversy

The defining management shift was Doppler replacing the needle. Mari's 2000 cohort showed MCA-PSV >1.5 MoM detected moderate-to-severe anaemia with 100% sensitivity (12% false-positive rate); the DIAMOND RCT (Oepkes 2006) then proved MCA-PSV was more sensitive and accurate than serial amniotic-fluid ΔOD450 (88% vs 76% sensitivity). Together they retired routine amniocentesis for surveillance — fewer invasive procedures, fewer iatrogenic sensitising events. The live controversy is the >1.5 MoM threshold after a first transfusion: post-IUT the cutoff loses specificity, and the Dodd 2018 trial showed Doppler is non-inferior to scheduled timing based on expected haematocrit decline — so most centres now time the first IUT by Doppler and subsequent IUTs by a hybrid of Doppler and predicted fall.

For prevention, the postpartum evidence is unequivocal (Cochrane: RR 0.04 for alloimmunisation), but routine antenatal prophylaxis (RAADP) is weaker than its widespread adoption implies — the Cochrane antenatal review (McBain 2015) graded the evidence low quality and found no statistically significant reduction in alloimmunisation, while the larger Turner 2012 meta-analysis did support benefit. This matters acutely in SA, where anti-D is chronically short and expensive. The defensible SA position is explicit rationing (below), not pretending supply is infinite.

The frontier is maternal FcRn blockade. Nipocalimab (UNITY, 2024) reduces transplacental IgG transfer; in 13 high-risk recurrent-HDFN pregnancies, 54% reached live birth at ≥32 weeks without any IUT and there were no cases of hydrops — against a historical benchmark of essentially 0%. This is a single-arm phase 2 result in a tiny, highly selected cohort; the placebo-controlled phase 3 (AZALEA) is the trial to watch before this changes practice. For now it is a referral-centre option for the worst recurrent cases, not standard care.

Reading the RAADP evidence honestly

Why does a programme adopted across the high-income world rest on "low-quality" evidence? Because the postpartum programme had already driven sensitisation so low that an antenatal RCT would need to be enormous to show an additional effect, and the trials that exist were under-powered for the hard outcome (alloimmunisation), powered instead for surrogates (fewer positive Kleihauers, fewer detectable sensitisations at delivery). McBain's Cochrane review is therefore not evidence that RAADP doesn't work — it is evidence that the trials can't resolve a small added benefit on top of an already-excellent postpartum programme. Turner's meta-analysis, adjusting for study quality, does find a reduction. The honest synthesis — and the one that justifies SA's rationing hierarchy — is that postpartum and event-driven anti-D are high-value and non-negotiable, while RAADP is lower-value at the margin and is the first thing to ration when supply is short.

Where the >1.5 MoM threshold genuinely fails

Beyond the post-IUT specificity loss, the threshold has two further failure modes. After ~35 weeks the MoM cut-offs over-call anaemia (more false positives), which is part of why management converts to delivery rather than another transfusion near term. And in anti-Kell, because anaemia is driven by erythroid suppression with a hyperdynamic but not yet hugely anaemic circulation, the velocity can lag the true severity — the rare but real "weekly Doppler that missed it" — which is why Kell pregnancies are surveilled more intensively and referred earlier than the titre alone would suggest.

Landmark trials & key evidence

Exam traps & red flags

• Forgetting non-D antibodies. Anti-c and anti-Kell cause severe disease. Screening only for anti-D, or giving anti-D to an already-sensitised woman, is a classic fail.
• Trusting the titre in anti-Kell disease. Kell suppresses erythropoiesis — anaemia outruns the titre and ΔOD450. Refer on antibody presence, not titre level.
• Reaching for amniocentesis to monitor. Post-DIAMOND, surveillance is MCA-PSV; an invasive test can itself worsen sensitisation.
• Giving anti-D where it cannot work. No benefit once anti-D antibodies are present; always confirm the Coombs is negative before a prophylactic dose.
• Route and timing errors. Rhesugam IV (anaphylaxis), into the buttock (poor absorption), or beyond 72 h without trying up to 10 days — all avoidable.
• Skipping the Kleihauer after large FMH. A large abruption, manual placental removal or caesarean may need more than the standard dose; one ampoule is not always enough. Scale ~125 IU per mL fetal red cells and re-check clearance.
• Mistaking passive anti-D for sensitisation. A low anti-D titre soon after a prophylactic dose is the dose, not immunisation — check whether anti-D was recently given before labelling a woman "sensitised".
• A single-exon cffDNA RHD result in an African population. The RHD pseudogene reads as "RHD present" — a false positive; the assay must be multi-exon (5/7/10) and pseudogene-aware, or the result is uninterpretable.
• Treating a mimic as alloimmune (or vice versa). Hydrops with a normal MCA-PSV or a clinically insignificant antibody is parvovirus, thalassaemia or a cardiac cause until proven otherwise — and missing treatable alloimmune anaemia is lethal.
• Over-correcting a hydropic fetus at the first IUT. Rapid full correction of a chronically anaemic failing circulation kills it — under-correct, then return in days.
• No neonatal handover. These infants risk severe jaundice, kernicterus and late hyporegenerative anaemia for weeks — discharge without paediatric follow-up is dangerous.

Worked viva — how to structure the answer

A typical stem: "A 31-year-old G3P2, both previous babies needed neonatal exchange transfusion, books at 14 weeks. Group O RhD-negative; antibody screen anti-D titre 1:32." A high-scoring answer runs:

1. Frame it and classify the antibody. "This is an established anti-D alloimmunisation in a sensitised multip with a critical titre (≥1:16) and two prior affected pregnancies — so this is high-risk, the disease will likely present earlier and more severely than before, and the titre is no longer my trigger; her history is."
2. Establish whether the fetus is even at risk. "I'd determine the fetal D status — ideally non-invasive cffDNA RHD genotyping, and in our population I'd ensure a multi-exon, pseudogene-aware assay because most D-negative Africans carry the RHD pseudogene; failing that, paternal genotype with the non-paternity caveat. If the fetus is D-negative, no fetal therapy is needed."
3. Set the surveillance plan. "If D-positive, refer to fetal medicine and start serial MCA-PSV early — from 16–18 weeks given the worsening history — looking for >1.5 MoM and early hydrops, not waiting on titres."
4. State the definitive plan and its technique. "A velocity >1.5 MoM prompts fetal blood sampling and, if anaemic, intrauterine intravascular transfusion of O RhD-negative, antigen-negative, irradiated, leucodepleted, CMV-negative, cross-matched packed cells to a haematocrit of ~0.4–0.45 — under-correcting if hydropic — with repeat IUTs at lengthening intervals as the marrow suppresses, then planned delivery at ~34–37 weeks after steroids with neonatology alerted."
5. Justify from evidence. "Mari and DIAMOND for MCA-PSV over amniocentesis; Dodd for timing repeat transfusions by predicted decline; postpartum-anti-D Cochrane for prophylaxis value."
6. Close the loop. "Neonatal handover for jaundice/exchange and late hyporegenerative anaemia, and counselling that disease recurs and worsens — for the worst recurrent cases, early referral so IVIG/plasma exchange or FcRn blockade can be considered before hydrops."

Anti-D in a shortage (the SA reality)

When anti-D stock is constrained, ration by impact on sensitisation prevention, in this order (SASOG/NDoH):

1. Withhold RAADP first — its marginal benefit is the weakest (see McBain).
2. Withhold postpartum anti-D only if the family is complete and sterilisation is done, after full informed consent.
3. Restrict miscarriage prophylaxis to ≥12 weeks of gestation.

Always minimise iatrogenic FMH: avoid manual removal of the placenta, let the placental cord side bleed freely, match blood carefully to avoid transfusing RhD-positive products, and determine the baby's RhD status at birth so anti-D is never wasted on the mother of an RhD-negative infant. If no stock exists for a woman with a clear indication, document it, counsel her, and arrange an indirect Coombs at 3 months to detect seroconversion.

The reasoning behind the hierarchy is the evidence above: postpartum and event-driven anti-D are high-value (Crowther RR 0.04) and protect the next pregnancy directly; RAADP buys a small marginal reduction on top of an already-good programme (McBain low-quality), so it is the rational first cut. The single most stock-sparing intervention is fetal/neonatal RHD typing — every D-negative infant correctly identified saves a maternal dose, and where a multi-exon-validated cffDNA pathway exists, antenatal fetal typing can spare anti-D in the ~40% of these pregnancies carrying a D-negative fetus.

Evidence anchors

• Mari G, et al. Noninvasive diagnosis by Doppler ultrasonography of fetal anemia. N Engl J Med 2000;342:9–14
• Oepkes D, et al. (DIAMOND) Doppler ultrasonography versus amniocentesis to predict fetal anemia. N Engl J Med 2006;355:156–164
• Dodd JM, et al. Fetal MCA Doppler to time intrauterine transfusion: a randomized trial. Ultrasound Obstet Gynecol 2018;51:306–312
• Vaughan JI, et al. Inhibition of erythroid progenitor cells by anti-Kell antibodies in fetal alloimmune anemia. N Engl J Med 1998;338:798–803
• Crowther CA, Middleton P. Anti-D administration after childbirth for preventing Rhesus alloimmunisation. Cochrane Database Syst Rev 1997 (CD000021)
• McBain RD, Crowther CA, Middleton P. Anti-D administration in pregnancy for preventing Rhesus alloimmunisation. Cochrane Database Syst Rev 2015;9:CD000020
• Turner RM, et al. Routine antenatal anti-D prophylaxis: meta-analyses. PLoS One 2012;7:e30711
• Moise KJ, et al. (UNITY) Nipocalimab in early-onset severe HDFN. N Engl J Med 2024;391:526–537
• Komatsu Y, et al. AZALEA phase 3 nipocalimab trial design. Am J Perinatol 2024;42:842–853
• SASOG/BetterObs "Rh Disease and Alloimmunisation 2.0" (v2, 2022): SA anti-D dosing (250 IU <20 wk, 500 IU ≥20 wk, RAADP 1500 IU at 28–30 wk), Rhesugam 100 µg = 500 IU IMI, Kleihauer for FMH, critical titre 1:16, MCA-PSV from 16 wk, and the shortage-rationing hierarchy.