Evaluate and manage the couple with recurrent pregnancy loss, applying current ESHRE, RCOG and ASRM guidance

In one line

Recurrent pregnancy loss is a clinical problem of probability, not pathology — the commonest cause of any single loss is embryonic aneuploidy and roughly half of couples have no cause found at all, so the consultant's job is to run a short, evidence-based work-up that finds the few genuinely treatable conditions (antiphospholipid syndrome above all), to not offer the long list of unproven and harmful "treatments" the field is littered with, and to give the couple an honest, individualised prognosis with supportive early-pregnancy care.

Much of the management here is in knowing what not to do. The Intermediate groundwork on a single sporadic miscarriage — its mechanism, expectant/medical/surgical management — is assumed; revise it at recurrent pregnancy loss basics and spontaneous miscarriage. This chapter defends a work-up and a treatment plan from the primary literature, against guidelines that do not fully agree even on the definition.

Mechanism & pathophysiology

Each individual miscarriage is overwhelmingly a chromosomal event in the embryo, not a maternal disease. Around half of all clinically recognised miscarriages carry a fetal chromosomal abnormality — usually a sporadic autosomal trisomy, monosomy X or triploidy arising from a non-disjunction error at meiosis — and the proportion rises steeply with maternal age as oocyte aneuploidy accumulates. This is why recurrent loss is, for most couples, recurrent bad luck: independent aneuploid events stacking up by chance, their probability set largely by maternal age. It is also why a karyotypically normal (euploid) loss is more informative than an aneuploid one — a euploid miscarriage is the one more likely to reflect a maternal or uterine factor you can act on.

Layered on top of that baseline are the minority of genuinely recurrent mechanisms, and they are worth understanding mechanistically because the mechanism dictates whether treatment helps.

Antiphospholipid syndrome (APS) is the one acquired cause with a proven, mechanistic treatment effect. Antiphospholipid antibodies — lupus anticoagulant, anticardiolipin and anti-β2-glycoprotein-I — bind β2-glycoprotein-I displayed on the syncytiotrophoblast surface and at the maternal–fetal interface. The injury is not simply intervillous thrombosis (the old "placental infarction" model is incomplete). The antibodies impair trophoblast proliferation, invasion and hCG secretion, and they activate complement (C5a) and a local inflammatory placentopathy that disrupts spiral-artery remodelling. This matters for the treatment argument: heparin's benefit in obstetric APS is plausibly as much anti-complement and anti-inflammatory at the trophoblast as it is anticoagulant — which is why prophylactic, not therapeutic, dosing is the obstetric standard, and why the drug that "thins blood" helps a disease that is not primarily about clots. The full pharmacology and the thrombotic-versus-obstetric phenotype split are developed in sle-and-antiphospholipid-syndrome; APS is the cause you must not miss because it is the cause you can treat.

The inherited-thrombophilia story is the instructive opposite. Factor V Leiden, prothrombin G20210A, protein C/S and antithrombin deficiency have a weak and largely epidemiological association with late loss, and the mechanistic logic ("maternal hypercoagulability → placental thrombosis → loss") was always shakier than it looked, because early loss is an implantation/embryonic event, not a uteroplacental thrombotic one. The field spent fifteen years anticoagulating these women on that logic — and the randomised evidence (ALIFE2, below) has now shown it does not work. The pathophysiology lesson is that a plausible mechanism is not evidence of a treatable one.

Genetic causes split into two layers. The common layer is the sporadic embryonic aneuploidy above. The rare-but-actionable layer is a parental structural chromosome rearrangement — most often a balanced reciprocal or Robertsonian translocation carried by one partner (found in ~2–5% of RPL couples). The carrier is phenotypically normal, but at meiosis produces a high proportion of unbalanced gametes, generating embryos with partial trisomies/monosomies that miscarry. This is the one genetic finding that changes counselling: it gives a recurrence mechanism and opens a specific intervention (genetic counselling ± pre-implantation genetic testing for structural rearrangements, PGT-SR).

Uterine factors act by distorting implantation and placentation. The septate uterus — a failure of resorption of the midline müllerian septum, leaving a fibrous, poorly vascularised septum — is the canonical anatomical association with mid-trimester loss, although (TRUST, below) its treatment is now contested. Submucous fibroids and intrauterine adhesions (Asherman) similarly compromise the implantation site. Cervical insufficiency is a distinct entity causing painless mid-trimester loss/extreme preterm birth and is mechanistically and managerially separate from first-trimester RPL.

Endocrine and emerging mechanisms round it out: overt hypothyroidism and poorly controlled diabetes are biologically plausible and treatable; chronic endometritis — persistent low-grade endometrial inflammation, often with plasma-cell infiltrate on CD138 staining — is an emerging association where the causal and treatment evidence is still maturing. Each of these is a candidate for the work-up, but the honest mechanistic summary is that for the majority of couples no single mechanism is found, and the dominant driver across the population is age-related embryonic aneuploidy.

Assessment

The assessment has two jobs that pull in opposite directions: find the treatable minority, and avoid over-investigating a couple whose losses are stochastic. A focused, guideline-anchored work-up does both. The first decision is when to investigate at all, and this is where the guidelines openly disagree (developed under Guidelines compared).

History. Number, gestation and nature of each loss (first- vs second-trimester; the pattern matters — late losses point harder at APS, cervix and uterine factors); whether any products were karyotyped; menstrual and subfertility history; thyroid, diabetes and autoimmune symptoms; thrombotic and obstetric history (VTE, prior pre-eclampsia/FGR/stillbirth — the obstetric-APS clinical criteria); drug, alcohol, smoking and weight history; consanguinity and family history of losses or translocations; and the couple's own emotional trajectory, because the intervention that most reliably "works" is supportive care.

Examination. General (BMI, signs of thyroid or autoimmune disease, hirsutism/acanthosis suggesting PMOS) and pelvic.

Investigations — and what each result means:

• Antiphospholipid panel is the load-bearing test: lupus anticoagulant (LAC), anticardiolipin IgG/IgM, anti-β2-glycoprotein-I IgG/IgM. A single positive does not diagnose APS — the assays shift transiently with pregnancy and thrombosis, so classification demands a repeat positive ≥12 weeks apart. The LAC is a functional clotting assay (dRVVT) and is uninterpretable once a woman is anticoagulated, so draw it before any heparin starts. Persistent positivity + the relevant obstetric morbidity = obstetric APS = a treatable diagnosis.
• Cytogenetics — and this is a real point of guideline divergence. The more informative modern test is karyotyping the products of conception (POC) of a further loss: a sporadic aneuploid result reassures (it explains the loss and points away from a parental cause), whereas a euploid loss raises the index of suspicion for a maternal/uterine factor. Parental peripheral-blood karyotyping detects the ~2–5% with a balanced translocation but has a low yield and high cost, so the current direction is to test the POC first and karyotype the parents selectively — e.g. when POC shows an unbalanced structural rearrangement, or there is a family history.
• Pelvic imaging for uterine anatomy: 3D transvaginal ultrasound is the first-line modern tool and, with saline infusion sonography, reliably distinguishes a septate from a bicornuate uterus (the critical distinction, because only the septum is intracavitary and potentially resectable). MRI is reserved for ambiguous müllerian anomalies. Plain 2D ultrasound under-calls the septum.
• Thyroid: TSH and free T4 to detect overt thyroid disease; TPO antibodies identify the euthyroid-autoimmune group — but note in advance that finding them does not mandate treatment (TABLET, below).
• Targeted, not reflexive, endocrine/metabolic tests: HbA1c/glucose if diabetes is suspected; prolactin if symptomatic; an assessment for PMOS (previously PCOS) where the phenotype fits.
• What you should not routinely order: inherited (hereditary) thrombophilia screening — factor V Leiden, prothrombin gene, protein C/S, antithrombin, MTHFR — is not recommended as part of an RPL work-up (weak association, and now no treatment benefit — ALIFE2). Routine TORCH serology, routine "natural killer cell" testing and the commercial reproductive-immunology panels are not evidence-based and should not be sent.

Two further investigations are worth a sentence each because candidates either over-order or forget them. Endometrial biopsy for chronic endometritis (CD138 immunostaining for plasma cells) is not yet a routine RPL test — order it only in the context of the emerging-evidence discussion below, not reflexively. And TORCH serology, karyomapping panels, and "implantation" or natural-killer-cell immune assays are not part of an evidence-based work-up at all; sending them generates anxiety, cost and false trails without changing management.

The interpretive thread to carry into management: a normal work-up is the commonest result and is itself a finding — it places the couple in the unexplained group, which carries a genuinely good cumulative prognosis with supportive care alone, and for whom the evidence base is mostly a list of treatments that don't work. The dominant prognostic variables are maternal age and the number of prior losses — a woman in her early thirties after three losses still has a clear majority chance of a live birth in her next pregnancy, whereas advancing maternal age and a rising loss count progressively lower it. Put a realistic, individualised number to that — rather than a vague reassurance or a counsel of despair.

Management

Organise the plan immediate → ongoing → long-term, and lead with the governing principle: the number of evidence-based treatments is small, and a large part of good management is confidently withholding the unproven ones.

Immediate — diagnosis-driven treatment for the treatable few:

Ongoing — supportive care, which is not a placebo to be dismissed: for the unexplained group, the intervention with the best observational support is early reassurance scanning and structured supportive care ("tender loving care") in a dedicated early-pregnancy/recurrent-loss clinic. Cumulative live-birth rates in unexplained RPL are good — the majority will carry a subsequent pregnancy to term with supportive care alone — and conveying that honestly is therapeutic. Continue any diagnosis-specific drug (aspirin/LMWH for APS, levothyroxine) and address modifiable lifestyle factors: smoking cessation, alcohol, caffeine moderation, and optimising weight (both extremes raise loss risk).

The progesterone question demands precision, because it is the single most mis-stated point in this topic. The evidence does not support progesterone for unexplained RPL: PROMISE gave unselected women with ≥3 unexplained losses vaginal progesterone from a positive test and found no benefit (live birth 65.8% vs 63.3%). What is supported is a narrower indication: PRISM tested progesterone in women with early-pregnancy bleeding, and although the trial was negative overall (75% vs 72%, P=0.08), the pre-specified subgroup with three or more previous miscarriages who were bleeding in the current pregnancy benefited (live birth 72% vs 57%, RR 1.28). So the defensible position — and the one ESHRE now adopts — is: progesterone for the woman with ≥3 prior losses and current bleeding, not for RPL in general. PROMISE and PRISM apply to different patients, and quoting each correctly — stating which population it studied and which patient it applies to — is what governs the prescribing decision and keeps the two trials from being conflated.

Long-term — counselling, prognosis and the next pregnancy: give a couple-specific prognosis (age and number of prior losses are the dominant variables), arrange pre-conception optimisation — folate, smoking cessation, weight, glycaemic and thyroid control, and a clear plan for which drugs continue into the next conception — and plan that pregnancy's care: early viability scan, continuation of any indicated therapy, and — in selected cases of mid-trimester loss with cervical insufficiency — discussion of cervical cerclage, which is a separate pathway from first-trimester RPL and is decided on cervical-length and obstetric-history grounds, not on the RPL work-up. The recurring counselling error to avoid is conflating unexplained with untreatable: the unexplained couple is precisely the one whose prognosis with supportive care is good, and saying so confidently is itself therapeutic.

What is NOT recommended — and why, because the reasoning is what justifies withholding each:

• Heparin/LMWH for inherited thrombophilia — ALIFE2 randomised exactly these women (≥2 losses + confirmed heritable thrombophilia) to LMWH or standard care and found no difference (72% vs 71%). Anticoagulating this group adds bleeding and injection burden for no gain.
• Aspirin and/or heparin for unexplained RPL — ALIFE (no benefit of aspirin±nadroparin over placebo) and EAGeR (preconception aspirin not significantly better for live birth, NS overall) close this door; RCOG states aspirin/LMWH "should not be used" in unexplained RPL.
• Levothyroxine for euthyroid TPO-antibody-positive women — TABLET showed no live-birth benefit (37.4% vs 37.9%); treat overt thyroid disease only.
• Septum resection as a reflex — TRUST found hysteroscopic metroplasty did not improve live birth over expectant management (31% vs 35%) in its (small) randomised population; the operation is no longer an automatic recommendation and must be an individualised, counselled decision.
• Corticosteroids, intravenous immunoglobulin (IVIG), intralipid, and lymphocyte/paternal immunisation — the "reproductive immunology" therapies. These lack RCT support for RPL (ESHRE confines any possible IVIG signal to ≥4 unexplained losses, low certainty), carry real harms and cost, and are widely marketed to desperate couples. Decline them, and explain why (developed under The evidence & the controversy).

Guidelines compared

The three major bodies agree on the management skeleton but diverge on the definition and on a few work-up details — and the divergence matters, because it changes who gets investigated.

The sharpest divergence is the definition: ESHRE and ASRM have moved to two losses (so a couple is investigated earlier), whereas RCOG retains three as the formal threshold but builds in clinical discretion to start at two. The practical reconciliation for a SA answer: investigate after two losses where there is a clinical reason to suspect a treatable cause (e.g. a late loss, a euploid loss, an autoimmune history), and certainly after three — and do not let the older "three consecutive" line delay finding an APS that is sitting there to be diagnosed.

The evidence & the controversy

Three threads define the modern Final-level discussion.

First, RPL is a field defined by negative trials, and that is the point. Almost every intervention that biological plausibility recommended has failed when finally tested: progesterone in the unselected (PROMISE), aspirin±heparin in the unexplained (ALIFE, EAGeR), LMWH in heritable thrombophilia (ALIFE2), levothyroxine in euthyroid TPO-positive women (TABLET), and septum resection (TRUST). Each of these trials closed a previously plausible intervention, and naming what each one shut down is the substance of the modern discussion. The unifying lesson — that a plausible mechanism does not license a treatment until an RCT says so — runs through the whole of recurrent loss.

Second, the commercial "reproductive immunology" industry is the live controversy, and it has an ethical edge. Around the genuine science of the maternal–fetal immune interface has grown a marketplace of unproven tests (uterine/peripheral natural-killer-cell assays, cytokine panels) and treatments (corticosteroids, IVIG, intralipid infusions, TNF-α blockers, paternal lymphocyte immunisation) sold — often expensively, often in the private sector — to couples after recurrent loss. The evidence base for these in RPL is absent or negative; some carry real harm (steroid and IVIG risks, the historical withdrawal of lymphocyte immunotherapy for safety). The defensible consultant stance is not contempt but a framework: explain that the immunology is real but the specific tests and treatments are unvalidated, that desperation is not a clinical indication, that offering an unproven intervention outside a trial is hard to justify, and that the most evidence-based "immunological" support is structured supportive care in a recurrent-loss clinic. This is a marketed, contested area, and the principled response is to decline it on those grounds.

Third, chronic endometritis is the emerging frontier where you should be appropriately tentative. There is a growing literature associating persistent low-grade endometrial inflammation (CD138-positive plasma cells on biopsy, sometimes with dysbiotic endometrial flora) with RPL and implantation failure, and small studies suggest antibiotic treatment may improve outcomes. This is an area to watch and to flag as evolving — not yet a standard test or treatment; presenting it as established overreaches the evidence. The honest framing is "biologically interesting, evidence immature, not yet routine."

Landmark trials & key evidence

A worked piece of arithmetic, from PRISM's headline subgroup: in women with ≥3 prior losses who are bleeding, live birth rose from 57% to 72%, an absolute risk increase of 15 percentage points (ARR 0.15), so the number needed to treat ≈ 1/0.15 ≈ 7 — about seven such women treated with a cheap, safe vaginal pessary to gain one extra live birth. Contrast that with PROMISE's unselected population, where the same drug moved live birth by a non-significant 2.5 points and the confidence interval crossed 1 — same intervention, opposite conclusion, because the population was different. Progesterone works in this patient and not that one, and PRISM and PROMISE are the two trials that draw the line between them — the difference between treating the right woman and treating an entire diagnosis on a single positive subgroup.

Worked viva — how to structure the answer

A typical stem: "a 34-year-old has had three first-trimester miscarriages; she and her partner are otherwise well. How do you investigate and manage her?" A high-scoring answer runs:

1. Frame the probability. "Most individual miscarriages are sporadic embryonic aneuploidy, and around half of couples have no cause found — so my work-up is short and targeted at the treatable minority, chiefly antiphospholipid syndrome, while I avoid over-investigation."
2. Run the focused work-up. "Antiphospholipid panel — lupus anticoagulant, anticardiolipin, anti-β2-glycoprotein-I — drawn before any anticoagulant and repeated at 12 weeks to confirm persistence; a 3D transvaginal ultrasound to assess uterine anatomy and distinguish a septate from a bicornuate uterus; TSH and free T4 for overt thyroid disease; and karyotype of the products of any further loss, with parental karyotyping reserved for an unbalanced result or a family history. I would not screen for inherited thrombophilia."
3. Treat what you find, by mechanism. "If she has persistent antiphospholipid antibodies and obstetric morbidity, that is obstetric APS — low-dose aspirin plus prophylactic LMWH. Overt hypothyroidism gets levothyroxine. A parental translocation gets genetic counselling and the option of PGT-SR."
4. Be precise about progesterone. "I would not give progesterone for unexplained recurrent loss — PROMISE showed no benefit. But if she has three or more prior losses and bleeds in the next pregnancy, that is the PRISM subgroup, and vaginal progesterone 400 mg twice daily to 16 weeks is justified."
5. State what you will not do, and why. "No heparin for inherited thrombophilia — ALIFE2 was negative. No levothyroxine for TPO antibodies alone — TABLET was negative. No corticosteroids, IVIG or intralipid — these are unproven and potentially harmful, and I'd counsel her against the commercial reproductive-immunology offerings she may encounter."
6. Close with prognosis and support. "If the work-up is normal — the commonest outcome — her cumulative chance of a live birth is good, and structured supportive care with an early reassurance scan in a recurrent-loss clinic is the most evidence-based 'treatment' I can offer. In our setting I'd manage this at a regional/tertiary recurrent-loss clinic, with district follow-up of the ongoing pregnancy."

South African context

The work-up is mostly cheap and deliverable, but a few realities shape practice. The antiphospholipid panel (LAC, aCL, anti-β2GPI) is available through NHLS and is the test that matters; insist on it and on the 12-week repeat, because a single transient positive over-diagnoses APS and commits a woman to unnecessary injections. 3D ultrasound and MRI for müllerian assessment are unevenly available — at district level, careful 2D transvaginal ultrasound with saline infusion sonography goes a long way, and ambiguous anatomy is a reason to refer rather than to operate. Karyotyping (parental and POC) is a constrained, tertiary-level resource, which is an additional argument for the modern POC-first, parents-selectively strategy rather than reflex parental karyotyping. Inherited-thrombophilia testing is exactly the low-value test to avoid sending in a resource-limited service — it costs money, the association is weak, and ALIFE2 has shown that even a positive result does not earn treatment. Enoxaparin, levothyroxine, low-dose aspirin and vaginal progesterone are all on the SAMF/EML formulary, so the genuinely indicated treatments are affordable. Structurally, recurrent-loss assessment belongs at regional→tertiary level (the antibody interpretation, imaging and counselling are not all district resources), with the subsequent pregnancy's supportive antenatal care shared back to district.

Exam traps & red flags

• Anticoagulating inherited thrombophilia. The commonest outdated answer. Heparin/LMWH for factor V Leiden or other heritable thrombophilia does not improve live birth (ALIFE2) — and you should not even be screening for it routinely.
• Giving progesterone for unexplained RPL. PROMISE was negative. The only defensible progesterone indication is ≥3 prior losses and current early-pregnancy bleeding (PRISM subgroup). Conflating the two trials loses the mark.
• Treating TPO antibodies in a euthyroid woman. Levothyroxine helps overt hypothyroidism, not euthyroid autoimmunity (TABLET). Don't start it on antibodies alone.
• Reflex septum resection. TRUST showed no benefit over expectant management; metroplasty is now a counselled, individualised decision, not an automatic one — and you must first distinguish a septate (resectable) from a bicornuate (not) uterus, which needs 3D imaging, not 2D.
• Diagnosing APS on a single positive. Antibody titres shift with pregnancy and thrombosis — classification needs a repeat positive ≥12 weeks apart, and the LAC must be drawn before anticoagulation or it is uninterpretable. Over-calling APS commits a woman to needless injections.
• Offering reproductive-immunology therapies. Corticosteroids, IVIG, intralipid, NK-cell testing and lymphocyte immunotherapy are unproven (and some harmful) in RPL. Offering them outside a trial — especially as a paid private service to a desperate couple — is poor practice; declining them with a clear explanation is the right answer.
• Over-investigating after one loss, or refusing to investigate until three. The modern position is graded: investigate after two losses where a treatable cause is plausible. Hiding behind "three consecutive" can delay diagnosing an APS that is there to be found.
• Forgetting the karyotype of the products. A euploid loss reframes the work-up (points to a maternal/uterine factor); an aneuploid loss reassures. Not requesting POC cytogenetics where available wastes the most informative test.
• Dismissing supportive care as nothing. For the unexplained majority, early reassurance scanning and structured support in a recurrent-loss clinic is the most evidence-based offering and carries a good cumulative live-birth rate — present it as treatment, not as a consolation.

Evidence anchors

• ESHRE guideline: recurrent pregnancy loss — an update in 2022, Human Reproduction Open 2023
• RCOG Green-top Guideline No. 17, Recurrent Miscarriage (4th edition), Regan et al., BJOG 2023
• PROMISE — Coomarasamy et al., progesterone in unexplained recurrent miscarriage, N Engl J Med 2015
• PRISM — Coomarasamy et al., progesterone in early-pregnancy bleeding, N Engl J Med 2019
• EAGeR — Schisterman et al., preconception low-dose aspirin and pregnancy outcomes, Lancet 2014
• ALIFE — Kaandorp et al., aspirin plus heparin or aspirin alone in recurrent miscarriage, N Engl J Med 2010
• ALIFE2 — Quenby et al., heparin for recurrent miscarriage and inherited thrombophilia, Lancet 2023
• TABLET — Dhillon-Smith et al., levothyroxine in TPO-antibody-positive euthyroid women, N Engl J Med 2019
• TRUST — Rikken et al., septum resection vs expectant management in septate uterus, Human Reproduction 2021
• South Africa NDoH Maternity Care Guidelines / SAMF / EML — enoxaparin, levothyroxine, low-dose aspirin and vaginal micronised progesterone are formulary-available; NHLS provides the antiphospholipid panel; 3D ultrasound, MRI and karyotyping are uneven, tertiary-weighted resources; recurrent-loss assessment is a regional→tertiary undertaking with district-level antenatal follow-up.