Appraise the physiological effects and clinical uses of progesterone in obstetrics and gynaecology

Clinical overview

The objective is to appraise, not "describe." With the physiology and the principal indications taken from the Intermediate chapter on progesterone in pregnancy, the harder task is to weigh the evidence behind a hormone that is cheap, ubiquitous, perceived as harmless and therefore reflexively prescribed. The consultant discipline is to separate the few subgroups in which progesterone is genuinely disease-modifying from the larger penumbra of unproven or negative uses, to read the pivotal trials critically (subgroup signals, intention-to-treat dilution, relative versus absolute effect), and to translate that into resource-conscious practice under the National Integrated Maternal and Perinatal Care (NIMPC) Guidelines (NDoH, 2024).

The easily-forgotten half of the objective is the gynaecological and antagonist side: progesterone as luteal-support agent, the distinction between bio-identical progesterone and the synthetic progestogens of contraception, and the clinical use of anti-progesterone action — mifepristone for medical evacuation and the cautionary tale of selective progesterone-receptor modulators (SPRMs) in fibroid therapy. The thread running through all of it is the same consultant discipline: which molecule, which route, which woman, and what does the best trial actually show?

Core knowledge

This section assumes the receptor biology and the luteoplacental physiology taught at Intermediate (progesterone in pregnancy) and builds the advanced layer — the mechanisms that justify a prescribing decision.

Mechanism worth re-stating at Final level

Progesterone acts through nuclear receptors (PR-A and PR-B isoforms) and faster membrane-associated receptors. The crucial mechanistic nuance is functional progesterone withdrawal: in humans, term and preterm labour are triggered not by a fall in serum progesterone (unlike many species) but by a local shift in the PR-A:PR-B ratio and coregulators that renders the myometrium progesterone-resistant despite high circulating levels. This is why serum progesterone levels do not guide supplementation — there is no validated replacement threshold — and why the preterm-birth hypothesis is one of reinforcing an endogenous brake, not correcting a deficiency. Vaginal micronised progesterone exploits a uterine first-pass effect, giving high myometrial concentrations at low systemic levels; that pharmacology, not a serum target, underlies the route choice.

The luteoplacental shift (corpus luteum until ~7–9 weeks, then placental syncytiotrophoblast) is assumed. Its corollary: exogenous progesterone has its strongest physiological rationale before the shift (luteal-phase support after ART, where stimulation/down-regulation impairs the corpus luteum) and a weaker one after — one reason most obstetric regimens stop around 16 weeks.

The pharmacology that decides route and molecule

"Progesterone" is a trap word because the clinical objects behind it are pharmacologically distinct, and the trial evidence is preparation-specific — you cannot read a result for one and prescribe another.

• Micronised progesterone is bio-identical (the same molecule the ovary and placenta make). Given orally it suffers heavy hepatic first-pass metabolism to 5β-pregnanolone and allopregnanolone — neurosteroids that cause the somnolence and dizziness women describe, and which make oral dosing pharmacokinetically unreliable for a uterine target. Given vaginally it bypasses the liver and concentrates in the uterus by a local "first-uterine-pass" effect, achieving high endometrial/myometrial tissue levels at modest serum concentrations. This dissociation of serum from tissue level is precisely why a serum progesterone reading is uninterpretable on vaginal therapy and why the obstetric trials (PRISM, PREGNANT, OPPTIMUM) almost all used the vaginal route.
• Dydrogesterone is a retroprogesterone — orally active, selective for the progesterone receptor, with negligible androgenic, oestrogenic or glucocorticoid activity, and it does not cross-react with the common immunoassay for progesterone. It is the molecule that makes an oral regimen pharmacologically credible (LOTUS programme below).
• 17-hydroxyprogesterone caproate (17-OHPC) is a synthetic caproate ester of a hydroxylated progesterone, formulated for slow intramuscular release — a fundamentally different molecule from micronised progesterone, which is one reason the vaginal and intramuscular evidence streams diverge and must never be pooled in your reasoning.
• The contraceptive progestins (levonorgestrel, etonogestrel, medroxyprogesterone acetate, drospirenone) are distinct synthetic molecules with their own receptor cross-reactivities (androgenic, anti-mineralocorticoid, glucocorticoid). Their safety datasets — bone-mineral density on depot medroxyprogesterone, VTE class effects — do not transfer to obstetric micronised progesterone, a confusion that surfaces constantly in counselling.

The consultant link to make explicit: route changes the evidence base, molecule changes the safety profile — these must be kept separate rather than treating "progesterone" as one thing.

Assessment — appraising the pivotal evidence

The question is not "what is progesterone for" but "how good is the evidence, and where does the headline result mislead?" Each subtype of use below has a different evidentiary footing, and the judgement call differs accordingly.

Early-pregnancy bleeding and recurrent loss: PRISM and PROMISE, read properly

PROMISE (first-trimester vaginal progesterone in unexplained recurrent miscarriage) was negative for live birth: progesterone is not a treatment for recurrent loss per se. The subtlety lives in PRISM (NEJM 2019; n = 4,153 with early-pregnancy bleeding). Overall live birth was 75% with progesterone vs 72% with placebo — relative rate 1.03 (95% CI 1.00–1.07), p = 0.08, not significant — so no benefit in unselected threatened miscarriage. But a pre-specified subgroup graded by prior losses became meaningful at three or more previous miscarriages: 72% vs 57%, RR 1.28 (95% CI 1.08–1.51) — an absolute live-birth difference of ~15 percentage points.

The discipline in appraising this is to read the whole gradient, not just the positive cell. PRISM's pre-specified subgroups by prior-loss count form a clean dose–response:

• No previous miscarriage: 74% vs 75%, RR 0.99 (95% CI 0.95–1.04) — no effect, and the point estimate is on the wrong side of unity.
• One or two previous miscarriages: 76% vs 72%, RR 1.05 (95% CI 1.00–1.12) — a small effect whose CI just touches 1.
• Three or more previous miscarriages: 72% vs 57%, RR 1.28 (95% CI 1.08–1.51) — a large effect.
• Test for interaction across the three strata: p = 0.007 — a statistically significant gradient, not noise.

A positive subgroup nested in a negative trial is normally treated with suspicion (multiplicity, post-hoc dredging, regression to the mean). PRISM survives that scrutiny for three concrete reasons: the subgroup was pre-specified; the effect is biologically gradated (monotonic across prior-loss count, with a significant interaction test — the strongest single defence against a chance subgroup); and it is externally coherent — the ESHRE recurrent-pregnancy-loss guideline (update 2022, published 2023) re-analysed pooled PRISM/PROMISE data and adopted the same bleeding-plus-prior-loss signal. Compute the absolute effect to keep the magnitude honest: at ≥3 losses the ARR is 72% − 57% = 15 percentage points, giving an NNT of 1/0.15 ≈ 7 women treated for one extra live birth — a genuinely large effect for a cheap, safe drug. In the unselected trial the ARR is ~3 points (NNT ≈ 33), which is why the overall result is unconvincing.

There is a health-economic layer too. The PRISM cost-effectiveness analysis found an incremental cost-effectiveness ratio of £3,305 per additional live birth overall, but in the ≥1-prior-miscarriage subgroup progesterone was cost-saving (the averted miscarriages and their downstream management more than paid for the drug). This is the formal basis on which a cheap intervention with a modest absolute effect becomes defensible policy — and the argument that lets NICE pragmatically lower the treatment threshold to one prior loss when the robust efficacy signal was at three. That threshold mismatch is the live debate: erring toward access is reasonable given a cheap, safe drug and a cost-saving signal, but it should be stated plainly that the strongest efficacy evidence sits at ≥3 losses and that certainty is being traded for reach.

NICE NG126 (updated August 2023) translates this into a recommendation: vaginal micronised progesterone 400 mg twice daily for a woman with a scan-confirmed intrauterine pregnancy, bleeding and a previous miscarriage, continued to 16 completed weeks if a heartbeat is seen. It supplements, not replaces, the recurrent-loss work-up (antiphospholipid screen, uterine assessment, thyroid and parental karyotype where indicated).

Preterm birth: EPPPIC as the reference appraisal

Beyond "short cervix and prior spontaneous preterm birth," the argument rests on EPPPIC (Lancet 2021), the individual-participant-data meta-analysis of 31 trials, 11,644 women, 16,185 offspring — a model of how IPD beats aggregate meta-analysis (uniform outcome definitions, proper subgroup interaction tests, the ability to test effect-modifiers like cervical length at the individual level):

• Vaginal progesterone in singletons reduced birth before 34 weeks: RR 0.78 (95% CI 0.68–0.90), absolute benefit greatest where baseline risk was highest — short cervix. Critically, the benefit held whether or not there was a prior preterm birth, so the short cervix, not the obstetric history, is the operative risk marker.
• 17-OHPC in singletons: RR 0.83 (95% CI 0.68–1.01) — CI crosses 1, no clear benefit.
• Multifetal pregnancy: neither agent reduced preterm birth; 17-OHPC was associated with more PPROM before 34 weeks (RR 1.59, 95% CI 1.15–2.22) — a harm signal making "progesterone in twins" a wrong answer.
• Oral progesterone: a nominal signal (RR 0.60, 95% CI 0.40–0.90) but resting on just 2 trials and 181 women — too sparse to recommend; do not equate it with the robust vaginal data, and note this is oral micronised, not dydrogesterone, so even the molecule is unsettled here.

The short-cervix indication itself rests on the PREGNANT trial (Hassan, Ultrasound Obstet Gynecol 2011): in women with a sonographic cervix of 10–20 mm, vaginal progesterone gel cut birth before 33 weeks from 16.1% to 8.9% — RR 0.55 (95% CI 0.33–0.92), a 45% relative reduction, with fewer cases of respiratory distress syndrome (the ≤25 mm threshold used in practice generalises this signal across the wider short-cervix range). The arithmetic is worth carrying: ARR = 16.1% − 8.9% = 7.2 percentage points, NNT = 1/0.072 ≈ 14 women screened-and-treated to prevent one birth before 33 weeks — and that NNT is only achievable if you have a cervical-length screening programme to find the short cervix in the first place, which is the real resource question in SA.

Layer onto this OPPTIMUM (Lancet 2016), the largest single RCT, which found no overall reduction in preterm birth or the neonatal composite (obstetric OR 0.86, 95% CI 0.61–1.22) but gave the reassuring datum that vaginal progesterone has no adverse effect on childhood neurodevelopment at two years — the safety anchor for counselling. The apparent tension between OPPTIMUM (neutral) and PREGNANT/EPPPIC (positive) is itself an appraisal point: OPPTIMUM enrolled a heterogeneous high-risk population (mixed risk markers, not a pure short-cervix group), diluting any effect concentrated in the short-cervix subgroup, and adherence was imperfect. The coherent reading: vaginal progesterone earns its place in the singleton with a short cervix (commonly ≤25 mm) or a prior spontaneous preterm birth, integrated with the cervical-length-and-cerclage decision; it does nothing useful in twins; and the better the cervical-length phenotyping, the larger the observed effect.

The 17-OHPC collapse

Intramuscular 17-hydroxyprogesterone caproate rested on the single Meis 2003 trial. The mandated confirmatory study, PROLONG, failed to reproduce any benefit, and on 5 April 2023 the FDA withdrew approval of 17-OHPC (Makena and generics) entirely — a worked example of regulatory accountability (accelerated approval on a surrogate, confirmatory obligation, negative result, withdrawal). The deeper appraisal lesson is why Meis and PROLONG disagreed: Meis enrolled an exceptionally high-risk, predominantly African-American US population with a very high baseline preterm-birth rate, whereas PROLONG's population had a lower event rate — so the original effect may have reflected an extreme-risk subgroup that did not generalise, the same trap PRISM avoids by showing a gradient. Where progesterone is indicated for preterm prophylaxis, vaginal micronised progesterone is the agent, not 17-OHPC — and in SA this is moot anyway, because 17-OHPC has no place in the formulary pathway.

Luteal-phase support and the dydrogesterone debate

After IVF the corpus luteum is functionally impaired by the supraphysiological steroid milieu and GnRH-agonist down-regulation, so progesterone support is standard. The Final controversy is route and molecule: the LOTUS I and II phase-III programme (together >2,000 women) showed oral dydrogesterone non-inferior to micronised vaginal progesterone for ongoing pregnancy and live birth in fresh-cycle IVF, with comparable safety despite hepatic first-pass — and the integrated patient-data analysis suggested if anything a numerically higher live-birth rate with dydrogesterone. This is a real challenge to the "vaginal is always best" reflex, driven by adherence and tolerability (women strongly prefer an oral tablet to a vaginal pessary, and adherence is a live-birth determinant). The critical appraisal point: the obstetric miscarriage evidence (PRISM/NICE) is specifically for the vaginal micronised preparation and must not be silently substituted with oral dydrogesterone, whereas for luteal support after IVF the two are reasonable alternatives. A subtlety to add: LOTUS studied fresh cycles; frozen embryo transfer in a programmed (artificial) cycle is a different luteal environment (no corpus luteum at all), where adequate progesterone exposure is even more critical and serum-progesterone-guided "rescue" supplementation is an active research area — do not over-extrapolate fresh-cycle equivalence to every transfer.

Where progesterone has no evidentiary footing

Closing the appraisal honestly means naming the negatives as confidently as the positives:

• Unexplained recurrent miscarriage without current bleeding — PROMISE negative.
• Unselected threatened miscarriage (bleeding, no prior loss) — PRISM's no-prior-loss stratum had RR 0.99.
• Any twin or higher-order pregnancy — EPPPIC null for vaginal progesterone and a PPROM harm signal for 17-OHPC.
• A "low" serum progesterone as a treatment trigger — no validated threshold; the marker is biologically uninterpretable for the reasons in Core knowledge.
• Threatened preterm labour with intact membranes as acute tocolysis — progesterone is a maintenance/prophylactic brake, not an acute tocolytic; do not reach for it in established preterm labour.

Management — disciplined indications and the antagonist side

When progesterone earns its place

1. Early-pregnancy bleeding plus ≥1 previous miscarriage — vaginal micronised progesterone 400 mg twice daily to 16 weeks (NICE NG126), after transvaginal scan has excluded ectopic and confirmed an ongoing intrauterine pregnancy. Counsel honestly that absolute benefit is modest in those with one or two prior losses and substantial (ARR ~15 points, NNT ~7) only at three or more.
2. Singleton, short cervix (≤25 mm) or prior spontaneous preterm birth — vaginal progesterone into the early third trimester, alongside cervical-length surveillance and the cerclage decision. (See the SA dosing note below — the EML-review dose and the registered-product dose differ.)
3. Luteal-phase support after ART — vaginal micronised progesterone or oral dydrogesterone, the latter a legitimate first-line on adherence grounds.

Named regimens and how they differ

Exact regimens matter and so does the reasoning for the difference, much as magnesium has its Pritchard/Zuspan/Sibai variants. For progesterone the variation is by indication, route and the SA-vs-international choice:

The two SA lines deserve emphasis because they are a genuine SA divergence: the NDoH EML adult review (Oct 2019) recommended 100 mg vaginally daily for short cervix ≤25 mm in a singleton, while the SAHPRA-registered Utrogestan label is 200 mg nightly from ~20–34 weeks — and Utrogestan is the only progesterone with a SAHPRA registration for preterm-birth prevention. A consultant should know both numbers, know that the public-sector/EML figure and the product label differ, and prescribe to the formulary and stock realities of the facility rather than reflexively to the international 400 mg BD figure (which is the miscarriage dose, not the preterm dose — a classic conflation).

Practical prescribing and counselling points

• Counsel the modest absolute benefit. For bleeding-plus-one-prior-loss the absolute gain is small; for the short cervix the NNT (~14) depends on actually having found the short cervix. Over-promising erodes trust.
• Tolerability and route. Vaginal preparations can cause discharge, irritation and leakage; oral micronised progesterone causes drowsiness/dizziness (neurosteroid metabolites) — dose at night. These are the real adherence levers.
• Do not measure serum progesterone to titrate an established vaginal regimen — serum and tissue levels are dissociated.
• Stop points matter: 16 weeks for the miscarriage indication; into the early third trimester (label: ~34 weeks) for the preterm indication. Continuing indefinitely "to be safe" is not evidence-based.

South African context

In SA, confirm EML/facility stock and the registered preparation before promising therapy, and remember the system geometry: cervical-length screening and ART luteal support are largely regional/tertiary and private-sector activities, so the "screen-and-treat-the-short-cervix" NNT is only realisable where transvaginal cervical scanning exists. Manage within the district → regional → tertiary referral chain of the NIMPC Guidelines. Above all, progesterone has no bearing on the haemorrhage/hypertension/sepsis triad that dominates Saving Mothers mortality — it must never displace those priorities, and reaching for progesterone before securing the basics is misjudged.

The antagonist and modulator side

• Mifepristone is a progesterone-receptor antagonist (and a glucocorticoid-receptor antagonist); with misoprostol it is the standard regimen for early pregnancy loss and termination (commonly mifepristone 200 mg then misoprostol) — clinical proof that blocking PR is sufficient to end an early pregnancy, the mirror image of the support indications. Its efficacy is the cleanest demonstration that progesterone signalling is necessary to maintain early pregnancy, which is the same biology the support indications try (mostly unsuccessfully) to exploit.
• Selective progesterone-receptor modulators (SPRMs) — ulipristal acetate for fibroids — are a pharmacovigilance cautionary tale: after severe drug-induced liver injury including transplant-requiring cases, the EMA suspended ulipristal for uterine fibroids in March 2020 and its use is now tightly restricted (retained for emergency contraception as a single dose, where the cumulative hepatic exposure is negligible). The teaching point is dose-and-duration dependence of an idiosyncratic hepatic signal: an effective fibroid drug undone by months of exposure, while the single-dose contraceptive use is unaffected.
• Bio-identical progesterone is not a contraceptive progestogen. The synthetic progestins (levonorgestrel, etonogestrel, medroxyprogesterone acetate) are distinct molecules; do not extrapolate contraceptive-progestin safety data (e.g. bone density with depot medroxyprogesterone, or the SAHPRA 2025 clarification of medroxyprogesterone benefits/risks) to obstetric micronised progesterone — a confusion that surfaces in counselling and in high-risk contraceptive prescribing.

Landmark trials & key evidence

The studies below define the evidence base — direction of effect, magnitude, and what each one changed in practice. The numbers below are the published primary findings.

Red flags / pitfalls

• Do not let progesterone delay diagnosis of ectopic pregnancy — locate the pregnancy first; a "treat the bleeding" reflex can mask a tubal ectopic to rupture.
• Do not prescribe for unselected threatened miscarriage or recurrent loss without current bleeding — PRISM was overall negative (no-prior-loss stratum RR 0.99) and PROMISE was negative.
• Do not use vaginal progesterone for preterm prevention in multiple pregnancy — EPPPIC shows no benefit, and 17-OHPC raised PPROM in twins.
• Do not measure serum progesterone to justify or titrate treatment — no validated threshold; functional withdrawal is local, and serum/tissue levels are dissociated on vaginal therapy.
• Do not assume 17-OHPC works — PROLONG was negative and the FDA withdrew it (2023); prefer vaginal micronised progesterone.
• Do not over-read a positive subgroup — but do not dismiss PRISM's ≥3-losses signal either: pre-specified, gradated, significant interaction test (p = 0.007) and externally replicated. The credit lies in articulating why this subgroup is credible when most are not.
• Do not confuse the doses. 400 mg BD vaginally is the miscarriage regimen; the preterm regimen is 100 mg (SA EML review) or 200 mg nightly (SAHPRA label) — different indication, different dose.
• Do not substitute oral dydrogesterone for vaginal progesterone in the miscarriage indication — the PRISM/NICE evidence is preparation-specific; the molecules are interchangeable only for luteal support.
• Do not use progesterone as an acute tocolytic in established preterm labour — it is a maintenance brake, not an acute uterine relaxant.
• Do not conflate progesterone with contraceptive progestins or with the antagonist mifepristone — three different clinical objects with three different safety datasets.

Worked viva — how to structure the answer

A Final stem is rarely "tell me about progesterone." It is a prescribing decision with a trap — "a 32-year-old at 7 weeks with bleeding and one previous miscarriage asks if there is anything to save the pregnancy," or "a woman with a previous 28-week delivery is found to have a 19 mm cervix at 22 weeks — what do you offer?" A high-scoring answer runs:

1. Frame the question as appraisal, not recitation — "Progesterone is cheap, safe and over-prescribed; the skill is to give it only where a good trial shows benefit, and to name the right preparation and dose."
2. Place the woman in the evidence — for the bleeder: "scan first to confirm an intrauterine pregnancy and exclude ectopic; she has bleeding and a prior loss, so she falls inside NICE NG126 — vaginal micronised progesterone 400 mg twice daily to 16 weeks." For the short cervix: "singleton with a 19 mm cervix — vaginal progesterone is indicated; in SA I would use the registered Utrogestan 200 mg nightly (the EML review modelled 100 mg), continued into the early third trimester, alongside the cervical-length and cerclage decision."
3. Quote the pivotal trial and appraise it — PRISM (overall negative, but a pre-specified, gradated, significant-interaction subgroup at ≥3 losses, NNT ~7; cost-saving in prior-loss women), or PREGNANT/EPPPIC for the short cervix (RR 0.55 / RR 0.78, benefit independent of obstetric history).
4. State the honest limits — modest absolute benefit at one prior loss; no role in twins, unselected threatened miscarriage, recurrent loss without bleeding, or a "low" serum level.
5. Name the SA reality — registered preparation and stock, the screening infrastructure the NNT depends on, the referral chain, and that progesterone never displaces the haemorrhage/hypertension/sepsis priorities.
6. Close on the antagonist mirror — show command of the whole receptor axis by noting that blocking PR (mifepristone) ends an early pregnancy, the converse of these support indications.

Evidence anchors

• PRISM trial — A Randomized Trial of Progesterone in Women with Bleeding in Early Pregnancy (NEJM 2019). Overall negative (RR 1.03, 95% CI 1.00–1.07; p = 0.08); pre-specified gradient — no prior loss RR 0.99, 1–2 losses RR 1.05, ≥3 losses 72% vs 57% (RR 1.28, 95% CI 1.08–1.51); interaction p = 0.007.
• PRISM economic evaluation — cost-effectiveness of progesterone in early-pregnancy bleeding (BJOG 2020). ICER £3,305 per additional live birth overall; cost-saving in the ≥1-prior-miscarriage subgroup — the policy basis for the lowered NICE threshold.
• NICE NG126 — Ectopic Pregnancy and Miscarriage (updated Aug 2023). Vaginal micronised progesterone 400 mg twice daily for bleeding + previous miscarriage, continued to 16 weeks if heartbeat confirmed.
• EPPPIC IPD meta-analysis (Lancet 2021). 31 trials/11,644 women; vaginal progesterone singleton PTB <34w RR 0.78 (0.68–0.90), benefit independent of prior sPTB; 17-OHPC RR 0.83 (0.68–1.01); twins RR 1.01 (0.84–1.20, no benefit); 17-OHPC ↑PPROM in twins (RR 1.59, 1.15–2.22).
• OPPTIMUM study (Lancet 2016). No overall preterm-birth benefit (obstetric OR 0.86, 95% CI 0.61–1.22); reassuring 2-year neurodevelopment.
• PREGNANT trial — vaginal progesterone gel for a sonographic short cervix (Hassan, Ultrasound Obstet Gynecol 2011). Birth <33 weeks 8.9% vs 16.1%, RR 0.55 (95% CI 0.33–0.92); SA centres contributed.
• PROMISE trial — progesterone in unexplained recurrent miscarriage (NEJM 2015). Negative: live birth 65.8% vs 63.3%, RR 1.04 (95% CI 0.94–1.15) — no benefit without a current-bleeding criterion.
• PROLONG trial — 17-OHPC for recurrent preterm birth (Blackwell, Am J Perinatol 2020). Confirmatory RCT; failed to reproduce Meis 2003 benefit; led to the 2023 FDA withdrawal of 17-OHPC.
• ESHRE Guideline: Recurrent Pregnancy Loss (update 2022). Adopts the PRISM bleeding-plus-prior-loss signal; cautious otherwise.
• SMFM statement on FDA withdrawal of 17-OHPC (AJOG 2023). Context for PROLONG and the April 2023 withdrawal of Makena/generics.
• LOTUS I — dydrogesterone vs micronised vaginal progesterone for luteal support (PubMed 2017). Oral dydrogesterone non-inferior for live birth.
• LOTUS II — oral dydrogesterone vs intravaginal progesterone gel for luteal support (Griesinger, Human Reproduction 2018). Non-inferior: live birth 34.4% vs 32.5% (adjusted difference 1.9%, 95% CI −4.0–7.8); the two-trial programme (>2,000 women) supports oral dydrogesterone as a validated luteal-support option.
• EMA — suspension of ulipristal acetate for uterine fibroids (2020). Drug-induced liver injury; pharmacovigilance exemplar.
• SAHPRA-approved Utrogestan product information (micronised progesterone, June 2024). The SA-registered preparation; 200 mg vaginally at night, ~week 20–34, for short cervix/prior sPTB in a singleton — the only progesterone registered for this indication in SA.
• NDoH Essential Medicines List — progesterone to prevent preterm delivery, adult review (Oct 2019). SA public-sector basis: 100 mg vaginally daily, short cervix ≤25 mm, singleton.
• South African National Integrated Maternal and Perinatal Care Guidelines (NDoH, 2024). SA source of truth for antenatal practice, levels of care and referral.