Critically appraise the evidence-based management of preterm labour and PPROM

In one line

Preterm labour and PPROM are two faces of the same final common pathway, and almost nothing we do stops the birth — so the craft is not tocolysis but buying the ~48 hours of pregnancy in which antenatal corticosteroids and magnesium neuroprotection actually change the baby's outcome, while deciding, condition by condition, when prolonging the pregnancy stops helping and starts harming.

This chapter assumes the diagnostic and screening groundwork in preterm birth & PPROM basics and cervical cerclage; it spends its words on the appraisal, the thresholds and the judgement calls. The prophylaxis story (progesterone, cerclage, cervical-length surveillance) is the prevention arm; this objective is about the woman who is already contracting or has already ruptured.

Why this matters in South Africa

Prematurity is the leading direct cause of neonatal death in the SA Perinatal Problem Identification Programme data, and the modifiable deaths cluster in two places that have nothing to do with heroic neonatal care: steroids not given (or given too late to act) and the very-preterm baby born in a facility with no ventilator or no retrieval pathway. The management of this disease is almost entirely logistical and protocol-driven — getting the steroid in, getting the magnesium running, and getting mother-and-fetus to the right level of care before birth — because the things that move the outcome are cheap, EML-listed and time-critical, while tocolysis, which feels like "doing something", changes no perinatal outcome at all. Early-onset and very-preterm presentations additionally collide with the SA reality that neonatal ICU is scarce and tiered, so the "prolong the pregnancy" calculus a high-income unit takes for granted is, in a district hospital, usually a stabilise-and-transfer-in-utero decision instead. The two competencies that matter are running the time-critical protocol and making the right system-level call about where the woman should deliver.

Pathophysiology — why the subtypes behave differently

Preterm birth is a syndrome, not a diagnosis: a single clinical endpoint reached by several distinct mechanisms, and recognising which pathway is operating changes the management. The basic two-stage "decidual activation → uterotonin release → cervical ripening + membrane weakening" sequence is assumed from Intermediate; classifying the subtypes builds on it.

• Infection/inflammation-driven (intra-amniotic infection or sterile inflammation). Ascending microbial invasion of the amniotic cavity (or a sterile "danger-signal" inflammatory response) activates Toll-like receptors, drives IL-6/IL-8 and prostaglandin release, and matrix metalloproteinases degrade the chorioamniotic membranes. This is the dominant mechanism at the earliest gestations — the more preterm the birth, the more likely infection is driving it. The clinical consequence is decisive: this subtype carries the fetal inflammatory response syndrome (FIRS), which is independently linked to cerebral palsy and bronchopulmonary dysplasia, and it is the subtype where prolonging the pregnancy can harm the fetus even before overt chorioamnionitis appears. Mechanism → action: a raised amniotic IL-6 or a positive Gram stain on amniocentesis predicts imminent delivery and poor neonatal outcome regardless of tocolysis — which is exactly why tocolysis is futile here and why occult intra-amniotic infection is the thing you are always trying to exclude.
• Vascular / ischaemic (decidual haemorrhage, abruption). Thrombin generated at the decidual–placental interface is a powerful uterotonic and also degrades membranes (thrombin → protease-activated receptor signalling). This is the mechanism behind preterm labour or PPROM that follows an abruption or presents with antepartum bleeding. The consequence: bleeding-associated preterm labour is tocolysis-resistant and often progresses fast, and a concealed abruption is a contraindication to prolonging the pregnancy — the bleeding subtype is the one most likely to deteriorate to a fetal-distress or DIC emergency.
• Mechanical / cervical (uterine overdistension and cervical insufficiency). Multiple pregnancy and polyhydramnios stretch the myometrium (stretch → gap-junction and oxytocin-receptor upregulation); a structurally or functionally cervical insufficiency shortens and funnels. This is the subtype the prevention arm (progesterone, cerclage, cervical-length surveillance) actually targets, and the one where a short cervix on this admission predicts recurrence and shapes the next-pregnancy plan.
• Idiopathic / physiological-pathway activation. A proportion have premature activation of the normal labour cascade with no identifiable trigger — the subtype most likely to be over-diagnosed as "threatened preterm labour" and over-treated.

Two mechanism→consequence links anchor the rest of the chapter. First, PPROM and intact-membrane preterm labour are the same syndrome stratified by membrane integrity, and membrane status flips the antibiotic decision completely (the macrolide helps once membranes are ruptured and harms when they are intact — see the evidence section). Second, the gestational gradient of aetiology explains the management philosophy: because infection dominates at the earliest gestations, the very-preterm presentation is the one where (a) you most need to exclude occult chorioamnionitis, (b) tocolysis is most futile, and (c) the neuroprotective interventions (steroids, magnesium) have the largest absolute benefit — so the earlier the gestation, the more the plan is "give the proven interventions and get to neonatal ICU", not "stop the labour".

Assessment

The first job is to confirm you are dealing with what you think you are — over-diagnosis of "threatened preterm labour" fills antenatal wards and exposes women to tocolytics they never needed. The basic clinical picture (contractions, speculum findings) is assumed from Intermediate; the discrimination is what matters — separating true imminent labour from threatened labour, and detecting the occult infection that overrides everything.

• Is this really labour? Regular contractions plus cervical change (SASOG: dilatation >2 cm at the internal os, cervix <1 cm). Contractions without cervical change are threatened labour — observe, do not commit to tocolysis or steroids on contractions alone. The positive predictive value of symptoms alone is poor: most women admitted with "threatened preterm labour" are not delivering within a week, so the next two tests are there to avoid intervening in the majority who will not deliver.
• Transvaginal cervical length is the discriminator. A TVS cervical length ≥30 mm is a strong negative predictor — most such women are not in labour and can be considered for discharge. Combining a short cervix with fetal fibronectin sharpens triage further: a negative fFN has a high negative predictive value for delivery within 7–14 days, so its value is ruling out imminent birth and avoiding unnecessary transfer/steroids, not ruling it in. These tests mislead in defined situations: fFN is invalidated by recent intercourse, a digital examination, vaginal bleeding or lubricant in the preceding 24–48 h (false positives), so it must be taken before you touch the cervix and is uninterpretable if the woman is already frankly bleeding. A positive fFN is not a reason to act — its PPV is modest; the test earns its keep as a rule-out.
• Is the membrane ruptured? Sterile speculum first: pooling is the gold standard. If equivocal, an amniotic protein test (IGFBP-1/PAMG-1) helps, but interpret it against pre-test probability — a positive biomarker with no clinical pooling in a woman with no history of fluid loss is more likely a false positive than true PPROM, and a negative result with a convincing history of a gush does not exclude rupture (intermittent leak, hindwater rupture). Never do a digital examination in suspected PPROM (it shortens latency and seeds infection). The reflex "nitrazine/ferning" tests are too unreliable to hang a diagnosis on and are not the SA standard.
• Is there occult infection driving it? This is the single most important judgement here because it flips the plan from prolong to deliver. Overt chorioamnionitis (the Gibbs criteria — maternal pyrexia ≥38 °C plus two of: maternal tachycardia, fetal tachycardia, uterine tenderness, offensive liquor, maternal leucocytosis) is easy; the danger is subclinical intra-amniotic infection, which presents only as unexplained persistent fetal or maternal tachycardia, a creeping CRP/WCC trend, or labour that simply will not be tocolysed. A single normal CRP does not exclude it — the trend matters. In refractory cases at a tertiary unit, amniocentesis for amniotic glucose (low), Gram stain, culture and IL-6 is the definitive way to confirm occult infection, and a positive result mandates delivery.
• Stage and stratify the fetus. Anchor the gestation precisely (best estimate from the earliest scan; never re-date in the third trimester). Establish EFW, presentation (preterm breech and transverse lie are common and change the delivery plan), liquor volume (oligohydramnios in PPROM predicts shorter latency, cord accidents and pulmonary hypoplasia at early gestations), and CTG from viability. In PPROM the liquor volume and the gestation at rupture together drive the prognosis: rupture before ~22–24 weeks carries a real risk of lethal pulmonary hypoplasia and limb contractures, and that prognosis must be in the counselling.

Differential diagnosis and the mimics that change management

• Threatened versus established preterm labour — the commonest "diagnosis" that should not be one; the cost of getting it wrong is unnecessary tocolysis, steroids and transfer (steroid exposure is not free — see late-preterm hyperglycaemia below).
• PPROM versus other causes of vaginal wetness — urinary incontinence, heavy physiological or infective discharge, loss of the operculum, and semen. A confident "ruptured" label commits the woman to expectant management, antibiotics and a delivery deadline she may not need.
• Abruption presenting as preterm labour — painful bleeding with a tense, tender uterus and a non-reassuring CTG is abruption until proven otherwise; it is a deliver situation, not a tocolyse one, and tocolysis here is dangerous.
• Chorioamnionitis versus uncomplicated PPROM — the mimic that matters most, because the management is opposite (deliver vs expectant). Persistent maternal/fetal tachycardia with a tender uterus is sepsis until excluded.
• Acute surgical/medical abdomen masquerading as labour — appendicitis, pyelonephritis, ovarian torsion and red degeneration of a fibroid all cause uterine irritability and contractions; treating the trigger (e.g. the pyelonephritis) is what stops the "labour".
• Cervical insufficiency presenting as painless dilatation with bulging membranes in the mid-trimester is mechanistically and managerially distinct (rescue cerclage may be considered) from contraction-driven labour.

Management

Structure management as immediate → ongoing → long-term. The interventions that move neonatal outcomes are steroids and magnesium; tocolysis only buys the window for them. The subtype-specific divergence carries the detail — intact-membrane labour, PPROM and the very-preterm presentation are managed differently — as do the named regimens and their exact differences.

Immediate — give the interventions that work, in the right window.

Corticosteroids — the regimen, the mechanism of the window, and the contested margins. Betamethasone and dexamethasone are clinically equivalent for the standard indication; the SA EML stocks dexamethasone widely and it is acceptable. The biology of the window explains the timing: glucocorticoids induce type-II pneumocyte surfactant synthesis and accelerate structural lung maturation, and the maximal effect is seen from ~24 h after the first dose to ~7 days after the last — which is precisely why tocolysis only needs to buy ~48 hours, and why a course given <24 h before birth still confers benefit (start it even if delivery looks imminent). Two genuinely contested margins: (1) repeat/rescue courses — a single rescue course is reasonable if >7 days have elapsed and birth is again imminent before 34 weeks, but multiple repeated courses reduce birthweight and head circumference, so they are not given liberally; (2) the late-preterm margin (34–36 weeks), where ALPS showed a real reduction in respiratory morbidity but at the cost of a clinically meaningful increase in neonatal hypoglycaemia — a trade-off that makes late-preterm steroids a considered decision (with neonatal glucose monitoring), not a reflex, and one SA practice applies selectively rather than universally.

Magnesium neuroprotection — three regimens that differ, and how. The principle is the same as magnesium for eclampsia prophylaxis (assumed from the pre-eclampsia chapters), but the neuroprotection regimens differ in loading dose and whether a maintenance infusion is used:

• BEAM/Rouse regimen — 6 g IV load over 20–30 min, then 2 g/h maintenance, up to ~12 h, retreated if delivery is again imminent. This is the regimen in the largest neuroprotection trial.
• ACTOMgSO4/Crowther regimen — 4 g IV load, then 1 g/h for up to 24 h.
• SA/NDoH practical regimen — 4 g IV load over 15–20 min, then 1 g/h until birth or 24 h (the lower-dose, lower-resource-friendly schedule, and the same infusion that doubles as eclampsia prophylaxis if the woman also has pre-eclampsia).

The differences are loading dose (6 g vs 4 g) and maintenance rate (2 g/h vs 1 g/h) and ceiling duration; there is no head-to-head proof that the higher-dose regimen is superior, so SA uses the lower-dose schedule it already stocks and knows how to monitor. Whichever regimen, the surveillance is identical and clinical: patellar reflexes, respiratory rate (hold if <12/min) and urine output (magnesium is renally cleared — oliguria causes accumulation), with 10% calcium gluconate 10 mL IV drawn up as the antidote. Do not co-load magnesium and nifedipine carelessly — both lower BP and there is an additive neuromuscular-blockade signal.

Tocolysis — the agent choice is a side-effect argument, not an efficacy one. Two appraisal points govern its use. First, tocolysis does not improve perinatal outcome — it delays birth by up to ~48 hours, no more, and that is its only justification: to deliver steroids and arrange in-utero transfer. If steroids are already on board and the fetus is reassuring, the rationale for tocolysis evaporates; and maintenance tocolysis (continuing nifedipine for days after the initial 48 h) confers no benefit and is not done (APOSTEL-II). Second, nifedipine and atosiban produce equivalent perinatal outcomes (APOSTEL III), so SA uses nifedipine because it is oral, EML-listed and cheap, reserving atosiban (expensive, scheme-funded, the oxytocin-receptor antagonist with the cleanest side-effect profile) for the woman who can take neither nifedipine nor indomethacin. The named-agent cautions: never give nifedipine if SBP ≤100 or DBP ≤60; indomethacin is a <32-week-only agent (ductus arteriosus constriction and oligohydramnios beyond that); and betamimetics (salbutamol/ritodrine) are essentially abandoned for tocolysis because of maternal cardiovascular harm (tachyarrhythmia, pulmonary oedema, hyperglycaemia).

For PPROM, the calculus is different. Give a macrolide — erythromycin 250 mg QDS for 10 days (or until labour) — which prolongs latency and reduces neonatal morbidity; explicitly avoid co-amoxiclav (necrotising enterocolitis). Steroids and magnesium as above. Do not routinely tocolyse PPROM — there is no latency or outcome benefit and you risk masking sepsis. Then manage expectantly to ~37 weeks in the absence of infection, abruption or non-reassuring fetal status, with serial temperature, FHR, CRP/WCC and fetal surveillance. The advanced nuances: (1) the earlier the rupture, the more you weigh latency against the rising risk of occult chorioamnionitis — at very early gestations the expectant period is long but the infection risk is high, and the woman must be counselled on pulmonary hypoplasia and limb-contracture risk if rupture is pre-viable; (2) outpatient expectant management of stable PPROM is practised in some well-resourced settings but is generally unsafe in the SA context where rapid readmission for sepsis cannot be guaranteed — keep these women in a facility that can deliver; (3) GBS-positive PPROM needs the GBS regimen layered on, not the erythromycin instead of it.

Subtype-specific overrides. Bleeding-driven (abruption) preterm labour is not tocolysed and is delivered if the fetus or mother is compromised. Infection-driven labour (overt or proven occult chorioamnionitis) is delivered with broad-spectrum antibiotics regardless of gestation — tocolysis is contraindicated. Mechanical/multiple-pregnancy preterm labour is managed as above but with a low threshold for transfer because of the higher complication rate and the malpresentation problem.

Ongoing — when to stop prolonging. Deliver regardless of gestation for chorioamnionitis, abruption, cord prolapse or a non-reassuring CTG. In the SA system this is where structure decides care: a district hospital stabilises and transfers in-utero (steroids started, magnesium running, tocolysis to cover transfer) to a regional/tertiary unit with neonatal capacity — the preterm baby is far better moved before birth than after, because neonatal retrieval of an already-born extremely-preterm infant carries far worse outcomes than antenatal transfer. Below ~26–27 weeks, management is a shared, individualised, paediatrician-involved decision (active vs comfort-focused care) anchored honestly in local neonatal survival data, not a protocol or a textbook survival curve from a high-income NICU.

Long-term. Group B Streptococcus intrapartum prophylaxis for confirmed preterm labour; mode of delivery is obstetric (preterm vaginal birth is not contraindicated, but malpresentation is common and a preterm breech is a caesarean discussion). Counsel on recurrence (a prior spontaneous preterm birth is the strongest single risk factor — roughly a 2–3× increase, higher and earlier the earlier this birth was) and the prevention plan for the next pregnancy: early-pregnancy cervical-length surveillance, vaginal progesterone for a short cervix or a prior spontaneous preterm birth, and cerclage for the woman with cervical insufficiency or a progressively shortening cervix despite progesterone. The postnatal debrief should be explicit about the subtype identified this time, because it shapes the next plan (an infection-driven 26-weeker and a cervical-insufficiency 22-weeker get different prevention strategies).

The evidence & the controversy

The biggest shift in a generation is that antibiotics for intact-membrane preterm labour are now contraindicated, not optional. ORACLE II found no neonatal benefit; its 7-year follow-up (the ORACLE Children Study) then showed harm — more functional impairment and roughly a doubling of cerebral palsy with erythromycin (OR 1.93) and co-amoxiclav (OR 1.69). This is the exact opposite of PPROM, where ORACLE I showed erythromycin helps. The discriminator is membranes: give the macrolide for ruptured membranes, withhold antibiotics for intact-membrane labour unless there is GBS prophylaxis or proven infection. Confusing these two is a common error on this topic.

Magnesium neuroprotection is one of obstetrics' clearest, cheapest wins — but the gestational ceiling is genuinely contested, and the trial-by-trial story explains why. The two pivotal RCTs reached apparently opposite primary results that resolve in the meta-analysis. ACTOMgSO4 (Crowther 2003, <30 weeks) did not reach statistical significance on its primary outcome but trended toward less death and cerebral palsy. BEAM (Rouse 2008, 24–31 weeks, n=2241) also showed no difference on its composite primary outcome (death-or-moderate/severe-CP 11.3% vs 11.7%), but in its pre-specified secondary analysis moderate-to-severe cerebral palsy fell from 3.5% to 1.9% (RR 0.55, 95% CI 0.32–0.95) — the result that, pooled with the others, drove the practice change. Doyle's Cochrane synthesis (5 trials, 6145 infants) gives a cerebral-palsy RR 0.68 with an NNT of 63. The unresolved question is the upper bound: the benefit is unequivocal below 30 weeks, but MAGENTA (Crowther 2023, 30–34 weeks) found survival-without-cerebral-palsy essentially identical (97.2% vs 97.6%), which is why NICE offers magnesium 24+0–29+6 and only considers it to 33+6, and SA caps it around 32 weeks. The marginal neonate at 31–33 weeks is precisely where guidelines legitimately diverge, and the absolute benefit shrinks as the baseline CP risk falls with advancing gestation.

Worked NNT (magnesium, BEAM secondary outcome). Moderate/severe CP 3.5% (placebo) → 1.9% (magnesium). ARR = 3.5% − 1.9% = 1.6% = 0.016. NNT = 1 ÷ 0.016 ≈ 63 to prevent one case of moderate/severe cerebral palsy — which matches the Cochrane pooled NNT, for a cheap, EML-listed, single-dose-window intervention.

For late-preterm PPROM the pendulum has swung decisively to expectant management. PPROMT (n=1839, 34+0–36+6 weeks) found immediate delivery did not reduce neonatal sepsis (the feared complication) and actively caused more respiratory distress, more ventilation and longer NICU stays — prematurity iatrogenically inflicted. The reflex to "deliver because the membranes are gone" is the error: in the absence of infection or fetal compromise, waiting to ~37 weeks is the evidence-based answer, and PPROMT supports it.

Antenatal steroids: settled benefit, two live controversies. The Cochrane review (neonatal death RR 0.78, RDS RR 0.71) is among the strongest in the field, and WHO ACTION-I extended that proof directly into low-resource African and Asian settings — establishing SA external validity, since it showed a survival benefit without the feared increase in maternal infection. The two live arguments are (1) repeat/rescue courses — a single rescue course is reasonable if >7 days have passed and birth is again imminent, but multiple courses risk reduced birthweight and head circumference, so they are not given liberally; and (2) the late-preterm margin — ALPS (Gyamfi-Bannerman 2016, 34+0–36+5 weeks, n=2827) showed betamethasone reduced the respiratory primary outcome from 14.4% to 11.6% (RR 0.80, 95% CI 0.66–0.97) but increased neonatal hypoglycaemia from 15.0% to 24.0% (RR 1.60, 95% CI 1.37–1.87). That is a genuine trade-off: a respiratory benefit bought with a hypoglycaemia harm, in babies who will mostly do well anyway — which is why late-preterm steroids are given selectively, with neonatal glucose monitoring, rather than reflexively, and why a fetus already mature or already committed to immediate delivery for another reason gains little.

Worked numbers (ALPS). Respiratory outcome: ARR = 14.4% − 11.6% = 2.8% → NNT = 1 ÷ 0.028 ≈ 36 to prevent one respiratory event. Hypoglycaemia: ARI = 24.0% − 15.0% = 9.0% → NNH = 1 ÷ 0.09 ≈ 11 to cause one episode of neonatal hypoglycaemia. The NNT (36) being larger than the NNH (11) captures the controversy: you treat ~36 to help one lung but harm ~1 glucose in every ~11 — defensible only with neonatal glucose surveillance and a clear indication.

Tocolysis: on appraisal, it changes nothing that matters. APOSTEL III established nifedipine–atosiban equivalence (composite adverse perinatal outcome 14% vs 15%, RR 0.91) — so the choice is cost and side-effects, not efficacy. APOSTEL-II then showed that maintenance nifedipine after the initial 48 h does not reduce adverse perinatal outcome and, on 2-year follow-up, raised a fine-motor-problem signal — closing the door on prolonged tocolysis. Tocolyse once, briefly, to land the steroid course and effect transfer, then stop.

Landmark trials & key evidence

The antibiotic evidence splits by membrane status, the magnesium and steroid evidence by gestational window, and the delivery-vs-expectant evidence by gestation — each trial maps onto a specific clinical situation.

Worked viva — how to structure the answer

Examiners give a stem like "a 28-year-old, 29 weeks, 30 minutes of regular painful contractions, cervix 3 cm dilated, membranes intact, CTG reassuring, no fever." A high-scoring answer runs:

1. Frame it — "This is established preterm labour at 29 weeks with intact membranes; my goal is not to stop the birth but to buy the 48-hour window for the interventions that change the baby's outcome — steroids and magnesium — and to deliver her in a unit with neonatal ICU."
2. Confirm and exclude the override — confirm labour (cervical change present), and actively exclude occult chorioamnionitis (maternal/fetal tachycardia, CRP/WCC trend) and abruption, because either flips the plan to deliver.
3. Give what works, in the window — antenatal corticosteroids now (betamethasone 12 mg, repeat 24 h), magnesium neuroprotection (4 g load, 1 g/h — this gestation is squarely within the window), and brief tocolysis with nifedipine for the sole purpose of completing steroids and arranging transfer. No antibiotics (intact membranes, no infection, no GBS indication yet).
4. Decide the place — "If I am at a district hospital without neonatal ICU, I stabilise and transfer her in-utero, with steroids in and magnesium running, escorted."
5. Justify from evidence — the membrane-status antibiotic split (ORACLE I vs the ORACLE Children Study), magnesium neuroprotection (BEAM/Cochrane, NNT 63 below ~30 weeks), tocolysis-equivalence-and-futility (APOSTEL III/II), and — if the gestation were 34–36 weeks — the ALPS steroid trade-off.
6. Contrast with PPROM — show you know the calculus differs: macrolide, no routine tocolysis, expectant to ~37 weeks, PPROMT for the late-preterm case.
7. Close the loop — GBS prophylaxis in labour, obstetric mode of delivery, recurrence counselling and the next-pregnancy prevention plan (cervical-length surveillance + vaginal progesterone), debriefed by subtype.

Exam traps & red flags

• Giving antibiotics in intact-membrane preterm labour. Unless there is infection or GBS prophylaxis, antibiotics are harmful here (ORACLE Children Study — cerebral palsy). Reserve the macrolide for PPROM.
• Using co-amoxiclav for PPROM. Erythromycin, not co-amoxiclav — the latter causes necrotising enterocolitis (ORACLE I).
• Tocolysing to "win the baby more time" once steroids are in. Tocolysis buys ~48 h and changes no perinatal outcome; its only job is the steroid window and transfer. Prolonging beyond that exposes the mother to drug effects for no fetal gain, and is dangerous if it masks chorioamnionitis. Maintenance tocolysis is not done (APOSTEL-II — no benefit, a fine-motor signal).
• Treating a positive fetal fibronectin as a reason to act. fFN earns its keep as a rule-out (high NPV); its PPV is modest, and it is invalidated by recent intercourse, bleeding, lubricant or a digital exam — take it before you touch the cervix or not at all.
• Digital vaginal examination in suspected PPROM. It shortens latency and introduces infection — speculum only.
• Delivering late-preterm PPROM reflexively. Without infection or fetal compromise, immediate delivery causes more respiratory morbidity, not less sepsis (PPROMT).
• Missing occult chorioamnionitis. It is the one diagnosis that flips the whole plan to deliver now regardless of gestation; persistent maternal/fetal tachycardia, a creeping CRP trend, or labour that will not tocolyse is intra-amniotic infection until proven otherwise — and a single normal CRP does not exclude it.
• Indomethacin at or beyond 32 weeks. Ductus arteriosus constriction and oligohydramnios — it is an under-32-week-only agent.
• Reflex late-preterm steroids without weighing hypoglycaemia. ALPS bought a respiratory benefit at the cost of more neonatal hypoglycaemia (NNT ~36 vs NNH ~11); give selectively, with neonatal glucose monitoring, not as a reflex.
• Expectant management of the very preterm fetus at a district hospital. Below ~32 weeks the safe move is stabilise (steroids, magnesium, tocolysis to cover transfer) and transfer in-utero to a unit with neonatal ICU.
• Forgetting magnesium toxicity surveillance. Reflexes, respiratory rate and urine output; 10% calcium gluconate is the antidote.
• Counselling pre-viable PPROM without the pulmonary-hypoplasia conversation. Rupture before ~22–24 weeks risks lethal pulmonary hypoplasia and limb contractures — that prognosis belongs in the counselling, not just the latency figure.

Evidence anchors

• SASOG/BetterObs — Preterm Labour with Intact Membranes v2.0 (2023)
• NICE NG25 — Preterm labour and birth (recommendations, 2022 update)
• NICE QS135 statement 6 — magnesium sulfate 24+0 to 29+6 weeks
• RCOG Green-top Guideline No. 73 — suspected PPROM from 24+0 weeks
• NDoH Guidelines for Maternity Care in South Africa, 4th ed.
• ORACLE I — broad-spectrum antibiotics for PPROM (Lancet 2001)
• ORACLE Children Study II — 7-year outcomes, intact-membrane preterm labour (Lancet 2008)
• BEAM — magnesium sulphate for prevention of cerebral palsy (NEJM 2008)
• MAGENTA — magnesium sulphate at 30–34 weeks for neuroprotection (JAMA 2023)
• Cochrane — MgSO₄ for fetal neuroprotection (Doyle 2009)
• Cochrane — antenatal corticosteroids (McGoldrick 2020)
• ALPS — antenatal betamethasone for late preterm delivery (NEJM 2016)
• WHO ACTION-I — antenatal dexamethasone in low-resource countries (NEJM 2020)
• APOSTEL III — nifedipine vs atosiban (Lancet 2016)
• PPROMT — immediate delivery vs expectant management, late-preterm PPROM (Lancet 2016)
• Romero — vaginal progesterone for short cervix, IPD meta-analysis (AJOG 2018)