The intervention that does the real work — vaginal progesterone for a sonographically short cervix — is also one of the cheapest and least intrusive medical recommendations in this catalogue: a generic capsule at bedtime, a few hundred dollars over the whole course, no needles, no surgery in the typical case. The evidence is solid where the protocol is tight (universal cervical-length screening plus progesterone for short cervix); it gets messier at the edges (the women whose cervix looks fine but whose last baby came early; the cerclage decision; what to do for the racial gap that pharmacology won't close). Most of what you read about preventing preterm birth — bed rest, periodontal scaling, the injectable progesterone shot — either never worked or stopped working under harder replication. This is the residue that did.
Preterm birth isn't one disease — it's a final common picture that several different things can drive a pregnancy toward. The big five: subclinical infection that creeps up from the vagina into the membranes; a piece of placenta that bleeds; a uterus stretched too thin by twins or extra fluid; a body marinated long enough in stress hormones that the brain's labour cascade flips on; and a cervix — the gate at the bottom of the uterus — that quietly gives way before it should.
The cervix matters because it's the mechanical gatekeeper for most of the above. As the second trimester progresses, a healthy cervix stays long and closed; an at-risk one shortens, the internal opening starts to funnel, and the barrier between the baby and the vaginal flora thins. Once that's happening, several of the other pathways accelerate at once. Which is why one cheap measurement — the length of the cervix on a transvaginal ultrasound at 18–24 weeks — is, on its own, the single most powerful predictor we have of whether the pregnancy will deliver early Iams et al. 1996.
The other piece of the mechanism story is progesterone, which is the chemical signal that keeps the uterus quiet through pregnancy. It calms the muscle, suppresses the labour prostaglandins, and stabilises the cervical scaffolding. A local progesterone signal that fades early — even when the woman's blood levels look fine — appears to be one of the triggers for the contractions and cervical ripening that end pregnancies prematurely. A capsule of natural micronised progesterone placed in the vagina at bedtime gets to the uterine wall through the local blood supply, top up that signal where it matters, and barely registers anywhere else in the body. That's the entire pharmacological rationale: replace what the gate was supposed to be making for itself.
What we actually know works
Three things have survived a generation of trials. Cervical-length screening predicts preterm birth better than any other single measurement. Vaginal progesterone, started when the cervix is found short, prevents a meaningful slice of those preterm births. Cervical cerclage — a stitch around the cervix — helps a narrower group still, mainly women with both a prior preterm birth and a short cervix in this pregnancy. The other interventions you'll read about have either never worked or no longer work.
The vaginal-progesterone story has two anchor trials. Fonseca in 2007 randomised 250 women with a very short cervix (≤15 mm at 20–25 weeks) to a nightly 200 mg capsule or placebo; deliveries before 34 weeks fell from 34% to 19% Fonseca et al. 2007. The PREGNANT trial (Hassan et al. 2011) studied women between 10 and 20 mm and used 90 mg of progesterone gel; deliveries before 33 weeks fell from 16% to 9%, and serious breathing trouble in the newborns dropped by 61% Hassan et al. 2011. The 2018 patient-level meta-analysis pooled five trials and found a roughly 40% reduction in delivery before 33 weeks, the same in major newborn complications, and no signal of harm to mother or child Romero et al. 2018. The 2021 EPPPIC collaboration confirmed the picture across 31 trials and 11,644 women EPPPIC 2021.
It is not all one-sided. The biggest single trial of vaginal progesterone in a broader high-risk population — OPPTIMUM (Norman et al. 2016), in 1,228 women — was null on its primary outcomes Norman et al. 2016. The skeptical reading is that the earlier signals don't hold in unselected high-risk women; the prevailing reading, and the one the meta-analyses support, is that the benefit is concentrated in women with a short cervix and that OPPTIMUM's broader recruitment diluted it. Either way the headline holds for the short-cervix subgroup, which is the one a screening program actually flags.
The cerclage evidence is narrower. Owen 2009 randomised 302 women who had both a prior preterm birth and a cervix that shortened below 25 mm in this pregnancy; cerclage cut delivery before 35 weeks from 42% to 32%, with the clearest benefit when the cervix fell below 15 mm Owen et al. 2009. The patient-level meta-analysis confirmed cerclage works in this group — and does not help when the woman has a short cervix but no history of prior preterm birth Berghella et al. 2017.
Two interventions worth naming because so many women have heard of them and they have not survived: the weekly hydroxyprogesterone caproate injection (Makena, "the P-shot"), and the cervical pessary. The 2003 Meis trial made the case for the progesterone shot in women with prior preterm birth; the 2019 confirmatory trial (PROLONG) failed to replicate any effect, and the FDA withdrew Makena's approval in April 2023 Blackwell et al. 2020, FDA 2023. The cervical pessary had one positive Spanish trial and a string of negative follow-ups; it is not part of current U.S. guidelines outside research ACOG 2021.
Why the gestational week is the unit that matters
The reason to take any of this seriously is what's on the other side of an early birth. The thing to understand is that "preterm" is not one outcome — it's a cliff, and the slope changes steeply with each week back from term. A baby born at 36 weeks goes home with you and is mostly fine. A baby born at 32 weeks spends weeks in the neonatal intensive care unit and most of them come out well. A baby born at 28 weeks faces a serious chance of breathing trouble, brain bleeds, and developmental disability — and a real chance of not surviving. At 24 weeks the conversation in the delivery room is no longer about whether to resuscitate but whether to.
The Norwegian national registry put numbers on the gradient. Compared with babies born at term, the relative risk of cerebral palsy was 78.9 at 23–27 weeks, 31.7 at 28–30 weeks, and 13.0 at 31–33 weeks. Even at 34–36 weeks — late preterm, the slice of preterm births most people don't worry about — the registry found elevated rates of intellectual disability, schooling deficits, and disability benefits at age 30 Moster et al. 2008. The Swedish cohort of 670,000 children followed into young adulthood showed a graded increase in all-cause mortality through age 36 with each week of earlier birth — the gradient is not just an infancy phenomenon, it carries forward Crump et al. 2011. Among the smallest survivors (under 26 weeks), the British EPICure cohort found that at age 6, about 22% had severe disability and another 24% had moderate Marlow et al. 2005.
The mathematics of the prevention is the same gradient run in reverse. Every week the cervix holds past 28 weeks roughly halves the absolute risk of major disability. A nightly capsule that buys, on average, two or three weeks of gestation in a short-cervix pregnancy is not a small thing — it is, statistically, a different child's life. There is also the woman's side of it. The mother who spends six weeks of three-hour pumping cycles next to an isolette has measurably elevated rates of postpartum PTSD, depression, and partnership strain compared to mothers who took the baby home in the normal way. That part is not the headline of the trial but it is real Saigal & Doyle 2008.
What to actually do
The current ACOG and Society for Maternal-Fetal Medicine guidance has settled into a small number of clear moves ACOG 2021. Take them in order at the visits where they belong.
Two things to know about the progesterone in particular. First, it works in proportion to how diligently it is used. Studies that allowed lower adherence found smaller effects. Set a reminder. Second, the prescription is for natural micronised progesterone, not the synthetic injection that used to be called Makena — those are different drugs with different evidence bases and the injection no longer has FDA approval FDA 2023.
When the standard moves don't apply
The interventions here are unusually safe in the populations they're indicated for. Most of the caveats are not about real risk to the woman taking the medication — they're about not reaching for the wrong tool, or one whose evidence has since collapsed.
The thing none of this is: a substitute for first-trimester obstetric care. The whole framework rests on the woman being in care early enough that the screening, the medication, and the surveillance schedule are set up before they're needed. Late entry to prenatal care is itself one of the larger preventable risk factors for the preterm-birth distribution.
Who needs this most
The screening question applies to every pregnant woman; the risk profile is not uniform.
Prior spontaneous preterm birth is the strongest single risk factor. One prior preterm birth before 37 weeks carries a recurrence risk of about 16% — roughly double the population baseline. Two prior preterm births roughly double that again Iams et al. 1998. If you've had a prior preterm birth, your next pregnancy is managed with serial cervical-length checks every 1–2 weeks from 16 to 24 weeks, vaginal progesterone usually from 16 weeks regardless of cervical length, and cerclage if the cervix shortens — the prevention budget is spent more aggressively here.
Black women in the United States have a preterm birth rate of 14.6%, versus 9.4% for white women — a gap that persists after adjustment for income, education, and other markers of socioeconomic status Walani 2020. The mechanisms are thought to be cumulative-stress mediated rather than genetic, and the drugs in this article do not close it. What does shift it, in smaller amounts, is access to early and consistent prenatal care, group-based prenatal programs (CenteringPregnancy), and doula support during pregnancy and labour. These don't replace the medical interventions; they sit alongside them.
Cervical history. Prior cervical conisation, multiple LEEP (loop) procedures for abnormal Pap smears, or repeated dilation-and-curettage procedures all raise the risk of cervical insufficiency. Müllerian anomalies (a uterus with a developmental variant) and the rare in-utero DES exposure (women born before 1971 in some places) also elevate risk. Any of these should be on the obstetrician's chart at the first prenatal visit; serial cervical-length monitoring is reasonable even without prior preterm birth.
Other things that genuinely raise risk. Smoking during pregnancy roughly doubles the rate — quitting at any point in pregnancy reduces risk, with the largest effect before pregnancy or in the first trimester Cnattingius 2010. Cocaine and methamphetamine elevate risk substantially. Interpregnancy interval under 18 months — and especially under 6 — raises risk. Underweight pre-pregnancy (BMI under 18.5) raises risk. Multifetal pregnancy: around 60% of twins and 93% of triplets deliver before 37 weeks, by mechanisms (uterine overdistension) that the cervical-length toolkit doesn't fully address.
Things that didn't pan out. Periodontal disease is statistically associated with preterm birth, but treating gum disease during pregnancy does not reduce preterm birth. Screening unselected pregnancies for bacterial vaginosis does not reduce preterm birth. These are real risk factors at the population level; intervening on them in late pregnancy doesn't translate to fewer preterm deliveries.
The things widely repeated that are wrong
"Bed rest prevents preterm labour." It doesn't. Multiple trials and a Cochrane review show no benefit, and possible harm — blood clots, muscle deconditioning, depression, lost income. It is no longer a guideline-supported recommendation, but it persists in clinical practice, particularly outside maternal-fetal-medicine centers ACOG 2021.
"The progesterone shot prevents preterm birth." Used to be the standard answer for women with prior preterm birth, on the strength of the 2003 Meis trial. The 2019 confirmatory trial — larger, multinational — found no effect. The FDA withdrew approval of Makena in April 2023. The natural micronised vaginal progesterone capsule and the synthetic injectable caproate are pharmacologically different drugs; the case for one did not transfer to the other Blackwell et al. 2020, FDA 2023.
"If you've had a prior preterm birth but the cervix looks normal, there's nothing to do." The conservative reading of the evidence here is that vaginal progesterone is reasonable in this subgroup; the 2021 meta-analysis showed benefit but the largest single trial (OPPTIMUM) was null EPPPIC 2021, Norman et al. 2016. Most U.S. maternal-fetal-medicine specialists offer it; "no, the cervix is fine, you're done" is not the consensus answer.
"Cervical length screening catches almost all preterm births." It doesn't. Most preterm births happen in women whose mid-trimester cervix looked normal, because plenty of the pathways to preterm birth don't run through cervical shortening (placental abruption, infection-triggered labour, multifetal overdistension). Cervical length is a powerful predictor of the births it does catch, but it isn't most of them Esplin et al. 2017. The screening is still worth doing because the women it does flag are the women progesterone helps most.
"Cerclage helps anyone with a short cervix." Only in combination with a prior preterm birth. In women with a short cervix but no prior preterm birth, the patient-level meta-analysis shows no benefit; the procedure carries real risk; it's the wrong call Berghella et al. 2017.
"Periodontal cleaning during pregnancy prevents preterm birth." Tried in large randomised trials; doesn't work. Gum disease is still associated with preterm birth at the population level — the association is real but the intervention isn't the lever.
Where the chain breaks in practice
Most of the failures aren't biological. They are the gap between what the guidelines say and what happens in the room.
- The cervical-length measurement gets skipped at 20 weeks. Many U.S. clinics still do transabdominal-only anatomy scans; some bill the patient extra for transvaginal screening and don't push it. Ask in advance. If the answer is "we don't do that here," it is fair to ask whether they can refer you for one, or to bring it up before the appointment so they have time to plan.
- A short cervix is found and progesterone isn't started promptly. The shortening is already happening; the window between finding it and stabilising the cervix matters. Days, not weeks. If the result comes back short, the prescription should come the same day or the next.
- Adherence drifts past 28–32 weeks. The woman feels fine; the nightly capsule is the most easily forgotten thing in a day; the perceived urgency fades right as the third trimester begins. The trial data are clearest when women take it. Don't quit at 32 weeks. Stop at 36, or per your obstetrician's instruction.
- Cerclage is offered outside its indication. Cerclage for short cervix without prior preterm birth doesn't work; cerclage for a known cervical incompetence with the right history does. If a cerclage is being offered, the indication should be specific — and the alternative (vaginal progesterone alone) should be on the table for comparison Conde-Agudelo et al. 2018.
- The injectable progesterone is still prescribed. Months after the FDA withdrawal, some practices still default to Makena out of habit. If it's offered, ask why — the evidence and the regulatory status have moved.
- Bed rest is recommended anyway. Common, doesn't help, can hurt. Ask what specifically the rest is meant to achieve.
- The wrong baseline. A woman with a prior preterm birth who isn't asked about it at the first prenatal visit will not be enrolled in the more aggressive surveillance schedule. Volunteer the history; bring records if you have them.
What changes when the intervention works
The mechanism is gestational weeks. Each week the cervix holds past about 28 is, in disability-and-mortality terms, a different child. Past 32 the curve flattens out fast. Past 36 you are essentially back at term-baby outcomes. The intervention's job is to drag the delivery date later — sometimes by weeks, occasionally by months — into the part of the curve where the consequences are no longer the kind a family rearranges itself around for a decade.
What that looks like in a life: a delivery that happens on schedule, with a baby that cries and breathes and goes home with you. A first year that is sleep-deprived in the ordinary way, not the bedside-vigil way. Photos taken at home, not against the green walls of a NICU. Developmental milestones landing on a normal schedule, because the third trimester of brain development happened where it was meant to — inside the uterus — instead of being interrupted at 28 weeks and finished in an isolette under fluorescent light Moster et al. 2008.
Within months: the postpartum recovery without the prolonged hospital separation and its documented hit to maternal mental health Saigal & Doyle 2008. Within a year: a child whose developmental trajectory is on the normal track rather than under specialist surveillance. Across childhood: the absence of cerebral palsy, intellectual disability, chronic lung disease, retinopathy — outcomes whose absolute risk is roughly halved by every week of gestation gained between 28 and 32 weeks Crump et al. 2011. Into adulthood: the long-tail benefit the Norwegian registry quantified — the fact that gestational age at birth is still showing up in the data at age 30, in everything from education to employment to all-cause mortality. The cheap nightly capsule the woman placed before sleep through her second and third trimester is doing arithmetic that won't be fully visible for thirty years.
The honest qualifier: in the short-cervix subgroup, you need to treat roughly 14 women to prevent one preterm birth before 33 weeks. Most women on the protocol weren't going to deliver early. Some women on the protocol still deliver early — the intervention reduces the probability, not the universe of outcomes. The arithmetic is in the population, not in any individual case. But the population includes you.
Related ground worth walking next: preeclampsia prevention, since aspirin handles a different pathway to early delivery; group prenatal care and doula support, which carry the closest thing to evidence for shifting outcomes the drugs don't move; antenatal corticosteroids, the acute rescue when preterm labour is already happening — it is not prevention, but it is the next thing that buys you outcomes once prevention has not held; smoking cessation during pregnancy, the largest modifiable risk factor most women can act on directly; and cervical conisation and LEEP procedures, the elective cervical surgeries that change preterm-birth risk in future pregnancies and are worth knowing about before they happen.
Substance and claimed effects
Preterm birth — delivery before 37 completed weeks of gestation — is the leading cause of neonatal mortality worldwide and a top cause of long-term neurodevelopmental disability among survivors Goldenberg et al. 2008, Walani 2020. Globally about 13.4 million babies are born preterm each year (≈10% of live births); roughly 900,000 die in 2019 from direct preterm complications Walani 2020. The U.S. rate sits stubbornly near 10.4% (2022 vital statistics). This entry covers the modern, evidence-supported toolkit aimed at spontaneous preterm birth (sPTB) in singleton pregnancies: (1) universal mid-trimester transvaginal cervical-length (TVCL) screening, (2) vaginal progesterone for sonographically short cervix, (3) history-indicated and ultrasound-indicated cervical cerclage, and (4) management of women with prior spontaneous preterm birth. Claimed effects span (a) reduction in spontaneous preterm birth itself (<37, <34, <28 weeks), (b) reduction in neonatal mortality and major morbidity (respiratory distress, intraventricular hemorrhage, necrotizing enterocolitis, NICU admission), and (c) downstream reductions in childhood disability — cerebral palsy, cognitive impairment, chronic lung disease — and adult mortality and disability that track gestational age at birth Moster et al. 2008, Crump et al. 2011. Effects on maternal health are secondary: avoidance of preeclampsia-driven indicated preterm birth via low-dose aspirin Roberge et al. 2017 and avoidance of the postpartum psychological burden of a NICU stay. Out-of-scope: tocolysis once labour is established, antenatal corticosteroids and magnesium sulfate for fetal neuroprotection (acute rescue, not prevention), and indicated preterm delivery for maternal disease (these are different decision frameworks).
Evidence by addressing question
Mechanism
Spontaneous preterm birth is a syndrome, not a single disease. At least five overlapping pathways converge on a final common picture of cervical ripening, decidual activation, and uterine contractions before term: (i) intrauterine inflammation/infection (often subclinical ascending genitourinary), (ii) decidual hemorrhage (placental abruption), (iii) uterine overdistension (multiples, polyhydramnios), (iv) maternal stress activating the corticotropin-releasing-hormone axis, and (v) cervical insufficiency — a structural failure of the cervix to retain the pregnancy load Goldenberg et al. 2008. The mid-trimester cervix is the final mechanical gatekeeper for several of these pathways: as the internal os shortens and funnels, the barrier function falls, exposing membranes to vaginal flora and predisposing to chorioamnionitis. This is why cervical length, measured by transvaginal ultrasound at 18–24 weeks, is the single most powerful clinical predictor of spontaneous preterm birth in both unselected and high-risk populations Iams et al. 1996.
The biological rationale for progesterone is that progesterone maintains uterine quiescence by down-regulating myometrial gap-junction formation, suppressing prostaglandin synthesis, and stabilising the cervical extracellular matrix. A functional or local progesterone withdrawal — even when serum levels remain high — is thought to trigger the contractile and ripening cascade. Exogenous vaginal progesterone reaches the uterus through first-pass uterine vascular transfer, raising local concentrations to therapeutic ranges with negligible systemic effect. The intramuscular synthetic analogue 17α-hydroxyprogesterone caproate (17-OHPC) acts systemically and has a different pharmacological profile; the divergence between the natural micronised vaginal preparation and the synthetic intramuscular one matters clinically, as PROLONG later showed Blackwell et al. 2020.
Evidence
Cervical length screening. The 1996 Preterm Prediction Study established the dose-response between mid-trimester TVCL and sPTB risk: women in the bottom decile (≤25 mm at 24 weeks) had a six-fold increase in risk of delivery before 35 weeks compared with women at the median (35 mm); risk rose monotonically as length decreased Iams et al. 1996. The Crane & Hutchens 2008 systematic review of TVCL in high-risk asymptomatic women showed sensitivity of 35.7% and specificity of 92% at the <25 mm threshold for delivery <35 weeks, with positive likelihood ratio ≈4.3 Crane & Hutchens 2008. In nulliparous women, the nuMoM2b cohort confirmed serial TVCL predicts sPTB but with modest population-level positive predictive value, since most preterm births occur in women without a short cervix Esplin et al. 2017.
Vaginal progesterone, short cervix, no prior PTB. Fonseca et al. 2007 randomised 250 women with TVCL ≤15 mm at 20–25 weeks to vaginal progesterone 200 mg nightly or placebo; preterm birth <34 weeks fell from 34.4% to 19.2% (RR 0.56, 95% CI 0.36–0.86) Fonseca et al. 2007. The PREGNANT trial (Hassan et al. 2011) studied women with TVCL 10–20 mm at 19–24 weeks; vaginal progesterone gel 90 mg daily cut preterm birth <33 weeks from 16.1% to 8.9% (RR 0.55, 95% CI 0.33–0.92) and reduced respiratory distress syndrome by 61% Hassan et al. 2011. The 2018 individual-patient-data meta-analysis (Romero et al.) pooled five trials (974 women with TVCL ≤25 mm): vaginal progesterone reduced preterm birth <33 weeks (RR 0.62, 95% CI 0.47–0.81), composite neonatal morbidity/mortality (RR 0.59), respiratory distress syndrome (RR 0.47), birth weight <1500 g (RR 0.52), and NICU admission (RR 0.68); no signal for adverse maternal or child outcomes through follow-up Romero et al. 2018. The 2021 EPPPIC individual-participant meta-analysis (31 trials, 11,644 women) confirmed the effect of vaginal progesterone in women with short cervix and/or prior PTB and found a similar magnitude of effect for women with prior PTB regardless of cervical length EPPPIC 2021.
The OPPTIMUM dissent. Norman et al. 2016 randomised 1,228 high-risk women (prior PTB or short cervix or positive fetal fibronectin) to vaginal progesterone 200 mg or placebo and found no significant difference in the composite of obstetric (fetal death or PTB <34 weeks), neonatal (composite morbidity), or two-year childhood outcomes Norman et al. 2016. OPPTIMUM is the most consistently cited skeptic-case evidence; the EPPPIC meta-analysis incorporates it and still finds a net benefit EPPPIC 2021, but the heterogeneity is real and partly explains the muted recommendation strength outside the short-cervix subgroup.
17α-hydroxyprogesterone caproate (17-OHPC). Meis et al. 2003 randomised 463 women with prior spontaneous PTB to weekly intramuscular 17-OHPC 250 mg or placebo from 16–20 weeks; PTB <37 weeks dropped from 54.9% to 36.3% (RR 0.66, 95% CI 0.54–0.81) and <32 weeks from 19.6% to 11.4% Meis et al. 2003. The FDA granted accelerated approval to Makena (17-OHPC) in 2011. The confirmatory PROLONG trial (Blackwell et al. 2020) — multinational, 1,708 women, same indication — found no benefit on either PTB <35 weeks (11.0% vs 11.5%) or neonatal composite morbidity Blackwell et al. 2020. After multi-year regulatory wrangling, the FDA withdrew Makena's approval in April 2023 FDA 2023. The 2021 EPPPIC IPD meta-analysis found weaker, less consistent effects for 17-OHPC than for vaginal progesterone in singleton pregnancies EPPPIC 2021.
Cervical cerclage. Owen et al. 2009 randomised 302 women with prior spontaneous PTB <34 weeks and TVCL <25 mm before 24 weeks to McDonald cerclage or no cerclage; PTB <35 weeks dropped from 42% to 32% (OR 0.55, 95% CI 0.31–0.97) with a clearer benefit at <25 mm and especially at <15 mm Owen et al. 2009. Berghella et al. 2017's IPD meta-analysis showed that ultrasound-indicated cerclage in singleton women without prior PTB does not reduce preterm birth — the protection appears specific to women with both risk factors Berghella et al. 2017. Conde-Agudelo et al. 2018's indirect-comparison meta-analysis found vaginal progesterone and cerclage of similar efficacy in women with both prior PTB and short cervix, favouring progesterone for tolerability and cost Conde-Agudelo et al. 2018.
Cervical pessary. The Spanish PECEP trial (Goya et al. 2012) found pessary reduced PTB <34 weeks from 27% to 6% in short-cervix women Goya et al. 2012, but subsequent replication trials (e.g. ProTWIN, STOPPIT-2) failed to confirm; current ACOG/SMFM guidance does not recommend pessary outside research settings ACOG 2021.
Aspirin. Low-dose aspirin (81–162 mg) initiated before 16 weeks in women at high preeclampsia risk reduces preterm preeclampsia by ≈60% and indirectly reduces indicated preterm birth Roberge et al. 2017. This is a separate prevention pathway operating on the placental rather than the cervical axis.
Practice / clinical consensus
ACOG Practice Bulletin 234 (2021) and SMFM consultations endorse: (a) universal TVCL screening at 18 0/7–22 6/7 weeks in singleton pregnancies, (b) vaginal progesterone (200 mg micronised capsule or 90 mg gel nightly) from diagnosis of a short cervix (≤25 mm) until 36 weeks, (c) history-indicated cerclage at 13–14 weeks for women with three or more prior second-trimester losses or PTBs, (d) ultrasound-indicated cerclage between 16 and 23+6 weeks for women with prior sPTB and TVCL <25 mm, and (e) consideration of physical-exam-indicated (rescue) cerclage for asymptomatic cervical dilation in the second trimester ACOG 2021. After the PROLONG trial and FDA Makena withdrawal, ACOG and SMFM updated their guidance: 17-OHPC is no longer recommended; vaginal progesterone is the progesterone of choice for both short-cervix and prior-PTB indications SMFM 2016, FDA 2023. WHO 2022 recommendations align broadly with these positions in their preterm-prevention chapter, with attention to context-dependent feasibility in low- and middle-income settings WHO 2022.
Contraindications and safety
Vaginal progesterone is extremely well tolerated; large RCTs and the Romero IPD meta-analysis found no signal for maternal adverse events, congenital anomalies, or two-year child neurodevelopmental impairment Romero et al. 2018. Common side-effects are vaginal discharge and mild irritation. Cerclage carries small risks of membrane rupture (≈1–9%), infection, cervical laceration at delivery, and (rarely) cervical stenosis; benefit-risk is favourable in the indicated population and unfavourable outside it Berghella et al. 2017, Owen et al. 2009. 17-OHPC carried injection-site reactions, occasional gestational diabetes signal, and (after PROLONG) no demonstrable efficacy in singleton pregnancies Blackwell et al. 2020. Aspirin 81–162 mg daily from before 16 weeks has a small bleeding risk that is outweighed by the preeclampsia/PTB benefit in indicated women Roberge et al. 2017.
Misconceptions
(1) "Bed rest prevents preterm birth." It does not; multiple trials show no benefit and possible harm (thromboembolism, deconditioning) ACOG 2021. (2) "If progesterone failed in OPPTIMUM it doesn't work." The IPD meta-analyses including OPPTIMUM still demonstrate net benefit in the short-cervix subgroup Romero et al. 2018, EPPPIC 2021. (3) "All progesterone is the same." Vaginal micronised progesterone and intramuscular 17-OHPC are pharmacologically distinct and have diverged in trial outcomes Blackwell et al. 2020, EPPPIC 2021. (4) "Cerclage helps anyone with a short cervix." No — it benefits women with both prior PTB and short cervix; in women without prior PTB it is ineffective and exposes them to procedural risk Berghella et al. 2017. (5) "Cervical length screening creates unnecessary anxiety." Universal TVCL is cost-effective in U.S. modelling when paired with vaginal progesterone; the screening pathway is brief, the threshold is binary, and the intervention is low-burden.
Audience and population variability
Risk varies substantially by demographic and clinical context. Prior spontaneous PTB confers the strongest single risk: women with one prior sPTB <37 weeks have a 16% recurrence risk; with one prior <28 weeks the recurrence risk for <28 weeks alone is ≈10%; two prior sPTBs roughly doubles recurrence again Iams et al. 1998. Black women in the United States have preterm birth rates of 14.6% versus 9.4% in non-Hispanic white women — a disparity that holds even after adjustment for socioeconomic status, suggesting structural and chronic-stress mechanisms in addition to biology Walani 2020. Other strong risk factors: short interpregnancy interval (<18 months), low pre-pregnancy BMI, in-utero DES exposure, prior cervical conisation/LEEP, müllerian anomalies, multifetal pregnancy (60% of twins deliver <37 weeks), assisted reproductive technology, smoking during pregnancy, periodontal disease (association, intervention does not reduce risk), and uncontrolled chronic conditions (hypertension, diabetes). Vaginal progesterone's efficacy is most consistent in women with a sonographic short cervix; in EPPPIC the singleton-prior-PTB subgroup also benefited, but heterogeneity remained. Cerclage's strongest evidence is in women with both prior PTB and short cervix. Aspirin's preeclampsia/PTB-prevention effect is concentrated in women with one or more high-risk factors (chronic hypertension, prior preeclampsia, pre-gestational diabetes, autoimmune disease, multifetal pregnancy) Roberge et al. 2017.
Failure modes and practicalities
The intervention chain breaks at four predictable points: (1) the 20-week anatomy scan does not include TVCL (still common in lower-resource U.S. practices and globally) — transabdominal screening misses about half of short cervices; (2) a short cervix is found but progesterone is not started promptly — every day of delay matters when the cervix is already shortening; (3) progesterone is prescribed but adherence falters around weeks 28–34 when the woman feels fine and the nightly insertion becomes burdensome; (4) cerclage is offered too late, beyond 23+6 weeks, or rescue cerclage is offered without explicit benefit-risk discussion. Practical realities: vaginal progesterone is roughly $40–80/month generic (micronised capsules) and $300+/month branded gel in the U.S.; insurance coverage is broad post-Hassan but not universal. Cerclage is typically covered as a surgical procedure. Universal TVCL screening adds 5–10 minutes to the anatomy ultrasound and is reimbursed separately. Adherence to nightly insertion improves with patient education on the mechanism and on the steeply concentrated efficacy window (16–36 weeks).
History
Cervical insufficiency was first described by Lash and Lash in 1950 as a structural anomaly amenable to surgical reinforcement; Shirodkar (1955) and McDonald (1957) developed the modern transvaginal cerclage procedures still in clinical use. Iams et al. 1996 brought cervical length screening from research into mainstream obstetrics with the Preterm Prediction Study's quantification of the length–risk dose response Iams et al. 1996. Da Fonseca's progesterone trial (2003) and the Hassan PREGNANT trial (2011) established vaginal progesterone as a genuine pharmacological prevention; the field's optimism peaked between 2011 and 2015. PROLONG (2019) and the subsequent FDA withdrawal of Makena in 2023 marked the first major retraction of a long-standing FDA-approved obstetric prevention drug, recalibrating the field's expectations of effect-size durability under more rigorous replication Blackwell et al. 2020, FDA 2023.
Stakes
Preterm birth carries dose-dependent consequences across the gestational-age spectrum. Extremely preterm (<28 weeks): even in well-resourced settings, 30–40% mortality, and among survivors roughly 20% have severe disability (cerebral palsy, blindness, deafness, profound cognitive impairment) and a further 25–30% have moderate impairment EXPRESS Group 2009, Wood et al. 2000, Marlow et al. 2005. Very preterm (28–31 weeks): mortality <5% in tertiary NICUs, moderate-to-severe neurodevelopmental impairment ≈10–15%. Moderately/late preterm (32–36 weeks): substantially better short-term outcomes but a residual long-term gradient. The Norwegian registry follow-up (Moster et al. 2008) found relative risks of cerebral palsy of 78.9 at 23–27 weeks, 31.7 at 28–30 weeks, and 13.0 at 31–33 weeks — and even 34–36 weeks carried elevated risk of cerebral palsy, intellectual disability, schooling deficits, and disability pension by age 30 Moster et al. 2008. Crump et al. 2011 in 670,000 Swedes found a graded increase in all-cause mortality through young adulthood with each week-decrement in gestational age, persisting into the late-preterm range Crump et al. 2011. Maternal stakes are secondary but real: emergency cesarean, postpartum hemorrhage, prolonged NICU presence with documented increases in maternal PTSD, depression, and partnership strain.
Payoff
For a woman with a sonographic short cervix on the anatomy scan, vaginal progesterone reduces the probability of delivering before 33 weeks by ≈40% in absolute risk-ratio terms and the probability of major neonatal morbidity composite by ≈40% Romero et al. 2018. The downstream gradient is steep: each week of gestation gained between 24 and 32 weeks roughly halves the absolute risk of major disability. For the woman with prior PTB and short cervix, cerclage plus vaginal progesterone produces additive risk reduction and the absolute payoff is a child of a meaningfully different life trajectory Conde-Agudelo et al. 2018. The number-needed-to-treat in the short-cervix subgroup is roughly 14 to prevent one preterm birth <33 weeks and roughly 25 to prevent one case of major neonatal morbidity — large effect sizes by preventive-medicine standards.
The credibility range
Optimist case. The story is unusually clean for an obstetric intervention. TVCL is cheap, fast, and quantitative, with a well-characterised dose-response established three decades ago and replicated everywhere it has been measured. Vaginal progesterone has positive trials from da Fonseca, Hassan, and others, an individual-patient-data meta-analysis showing benefit on hard outcomes including neonatal mortality/major morbidity, and two-year and longer follow-ups showing no harm Romero et al. 2018. The intervention is biologically plausible (local progesterone withdrawal model), low-cost, low-burden, and additively combinable with cerclage in the strict-indication subgroup. Universal TVCL plus vaginal progesterone for the ≈2% of women found with a short cervix is the highest-ROI obstetric prevention since folic-acid fortification: estimates from U.S. cost-effectiveness models suggest savings of $11–19 million per 100,000 pregnancies and ≈1,000 quality-adjusted life-years gained.
Skeptic case. OPPTIMUM, the largest single trial of vaginal progesterone in high-risk women, was null Norman et al. 2016. PROLONG, the confirmatory trial of 17-OHPC, was null and resulted in regulatory withdrawal Blackwell et al. 2020, FDA 2023 — proof that obstetric prevention effect sizes can vanish under more rigorous replication. The Hassan trial was industry-sponsored. Effect sizes for vaginal progesterone in the broader (non-short-cervix) prior-PTB subgroup are heterogeneous in EPPPIC EPPPIC 2021. Universal TVCL screening shifts a small absolute number of preterm births (most preterm births still occur in women with a normal mid-trimester cervix); the screening is a low-PPV test against the population background risk Esplin et al. 2017. Cerclage is overused in clinical practice relative to its evidence base. And the deepest disparity driver — Black-white disparity in U.S. preterm rates — barely moves with any of these interventions because the upstream causes (chronic stress, structural racism, prenatal-care access) are not pharmacological.
Author's call. For singleton pregnancies, two elements of the toolkit are settled enough to recommend almost categorically: (a) universal mid-trimester TVCL at the anatomy scan, and (b) vaginal progesterone 200 mg nightly from short-cervix diagnosis to 36 weeks. Effect sizes survive IPD meta-analysis even with OPPTIMUM included; biological plausibility is high; harm signal is absent; cost and burden are low. For women with prior spontaneous PTB, vaginal progesterone is the progesterone of choice (17-OHPC is dead post-PROLONG); ultrasound-indicated cerclage is added when TVCL drops below 25 mm. The remaining uncertainties — pessary, vaginal progesterone in the prior-PTB-only subgroup with normal cervix, the upstream social determinants — are real but should not delay action on the established core. Controversy score sits at the middle of the range because the field is genuinely actively debating the prior-PTB question and OPPTIMUM's interpretation, even though the short-cervix-plus-progesterone core is solid.
Stakeholder and incentive map
- Manufacturers. The off-patent micronised vaginal progesterone market is fragmented; Crinone (Allergan/AbbVie) holds the branded vaginal gel approval. AMAG (and successor) marketed Makena (17-OHPC) at ≈$700/dose for years before PROLONG and the FDA withdrawal. The branded/generic split shaped early 2010s practice patterns more than evidence did.
- Professional bodies. ACOG and SMFM (U.S.), RCOG (UK), FIGO, WHO. SMFM's serial consultations have become the de facto American practice anchor; its post-PROLONG pivot away from 17-OHPC was rapid and public.
- Investigators. The "preterm birth network" — Romero (NIH/Detroit), Iams (Ohio State), Nicolaides (King's College London), Berghella (Jefferson), Goldenberg (Columbia) — has shaped the field for three decades. Some have served on industry advisory boards; the IPD-meta-analysis authorship overlaps with the original-trial authorship.
- Patient advocacy. March of Dimes (U.S.) lobbies for universal screening and progesterone access; was a major proponent of keeping Makena available during the FDA process. Tommy's (UK) similarly.
- Regulatory. FDA's Makena withdrawal was the rare instance of an obstetric drug failing confirmatory trial and being withdrawn — a precedent that recalibrates the field's tolerance for accelerated approval based on one positive trial.
- Skeptic counter-incentive. Evidence-based-obstetrics commentators (e.g. Caughey, Greene) have pushed back against universal screening and aggressive cerclage practice; insurers occasionally resist coverage outside guideline indications.
Population variability
- Prior PTB. Recurrence risk after one spontaneous PTB <37 weeks is ≈16% (vs ≈8% baseline); after one <28 weeks, ≈10% recurrence at <28 weeks specifically; after two prior, recurrence roughly doubles again Iams et al. 1998.
- Race/ethnicity (U.S.). Non-Hispanic Black women: 14.6%. Non-Hispanic white: 9.4%. American Indian/Alaska Native: 12.5%. Asian: 9.0%. The gap persists after adjustment for income, education, parity. Mechanisms thought to be cumulative-stress mediated (allostatic load, chronic inflammation) rather than monogenic Walani 2020.
- Multifetal pregnancy. ~60% of twins, ~93% of triplets deliver <37 weeks. Progesterone has not shown benefit in multifetal pregnancy EPPPIC 2021; cerclage is generally not recommended.
- Cervical history. Prior conization, LEEP, multiple D&Cs, müllerian anomalies, DES exposure all elevate risk and inform individualised TVCL surveillance schedules.
- Smoking and substance use. Smoking during pregnancy ≈doubles PTB risk; cessation any time reduces risk, with the largest effect from cessation before pregnancy or in the first trimester Cnattingius 2010. Cocaine, methamphetamine substantially elevate risk.
- Infection. Untreated bacterial vaginosis modestly increases risk; screening-and-treatment in unselected populations has not reduced PTB. Asymptomatic bacteriuria treatment does reduce pyelonephritis but its effect on PTB is small. Periodontal treatment in pregnancy does not reduce PTB.
- Interpregnancy interval. <18 months elevates risk; <6 months substantially so.
- BMI. Underweight (BMI <18.5) elevates risk; obesity has a more complex relationship (higher indicated PTB via preeclampsia, similar or lower sPTB).
- Age. Adolescents (<17) and women ≥40 have modestly elevated risk; the U-shape is small relative to other risk factors.
Knowledge gaps
The major open questions: (1) Does vaginal progesterone benefit women with prior PTB but normal mid-trimester cervix? EPPPIC suggests yes; OPPTIMUM and PROLONG complicate the picture; ongoing trials may resolve it EPPPIC 2021, Norman et al. 2016. (2) What explains the racial disparity in U.S. PTB rates and what intervention shifts it? Group prenatal care (CenteringPregnancy), doula support, and structural-determinants interventions have small to moderate signals; pharmacology alone does not close the gap. (3) Is there a place for combined vaginal progesterone + cerclage versus cerclage alone in women with both indications? Limited head-to-head data; clinical practice often combines them. (4) Could biomarker-based prediction (cell-free RNA, fetal fibronectin, microbiome signatures) refine risk stratification beyond TVCL? Active research area, no clinically deployable test yet. (5) Are there modifiable upstream contributors to functional progesterone withdrawal — stress, inflammation, microbiome — that could be targeted? Mechanistic plausibility but no trial evidence yet. (6) Long-term (school-age, adult) outcomes of vaginal-progesterone-exposed children are reassuring through follow-up of ≈2 years but extended cohorts will continue accumulating data Romero et al. 2018.
Brief said: "Identifying and reducing the risk of delivery before 37 weeks through cervical length screening, progesterone, and management of prior preterm history. Its effects on neonatal outcomes, maternal health, and long-term child development." Covered all five named threads end to end.
Hard scoping calls:
- Antenatal corticosteroids, magnesium sulfate, tocolytics were kept out of scope. They are acute-rescue interventions once preterm labour is established, not prevention. Flagged in
out-of-scopeas the next thing to read. - Aspirin for preeclampsia prevention is included briefly because it indirectly prevents indicated preterm birth — but is also flagged as warranting its own entry rather than being expanded here. (Future-link candidate:
aspirin-pregnancyorpreeclampsia-prevention.) - 17-OHPC ("the P-shot," Makena) covered in misconceptions and contraindications even though the drug is now withdrawn. The reason: many readers will have heard of it from older obstetric guidance, and several U.S. practices were still defaulting to it months after the April 2023 FDA withdrawal. Worth saying clearly.
- Cervical pessary mentioned briefly under evidence but not given its own treatment. The replication failures (ProTWIN, STOPPIT-2) make a long defence redundant; current ACOG guidance settles it.
- U.S. racial disparity named explicitly in
audienceandknowledge gaps. The author's-call discipline matters here: the pharmacological prevention this article covers does not close the Black-white gap, and saying so honestly is part of not overselling the toolkit. - Multifetal pregnancy mentioned but not expanded. The evidence picture (progesterone null, cerclage generally unhelpful) deserves its own entry — flag as future-link candidate
twin-pregnancy-management.
Audience scoping: audience.gender set to female; no age band set, because advanced maternal-age pregnancy is real and the topic applies across the full reproductive age range.
Rating difficulties:
longevityat 4 reflects the substance's effect on child lifelong mortality and disability, not the mother. This is an unusual case where the dimension is scored against the dyad rather than the reader proper. Argued in the justification.health_short_termat 3 is a compromise — for the mother the direct short-term wellness effect is small (avoiding emergency delivery complications, NICU separation); for the dyad it is large. Landed at 3 as the dyad-honest reading.controversyat 3 reflects active expert debate over OPPTIMUM, the prior-PTB-normal-cervix subgroup, and post-PROLONG practice gaps — not over the short-cervix-plus-progesterone core, which is settled.evidenceat 4 rather than 5 because the 17-OHPC arm of the framework collapsed under PROLONG. A category that loses a long-standing FDA-approved drug under replication isn't a 5 on field maturity, even if the surviving core (vaginal progesterone for short cervix) is well-supported.applicabilityat 3 per the women's-health convention.
Dream narrative: overall score below 40 — narrative was optional but written because the topic genuinely supports the relief lever (the NICU vigil that doesn't happen). The dek and tagline were written from it; the marketing-words ban was lifted in proportion to the score (light crank, not full).
Future-link candidates:
antenatal-corticosteroids— the lung-maturation rescue in established preterm labour.aspirin-pregnancyorpreeclampsia-prevention— the placental-axis prevention arm.group-prenatal-care— the social-determinants intervention that has the closest thing to evidence for shifting outcomes drugs don't move.twin-pregnancy-management— multifetal-specific decision framework.smoking-cessation-pregnancy— the largest modifiable risk factor.cervical-conisation-leep— the elective cervical procedures whose preterm-birth implications women should know about before they happen.
Separate-entry candidates: none arose during this write beyond the future-link list above — the brief's scope is the single substance ("preterm birth prevention as a toolkit") and the toolkit holds together.
Preterm Birth Prevention
Generic vaginal micronised progesterone runs ~$40-80/month for ~5 months ($200-400 total course); transvaginal cervical-length measurement adds 5-10 minutes to a covered anatomy ultrasound. Cerclage is a surgical procedure covered by insurance when indicated.
Each gestational week gained between 24 and 36 weeks roughly halves the absolute risk of major neonatal disability and death; Crump et al. 2011 showed a graded all-cause mortality gradient through young adulthood, and Moster et al. 2008 documented cerebral-palsy relative risks of 78.9 at 23-27 weeks vs term. The intervention reduces lifelong disability burden in the child rather than the mother — large effect on the dyad.
Nightly vaginal insertion of a progesterone capsule from short-cervix diagnosis (around 20 weeks) through 36 weeks is a mild but persistent lifestyle addition; cerclage adds one outpatient procedure and pelvic rest for a period. Cervical-length screening itself is a single ultrasound.
Universal transvaginal cervical-length screening rests on the 1996 Preterm Prediction Study and consistent replication; vaginal progesterone for short cervix is supported by the Fonseca 2007 and PREGNANT (Hassan 2011) RCTs and confirmed by the Romero 2018 IPD meta-analysis and EPPPIC 2021. ACOG, SMFM, and WHO are aligned. Not 5: OPPTIMUM (Norman 2016) was null in the high-risk-mixed population and the PROLONG failure of 17-OHPC (Blackwell 2020) erased one limb of the framework.
For the mother-newborn pair, preventing preterm delivery substantially reduces immediate post-delivery complications (NICU separation, postpartum hemorrhage, emergency cesarean) and lets the woman experience the term-birth recovery rather than the NICU-vigil one. IPD meta-analysis showed composite neonatal morbidity/mortality dropped 41% with vaginal progesterone in short-cervix women (Romero et al. 2018).
NICU stays are a documented risk factor for maternal PTSD, depression, and partnership strain; avoiding the prolonged hospital separation and the trauma of an early delivery has a real, if secondary, effect on maternal mental health in the months after birth.