Re‐analysis of the PREGNANT trial confirms that vaginal progesterone reduces the rate of preterm birth in women with a sonographic short cervix

Two randomized controlled trials reported that vaginal progesterone reduces preterm birth in women with a short cervix1, 2. Progesterone use in women with a mid-trimester cervical length of 10–20 mm was associated with reduced preterm birth < 33 weeks and respiratory distress syndrome (RDS). Results of the PREGNANT trial2 were reviewed by the US Food and Drug Administration (FDA) for a New Drug Application. The FDA subsequently issued a document concluding that the data ‘[did] not support the efficacy of progesterone gel.’3. Given the discrepancy between the FDA findings and the PREGNANT trial, the PREGNANT principal investigator provided the original data and requested an independent re-analysis of the PREGNANT trial, which is reported here. Methodological details are supplied in the supporting information online (Appendix S1). Table S1 describes baseline demographic and treatment characteristics of 458 women randomized to receive progesterone (n = 235) or placebo (n = 223). These 458 women comprised the intention-to-treat (ITT) analysis set. Women allocated to progesterone had a significantly lower rate of preterm birth < 33 weeks than women administered placebo (8.9% vs 16.1%; relative risk (RR) = 0.55 (95% CI, 0.33–0.92), P = 0.02), even after adjustment for study site and risk strata (RR = 0.54 (95% CI, 0.33–0.89), P = 0.01). The progesterone group also experienced significantly fewer preterm births < 28 weeks and < 35 weeks (Table 1). Neonates in the progesterone group experienced significantly less RDS than those born to mothers given placebo, an effect that persisted after adjustment (RR = 0.40 (95% CI, 0.17–0.94), P = 0.03). The unadjusted risk for any morbidity/mortality event was significantly lower in neonates born to mothers in the progesterone group (RR = 0.57 (95% CI, 0.33–0.99), P = 0.04), as were composite morbidity/mortality scores ‘0–4 scale without NICU’ and ‘0–6 scale without NICU’ (both P = 0.05). Progesterone therapy was associated with a significant reduction in birth weight < 1500 g (RR = 0.47 (95% CI, 0.26–0.85), P = 0.01). Results for all neonatal outcomes are given in Table 1. Logistic regression for preterm birth < 33 weeks in the ITT analysis controlled for pooled study site and other risk factors, but the effect of treatment remained essentially unchanged (odds ratio = 0.48 (95% CI, 0.25–0.91), P = 0.02). Results of the ‘treated-patient analysis set’ and ‘compliant analysis set’ are reported in Tables S2 and S3. Re-analysis of the PREGNANT trial by independent academic investigators has replicated in every respect the original results2, which are also consistent with an individual patient data meta-analysis based on five randomized trials including PREGNANT4. We calculate the risk difference (RD) in the meta-analysis for early preterm delivery to be RD = −0.09 (95% CI, −0.14 to 0.04), with number needed to treat = 11 (95% CI, 7–25). One observation in the FDA report merits further comment5. The Cochran–Mantel–Haenszel (CMH) test used in the original analysis2 was considered inappropriate owing to ‘insufficient sample size’ in each stratum and ‘inconsistent treatment effects among strata’5. We disagree. The CMH test in SAS used by us is equivalent to the Mantel–Haenszel conditional test, both providing identical results and a valid stratified analysis even when sample sizes within strata are small6. Given the persistently high prevalence of preterm delivery, there is substantial evidence that administration of vaginal progesterone to women with an asymptomatic sonographic short cervix promises considerable public health benefit. This study was performed through subcontract N01HD23342 between Wayne State University and Yale University, funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health. We are grateful to Livia Kidd for technical assistance in the preparation of this manuscript. The authors declare no conflicts of interest. L. A. McKay†‡, T. R. Holford†‡ and M. B. Bracken*†‡§ †Yale School of Public Health, New Haven, CT, USA; ‡Yale Center for Perinatal, Pediatric and Environmental Epidemiology, New Haven, CT, USA; §School of Medicine, Yale University, New Haven, CT, USA *Correspondence. (e-mail: [email protected]) Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.