Progesterone vaginal gel for the reduction of recurrent preterm birth: primary results from a randomized, double-blind, placebo-controlled trial




Preterm birth is the leading cause of perinatal morbidity and mortality worldwide. Treatment of preterm labor with tocolysis has not been successful in improving infant outcome. The administration of progesterone and related compounds has been proposed as a strategy to prevent preterm birth. The objective of this trial was to determine whether prophylactic administration of vaginal progesterone reduces the risk of preterm birth in women with a history of spontaneous preterm birth.


This randomized, double-blind, placebo- controlled, multinational trial enrolled and randomized 659 pregnant women with a history of spontaneous preterm birth. Between 18 + 0 and 22 + 6 weeks of gestation, patients were assigned randomly to once-daily treatment with either progesterone vaginal gel or placebo until either delivery, 37 weeks' gestation or development of preterm rupture of membranes. The primary outcome was preterm birth at ≤ 32 weeks of gestation. The trial was analyzed using an intent-to-treat strategy.


Baseline characteristics were similar in the two treatment groups. Progesterone did not decrease the frequency of preterm birth at ≤ 32 weeks. There was no difference between the groups with respect to the mean gestational age at delivery, infant morbidity or mortality or other maternal or neonatal outcome measures. Adverse events during the course of treatment were similar for the two groups.


Prophylactic treatment with vaginal progesterone did not reduce the frequency of recurrent preterm birth (≤ 32 weeks) in women with a history of spontaneous preterm birth. The effect of progesterone administration in patients at high risk for preterm delivery as determined by methods other than history alone (e.g. sonographic cervical length) requires further investigation. Copyright © 2007 ISUOG. Published by John Wiley & Sons, Ltd.


Preterm birth is a major challenge in obstetrics. It occurs in up to 15% of pregnancies in the developed world1–3 and its frequency is even higher in developing countries3. The prognosis of preterm neonates is related to gestational age at birth. These infants are at greater risk of developing respiratory distress syndrome, sepsis, intraventricular hemorrhage, necrotizing enterocolitis and severe neurological deficits, as well as a variety of other disorders and developmental disabilities4–6. Preterm birth is the antecedent cause of at least 36% of infant deaths and 50% of neurodevelopmental disabilities5, 7. It is also a leading cause of health care expenditure in the perinatal period and throughout the life of infant survivors8–10.

Approximately 25% of preterm births are the direct result of a medical decision to initiate delivery because of maternal or fetal indications. The majority, however, occur spontaneously due to preterm labor with intact membranes or preterm premature rupture of membranes11. Tocolysis to delay delivery has not been associated with improved neonatal outcome12. Therefore, intervention aimed at preventing the onset of preterm labor may be of greater benefit in improving perinatal outcome than is treatment implemented after the development of symptoms of premature parturition.

For more than 30 years, progesterone administration has been advocated for the prevention of preterm birth in women considered to be at high risk13, 14. The findings of previous clinical trials have encouraged investigators to explore the use of progestins/progesterone as a prophylactic. Two randomized clinical trials, one investigating the prophylactic use of natural vaginal progesterone suppositories and the other investigating weekly injections of 17α-hydroxyprogesterone caproate, have yielded encouraging results because progesterone/progestin administration has been associated with a significant reduction in the rate of preterm birth (defined as < 37 weeks) in women at risk of a previous preterm birth (P = 0.03 and P < 0.001, respectively)15, 16. Because prophylactic intervention during pregnancy entails prolonged administration of progesterone, less invasive forms of administration are preferred17–19. Unfortunately, progesterone administered orally has variable absorption, is subject to a first-pass hepatic metabolism and may cause central nervous system sedation17, 20. Vaginal administration of progesterone avoids first-pass metabolism prior to reaching the genital tract and provides endometrial concentrations 10 times higher than does the intramuscular formulation, making it the preferable option17, 21–25.

Although the use of progesterone/progestin to treat all women at risk for preterm birth does not have unqualified support at present, the American College of Obstetricians and Gynecologists (ACOG) has recommended that this approach be considered; the ACOG Committee on Obstetric Practice suggests restricting the use of progesterone/progestin to women with a documented history of spontaneous preterm birth26. Additional information on the optimal route of progesterone/progestin delivery and its long-term safety may allow for its broader use, and the Committee advocates further study26. ACOG and other investigators have indicated the need for adequately designed, randomized trials in larger populations to demonstrate whether progestin administration results in a decline in preterm births2, 26, 27. Furthermore, the ACOG Committee suggests further study to evaluate the use of progesterone/progestin in patients with other high-risk factors, such as a short sonographic cervical length26. Short cervical length has been shown to be a better predictor of preterm birth than is obstetric history alone28, 29.

The primary objective of this prospective, randomized, double-blind, placebo-controlled, multinational study was to assess the efficacy and safety of progesterone vaginal gel compared with placebo in decreasing the rate of preterm birth (≤ 32 + 0 weeks) among women with a prior history of spontaneous preterm birth.



The study protocol was approved by the institutional review board at each site and recruitment began in April 2004. Subjects were enrolled from 53 medical centers on five continents, with the last subject being delivered on 8 January 2007.

Pregnant women were eligible to enter the trial if they were between 18 and 45 years of age with an estimated gestational age between 16 + 0 and 22 + 6 weeks, and had a history of spontaneous singleton preterm birth at between 20 + 0 and 35 + 0 weeks of gestation in the immediately preceding pregnancy, confirmed by review of medical records. Patients were also required to: (1) understand English or a common local language; (2) provide written informed consent; (3) demonstrate understanding of the purpose of the study; and (4) agree to adhere to the study protocol.

Exclusion criteria included: (1) a history of adverse reaction to progesterone or any component of the formulation; (2) progesterone treatment within 4 weeks before enrollment; (3) treatment for a seizure disorder, a psychiatric illness or chronic hypertension at the time of enrollment; (4) a history of acute or chronic congestive heart failure, renal failure, uncontrolled diabetes mellitus, an active liver disorder, HIV infection with a CD4 count of < 350 cells/mm3 and requiring multiple antiviral agents; (5) placenta previa; (6) a history or suspicion of breast or genital tract malignancy; (7) a history or suspicion of thromboembolic disease; (8) a Müllerian duct anomaly; (9) enrollment in another investigational study within 1 month of screening for the present study; (10) a major fetal anomaly diagnosed by ultrasound or a known chromosomal disorder; (11) multifetal gestation; (12) a cervical cerclage or planned cervical cerclage placement during the current pregnancy; (13) preterm labor, preterm premature rupture of membranes, clinical chorioamnionitis or vaginal bleeding; (14) a history of previous preterm delivery without spontaneous preterm labor (indicated preterm delivery); (15) unwillingness or inability to comply with study procedures as explained by the clinical trial personnel.

Randomization and treatment

The study drug was packaged, labeled and shipped to the participating study sites by Aptuit, Inc. (Mount Laurel, NJ, USA). Labeling was carried out according to the 1 : 1 randomization scheme provided by Quintiles, Inc (Kansas City, MO, USA). An SAS (SAS Institute Inc., Cary, NC, USA) procedure for variable block size was used to generate the randomization schedule stratified by study site. The study patients, obstetric care providers, study investigators, study coordinators and study monitors were blind to the exposure status (progesterone or placebo) of all study patients.

The study drug (Prochieve® 8%/Crinone® 8% progesterone gel) and placebo (Replens®) were provided by Columbia Laboratories, Inc. (Livingston, NJ, USA). This company sponsored the study and was the funding source. The drug preparation is a bioadhesive formulation of progesterone for vaginal application, consisting of a polycarbophil-based gel that contains 8% (wt/wt) progesterone. A prefilled, single-use, disposable plastic applicator delivers the dose of 1.125 g of gel containing 90 mg of progesterone30, 31. The placebo consisted of an identical bioadhesive delivery system, but without progesterone. Patients were instructed to self-administer the full applicator of vaginal gel at approximately the same time daily, preferably in the morning. Patients received a 2-week supply of the allocated treatment at the time of randomization and at every 2-week study visit thereafter. Subjects received an extra 1-week drug supply when treatment was initiated, in case they could not attend the next regularly scheduled appointment. All drug supplies (or their packaging) were to be brought in at each visit at all study centers. Compliance with the study medication was determined from returned empty wrappers and unused wrappers. Percent compliance was assessed as total treatment duration compliance: (total applicators used/total days of treatment) × 100. Total dosing days was defined as the interval from enrollment to either the date of preterm premature rupture of membranes, the date of preterm birth without accompanying ruptured membranes, or 37 + 0 weeks' gestation.


Patients with a documented history of spontaneous preterm birth were screened between 16 + 0 and 22 + 6 weeks of gestation by the investigator or study coordinator. Gestational age was based on the last menstrual period and sonographic fetal biometry combined using a previously described algorithm32. Each patient had at least one ultrasound examination prior to randomization to confirm gestational age and to rule out major fetal anomalies, and a transvaginal scan to determine sonographic cervical length. Subjects meeting the study criteria were enrolled by the investigator between 18 + 0 and 22 + 6 weeks to receive study medication. Following randomization, daily treatment was initiated by the patient and continued until either 37 + 0 weeks' gestation, occurrence of preterm rupture of the membranes or preterm delivery. Each patient was interviewed every 2 weeks for occurrence of any adverse events and resupplied with study drug/placebo. At 28 weeks of gestation, all subjects had a second transvaginal ultrasound examination to determine cervical length.

Preterm labor was defined as six or more uterine contractions per hour accompanied by cervical dilation greater than 2 cm, cervical effacement greater than 80%, or documented change in cervical effacement greater than 50%33. If a patient developed spontaneous preterm labor during the study period, she was treated and kept in the study until delivery, in the absence of preterm rupture of membranes. The treatment protocol for spontaneous preterm labor was instituted at the discretion of the treating physician. It may have included hospitalization, hydration, bed rest, tocolysis and/or steroids. The study drug was to be continued during treatment for preterm labor. If the patient developed preterm rupture of membranes during the study period, the study drug was discontinued.

At delivery, data pertaining to the timing of labor onset and delivery, along with maternal and neonatal complications, were documented. Infant outcome parameters of morbidity and congenital anomalies were assessed during the initial postpartum hospitalization period.

The primary efficacy outcome measure was delivery at ≤ 32 + 0 weeks' gestation. An additional efficacy outcome was the frequency of neonatal morbidity during initial hospitalization. Measures of neonatal outcome included APGAR scores, birth weight, head circumference, hospital days, occurrence of medical complications and presence of congenital anomalies. Secondary efficacy outcomes included preterm birth at ≤ 28, ≤ 35 and < 37 weeks of gestation, hospital admission for preterm labor, neonatal hospital days from delivery to initial discharge, interval to delivery in patients who received tocolytic therapy for preterm labor and change in cervical length from randomization to 28 weeks of gestation. An additional secondary outcome assessed treatment effect based on cervical length at enrollment. At a subset of centers (17 of 53), a planned subinvestigation assessing cervical length as an indication for treatment was performed by enrolling additional patients with a cervical length ≤ 25 mm but without a history of preterm birth, and using a separate randomization scheme. This subinvestigation was not to exceed 10% of the total trial population. This subset of patients with a sonographically short cervix was excluded from the analysis of this report. The rationale for this was that this clinical trial focused on patients at high risk for preterm delivery based on maternal history alone.

Statistical analysis

All the efficacy endpoints were analyzed using an intent-to-treat (ITT) approach. The ITT population was defined as patients randomized based on a history of previous preterm birth and who had efficacy data (delivery date).

Data were analyzed with the SAS 9.1 software package (SAS Institute Inc.). Baseline characteristics and outcome data were compared between treatment groups using a chi-square or Fisher's exact test for categorical variables and using ANOVA for continuous variables. The duration of pregnancy in the placebo and intervention groups was evaluated using survival analysis (life-table analysis and the Kaplan–Meier method). A P-value of < 0.05 was considered statistically significant.

A two-sided significance level of 5% and a power of 90% for the primary efficacy outcome was used to determine the required sample size of 636 subjects, but as a result of an early protocol amendment, the power decreased to 88%. Assumptions about the drug and placebo effect rates were based on previous recurrent preterm birth rates reported by Meis et al.16 (19.6% placebo vs. 11.4% progesterone, < 32 weeks) and by da Fonseca et al.15 (18.6% placebo vs. 2.8% progesterone, < 34 weeks). We estimated a 20% rate of preterm delivery at ≤ 32 weeks in the placebo group, with expectations of a reduction to 10% in the treatment group. An interim analysis was not planned or performed.


A total of 711 women gave written informed consent, of whom 42 were not randomized. The most common reasons for non-randomization were a planned cerclage, maternal complications (e.g. hypertension) in pregnancy and inability to document a previous spontaneous preterm birth at the pre-specified gestational age. The study protocol required that the history of a previous preterm birth be verified by the medical record of the previous pregnancy. In some instances, such medical records could not be obtained or the information in the medical record contradicted the patient's recollection. A total of 669 patients were considered eligible for enrollment into the study, with 659 being randomized to the two treatment groups indicated for maternal history alone (Figure 1); nine patients enrolled into the planned subinvestigation for short cervix alone and one qualifying patient lost to follow-up prior to randomization were excluded from the analysis. Patients who took at least one dose of study medication and provided a delivery date were included in the ITT population. Patients without a delivery date were considered lost to follow-up.

Figure 1.

Trial profile.

The ITT population of previous preterm birth patients included 309 patients in the progesterone group and 302 patients in the placebo group. Baseline characteristics for this population are shown in Table 1. The groups were similar for maternal age, ethnicity, body mass index, parity, number of prior preterm births and number of spontaneous miscarriages. The mean ( ± SD) gestational age at randomization was 19.9 ( ± 2.1) and 20.1 ( ± 3.3) weeks for the progesterone and placebo groups, respectively. The frequency of one previous spontaneous preterm birth was similar in patients allocated to the placebo and progesterone groups (74.5% and 76.4%, respectively). Importantly, the frequency of two or more previous preterm births was therefore not different between the two groups (Table 1).

Table 1. Baseline characteristics of the study population
CharacteristicGroupMean difference* or odds ratio (95% CI)
Progesterone (n = 309)Placebo (n = 302)
  • *

    Calculation for mean difference = Progesterone—Placebo, with 95% CI for the difference (significant if zero not included in the range). There were no significant differences for any comparisons in mean difference or odds ratio between groups, P > 0.05. GA, gestational age.

Maternal age (years)
 Mean (SD)27.1 (5.8)27.3 (5.6)− 0.19 (−1.1 to 0.72)*
Race/ethnicity (n (%))
 Caucasian111 (35.9)99 (32.8)1.14 (0.82 to 1.6)
 African–American76 (24.6)85 (28.1)0.83 (0.58 to 1.2)
 Hispanic22 (7.1)14 (4.6)1.57 (0.79 to 3.14)
 Asian/Pacific Islander55 (17.8)60 (19.9)0.87 (0.58 to 1.3)
 Native American01 (0.3)
 Other45 (14.6)43 (14.2)1.02 (0.82 to 1.6)
Country of study site (n (%))
 United States200 (64.7)195 (64.6)1.0 (0.70 to 1.4)
 India54 (17.5)57 (18.9)0.91 (0.6 to 1.4)
 South Africa44 (14.2)40 (13.2)1.1 (0.68 to 1.7)
 Czech Republic7 (2.3)6 (2.0)1.14 (0.37 to 3.4)
 Chile/El Salvador4 (1.3)4 (1.3)1.0 (0.2 to 3.9)
Body mass index (mean (SD))26.6 (6.5)26.4 (7.1)0.23 (−0.8 to 1.2)*
Parity (mean (SD))1.5 (1.1)1.5 (1.1)− 0.02 (−0.19 to 0.15)*
Prior preterm births (n, mean (SD))1.3 (0.6)1.4 (0.7)− 0.06 (−0.17 to 0.04)*
> 1 prior preterm birth (n (%))73 (23.6)77 (25.5)0.90 (0.6 to 1.3)
Prior cervical surgery (n (%))22 (7.1)28 (9.3)0.75 (0.4 to 1.3)
> 1 spontaneous miscarriage (n (%))99 (32.0)100 (33.1)0.95 (0.67 to 1.3)
GA at randomization (weeks, mean (SD))19.9 (2.1)20.1 (3.3)− 0.14 (−0.57 to 0.29)*
Cervical length at randomization (cm, mean (SD))3.7 (0.7)3.7 (0.7)− 0.002 (−0.13 to 0.13)*

Maternal and neonatal outcomes are shown in Table 2. The primary outcome, rate of preterm birth at ≤ 32 + 0 weeks of gestation, was not significantly different between the study groups: 10.0% (n = 31) in the progesterone group and 11.3% (n = 34) in the placebo group. There was no significant difference in primary outcome according to country or region. The mean gestational age at delivery was 36.6 weeks for both the progesterone and the placebo groups. No other study outcomes differed between the progesterone and the placebo groups (Table 2) and survival curves for time to delivery were similar for both groups (Figure 2).

Figure 2.

Probability of patients remaining undelivered according to treatment group (placebo (▪, n = 302) or progesterone (□, n = 309)). The Kaplan–Meier method was used for calculation.

Table 2. Maternal and neonatal outcomes
OutcomeGroupMean difference* or odds ratio (95% CI)
Progesterone (n = 309)Placebo (n = 302)
  • *

    Calculation for mean difference = Progesterone—Placebo, with 95% CI for the difference (significant if zero not included in the range).

  • Data for mode of delivery were not obtained for every patient. GA, gestational age; NICU, neonatal intensive care unit; PPROM, preterm premature rupture of membranes.

GA at birth (weeks, mean (SD))36.6 (3.8)36.6 (4.2)0.0 (−0.64 to 0.64)*
Preterm birth (n (%))
 < 37 weeks129 (41.7)123 (40.7)1.08 (0.76 to 1.52)
 ≤ 35 weeks70 (22.7)80 (26.5)0.9 (0.61 to 1.34)
 ≤ 32 weeks31 (10.0)34 (11.3)0.9 (0.52 to 1.56)
 ≤ 28 weeks10 (3.2)9 (3.0)1.07 (0.38 to 2.96)
Intrauterine fetal demise (n (%))
 < 20 weeks00
 > 20 weeks5 (1.6)4 (1.3)1.22 (0.33 to 4.61)
PPROM (n (%))37 (12.0)38 (12.6)0.95 (0.58 to 1.53)
Admission for preterm labor (n (%))79 (25.6)75 (24.8)1.14 (0.38 to 3.37)
Tocolytic therapy for preterm labor (n (%))35 (11.3)31 (10.3)1.12 (0.67 to 1.86)
Latency period to delivery after tocolysis for preterm labor (days, mean (SD))30.0 (30.0)19.6 (19.8)10.3 (−2.4 to 23.0)*
Antepartum corticosteroid administration (n (%))72 (23.3)74 (24.5)0.94 (0.65 to 1.36)
Mode of delivery (n (%))
 Vaginal218 (71.0)216 (72.2)0.94 (0.66 to 1.34)
 Cesarean section89 (29.0)83 (27.8)1.06 (0.75 to 1.51)
Study medication compliance (%, mean (SD))96.2 (9.4)96.4 (7.8)− 0.2 (−1.5 to 1.2)*
Apgar score (median (SD))
 1 min7.7 (1.8)7.7 (1.6)0 (−0.3 to 0.3)*
 5 min8.9 (2.3)8.8 (1.2)0.1 (−0.2 to 0.4)*
Birth weight (g, mean (SD))2680 (710)2661 (738)19 (−96 to 135)*
NICU admission (n (%))54 (17.5)65 (21.5)0.75 (0.51 to 1.11)
Days in NICU per admission (n, mean (SD))14.2 (16.6)20.5 (30.7)− 6.2 (−15.2 to 2.8)*
Head circumference (cm, mean (SD))32.3 (3.34)32.5 (3.75)− 0.2 (−0.8 to 0.4)*
Respiratory distress syndrome (n (%))34 (11.0)36 (11.9)0.91 (0.56 to 1.50)
Intraventricular hemorrhage (n (%))6 (1.9)5 (1.6)1.18 (0.36 to 3.90)
 Grade 14 (1.3)4 (1.3) 
 Grade 21 (0.3)0 (0.0) 
 Grade 31 (0.3)0 (0.0) 
 Grade 40 (0.0)1 (0.3) 
Necrotizing enterocolitis (n (%))3 (1.0)5 (1.7)0.58 (0.14 to 2.46)
 Surgical0 (0.0)3 (1.0) 
 Clinical3 (1.0)2 (0.7) 
Neonatal death (< 28 days) (n (%))6 (1.9)7 (2.3)0.87 (0.29 to 2.60)

Compliance rates with the study medication were similar between groups: 96.2% for women in the progesterone group and 96.4% for women in the placebo group.

Congenital abnormalities potentially acquired in the second or third trimesters were observed in two neonates in each group: one case each of hip subluxation and pulmonary stenosis. Congenital anomalies related to a disturbance of first-trimester organogenesis, such as hypospadias, polydactyly and tetralogy of Fallot, which were diagnosed at birth, are not summarized in the table and attempts were made to exclude such cases at the time of enrollment by initial screening.

The frequency of adverse events during the course of therapy was similar between the progesterone and the placebo groups (81.3% vs. 83.2%), as was the occurrence of adverse events considered serious (39.6% vs. 42.7%, respectively). Complications of pregnancy leading to hospitalization, such as preterm labor, preterm birth and premature rupture of the membranes, accounted for 85% and 91% of the serious adverse events in the progesterone and placebo groups, respectively. Complaints about vaginal discharge occurred in 8.4% of patients allocated to progesterone and in 9.2% of patients allocated to placebo. Vaginal discharge was considered to be related to the study medication in 4.0% of the progesterone group and 4.4% of the placebo group. In the safety population 1.2% (4/321) of the progesterone group and 0.9% (3/316) of the placebo group had vaginal discharge that was considered serious; one patient discontinued therapy as a result. The rate of study discontinuation due to an adverse event was only 1.6% in the progesterone group and 0.9% in the placebo group. The most common adverse events leading to discontinuation in both the progesterone and the placebo groups were complications of pregnancy (such as hypertension, preterm labor/delivery and fetal death). Two patients, both in the progesterone group, discontinued therapy due to nausea.


This study is the largest randomized trial to date evaluating the efficacy of progesterone for the prevention of preterm birth in women with a history of spontaneous preterm birth. We found that the administration of vaginal progesterone did not reduce the rate of early preterm delivery (≤ 32 weeks) or decrease the frequency of neonatal morbidity and mortality.

The vaginal progesterone gel formulation containing 90 mg of natural progesterone was chosen for this trial for several reasons. Foremost was the desire to evaluate a readily available, commercially manufactured product that undergoes strict quality control in its formulation. The vaginal route was chosen due to its potential for greater patient satisfaction compared with intramuscular administration, and for improved efficacy through enhanced drug delivery to target tissues17–21. Studies of this formulation noted a 14-times greater increase in the ratio of endometrial-to-serum concentrations after vaginal dosing than after systemic, intramuscular administration (termed the ‘first-pass uterine effect’)18–20. The daily dose used in this study was chosen because 90 mg of progesterone in the commercially available product is equivalent to 600 mg of progesterone suppositories34. This equivalency is based on studies in early pregnancy in which progesterone has been used extensively for luteal phase support22. In addition, this dose (90 mg per day of progesterone) is similar to the one employed by da Fonseca et al. (100-mg suppositories) in a randomized clinical trial that reduced the frequency of preterm delivery15. We selected progesterone rather than a synthetic progestin because the former is a natural hormone with a well-established safety profile in pregnancy, while questions have been raised about the safety of synthetic compounds such as 17α-hydroxyprogesterone caproate35.

The only other similar study of comparable size and patient profile is the Meis et al. trial16, which investigated the efficacy of intramuscular 17α-hydroxyprogesterone caproate to prevent recurrent preterm birth in women with a history of prior preterm birth. This trial reported a significant reduction in the rate of preterm birth with treatment. This finding has led to an enthusiastic adoption of this strategy for the prevention of preterm birth, even though there are concerns about the high rate of preterm birth in the control group and the lack of conclusive data demonstrating a reduction in neonatal morbidity and mortality13. A recent survey of maternal–fetal medicine specialists in the United States demonstrated that two thirds of responding physicians currently prescribe 17α-hydroxyprogesterone caproate for patients with a history of prior preterm birth36.

Other randomized clinical trials evaluating the efficacy and safety of progestins in reducing the rate of preterm birth have had a small sample size37–40 and did not focus on similar outcomes, and this in turn has led to meta-analysis yielding conflicting results2, 13, 14, 41. Our trial focused on the prevention of early preterm birth (≤ 32 weeks) because there is consensus that this is a meaningful surrogate end-point (United States Food and Drug Administration (FDA) Advisory Committee, August 29, 2006). It is more likely that the reduction of early preterm birth (≤ 32 weeks) would be associated with a reduction in neonatal morbidity and mortality than would a reduction in preterm birth defined as < 37 weeks. Indeed, Dodd et al.13 concluded that there is insufficient evidence to advocate the routine use of progestins to reduce neonatal morbidity. These authors called for further studies on the subject.

Our results indicating that progesterone administration does not reduce the rate of spontaneous preterm birth (at ≤ 32 weeks, ≤ 35 weeks and < 37 weeks) are in contrast to those reported by Meis et al16. This is noteworthy because both studies focused on a similar patient population: women with a history of previous preterm birth. Several explanations may account for this discrepancy. First, the trials used different progestogen. Meis et al. used 17α-hydroxyprogesterone caproate, while the current trial used progesterone. However, it is unlikely that this accounts for the difference, because da Fonseca et al.15 found that vaginal progesterone reduced the rate of preterm delivery in a previous randomized clinical trial. Second, it is possible that the total weekly intravaginal progesterone dose of 630 mg administered in this trial (or 90 mg per day) may be less efficacious than the 250-mg weekly intramuscular dose of 17α-hydroxyprogesterone caproate42. This explanation is also unlikely, given the success rate of the da Fonseca et al. trial, which administered vaginally 100 mg per day, and given the relative potencies of these drugs. Third, there may have been differences in the patient population between the two trials. Yet, both trials focused on patients with a history of preterm birth. In our study, 24.6% of patients with a previous history delivered preterm at ≤ 35 weeks and 10.6% at ≤ 32 weeks. Thus, our population was clearly at risk. Similarly, the overall rate of preterm delivery (at ≤ 35 weeks and ≤ 32 weeks) in the Meis et al. trial was 25.3% and 14.2%, respectively16. Consequently, it seems that the discrepancy cannot be accounted for by the magnitude of the risk for the patients enrolled in the trial.

It is difficult to exclude the possibility that our placebo preparation had a treatment effect to reduce the occurrence of preterm delivery in our control group. However, the rate of preterm birth in the placebo group is consistent with that of other published studies on the recurrence risk for preterm birth15, 16, and any potential effect of the placebo on vaginal pH or bacterial flora, such as bacterial vaginosis, has been shown not to alter the rate of preterm birth in other randomized trials43, 44.

Other possibilities to explain a difference in these studies, while conceivable, are unlikely based on our current understanding of progestogens. Both progesterone and its analogs have complex mechanisms of action, binding nuclear and non-nuclear receptors and altering genomic and non-genomic functions in cells of target tissues. It has been suggested that progesterone and some of its analogs induce multiple physiological changes to inhibit the onset of premature parturition, including suppressing myometrial activity by inhibiting gap junction formation, enhancing the barrier to ascending infection by altered cervical mucous production, and improving resistance to cervical stromal degradation. Physiologically, progesterone and 17α-hydroxyprogesterone caproate can have similar effects on selected end points. For instance, both agents are known to cause a secretory transformation of the endometrium45, 46. However, there are some physiological differences, such as the inability of 17α-hydroxyprogesterone caproate to inhibit oxytocin-induced contractions in vitro47. Given experimental observations, it is not plausible that 17α-hydroxyprogesterone caproate is superior to the native hormone at enhancing progesterone receptor-mediated activity on the target tissues of the reproductive tract: the uterus and cervix. Furthermore, due to physiological differences observed from in-vitro data, progesterone and its analogs should be evaluated individually as drugs, with compilation of individual efficacy and safety profiles.

A recent reevaluation of the Meis et al.16 study data was performed by the United States Food and Drug Administration (FDA) to assess the efficacy and safety of 17α-hydroxyprogesterone caproate in this population of women with a history of preterm birth. The FDA was provided with the outcome data for the published study and in their analysis did not identify a significant reduction in the rate of preterm birth at < 32 weeks' gestation48. Importantly, the FDA was also provided with data from 150 randomized patients in a previous study of the same design by the same organization, which was terminated early due to a drug recall. Most striking was a comparison of the data from these two studies regarding the rate of preterm birth among placebo patients, as very different rates were observed (35.7% in the earlier unpublished trial and 54.9% in the published results)48. Another finding of concern identified by the FDA assessment was a potential safety signal of an increase in pregnancy loss due to miscarriage, after exposure to 17α-hydroxyprogesterone caproate48. This observation of increased fetal wastage has also been documented in a recent large retrospective controlled trial49. Therefore, the efficacy and safety of the progesterone analog 17α-hydroxyprogesterone caproate for the prevention of preterm birth requires further study. Practitioners should reconsider its routine use for prevention of recurrent preterm birth based on history alone.

The vaginal route of administration of natural progesterone utilized in the present study is unlikely to explain its lack of effect in preventing preterm birth. Other studies have reported a benefit from intravaginal dosing for the prevention of preterm birth. The randomized trial by da Fonseca et al.15 enrolled 142 patients with varied risk factors for spontaneous preterm birth, and found intravaginal natural progesterone to be efficacious in reducing the occurrence of preterm delivery. In that trial, the frequency of delivery at ≤ 34 weeks was 18.6% for the placebo group and 2.8% in the group allocated to daily 100-mg vaginal progesterone suppository treatment. The patient profile included women with a history of preterm birth, those requiring cerclage, and those with uterine anomalies or other high risk factors. Cervical length data were not provided for these patients. The data suggest that intravaginal natural progesterone can have a benefit if an appropriate target population is identified. Indeed, a landmark study by The Fetal Medicine Foundation recently revealed that intravaginal progesterone could reduce the rate of preterm birth in a population with a short cervical length50.

The two reports by da Fonseca et al.15, 50 are also pertinent to examining the timing of drug administration and its possible impact on efficacy. Both of these studies initiated treatment later (i.e. at 24 weeks) than did the present investigation, yet demonstrated a treatment benefit. Furthermore, the average gestational age at the start of treatment for the intervention group in the present trial was 19.9 ± 2.1 weeks compared with 18.9 ± 1.4 weeks for the trial of Meis et al.48. We believe this 1-week difference is clinically unremarkable. It is unlikely, therefore, that our negative findings are the result of a delay in initiating therapy.

Cervical status was not evaluated in the trial published by Meis et al. In the current study, by design, all patients received a baseline transvaginal cervical measurement and patients with early-onset cervical shortening, who were considered possible candidates for cerclage placement by the attending physician, were excluded from trial participation. Therefore, in the current trial, patients who may have been at greatest risk for preterm birth (those with both a history of preterm birth and early-onset cervical shortening) were potentially excluded from enrollment. Based on data from Iams et al.51, the 10th percentile of cervical length in a low-risk population is 26 mm. In our trial of high-risk patients, 4% (24 of 611) had a cervical length of ≤ 25 mm in the midtrimester, while approximately 60 patients would be expected to have this cervical length. This suggests that a proportion of patients with cervical length ≤ 25 mm were excluded as a result of the measurement of cervical length at baseline and the exclusion criteria for consideration of cerclage. We attempted to compensate for this expected occurrence with a planned subinvestigation of short cervical length that is reported in the accompanying article in this issue52. These patients with a short cervix and history of prior preterm birth may be the ones who most frequently benefit from progesterone therapy. The observed rate of delivery at ≤ 32 weeks of gestation was less than expected in the placebo group in our power calculation—again, possibly the result of the lower frequency of subjects with a short cervix. This lower incidence in the placebo group raises the possibility of a type-II error. However, we believe that our sample size was adequate given the small (12%) difference in the rates of the primary outcome between groups. Given the same pattern of recruitment and incidence rates for the placebo group, we estimate that a sample size of greater than 1000 subjects would have been necessary to better exclude the possibility of this type of error.

We therefore believe that the most likely explanation for our results is that patients with a history of previous spontaneous preterm delivery constitute a heterogenous population and that a subset of these patients may benefit from progestin prophylaxis (responders) while others would not (non-responders). It is possible that the patients included in previous positive trials, such as those of da Fonseca et al. and Meis et al., included a substantial number of responders, while the current trial did not. The recent observation that vaginal progesterone reduces the rate of preterm delivery in women with a short cervix and the accompanying secondary analysis of this trial52 supports the view that patients with a sonographic short cervix are potential responders.

No safety issues have been identified with the intravaginal formulation of natural progesterone utilized in this trial, or with prior experience when it was used as luteal support for in-vitro fertilization cycles53. Specifically, in the present investigation, no deliveries occurred at < 20 weeks' gestation in women exposed to natural progesterone, and adverse outcomes after 20 weeks were not different between natural progesterone and placebo groups. Due to the potential for benefit with a proven-safe intervention, natural progesterone is worthy of future investigations in other clearly defined at-risk pregnant populations.

Clinical consensus regarding therapeutic intervention in obstetrics is rare, because conditions such as preterm birth have multifactorial etiologies, and knowledge regarding exact pathophysiology in individual patients is often lacking. Furthermore, the paucity of adequately powered randomized trials in this discipline, as well as issues relating to study design, such as heterogeneity in patient selection, have resulted in a limited number of consistent observations justifying potential interventions13.

This trial has not confirmed the findings of earlier trials on the use of progestogens in women with a history of spontaneous preterm birth alone. Routine sonographic examinations of the uterine cervix may be the tool to identify patients who would benefit from progesterone administration to prevent preterm delivery. At present, the conflicting data have created uncertainty for utilizing progesterone or its analogs, and consensus remains elusive.


We would like to thank all the physicians who participated in the study and, in particular, Ken Muse, MD, for his efforts in soliciting sponsor support. We would also like to thank all who identified candidates for this trial, the nurses who were essential for its completion, and other support staff including the monitors and data managers who worked to ensure the integrity of the study. This trial was funded by Columbia Laboratories, Inc.

Conflict of interest

J. M. O'Brien is a consultant and has received honoraria from Cook Biotech, Inc. G. W. Creasy is an employee of Columbia Laboratories, Inc.

Additional collaborators

W. Hansen (Lexington, KY), M. Newman (Baton Rouge, LA), B. Rosenn (New York, NY), S. Dabak (Pune, India), L. Parker, (Winston-Salem, NC), J. Stern (Memphis, TN), L. Bayer-Zwirello (Boston, MA), L. Cousins (San Diego, CA), A. Kekre (Vellore, India), R. McDuffie, (Denver, CO), J. Schucker (Danville, PA), C. Barrera (Santiago, Chile), C. Goldberg (Tucson, AZ), A. Jiratko (Zlinska, Czech Republic), K. Swenson (Austin, TX), A. Evans (Lubbock, TX), G. Gross (St. Louis, MO), M. Short (Baltimore, MD), S. Sunderji (Toledo, OH), R. Artal (St. Louis, MO), M. Binstock (Bedford, OH), J. Hibbard (Chicago, IL), R. Kelly (Odessa, TX), X. Sandovol-Lopez (San Salvador, El Salvador), L. Smith (Livingston, NJ), M. Stitley (Morgantown, WV), E. Wang (Chicago, IL), M. Beall (Torrence, CA), J. Carvajal (Santiago, Chile), V. Rappaport (Albuquerque, NM), L. Wilkins-Haug (Boston, MA), B. Sibai (Cincinnati, OH).