Introduction
Spontaneous preterm birth (SPB) is a major risk factor for perinatal morbidity and mortality in twin gestations (more than half result in premature births). The frequency of twin gestations has increased over the years, from 19.8/1000 births in 1980 to 33.1/1000 births in 2012. The increased risk for adverse outcomes in twin gestations is largely attributed to preterm birth and the rates in twins are 5-6 times higher. A short cervix (<25 mm) is associated with early and very early preterm birth in twins, and 15% of women with twin gestations have a cervix ≤25 mm. This measurement has become the method of choice for screening asymptomatic pregnant women at risk for preterm birth <24 weeks of gestation both in twin and singleton pregnancies; this is based on a systematic review and metaanalysis.
Current options for the management of patients with a short cervix are: vaginal progesterone, cervical cerclage, and cervical pessary. A systematic review and metaanalysis showed that vaginal progesterone reduced the rate of preterm delivery in twins in cases of maternal short cervix by 30%, which was not statistically significant; however, a reduction was observed in the rate of neonatal morbidity and confirmed recently. Cervical cerclage has been reported to increase the frequency of adverse outcome in twin gestations. A cervical pessary to support the cervix in pregnant women with cervical insufficiency was introduced in 1960. Since then, several studies in pregnant women with cervical insufficiency have been published ; however, most were retrospective or case-control. In the study by Arabin et al, the use of the pessary in pregnant women with short cervical length (CL) on ultrasound reduced the risk of spontaneous birth, with a decrease being observed in the preterm birth rate in both singleton and twin pregnancies, in cases of maternal CL ≤25 mm.
A cervical pessary is a silicone ring with a smaller diameter to be fitted around the cervix and a larger diameter to fix the device against the pelvic floor. This effectively rotates the cervix toward the posterior vaginal wall and corrects the cervical angle.
We designed a multicenter, randomized controlled trial to evaluate the effect of cervical pessary on the early SPB rate in asymptomatic women with a short cervix (CL ≤25 mm) carrying twins (PECEP-Twins Trial).
Materials and Methods
Trial design
A prospective, open-label, multicenter, randomized clinical trial was conducted in 5 hospitals in Spain. The ethics committees of all participating hospitals approved the protocol. The trial was registered as ClinicalTrials.gov , number NCT01242410 .
Participants
Women with twin pregnancies undergoing routine ultrasonography at 18-22 weeks of gestation were given the option of transvaginal ultrasonographic CL measurement as a predictor of SPB. CL was measured according to the criteria of the Fetal Medicine Foundation. Despite large variations in gestational age at measurement, cut-off point for CL, and definition of preterm birth among countries, second-trimester CL is a strong predictor of preterm birth in women with a multiple pregnancy. Women with a CL ≤25 mm were invited to participate in the PECEP-Twins Trial. Exclusion criteria were major fetal abnormalities, painful regular uterine contractions, active vaginal bleeding, ruptured membranes, placenta previa, and history of cone biopsy or cervical cerclage in situ. Gestational age was determined from menstrual history and confirmed by measurement of fetal crown-rump length at a first-trimester scan carried out routinely at all participating hospitals. Recruitment began in January 2011 and was extended to July 2014.
Quality control of screening, handling of data, and verification of adherence to protocols at the different centers were performed on a regular basis by trial coordinators. Obstetricians who performed the scans had received extensive training and passed a practical examination administered by an expert to demonstrate their competence in cervical assessment. All images of the cases included in the trial and all cases of preterm birth were reviewed and discussed centrally. The central team in turn instructed the other centers in the use of the pessary.
Randomization and masking
After providing written informed consent, women were randomly allocated to cervical pessary insertion or expectant management at a 1:1 ratio. The randomization sequence was computer generated with variable blocks of 2 and 4, stratified for center and parity. The random-number lists were created by the Statistics Unit of the Vall d’Hebron Hospital Research Institute and implemented by the use of a central telephone. The allocation code was disclosed after the patient’s initials had been confirmed. The randomization sequence was not accessible to the investigators or the trial coordinator. Outcome assessors were blinded to the interventions. This study was open label since masking to intervention was not possible.
Interventions
Cervical and vaginal swabs were taken in all patients for Gram stain microbiologic studies (culture for trichomoniasis, Candida , group B streptococcus, and bacterial vaginosis). If clinical symptoms of infection were detected, empiric treatment according to our hospital protocol and Spanish Obstetrician and Gynecologist Society was provided and pessary insertion delayed by 1 week. Vaginal examination with a speculum was performed to observe cervical dilatation or visible membranes. Patients allocated to the pessary group had one inserted and were given detailed instructions on its subsequent management. Special emphasis was placed on the need to immediately report any adverse symptoms. Correct placement of the pessary was determined by transvaginal ultrasound and CL was measured. In women with a cervical pessary, sonographic visualization of CL is difficult owing to the shadow cast by the silicone on the cervix. We found that good visualization of the cervix is enabled by passing through the virtual space between the pessary and posterior vaginal wall and inserting the probe inside the pessary, which, if possible, touches on the external cervical os or anterior cervical lip. We propose this new technique for measuring and monitoring CL in women with a cervical pessary.
All these interventions were performed on the same day. The pessary was not removed when symptoms of infection occurred after pessary insertion; however, appropriate antibiotic (clindamycin) therapy was given.
EC-certified cervical pessaries for the indication of preventing SPB (CE0482/EN ISO 13485: 2003 Annexe/III of the Council Directive 93/42 EEC) were employed. A one-size pessary was used following the recommendations of Dr Arabin GmbH and Co KG: 65 × 25 × 32 mm (lower larger diameter, height, and inner diameter) purchased from Dr Arabin GmbH and Co KG; only in patients who had problems with the pessary did we consider changing the size of the pessary.
Both groups were seen every month until delivery. The following procedures were carried out: (1) transabdominal ultrasound for fetal biometry and assessment of fetal well-being; (2) maternal satisfaction questionnaire regarding pessary placement in the pessary group; (3) vaginal swab taken for study of Gram stain and microbiology (culture for trichomoniasis, Candida , group B streptococcus, and bacterial vaginosis); and (4) transvaginal ultrasound for CL measurement.
The pessary was removed during the 37th week of gestation. Indications for pessary removal <37th week were active vaginal bleeding, an episode of preterm labor with persistent contractions despite tocolysis (>5 contractions every 30 minutes, without reduction or end with tocolytic drug), or severe patient discomfort.
Preterm labor was defined as uterine contractions detected and CL shortening according to gestational age; in these cases, tocolysis and corticosteroids were administered. Atosiban was the first-line tocolytic owing to the twin nature of the pregnancy.
Patients whose pessary was removed (even on the day of insertion) remained in the trial (intention-to-treat analysis). The pessary was not initially removed if preterm rupture of membranes occurred: these patients were followed up at the hospital and, if labor began or chorioamnionitis was detected, the pessary was removed. Sexual intercourse was not prohibited in either groups.
Outcome measures
The primary outcome was SPB <34 weeks (238 days) of gestation. Secondary outcomes were: birthweight, intrauterine fetal demise, neonatal death, neonatal morbidity (intraventricular hemorrhage, respiratory distress syndrome, retinopathy of prematurity, necrotizing enterocolitis, proven or suspected sepsis, need for neonatal special care [neonatal intensive care unit, need for ventilation, phototherapy, antibiotics, or blood transfusion]), composite adverse outcomes (intraventricular hemorrhage, respiratory distress syndrome, retinopathy of prematurity, necrotizing enterocolitis, and proven or suspected sepsis), significant maternal adverse effects (heavy bleeding, cervical tear, uterine rupture, intolerance to pessary), SPB <37 weeks, SPB <28 weeks, spontaneous rupture of membranes <34 weeks, chorioamnionitis, hospitalization for threatened preterm labor <34 weeks (mean duration of hospital stay, use of tocolytic treatment type of tocolytic, days of treatment, dose) and incidence of vaginal infections. Placentas were analyzed in both groups. Chorioamnionitis was defined as acute inflammation of the extraplacental membranes, chorion, and amnion, by pathologic study after delivery.
Data analysis
Sample-size calculation was based on a reduction in the frequency of spontaneous delivery <34 weeks, from 50% in the expectant management group to 25% in the pessary group, with power of 80%. To detect this difference at a significance level of 5%, we needed to recruit 126 patients with CL of ≤25 mm.
Analysis was performed according to the intention-to-treat principle. The means and SD summarized baseline data for the pessary and expectant management groups ( Table 1 ). Comparisons between groups were made with the Mann-Whitney U test. Neonatal outcomes were analyzed as nonindependent data if they occurred in the same birth. Univariate comparisons of dichotomous data were made with Fisher exact test. P values for all hypotheses were 2-sided, and P values <.05 were considered statistically significant. The risk of SPB <34 weeks was quantified by the relative risk (RR) and 95% confidence interval (CI). Multivariate analysis was performed by logistic regression. The interval from randomization (pessary or not) to delivery was assessed using Kaplan-Meier analysis, where gestational age was the time scale, spontaneous delivery was the event, and elective deliveries were treated as censored. For purposes of this analysis, all pregnancies were considered to be no longer at risk at the start of the 34th week. All statistical analyses were performed with a software package (SPSS, Version 16; IBM Corp, Armonk, NY). No interim analysis was planned for this trial.
| Baseline characteristics of the study subjects a | Pessary (n=68) | No treatment (n=66) |
|---|---|---|
| Maternal Age (years) | 35.4 (3.6) | 35.9 (5.6) |
| Body mass index b | 24.3 (1.5) | 24.7 (2.0) |
| Obstetric history — no. (%) | ||
| Nulliparous | 31 (45.6%) | 29 (43.9%) |
| Parous with no previous preterm births | 26 (36.4%) | 25 (36.8%) |
| Parous with ≥ 1 previous preterm births | 11 (16.7%) | 12 (17.6%) |
| Cigarette smoking during pregnancy — no. (%) | 10 (14.7%) | 9 (13.6%) |
| Race — no. (%) c | ||
| White European | 38 (57.6%) | 41 (60.3%) |
| Latin American | 16 (24.2%) | 15 (22.1%) |
| Others | 12 (18.2%) | 12 (17.6%) |
| Gestational age at randomization (weeks) | 22.1 (0.8) | 22.5 (0.7) |
| Cervical length at randomization (mm) | 19·2 (3.5) | 19.6 (3.6) |
| Funnelling at randomization (yes) | 7 (10.3%) | 8 (12.1%) |
| Sludge at randomization (yes) | 1 (1.4) | 1 (1.5%) |
| Pregnancy after ART | 21 (30.9%) | 20 (30.3%) |
| Monochorionic Twins | 13 (19.1%) | 12 (17.6%) |
a Plus-minus values are means. p >0.05 for all between-group comparisons
b Body mass index: weight in kilograms divided by the square of height in meters
Materials and Methods
Trial design
A prospective, open-label, multicenter, randomized clinical trial was conducted in 5 hospitals in Spain. The ethics committees of all participating hospitals approved the protocol. The trial was registered as ClinicalTrials.gov , number NCT01242410 .
Participants
Women with twin pregnancies undergoing routine ultrasonography at 18-22 weeks of gestation were given the option of transvaginal ultrasonographic CL measurement as a predictor of SPB. CL was measured according to the criteria of the Fetal Medicine Foundation. Despite large variations in gestational age at measurement, cut-off point for CL, and definition of preterm birth among countries, second-trimester CL is a strong predictor of preterm birth in women with a multiple pregnancy. Women with a CL ≤25 mm were invited to participate in the PECEP-Twins Trial. Exclusion criteria were major fetal abnormalities, painful regular uterine contractions, active vaginal bleeding, ruptured membranes, placenta previa, and history of cone biopsy or cervical cerclage in situ. Gestational age was determined from menstrual history and confirmed by measurement of fetal crown-rump length at a first-trimester scan carried out routinely at all participating hospitals. Recruitment began in January 2011 and was extended to July 2014.
Quality control of screening, handling of data, and verification of adherence to protocols at the different centers were performed on a regular basis by trial coordinators. Obstetricians who performed the scans had received extensive training and passed a practical examination administered by an expert to demonstrate their competence in cervical assessment. All images of the cases included in the trial and all cases of preterm birth were reviewed and discussed centrally. The central team in turn instructed the other centers in the use of the pessary.
Randomization and masking
After providing written informed consent, women were randomly allocated to cervical pessary insertion or expectant management at a 1:1 ratio. The randomization sequence was computer generated with variable blocks of 2 and 4, stratified for center and parity. The random-number lists were created by the Statistics Unit of the Vall d’Hebron Hospital Research Institute and implemented by the use of a central telephone. The allocation code was disclosed after the patient’s initials had been confirmed. The randomization sequence was not accessible to the investigators or the trial coordinator. Outcome assessors were blinded to the interventions. This study was open label since masking to intervention was not possible.
Interventions
Cervical and vaginal swabs were taken in all patients for Gram stain microbiologic studies (culture for trichomoniasis, Candida , group B streptococcus, and bacterial vaginosis). If clinical symptoms of infection were detected, empiric treatment according to our hospital protocol and Spanish Obstetrician and Gynecologist Society was provided and pessary insertion delayed by 1 week. Vaginal examination with a speculum was performed to observe cervical dilatation or visible membranes. Patients allocated to the pessary group had one inserted and were given detailed instructions on its subsequent management. Special emphasis was placed on the need to immediately report any adverse symptoms. Correct placement of the pessary was determined by transvaginal ultrasound and CL was measured. In women with a cervical pessary, sonographic visualization of CL is difficult owing to the shadow cast by the silicone on the cervix. We found that good visualization of the cervix is enabled by passing through the virtual space between the pessary and posterior vaginal wall and inserting the probe inside the pessary, which, if possible, touches on the external cervical os or anterior cervical lip. We propose this new technique for measuring and monitoring CL in women with a cervical pessary.
All these interventions were performed on the same day. The pessary was not removed when symptoms of infection occurred after pessary insertion; however, appropriate antibiotic (clindamycin) therapy was given.
EC-certified cervical pessaries for the indication of preventing SPB (CE0482/EN ISO 13485: 2003 Annexe/III of the Council Directive 93/42 EEC) were employed. A one-size pessary was used following the recommendations of Dr Arabin GmbH and Co KG: 65 × 25 × 32 mm (lower larger diameter, height, and inner diameter) purchased from Dr Arabin GmbH and Co KG; only in patients who had problems with the pessary did we consider changing the size of the pessary.
Both groups were seen every month until delivery. The following procedures were carried out: (1) transabdominal ultrasound for fetal biometry and assessment of fetal well-being; (2) maternal satisfaction questionnaire regarding pessary placement in the pessary group; (3) vaginal swab taken for study of Gram stain and microbiology (culture for trichomoniasis, Candida , group B streptococcus, and bacterial vaginosis); and (4) transvaginal ultrasound for CL measurement.
The pessary was removed during the 37th week of gestation. Indications for pessary removal <37th week were active vaginal bleeding, an episode of preterm labor with persistent contractions despite tocolysis (>5 contractions every 30 minutes, without reduction or end with tocolytic drug), or severe patient discomfort.
Preterm labor was defined as uterine contractions detected and CL shortening according to gestational age; in these cases, tocolysis and corticosteroids were administered. Atosiban was the first-line tocolytic owing to the twin nature of the pregnancy.
Patients whose pessary was removed (even on the day of insertion) remained in the trial (intention-to-treat analysis). The pessary was not initially removed if preterm rupture of membranes occurred: these patients were followed up at the hospital and, if labor began or chorioamnionitis was detected, the pessary was removed. Sexual intercourse was not prohibited in either groups.
Outcome measures
The primary outcome was SPB <34 weeks (238 days) of gestation. Secondary outcomes were: birthweight, intrauterine fetal demise, neonatal death, neonatal morbidity (intraventricular hemorrhage, respiratory distress syndrome, retinopathy of prematurity, necrotizing enterocolitis, proven or suspected sepsis, need for neonatal special care [neonatal intensive care unit, need for ventilation, phototherapy, antibiotics, or blood transfusion]), composite adverse outcomes (intraventricular hemorrhage, respiratory distress syndrome, retinopathy of prematurity, necrotizing enterocolitis, and proven or suspected sepsis), significant maternal adverse effects (heavy bleeding, cervical tear, uterine rupture, intolerance to pessary), SPB <37 weeks, SPB <28 weeks, spontaneous rupture of membranes <34 weeks, chorioamnionitis, hospitalization for threatened preterm labor <34 weeks (mean duration of hospital stay, use of tocolytic treatment type of tocolytic, days of treatment, dose) and incidence of vaginal infections. Placentas were analyzed in both groups. Chorioamnionitis was defined as acute inflammation of the extraplacental membranes, chorion, and amnion, by pathologic study after delivery.
Data analysis
Sample-size calculation was based on a reduction in the frequency of spontaneous delivery <34 weeks, from 50% in the expectant management group to 25% in the pessary group, with power of 80%. To detect this difference at a significance level of 5%, we needed to recruit 126 patients with CL of ≤25 mm.
Analysis was performed according to the intention-to-treat principle. The means and SD summarized baseline data for the pessary and expectant management groups ( Table 1 ). Comparisons between groups were made with the Mann-Whitney U test. Neonatal outcomes were analyzed as nonindependent data if they occurred in the same birth. Univariate comparisons of dichotomous data were made with Fisher exact test. P values for all hypotheses were 2-sided, and P values <.05 were considered statistically significant. The risk of SPB <34 weeks was quantified by the relative risk (RR) and 95% confidence interval (CI). Multivariate analysis was performed by logistic regression. The interval from randomization (pessary or not) to delivery was assessed using Kaplan-Meier analysis, where gestational age was the time scale, spontaneous delivery was the event, and elective deliveries were treated as censored. For purposes of this analysis, all pregnancies were considered to be no longer at risk at the start of the 34th week. All statistical analyses were performed with a software package (SPSS, Version 16; IBM Corp, Armonk, NY). No interim analysis was planned for this trial.
