Introduction

Twin pregnancies are associated with an increased risk of perinatal morbidity and mortality primarily due to spontaneous preterm deliveries. The mean gestational age for delivery is 35.3 weeks and twins account for 23% of preterm births <32 weeks. A number of strategies have been proposed to prevent preterm deliveries: tocolytics, bed rest, hospitalization, home uterine activity monitoring, cerclage, and most recently, progesterone. Unfortunately, none have proven effective.

Recent metaanalyses and reviews suggest that transvaginal cervical length (TVCL) ultrasound in the second trimester is a powerful predictor of preterm birth among asymptomatic women. Indeed, TVCL has the highest positive and negative predictive values for determining the risk of spontaneous preterm delivery in twin pregnancies. It follows that TVCL assessment may allow identification of a subset of twin pregnancies that are better candidates for interventions intended to prevent prematurity.

Retrospective studies designed to assess the clinical utility of TVCL in the prevention of prematurity among twin pregnancies have yielded conflicting results. Michaels and colleagues compared the outcomes of twin pregnancies managed with transabdominal cervical length ultrasound to an historic control group managed without such ultrasound and both perinatal mortality and prematurity were decreased. Conversely, Gyamfi and colleagues compared the outcomes of twin pregnancies managed with TVCL to those of controls without exams and noted no differences in gestational age at delivery or the rate of preterm birth.

Given the encouraging predictive values of TVCL and the conflicting results from retrospective studies, we designed a randomized controlled trial to assess the utility of routine TVCL in the prevention of prematurity among twin pregnancies. The primary hypothesis was that twin pregnancies managed with the information gained from TVCL ultrasound would have a greater length of gestation and fewer preterm deliveries.

Materials and Methods

From December 2002 through June 2006, we performed a multicenter, randomized, controlled trial at 3 US military hospitals. The study was approved by the institutional review board and ethics committee at each hospital and with ClinicalTrials.gov ( NCT02372422 ). Informed consent was obtained on all subjects.

All twin pregnancies at <20 weeks were identified by ultrasound and referred for screening and possible recruitment. The subjects were either active duty members or military dependents. Because of frequent military moves, if individuals moved (learned of move after randomization) we planned to contact the new physician, inform them of the study, and ask for continued management according to our protocol (with subject consent) and obtain data for inclusion in the analysis.

Prior to randomization, subjects received counseling about the risks of prematurity and the possibility of a benefit to TVCL assessment in the prevention of preterm birth. Subjects were excluded if they had any of the following: (1) history of cervical insufficiency with plans for a history-indicated cerclage; (2) a history of possible cervical insufficiency with plans to perform serial TVCL; (3) a prior preterm delivery <28 weeks; (4) preexisting plans to relocate and deliver at a different hospital; (5) discovery of a lethal fetal congenital anomaly, confirmation of aneuploidy, monoamniotic twins, or twin-twin transfusion syndrome prior to enrollment; or (6) age <18 years. If not excluded, randomization was conducted with the use of a random number table in permuted blocks of 6 using opaque, sequentially numbered sealed envelopes at a central site. The study group was defined as the group assigned to have routine TVCL studies. A nurse who was trained in prematurity prevention counseled all participants in both groups and provided preterm labor educational handouts.

Scheduled visits occurred at 15-16, 19-20, 23-24, and at 27-28 weeks for all women. As was our practice, a gentle digital examination of the cervix was performed at each of these visits. Subjects randomized to the study group underwent TVCL either by one of the investigators or a trained sonographer and a set protocol was followed as described by Iams et al. All images were reviewed by the first author to ensure correct measurements were obtained. Based on the results of the TVCL, all women in the study group were managed according to a standard algorithm ( Table 1 ). In addition to the scheduled exams, TVCL were allowed if subjects presented with preterm labor.

Subjects in the control group were managed without the use of routine TVCL. In addition, use of this exam was discouraged in symptomatic women except for specific indications: symptoms of cervical insufficiency to include heavy vaginal discharge or bleeding at 16-22 weeks or for evaluation of bleeding to evaluate for placenta previa. Likewise, no transabdominal evaluation of the cervix was performed. No standardized protocol was established for the management of this group. The decision to perform a cerclage or alter activity status was left to the managing clinicians. Women who were placed at bed rest for preterm labor were also contacted weekly by a nurse. Subjects in the control group received a single blinded TVCL at 23-24 weeks to compare the mean cervical length between the 2 groups. Both the women and the managing physicians were blinded to the results and the exam was not performed by these physicians.

To minimize confounders, both groups were seen by the same providers and efforts were made to keep management identical except for the use of the TVCL. The diagnosis and management of preterm labor was standardized with the exception that women in the study group were allowed to have TVCL. The timing and use of steroids for asymptomatic individuals was based on either the TVCL protocol or was left to the discretion of the managing obstetrician. Planned deliveries <38 weeks were discouraged. Women who declined participation in the study or who were ineligible received care similar to the control group unless the reason for the exclusion required serial TVCL scans.

The primary outcome was gestational age at delivery. Since planned deliveries of twins were common at 38 weeks and to ensure that this did not affect the primary outcome, all subjects who delivered after this time were recorded as a delivery at 38 0/7 weeks.

We recorded data regarding demographics, risk factors, type of twin placentation as determined by ultrasound, and evidence of twin-twin transfusion. Additional maternal outcomes included the number of days of bed rest, number of hospital days, use of tocolytics, and administration of steroids for fetal lung maturity. Neonatal outcomes included birthweight, length of stay in the neonatal intensive care unit, and occurrence of a predefined composite measure of severe morbidity. When comparing neonatal outcome, twin A and twin B were compared separately. In addition to quantitative measures of gestational age, preterm births were categorized as spontaneous or medically indicated.

All analyses were performed by intention to treat. Interval data were compared with an unpaired t test or Mann-Whitney U test and categorical data were compared with the χ ² test or Fisher exact test depending on the cell size. For the primary outcome we planned to construct Kaplan-Meier survival curves for gestational age at delivery. The threshold for statistical significance for all tests was set at P < .05. All statistical procedures were performed using software (IBM SPSS Statistics for MAC, Version 22; IBM Corp, Armonk, NY).

For the primary outcome of gestational age at delivery, we estimated that 45 subjects would be needed in each group to detect an improvement from 36.5-38 completed weeks of gestation with 95% confidence and a power of 0.80. With institutional review board approval we added additional subjects to explore a secondary outcome, preterm delivery <35 weeks. We estimated that an additional 20 patients would be needed in each group to detect a 50% reduction in the number of deliveries <35 weeks with the same confidence and power.

Materials and Methods

From December 2002 through June 2006, we performed a multicenter, randomized, controlled trial at 3 US military hospitals. The study was approved by the institutional review board and ethics committee at each hospital and with ClinicalTrials.gov ( NCT02372422 ). Informed consent was obtained on all subjects.

All twin pregnancies at <20 weeks were identified by ultrasound and referred for screening and possible recruitment. The subjects were either active duty members or military dependents. Because of frequent military moves, if individuals moved (learned of move after randomization) we planned to contact the new physician, inform them of the study, and ask for continued management according to our protocol (with subject consent) and obtain data for inclusion in the analysis.

Prior to randomization, subjects received counseling about the risks of prematurity and the possibility of a benefit to TVCL assessment in the prevention of preterm birth. Subjects were excluded if they had any of the following: (1) history of cervical insufficiency with plans for a history-indicated cerclage; (2) a history of possible cervical insufficiency with plans to perform serial TVCL; (3) a prior preterm delivery <28 weeks; (4) preexisting plans to relocate and deliver at a different hospital; (5) discovery of a lethal fetal congenital anomaly, confirmation of aneuploidy, monoamniotic twins, or twin-twin transfusion syndrome prior to enrollment; or (6) age <18 years. If not excluded, randomization was conducted with the use of a random number table in permuted blocks of 6 using opaque, sequentially numbered sealed envelopes at a central site. The study group was defined as the group assigned to have routine TVCL studies. A nurse who was trained in prematurity prevention counseled all participants in both groups and provided preterm labor educational handouts.

Scheduled visits occurred at 15-16, 19-20, 23-24, and at 27-28 weeks for all women. As was our practice, a gentle digital examination of the cervix was performed at each of these visits. Subjects randomized to the study group underwent TVCL either by one of the investigators or a trained sonographer and a set protocol was followed as described by Iams et al. All images were reviewed by the first author to ensure correct measurements were obtained. Based on the results of the TVCL, all women in the study group were managed according to a standard algorithm ( Table 1 ). In addition to the scheduled exams, TVCL were allowed if subjects presented with preterm labor.

Subjects in the control group were managed without the use of routine TVCL. In addition, use of this exam was discouraged in symptomatic women except for specific indications: symptoms of cervical insufficiency to include heavy vaginal discharge or bleeding at 16-22 weeks or for evaluation of bleeding to evaluate for placenta previa. Likewise, no transabdominal evaluation of the cervix was performed. No standardized protocol was established for the management of this group. The decision to perform a cerclage or alter activity status was left to the managing clinicians. Women who were placed at bed rest for preterm labor were also contacted weekly by a nurse. Subjects in the control group received a single blinded TVCL at 23-24 weeks to compare the mean cervical length between the 2 groups. Both the women and the managing physicians were blinded to the results and the exam was not performed by these physicians.

To minimize confounders, both groups were seen by the same providers and efforts were made to keep management identical except for the use of the TVCL. The diagnosis and management of preterm labor was standardized with the exception that women in the study group were allowed to have TVCL. The timing and use of steroids for asymptomatic individuals was based on either the TVCL protocol or was left to the discretion of the managing obstetrician. Planned deliveries <38 weeks were discouraged. Women who declined participation in the study or who were ineligible received care similar to the control group unless the reason for the exclusion required serial TVCL scans.

The primary outcome was gestational age at delivery. Since planned deliveries of twins were common at 38 weeks and to ensure that this did not affect the primary outcome, all subjects who delivered after this time were recorded as a delivery at 38 0/7 weeks.

We recorded data regarding demographics, risk factors, type of twin placentation as determined by ultrasound, and evidence of twin-twin transfusion. Additional maternal outcomes included the number of days of bed rest, number of hospital days, use of tocolytics, and administration of steroids for fetal lung maturity. Neonatal outcomes included birthweight, length of stay in the neonatal intensive care unit, and occurrence of a predefined composite measure of severe morbidity. When comparing neonatal outcome, twin A and twin B were compared separately. In addition to quantitative measures of gestational age, preterm births were categorized as spontaneous or medically indicated.

All analyses were performed by intention to treat. Interval data were compared with an unpaired t test or Mann-Whitney U test and categorical data were compared with the χ ² test or Fisher exact test depending on the cell size. For the primary outcome we planned to construct Kaplan-Meier survival curves for gestational age at delivery. The threshold for statistical significance for all tests was set at P < .05. All statistical procedures were performed using software (IBM SPSS Statistics for MAC, Version 22; IBM Corp, Armonk, NY).

For the primary outcome of gestational age at delivery, we estimated that 45 subjects would be needed in each group to detect an improvement from 36.5-38 completed weeks of gestation with 95% confidence and a power of 0.80. With institutional review board approval we added additional subjects to explore a secondary outcome, preterm delivery <35 weeks. We estimated that an additional 20 patients would be needed in each group to detect a 50% reduction in the number of deliveries <35 weeks with the same confidence and power.