Objective
We sought to evaluate the change in cervical length (CL) as a predictor of preterm birth in asymptomatic twin pregnancies.
Study Design
We studied a historical cohort of 121 twin pregnancies with CL testing between 18-24 weeks who had a follow-up CL 2-6 weeks after the initial CL.
Results
A total of 19 patients had their CL decrease by ≥20% (shortened CL group) and 102 patients’ CL decreased by less, or not at all (stable CL group). The shortened CL group had a significantly higher rate of spontaneous preterm birth <28 weeks, <30 weeks, <32 weeks, and <34 weeks. This remained true even when excluding patients with a short CL (≤25 mm) on the repeated CL.
Conclusion
In twin pregnancies, a CL that decreases by 20% over 2 measurements is a significant predictor of very preterm birth, even in the setting of a normal CL. Serial CL measurements should be considered in twin pregnancies, starting <24 weeks.
The number and rate of twin births continues to increase, from 2.2% of all US live births in 1990 to 3.2% of all live births in 2005. Preterm birth is the most common morbidity associated with twin pregnancies. A shortened cervix identified by endovaginal ultrasound is strongly predictive of preterm birth in twin pregnancies. In one prospective series of 147 twin pregnancies, for women with a cervical length (CL) ≤25 mm at 24 weeks’ gestation, the odds ratio (OR) of preterm birth <32 weeks was 6.9 (95% confidence interval [CI], 2.0–24.2) compared to women with twin pregnancies and a CL >25 mm.
In addition to a single CL predicting preterm birth, the change in CL over time may also predict preterm birth. A reduction in CL over time has been associated with preterm birth in singleton pregnancies, however, there are fewer data regarding the change in CL in twin pregnancies. One study of 20 women with twin pregnancies examined CL every week beginning at 24 weeks and found that the patterns of cervical change over this time period differed between women who delivered preterm (32-35 weeks) and women who delivered >36 weeks. However, this study did not address the association of earlier cervical changes and preterm birth. Additionally, the greatest morbidity in twin pregnancies is very preterm birth (<32 weeks), as opposed to preterm birth 32-35 weeks. One study of 91 twin pregnancies that did examine CL beginning at 18 weeks showed no predictive value to the change in CL from 18-24 weeks, or from 24-28 weeks. In this study, if the cervix shortened 2.5 mm per week from 18-28 weeks, there was an increased incidence of birth <35 weeks. Data on the relationship between early cervical changes and very preterm birth <32 weeks would be very useful in counseling patients with twin pregnancies, especially in the setting of a CL >25 mm.
In our practice, we routinely assess serial CL measurements in women with twin pregnancies, beginning at 16-20 weeks. Using published criteria, we counsel these women with a short cervix (≤25 mm) that they are at increased risk of preterm birth. We also use the CL to help guide our decisions whether to administer corticosteroids to accelerate fetal lung maturation and whether to administer tocolytics. However, for patients with a CL that has shortened, but remains >25 mm, there are limited data to rely on when counseling patients and making management decisions. The purpose of this study was to estimate the association of a shortening CL and preterm birth in asymptomatic twin pregnancies.
Materials and Methods
A historical cohort of patients was obtained from twin pregnancies in our maternal-fetal medicine practice between 2005 and 2008. Institutional review board approval was obtained prior to conducting the study. All twin pregnancies cared for in our practice were included. In our practice, all women with twin pregnancies undergo an initial CL measurement between 18-24 weeks’ gestation, followed by a repeated CL within 2-6 weeks of the initial CL measurement, regardless of the initial measurement. In this study, we only included CL measurements performed in an office setting in asymptomatic women; all patients with measurements performed as a part of an evaluation for contractions, abnormal discharge, or pelvic pressure were excluded. Patients who had a repeated CL measurement <2 weeks from the initial CL were excluded, as they also likely represented symptomatic women. Gestational age was based on the known last menstrual period and confirmed by first-trimester ultrasound, or based on first-trimester ultrasound in all patients. All CL measurements were measured by 4- to 8-MHz transvaginal probes (LOGIQ a200 and Voluson 530 and 730 Expert; GE Healthcare, Milwaukee, WI) with an empty bladder with the optimal image defined according to the criteria reported by Iams et al. The shortest functional CL recorded was used as this has been found to be the most reproducible measurement. Monoamniotic twins were excluded, as were pregnancies with aneuploidy, twin-twin transfusion syndrome, or major fetal anomalies discovered before or after birth.
Patients and obstetricians were not blinded to the CL measurements. Patients with a short CL were typically evaluated for contractions, and given steroids with or without tocolytics if deemed appropriate. We do not prescribe bed rest or hospitalize patients with a short cervix, aside from those actually in preterm labor. During the time period in which we cared for these patients, it was not our practice to alter management based on a shortening CL, assuming the repeated measurement was >25 mm.
Change in CL was defined as the difference between the initial CL and the repeated CL measured 2-6 weeks later. In cases where patients had >1 CL measurement during the 2- to 6-week period after the initial CL measurement, the first repeated CL measurement was used for analysis. The primary outcome was spontaneous preterm birth <32 weeks. We also planned a priori to assess spontaneous preterm birth at <28 weeks, <30 weeks, and <34 weeks. Patients with an indicated preterm birth were excluded from analysis of spontaneous preterm birth at all later gestational ages. For example, if there was an indicated preterm birth at 31 weeks, the patient was not included in the analysis for spontaneous preterm birth <32 weeks, but was considered as not having a spontaneous preterm birth <30 weeks and <28 weeks.
Fisher exact test, Student t test, and multivariable analysis were used when appropriate (SPSS for Windows 16.0; SPSS Inc, Chicago, IL). In our practice, approximately 10% of twin pregnancies deliver spontaneously <32 weeks. Assuming that 20% of patients would have a significantly shorter repeated CL, 105 patients would be needed to have 80% power to demonstrate a 4-fold increase in the rate of spontaneous preterm birth <32 weeks with an alpha error of 5%.
Results
A total of 121 patients met the inclusion criteria. Based on receiver operating characteristic curve analysis, the CL was considered to have shortened significantly if the change in CL was ≥20% of the initial CL (area under the curve 0.713 ± 0.080; 95% CI, 0.556–0.870). We examined outcomes in women with a CL that shortened ≥20% (shortened CL group) compared to women with a CL that did not shorten, or shortened <20% (stable CL group).
There were 19 (15.7%) patients in the shortened CL group, and 102 (84.3%) patients in the stable CL group. Baseline characteristics are described in Table 1 and were similar between the 2 groups. The time between CL measurements did not differ between the 2 groups. As expected, the change in CL per day was significantly greater in the shortened CL group.
| Characteristic | Shortened CL group (n = 19) | Stable CL group (n = 102) | P |
|---|---|---|---|
| Age, y | 33.6 ± 5.0 | 34.8 ± 7.3 | .389 |
| White | 78.9 | 87.3 | .270 |
| Starting body mass index, kg/m 2 | 23.82 ± 4.99 | 23.79 ± 4.46 | .975 |
| In vitro fertilization, % | 63.2 | 71.6 | .585 |
| Multifetal pregnancy reduction | 10.5 | 11.8 | 1.000 |
| Prior preterm birth, % | 5.3 | 6.9 | .797 |
| Monochorionic, % | 15.8 | 10.8 | .531 |
| Initial CL, mm | 38.6 ± 9.0 | 40.9 ± 7.2 | .227 |
| Days between CL measurements | 24.7 ± 9.7 | 23.4 ± 8.4 | .539 |
| CL change/d, mm | −0.53 ± 0.34 | −0.01 ± 0.27 | < .001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
