Does cervical length at 13–15 weeks' gestation predict preterm delivery in an unselected population?
Abstract
Objective
To assess the role of early mid-trimester cervical length measurement as a predictor of spontaneous preterm birth in an unselected population.
Methods
In this prospective study, unselected, asymptomatic, Caucasian women with singleton pregnancies underwent standardized transvaginal ultrasonographic (TVS) cervical length measurement at 13–15 weeks' gestation as a screening test for preterm delivery (PTD). Women with multiple gestations, iatrogenic PTD, and previous cervical conization were excluded. The primary outcome measures were spontaneous PTD at < 37 and < 34 weeks. The correlation between cervical length and previous obstetric history was evaluated.
Results
A total of 2469 patients met the inclusion criteria. The mean gestational age at cervical assessment was 14 + 2 weeks. The mean gestational age at delivery was 40 + 0 weeks. The rate of spontaneous deliveries before 37 weeks' gestation was 1.7%. In 0.2% the delivery occurred before 34 weeks' gestation. The mean ± standard deviation cervical length for the entire population was 44.2 ± 5.4 mm. No difference was observed between cervical length in women that delivered at term and those that delivered either before 37 or before 34 weeks' gestation. Previous obstetric history (prior preterm birth, previous miscarriages and terminations, and parity) did not affect cervical length at 14 weeks of gestation.
Conclusions
Performed at 14 weeks' gestation, TVS measurement of the cervical canal length to predict spontaneous PTD is not a reliable screening procedure. Copyright © 2003 ISUOG. Published by John Wiley & Sons, Ltd.
Introduction
Preterm birth, defined as delivery before the completion of 37 weeks of gestation, is a clinical problem of major concern as it affects 7–11% of all infants, and accounts for more than two-thirds of neonatal mortality, morbidity and later sequelae, particularly if it takes place at a gestational age prior to 34 weeks1.
The mechanisms that underlie the syndrome of spontaneous preterm birth are multifactorial and not completely understood2. Dramatic reductions in perinatal and infant mortality have occurred during the last several decades, mainly attributable to the improvement in the quality of high-risk obstetric and neonatal care. However, this has not been associated with a reduction of preterm births, due to the combination of the poor positive predictive values of the available screening tests3, the low success rate of interventional modalities to prevent preterm delivery (PTD) in high-risk patients, and the increasing incidence of multiple pregnancies.
Many authors have focused their attention on the study of cervical changes in an attempt to predict PTD. Digital examination of the cervix has its shortcomings4, 5 as it is subjective6, inaccurate for evaluating the internal os7, non-specific8, and represents a risk factor for the introduction of vaginal microorganisms into the cervical canal9.
Transabdominal ultrasonography has limitations for the measurement of the cervix because the fetal presenting part may obscure the cervix10, and the full bladder may cause a lengthened measurement of the cervical canal11.
Transvaginal sonography (TVS) is the tool of choice for the imaging of the uterine cervix. TVS is simple, reproducible12, 13, acceptable to patients12, and is inexpensive. It also provides high-quality images of the cervix, enables the assessment of the cervical length and funneling14 as defined as dilatation of the internal cervical os, when present, and has little or no risk of carrying vaginal microorganisms into the cervical canal as the probe is placed in the vaginal fornix15.
By accurately identifying a short cervix, cervical TVS may be useful for selecting women with the greatest need for therapeutic strategies to prevent spontaneous preterm birth. Even though the efficacy of cervical cerclage is still under debate16, 17, there is evidence that the procedure should be performed as early as possible in gestation to achieve the highest rate of success.
In the literature there is evidence that the detection of a short cervix in the late second trimester by means of TVS is a strong predictor of PTD4, 18-22. Nevertheless, the role of this method as a screening tool to predict PTD before 16 weeks' gestation is still controversial4, 19, 23-25.
The purpose of this study was to determine whether there is a relationship between cervical length measured at 14 weeks' gestation in an unselected population of singleton pregnancies and the risk of spontaneous preterm birth.
Methods
From July 1991 to September 1997, 2550 consecutive, unselected, asymptomatic pregnant women attending the ultrasound laboratory of the Institute ‘Burlo Garofolo’ Department of Obstetrics and Gynecology in Trieste for early scanning of fetal malformations by TVS underwent cervical length measurement. The results of the anatomy scans for this study group have been presented previously26.
The ultrasonographic examinations were performed between 13 + 0 weeks and 15 + 6 weeks of gestation by a group of experienced operators, using an Acuson 128 XP10 ultrasound system equipped with a 5-MHz transvaginal probe (Acuson Inc., Mountain View, CA, USA). Gestational age was confirmed by measuring biparietal diameter and femur length. Each patient underwent transvaginal cervical assessment only once.
We measured only the cervical length because it is easy to obtain and is reproducible. Other parameters correlated with the prediction of PTD, such as cervical funneling, were not studied because they have not shown an advantage over cervical length27.
The technique was standardized in order to obtain reproducible measurements. (1) The women were asked to empty their bladder just before the examination. (2) The transducer was placed as close as possible to the cervix without pressure, to avoid any deformation or elongation of the cervical canal28. (3) The entire length of the cervical canal, identified by the sonolucent endocervical mucosa, was to be visualized in a longitudinal view of the cervix. (4) Calipers were used to measure the distance between the triangular area of echodensity at the external os and the T-shaped or V-shaped internal os (Figure 1). (5) To avoid the shortening effect of the physiological curve of the cervix29, when the cervical canal appeared to be curved, the measurement was divided into two segments, whose length was summed and recorded as the whole cervical length30 (Figure 2). (6) The assessment of the cervix was carried out over a period of approximately 3 min in order to note any dynamic changes in the cervix. (7) The shortest of at least three measurements was recorded.
Exclusion criteria for this study were: multiple pregnancy, elective PTD for maternal or fetal conditions, and history of cervical conization. Demographic characteristics, ultrasound findings, and data on pregnancy outcome were recorded in a computer database. The primary outcome measures were spontaneous PTD at < 37 and < 34 weeks, because of the higher rate of morbidity and mortality occurring in infants born before 34 weeks' gestation31.

Transvaginal ultrasonographic measurement of cervical length at 14 weeks' gestation (31 mm). AL, anterior lip; EO, external os; F, fetal part; IO, internal os; PL, posterior lip.

Transvaginal longitudinal scan of the cervix at 15 weeks' gestation (28 mm): the cervical canal is curved and the measurement is divided into two segments. AL, anterior lip; D1, distance 1 (13 mm); D2, distance 2 (15 mm); EO, external os; F, fetal part; IO, internal os; PL, posterior lip.
The correlation between cervical length and previous obstetric history was evaluated stratifying the population according to the following factors: history of one or more previous spontaneous PTD (between 23 and 37 weeks), two or more dilatation and curettage procedures for first-trimester (before 13 weeks) and second-trimester (between 13 and 22 weeks) spontaneous miscarriage or voluntary termination of pregnancy, and parity.
In evaluating previous obstetric history, we considered the difference between proportions for categorical variables using the Z-test with the significant confidence interval of 95%.
The cervical length measurements were expressed as mean, ± standard deviation (SD), range, and percentiles. The Kolmogorov–Smirnov test was used to determine whether the cervical measurements were normally distributed. Student's t-test was used to compare mean cervical length between subgroups. Statistical significance was set at P < 0.05.
Results
During the study period, 2550 Caucasian pregnant women underwent cervical assessment as a screening test for PTD. Twenty-one women (0.8%) were excluded from further evaluation due to multiple pregnancy and 60 (2.3%) were excluded as a result of elective delivery before 37 weeks' gestation for maternal or fetal conditions. No patients with a history of conization were seen. A total of 2469 singleton pregnancies met the inclusion criteria and were submitted for further analysis.
The mean ± SD gestational age at ultrasonographic examination was 14 + 2 weeks ± 4 days (range, 13 + 0 weeks to 15 + 6 weeks). The mean ± SD cervical length for the entire population was 44.17 ± 5.43 mm (range, 23–68 mm). Figure 3 shows that cervical length was normally distributed. The mean ± SD gestational age at delivery was 40 + 0 weeks ± 1 + 2 weeks (range, 23 + 2 weeks to 44 + 6 weeks).

Normal distribution (solid line) and frequency distribution (bars) of cervical length at 13–15 weeks' gestation (mean ± SD, 44.17 ± 5.43; n = 2469).
Table 1 describes the demographic and clinical characteristics of the whole population and of the patients stratified into three groups, according to the gestational age at delivery: at term, between 34 and 37 weeks' gestation, and before 34 weeks' gestation. The overall rate of spontaneous deliveries before 37 weeks' gestation was 1.7% (42/2469). In six cases (0.2%) delivery occurred before 34 weeks' gestation. There were ten women (0.4%) with at least one previous PTD, 85 (3.4%) with two or more previous first- and second-trimester miscarriages, 97 (3.9%) with two or more previous first- and second-trimester terminations of pregnancy, 1478 (59.9%) with at least one previous spontaneous term delivery, and 991 (40.1%) nulliparous women. Patients with PTD showed a higher rate of prior preterm birth (P < 0.05), previous miscarriages or terminations (not significant), and nulliparity (P < 0.05) than women who delivered at term.
Total | Term delivery | Delivery at 34–37 weeks | Delivery at < 34 weeks | Statistical significance | |
---|---|---|---|---|---|
n (%) | 2469 | 2427 (98.3) | 36 (1.5) | 6 (0.2) | |
Mean maternal age in years (± SD) and range | 31 (± 5), 17–46 | 31 (± 5), 17–46 | 30 (± 5), 23–43 | 35 (± 5), 32–42 | |
Mean parity (± SD) and range | 1 (± 1), 0–12 | 1 (± 1), 0–12 | 1 (± 1), 0–8 | 1 (± 1), 0–2 | |
Prior spontaneous preterm birth (%) | 10 (0.4) | 8 (0.3) | 1 (2.8) | 1 (16.7) | P < 0.05* |
≥ 2 miscarriages (%) | 85 (3.4) | 83 (3.4) | 2 (5.5) | 0 | NS* |
≥ 2 terminations of pregnancy (%) | 97 (3.9) | 93 (3.8) | 4 (11.1) | 0 | NS* |
≥ 1 term delivery (%) | 1478 (59.9) | 1465 (60.4) | 10 (27.8) | 3 (50) | P < 0.05* |
Nulliparous (%) | 991 (40.1) | 962 (39.6) | 26 (72.2) | 3 (50) | P < 0.05* |
- * Term deliveries versus all cases of preterm deliveries. NS, not significant; SD, standard deviation.
Table 2 shows the mean, SD, range, and percentiles of cervical canal measurement at 14 weeks' gestation in women who subsequently delivered at term and preterm. The mean ± SD cervical length in women with term delivery was 44 ± 5 mm. No difference was observed in either the whole group of patients with PTD, or in the two subcategories of women that delivered between 34 and 37 weeks and before 34 weeks of gestation.
Term delivery | All cases with PTD < 37 weeks | PTD 34–37 weeks | PTD < 34 weeks | Statistical significance | |
---|---|---|---|---|---|
n | 2427 | 42 | 36 | 6 | |
Mean cervical length | 44 | 44 | 45 | 44 | NS* |
± 1 SD | 5 | 6 | 6 | 7 | |
Range | 23–68 | 34–60 | 37–60 | 34–54 | |
5th percentile | 37 | 38 | 38 | 35 | NS* |
25th percentile | 40 | 39 | 40 | 39 | NS* |
50th percentile | 44 | 44 | 44 | 45 | NS* |
90th percentile | 51 | 52 | 52 | 50 | NS* |
- * Term deliveries versus all cases of PTDs. NS, not significant; PTD, preterm delivery; SD, standard deviation.
Tables 3 and 4 describe the effect of the studied risk factors (previous spontaneous PTD, two or more previous spontaneous first- and second-trimester miscarriages or terminations of pregnancy), and parity on the cervical length in the early mid-trimester. None of these factors was related to cervical length at 14 weeks' gestation in our population.
Whole group | Term delivery | PTD | Statistical significance | |
---|---|---|---|---|
History of previous PTD at 23–37 weeks | ||||
n | 10 | 8 | 2 | |
Mean cervical length | 44 | 45 | 41 | NS |
± 1 SD | 4 | 4 | — | |
Range | 37–49 | 37–49 | 38–43 | |
5th percentile | 37 | 39 | 38 | NS |
25th percentile | 43 | 44 | 39 | NS |
50th percentile | 44 | 45 | 40 | NS |
90th percentile | 48 | 48 | 42 | NS |
History of previous miscarriages at < 23 weeks | ||||
n | 85 | 83 | 2 | |
Mean cervical length | 44 | 44 | 50 | NS |
± 1 SD | 6 | 6 | — | |
Range | 30–68 | 30–68 | 40–60 | |
5th percentile | 37 | 35 | 41 | NS |
25th percentile | 40 | 40 | 45 | NS |
50th percentile | 44 | 44 | 50 | NS |
90th percentile | 50 | 50 | 58 | NS |
History of previous terminations at < 23 weeks | ||||
n | 97 | 93 | 4 | |
Mean cervical length | 44 | 44 | 45 | NS |
± 1 SD | 6 | 6 | 10 | |
Range | 23–60 | 23–58 | 38–60 | |
5th percentile | 36 | 36 | 38 | NS |
25th percentile | 40 | 40 | 39 | NS |
50th percentile | 45 | 45 | 41 | NS |
90th percentile | 52 | 51 | 55 | NS |
- NS, not significant; PTD, preterm delivery; SD, standard deviation.
Whole group | Term delivery | PTD | Statistical significance | |
---|---|---|---|---|
All nulliparous women | ||||
n | 991 | 962 | 29 | |
Mean cervical length | 44 | 44 | 45 | NS |
±1 SD | 6 | 5 | 5 | |
Range | 27–68 | 23–68 | 38–52 | |
5th percentile | 37 | 36 | 38 | NS |
25th percentile | 40 | 40 | 40 | NS |
50th percentile | 44 | 43 | 45 | NS |
90th percentile | 52 | 51 | 52 | NS |
History of ≥ 1 previous term delivery | ||||
n | 1478 | 1465 | 13 | |
Mean cervical length | 44 | 44 | 44 | NS |
± 1 SD | 5 | 6 | 6 | |
Range | 23–68 | 23–68 | 34–60 | |
5th percentile | 37 | 37 | 37 | NS |
25th percentile | 40 | 40 | 40 | NS |
50th percentile | 44 | 44 | 43 | NS |
90th percentile | 51 | 52 | 51 | NS |
- NS, not significant; PTD, preterm delivery; SD, standard deviation.
Discussion
Preterm birth is one of the most challenging problems in modern obstetrics. Despite the recent improvements in neonatal care, the rate of prematurity has even been increasing in the last two decades32. Major neonatal morbidity increases with decreasing gestational age and weight at birth33. Therefore, every effort to identify a predictive method for the early detection of women at risk for PTD, especially before 34 weeks of gestation31, and useful intervention to prevent preterm birth is justified.
Spontaneous preterm labor has multiple etiologies resulting in the activation of the common terminal pathway of parturition so primary cervical changes before term represent only part of the problem34. Cervical incompetence accounts for not more than 2% of late miscarriages or PTDs35. When suspected, this condition is commonly treated by the prophylactic application of a cervical cerclage, even though it has been demonstrated that this procedure is not effective in reducing the rate of PTD before 34 weeks' gestation or in improving the neonatal survival rate36. Furthermore, cervical cerclage is an invasive surgical intervention with a potential risk of morbidity37, 38. Hence, the need for a reliable method for the assessment of the uterine cervix is imperative, in order to distinguish patients that might benefit from the suture from those in whom it would be unnecessary.
Obstetric history has been shown to be correlated with PTD, but is of limited value in identifying patients at higher risk of preterm birth and who are eligible for preventive intervention. According to the Medical Research Council/Royal College of Obstetricians and Gynaecologists' multicenter randomized trial of cervical cerclage, 25 cerclage placements would need to be performed to prevent one PTD before 33 weeks' gestation38. Holbrook et al.39 reported that, even though a clinical scoring system was a better predictor of PTD than individual risk factors, its sensitivity (41%) and positive predictive value (24.6%) were suboptimal.
Many studies have shown that cervical changes seen by TVS are associated with an increased risk of preterm birth and that the length of the cervix is directly correlated with the duration of pregnancy: the shorter the cervix, the greater the likelihood of PTD18. It has been reported that a cervical length cut-off of 25 mm in the mid-trimester is a good predictor of PTD4, 18, 24. Patients with a cervical length of ≤ 15 mm on a mid-trimester scan had a 48%19 and 52%40 risk of early preterm birth (≤ 32 weeks' gestation). Heath et al.22 performed TVS to measure cervical length in 2567 singleton pregnancies attending for routine antenatal care at 22–24 weeks' gestation. The cervical length was ≤ 15 mm in 1.7% of 1252 pregnancies that were managed expectantly. Among these patients there were about 40% of women with spontaneous PTD at ≤ 34 weeks, 60% of those with PTD at ≤ 32 weeks, 80% of those with PTD at ≤ 30 weeks and 100% of those with PTD at ≤ 26 weeks.
There are very few reports on the reliability of ultrasonography in the early mid-trimester (before 16 weeks' gestation) as a predictor of PTD, and the results are contradictory (Table 5). In the studies of Naim et al.23, Berghella et al.4, and Andrews et al.24, the early detection of a short cervix was strongly associated with subsequent spontaneous preterm birth. Hassan et al.19 demonstrated that a shortened cervix at 14–19 weeks' gestation was a weaker predictor of PTD than if seen later in the mid-trimester. Opposite results were reported by Hasegawa et al.25 and Zorzoli et al.41, as they found no correlation between early cervical ultrasonographic measurement and gestational age at delivery. These studies are not homogeneous in terms of gestational age at which the scan was carried out, presence of risk factors, and study design.
Reference | GA at ultrasound (weeks) | Route of examination | Study size (n) | Study population | CL cut-off (mm) | Outcome: GA at delivery (weeks) | Results |
---|---|---|---|---|---|---|---|
Naim et al. (2002)23 | < 16 | Transvaginal | 154 | Low risk | < 30 | < 37 | A short cervix was associated with a 23% risk of PTD |
Berghella et al. (1997)4 | 14–17 | Transvaginal and translabial | 70 | High risk | < 25 (25th percentile) | < 35 | Shorter CL in patients with PTD (P < 0.0005) |
Andrews et al. (2000)24 | 15–20 | Transvaginal | 53 | High risk | ≤ 22 (10th percentile) | < 35 | PTD within 2 weeks of ultrasound in 25%, within 4 weeks in 75%, and < 35 weeks in 100% |
Hassan et al. (2000)19 | 14–24 | Transabdominal and transvaginal | 6877 | Unselected | ≤ 25 | ≤ 32 | Higher probability of PTD when a short cervix was detected at > 19–24 weeks, than at 14–19 weeks |
Hasegawa et al. (1996)25 | 8–12 | Transvaginal | 298 | Low risk | < 30 | < 37 | No difference observed in CL of patients with PTD |
Zorzoli et al. (1994)41 | 12 | Transvaginal | 154 | Unselected | < 37 | No difference observed in CL of patients with PTD |
- CL, cervical length; GA, gestational age; PTD, preterm delivery.
To the best of our knowledge, this is the first study in which a ‘one-shot’ cervical measurement at 13–15 weeks was correlated to the prediction of PTD. In our prospective study, 2469 unselected Caucasian pregnant women underwent TVS cervical assessment at 14 weeks' gestation as a screening test for PTD. In this population the mean gestational age at delivery was 40.0 weeks of gestation and the rate of PTD was 1.7%. Our results regarding obstetric history as a risk factor for PTD are consistent with other reports39, 42. The main finding of this study is that early measurement of cervical length was comparable in women who delivered at term and patients with PTD, regardless of the presence of risk factors.
In our opinion, the explanation for the contradictory results between reports in which cervical sonography in the early mid-trimester was found to be a good predictive tool for PTD4, 23, 24, and those where it was not19, 25, 41, including the present study, can be attributed both to methodological factors and study design.
First, in our population, cervical length at 14 weeks' gestation was comparable in women with and without risk factors. This result is in contrast to other studies conducted at later gestational ages, which demonstrated the association between previous mid-trimester losses or PTD, and late second-trimester cervical shortening12, 43, 44. One possible explanation is that in the present study the gestational age at which the sonogram was performed was far from the age at preterm birth. Therefore, we assume that in the early mid-trimester, past obstetric history is a better predictor of PTD than is the measurement of the cervix by ultrasonography. As already pointed out by Heath et al.22, the association between past obstetric history with PTD may be mediated through cervical changes that are apparent as cervical shortening only late in the second trimester.
Second, the presence of high risk factors, such as previous PTD and miscarriage, in the studied population affects the positive predictive value increasing the prevalence of PTD. Our unselected population was at very low risk for PTD, as there were very few patients with risk factors. Therefore our results may not be valid for high-risk populations. Furthermore, factors other than obstetric history are associated with changes in the length of the cervix. They are psychosocial and sociodemographic factors such as age, race, occupation, smoking, and stress12, 45, 46. The best comparison of cervical measurements between different studies is obtained by defining cut-off values for cervical length as standard deviations of a nomogram of mean cervical lengths of the appropriate population47.
Finally, during normal pregnancy, the cervix shortens progressively30, 48 and becomes straighter29. To avoid underestimation of the cervical length, in the presence of cervical curvature, the measurement should be the sum of the length of two or more segments, according to the technique reported by Gramellini et al.30.
Therefore, gestational age at ultrasonographic examination, standardization of the measurement of the cervical canal, size of the population, and presence or absence of risk factors are all crucial variables to be considered to avoid biases.
In conclusion, although TVS has been found to offer valuable diagnostic and prognostic information when used for assessing cervical changes in the late second trimester, or in cases at risk for PTD, or threatened preterm labor, according to our findings it is not reliable as a screening procedure in low-risk populations to predict PTD when performed at approximately 14 weeks' gestation.
Acknowledgements
The authors thank Anna Legovini for helpful discussion on statistical presentation of the data.