*Correspondence: Dr I. Hoesli, Universitäts Frauenklinik, Schanzenstr. 46, CH-4031 Basel, Switzerland.
Transvaginal ultrasonography has recently been shown to be an objective, reproducible and reliable method to assess the cervix and predict the risk of preterm delivery in high-risk pregnancies. Assessment of the cervix includes cervical length measurement (CLM) and measurement of dilatation of the internal os in a dynamic functional examination. There is an inverse correlation between cervical length and the frequency of preterm delivery. The high negative predictive value avoids unnecessary interventions such as tocolysis or cerclage in high-risk pregnancies. In contrast, a length of 25 mm or less at 28–30 weeks of gestation is associated with a significantly increased incidence of preterm delivery. Studies in women with high risk for preterm delivery, i.e. contractions, premature rupture of the membranes and history of preterm delivery, have shown a high sensitivity and a high positive predictive value, however in low-risk groups they have failed to show a high sensitivity. From large observational studies in low-risk populations we know that the 50th percentile of the cervical length is 35 mm at 24 weeks of gestation3. Advantages of CLM as a screening test include the fact that sonographical assessment of the cervix is a widely accepted and well-standardised method, which requires only a relatively short period of training. Disadvantages of screening are two factors, the first being the low sensitivity of the test and the low prevalence of preterm deliveries in a low-risk population, resulting in cut off values being set at a very low level (i.e. 5th percentile) in order to get acceptable specificity. Secondly, screening is only worthwhile if an effective preventive therapy is available. The debate about tocolysis and cerclage is not yet concluded. Therefore we would not currently recommend cervical length measurement as a screening tool—but as a routine method in high risk gravidas with or without symptoms. Further interest should be focused on scoring systems combining ultrasound with biochemical, endocrinological and maybe molecular cell methods such as the measurement of fetal DNA in maternal blood to prevent preterm deliveries in the general population.
Transvaginal ultrasonography has recently been shown by several randomised and non-randomised trials to be an objective, reproducible and reliable method to assess the cervix and so to be predictive for the risk of preterm labour in high-risk women. Furthermore, it plays an important role as a diagnostic test for differentiating between false and true labour in the context of preterm birth.
The rate of preterm deliveries (<37 weeks of gestation) has remained stable over the last decade, ranging between 6–8%1,2 in Europe and Australia and 9.6–11.6% in Canada and North America3,4. Risk factors include twin and higher order pregnancies due to female and male sterility treatment, an increase in maternal age and the spontaneous increase of twin pregnancies5,6. Uterine distension activates gap junctions, oxytocin receptors and prostaglandin synthetase and thus can lead to preterm uterine contractions and cervical shortening. Several different pathways leading to preterm delivery are involved and each marker has been evaluated in different studies7. Additionally Lo et al.8 reported an increased release of free fetal DNA into the maternal circulation in women with subsequent preterm labour and birth but not in women with preterm labour but term delivery (‘false’ preterm labour). Just recently we reported that erythroblasts in maternal blood are not significantly elevated in women with threatened prematurity or in preterm birth and are therefore not useful as a marker9. This confirms that the placenta provides a relatively impermeable barrier, since contractions do not lead to an increased influx of erythroblasts. The source of elevated levels of cell free DNA, which has been observed by ourselves and Lo's group, could be a reflection of an apoptotic process in the placenta preceding preterm labour.
Assessing Cervical Length
The last step before preterm birth occurs involves uterine contractions—not quite specific—and changes in the cervical length. As changes of cervical volume might even be a better predictor for preterm delivery than cervical length, we analysed cervical volume in symptomatic women with contractions and risk of preterm delivery. In an observation study with a control group, we used a 3D ultrasound approach for volume assessment, which, according to our hypothesis, should have been more accurate than volume measurement by 2D ultrasound. However, compared to cervical length measurements, volume measurement did not show a statistically significant improvement and so it was abandoned for further studies10 (Fig. 1).
Cervical length can be assessed by digital examination, but even serial digital examination cannot lead to a reduction in the preterm birth rate. One difference between digital and sonographical assessment is the visibility of the cervical canal and the internal os. Therefore, cervical length viewed sonographically will be almost twice as long as given by digital examination11. For sonographical assessment, experience in ultrasound and a standardised protocol are mandatory. Although perineal and translabial ultrasound can be performed with comparable accuracy, the transvaginal examination is superior as it gives a clear view even of the internal os. The examination is well accepted by pregnant women as demonstrated in a prospective study12. The technique is simple: a picture magnified to at least 75% of the screen including the complete overview of the cervical canal and the internal os can be stored, and the shortest of three measurements with is taken (Fig. 2). In addition to the cervical assessment, several advantages of the ultrasound guided approach have been identified, i.e. the possibility to evaluate the fetal structures or placenta. The inter- and intra-observer variability is lower with the transvaginal examination (4–10% and 5%, respectively) than the digital examination13,14. Errors can occur with pressure on the cervix by the probe, where the cervix is elongated iatrogenically. Besides the complete view, the correct measurement must provide an image at an equal distance from the surface of the anterior and the posterior lip to the cervical canal. In about 1% of cases, the assessment remains inadequate even if all procedures are followed correctly.
Several studies with different populations confirm that in the course of pregnancy the cervical length decreases14,15. As repeated measurements may be necessary, we created charts with the 5th, 50th and 95th centiles from 20 weeks up to 34 weeks of gestation from a low-risk population including more than 600 pregnant women with singleton pregnancy delivering at term (>37 weeks of gestation). These charts can be used for observing women at high risk for preterm delivery and for clearly identifying a significant deviation or decline in the centiles for these subjects (Fig. 3).
The dilatation of the internal os is a further marker, which can be assessed by ultrasound spontaneously or after transfundal pressure, standing or coughing. Relevant measurements are dilatation of more than 5 mm with or without bulging of the membranes.
Cervical Length as a Predictor of Preterm Birth
What are the benefits of choosing cervical length measurement instead of digital examination? Receiver operator curves indicate a higher sensitivity and a lower false positive rate for the length measurement in women at risk of preterm birth. They have indicated a significant relationship between the occurrence of preterm labour and ultrasonographical parameters, but no correlation to the results of a digital examination was found16. The same results were confirmed by Berghella et al. for the length and the internal os17. In a logistic regression analysis, both the digital examination and the sonographical measurement correlated inversely with the probability of preterm labour. There was a much better correlation for the ultrasound approach in the group of women with a history of previous preterm birth. A systematic review confirmed the good predictive values in women with preterm contractions18,19. Remaining limitations of the studies assessing cervical length measurement include the absence of standardisation of gestational age at the time of investigation, different cut off levels and different definitions of preterm birth (<32, <35 or <37 weeks of gestation).
Multiple pregnancies represent a significant risk group within the preterm birth cohort, accounting for about 20% of preterm births. Among women with twins, the rate of preterm delivery is significantly higher, it the cervix is already shortened (≤20 mm) between 15–24 weeks of gestation20. Cut off values have been established for triplets as well, although the number of women is small: cervical length ≤25 mm between 15–24 weeks of gestation is associated with a 51% rate of preterm delivery before 32 weeks of gestation21.
Compared to the high-risk groups with either a history of previous preterm delivery or multiple pregnancies, where the prevalence for prematurity is up to 31%, pregnant women with no risk factors have a low prevalence of preterm birth of about 4%. Iams et al. examined the cervical length in such a low risk population at 24 and 28 weeks of gestation and found an inverse correlation between the length of the cervix and risk of preterm birth in this population as well22. Taking the 10th centile as a cut-off value (25 mm), significantly more women with a value above this cut-off remained pregnant until the 37th week of gestation. Other and larger studies have confirmed the association between cervical length and the risk of preterm birth23. The predictive values were dependent on the time of measurement: evaluations at 19–24 weeks of gestation were more sensitive and specific for preterm birth than at 14–19 weeks of gestation. A cut off value of ≤15 mm corresponding to about the 3rd centile had a sensitivity of 8.2% and a specificity of 99.7% for early preterm delivery (≤32 weeks of gestation).
In summary, the recommendations for cervical length measurement as a screening test are based on the fact that sonographical assessment is a widely accepted and well-standardised method, which requires only a relatively short training period. Another advantage is its high negative predictive value, which was found in all studies including low- and high-risk women. This provides the opportunity to avoid unnecessary tocolysis and reduce time of hospitalisation, or to indicate lung maturation treatment more precisely based on the cervical length. Even the outcome of preventive interventions such as cerclage seems to depend on the cervical length24.
There are, however, two drawbacks with screening. Firstly, there is still a low sensitivity of the test relating to the low prevalence of preterm deliveries in a low-risk population. Therefore cut-off values have to be set at a very low level in order to get acceptable specificity. In a Finnish study comparing different cut-off points, as the cut off point was increased, the sensitivity increased but the specificity and the positive predictive value decreased25. A combined assessment of cervical length and dilatation of the internal os increased the sensitivity, but only up to 29%. Compared with a high-risk population, with a prevalence for preterm birth >30%, a sensitivity of 79%, and a positive predictive value of 67% is acceptable, but a sensitivity of 19–39% and a positive predictive value of 6–20% is not acceptable in low-risk populations with a prevalence for preterm birth <4%. Therefore the low sensitivity and the low positive predictive value limit the strength of cervical length measurement as a screening test alone. Our recommendations for cervical length measurement are summarised in Table 1.
Table 1. Recommendations for cervical length measurement.
• Established as predictor for preterm delivery especially preterm birth <32 weeks of gestation
• Equipment and exercise needed
• Early screening (second trimester) in selected high-risk groups
• Repetition of the length measurement as cervical length is a temporary marker
• Clinical usefulness increase in conjunction with other predictors (fetal fibronectin, E2, etc.)
• Parallel/stepwise multiple marker
• High negative predictive value reduce unnecessary interventions
• Appropriate timing and selection of interventions (tocolysis, cerclage, corticosteroids)
To prevent preterm deliveries in the general population, further interest should be focused on scoring systems combining ultrasound with biochemical, endocrinological and perhaps molecular cell methods such as the measurement of fetal DNA in maternal blood.