- Top of page
- MATERIALS and METHODS
Objective To examine the relationship between ultrasound-determined cervical status and pregnancy outcome in women ‘at-risk’ of spontaneous preterm delivery.
Design A prospective cohort study of 120 pregnant women considered to be ‘at-risk’ of spontaneous preterm delivery by their clinician. Transvaginal ultrasound of the cervix was used to assess overall cervical length, closed endocervical canal length, diameter and internal os dilatation in the second trimester. The main outcome measure was occurrence of spontaneous preterm birth (< 34 and < 37 weeks of gestation).
Results The overall preterm delivery rate (< 37 weeks gestation) in these women was 35% (n = 42) with 20% (n = 24) delivering < 34 weeks gestation. Of the 71 women with a normal cervix, 8 (11%) delivered < 34 weeks, whereas of the 49 women with an abnormal cervix, 16 (33%) delivered < 34 weeks (RR 2.90; 95% CI 1.35–6.24). Using linear regression, closed endocervical canal length of < 21 mm before 20 weeks is associated with delivery < 34 weeks in 95% of women, and with delivery < 37 weeks in 95% of women if the canal length is < 33 mm. Logistic regression showed closed endocervical canal length to be the only significant factor in those women who delivered < 34 weeks after controlling for possible confounders.
Conclusions A strong relationship is demonstrated between cervical status and pregnancy outcome, particularly the cervical findings before 20 and 24 weeks of gestation. The length of the closed portion of the endocervical canal is the best predictor. A beneficial effect of this approach to ‘at-risk’ women is the reduction in unnecessary interventions in those with normal cervical findings.
- Top of page
- MATERIALS and METHODS
Spontaneous preterm delivery continues to be a major problem in modern obstetrics. Preterm birth accounts for a large proportion of perinatal morbidity and mortality and remains costly in both financial and emotional terms. The incidence of preterm delivery is unchanged in Australia at 6–7% 1 and over half of these deliveries may be the result of spontaneous preterm labour. Effective prediction of spontaneous preterm delivery continues to be elusive.
In pregnancies which continue to term, cervical parameters assessed by transvaginal ultrasound remain fairly constant during the second trimester 2. In contrast, it has been demonstrated that the risk of preterm delivery is associated with cervical length reduction at 24 and 28 weeks 3.
The purpose of this study was to explore the role of cervical assessment by transvaginal ultrasound in the prediction of preterm delivery, with the emphasis on delivery before < 34 weeks of gestation. The women targeted were those considered ‘at risk’ of preterm delivery by their clinician based on their clinical history. This work also aimed to determine if the gestational age at which an abnormal cervical parameter became evident was significant in relation to pregnancy outcome.
MATERIALS and METHODS
- Top of page
- MATERIALS and METHODS
This prospective cohort study involved women with a singleton pregnancy who were referred for cervical assessment as part of their clinical management because their clinician considered them ‘at-risk’ of spontaneous preterm delivery based on their clinical history. Women who subsequently had an elective delivery before 37 weeks gestation were excluded.
Transvaginal ultrasound of the cervix was performed as previously reported 2 ensuring an empty bladder and no transducer pressure on the cervix. Scans were performed using either an Ultramark 4 or HDI 3000 system (ATL, Seattle, USA). The transvaginal scanning procedure was explained to each woman and informed consent for the study obtained. Parameters measured were overall cervical length along the endocervical canal from external to internal os and cervical diameter at midpoint of the cervix. When internal os dilatation was evident, this was measured plus the length of the closed portion of the endocervical canal. Initial cervical assessment with transvaginal ultrasound was made between 9.42 and 29.71 weeks of gestation. Further assessments varied from weekly to monthly according to the needs of the individual woman with consideration of such factors as cervical findings and history. The majority of examinations ceased by 32 weeks with the exception of a number of women with a cervical suture in situ. All results were reported to the responsible clinician and any interventions were instituted at their discretion. Institutional approval for this work was obtained from the Ethics Committee.
Data were analyzed using the statistical package SPSS for Windows Version 6 (SPSS Inc., Chicago, IL, USA). Comparisons for categorical data used χ2 and Fisher’s exact test when expected cell frequency was < 5. Relative risk (RR) and 95% confidence intervals (95% CI) are reported. The Student’s t-test was used for continuous data which were normally distributed and the Mann–Whitney U-test for nonparametric data. Multiple linear regression was used to examine the relationships of the gestational age at delivery (dependent variable) and the cervical parameters. Logistic regression was used to control for possible confounding variables.
- Top of page
- MATERIALS and METHODS
A total of 562 studies was performed on 120 women, an average of 4.7 scans per woman with a range of 1–17. Most of the women reported a past history of reproductive problems such as a previous preterm delivery prior to 34 weeks gestation (62), a second trimester miscarriage between 14 and 20 weeks gestation (16), three or more first trimester miscarriages and/or pregnancy terminations (9). A further number were women with past major cervical surgery such as surgical laser or cone biopsy (10), uterine anomalies (13) or miscellaneous problems such as second trimester bleeding (10). The women were categorized by reason for referral, though 43% (51) had two or more of these ‘risk’ factors.
An abnormal cervix was defined as a closed endocervical canal length of < 30 mm plus dilatation of the internal os of > 5 mm. Seventy-one women were found to have a normal cervix throughout their study period. Forty-nine women were found to have abnormal cervical parameters on their initial scan or developed an abnormal cervix as the pregnancy progressed. There were no differences between the groups in parity (25% nulliparous in both groups) or race (98% Caucasian). Overall, gestational age at delivery was 36.4 ± 5.1 weeks (mean ± SD) with 42 (35%) women delivered preterm after spontaneous labour, and 24 (20%) delivered before 34 weeks following spontaneous labour. In the women with normal cervical findings, gestational age at delivery was 37.8 ± 3.7 weeks whereas those women with abnormal cervical parameters delivered at 34.4 ± 6.0 weeks (Mann–Whitney U-test, P < 0.001). The preterm delivery rate of 49% (24/49) was higher in the abnormal cervix group when compared to 25% (18/71) in the normal cervix group (χ2 7.11; P < 0.008; RR 1.93; 95% CI 1.18–3.16). The incidence of delivery before 34 weeks gestation was 33% (16) in the abnormal cervix group in contrast to 11% (8) in the normal cervix group (χ2 8.29; P < 0.004; RR 2.90; 95% CI 1.35–6.24).
The timing of cervical change was examined to determine if cervical status at 20 and 24 weeks gestation predicted outcome. Eighty-one women had at least one study before 20 weeks of gestation and those whose cervix was classified as abnormal had a lower mean gestational age at delivery than those women with a normal cervix (30.8 ± 7.5 weeks vs. 37.4 3.9 weeks; Mann–Whitney U-test P < 0.001). This difference was also evident in the 100 women whose cervix had been assessed by 24 weeks of gestation ( Table 1).
Table 1. Relationship of cervical status as determined before 20 and 24 weeks’ gestation to gestation at delivery
| || No.||Gestation at delivery (weeks)|| P value|| Delivery < 34 weeks|| P value|
|Women studied by 20 weeks||81|| || || || |
|Normal by 20 weeks||65||37.4 ± 3.9||< 0.001*||9 (14%)||< 0.001**(RR 4.06; 95% CI 1.93 to 8.55)|
|Abnormal by 20 weeks||16||30.8 ± 7.5|| ||9 (56%)|| |
|Women studied by 24 weeks||100|| || || || |
|Normal by 24 weeks||71||37.9 ± 3.7||< 0.001*||9 (13%)||< 0.001 (χ2 12.4; RR 3.54; 95% CI 1.70 to 7.35)|
|Abnormal by 24 weeks||29||32.6 ± 6.9|| ||13 (45%)|| |
To ascertain the predictive abilities of these findings, sensitivity, specificity, positive and negative predictive values were calculated in relation to delivery before 34 weeks ( Table 2). Sensitivity improved with increasing gestational age at which the abnormality was detected, whereas both specificity and positive predictive values decreased.
Table 2. Sensitivity, specificity, positive and negative predictive values for delivery < 34 weeks’ gestation based on abnormal cervical findings overall and before 20 and 24 weeks’ gestation
| ||Cervix abnormal by 20 weeks (%)||Cervix abnormal by 24 weeks (%)||Abnormal cervix overall (%)|
|Positive predictive value||56||45||33|
|Negative predictive value||86||87||89|
The interventions utilized as a result of the ultrasound scans were various combinations of cervical suture (22), antibiotics (45), bed rest (35) and tocolytics (14). In total, only 18% (13/71) of the women with normal cervical measurements received an intervention compared to 84% (41/49) in the women with abnormal measurements (χ2 50.0; P < 0.001; RR 6.26; 95% CI 3.22–12.20). However, examination of gestational age at delivery in relation to interventions showed no difference in either the normal cervix group—intervention 36.7 ± 5.2 weeks; no intervention 38.1 ± 3.3 weeks (Mann–Whitney U-test, P = 0.57)—or the abnormal cervix group—intervention 34.4 ± 6.0 weeks; no intervention 34.3 ± 6.3 weeks (Mann–Whitney U-test, P = 0.88).
The various cervical parameters measured were examined individually to determine differences between those women who delivered before or after 34 weeks. As shown in Table 3, mean closed endocervical canal length was very significantly different whereas mean differences in internal os dilatation only just achieved statistical significance. The other parameters of overall cervical length and diameter were also assessed, but neither showed any difference between the two groups.
Table 3. The four cervical parameters measured (mean ± SD), grouped by delivery before or after 34 weeks’ gestation
| ||Delivered < 34 weeks||Delivered > 34 weeks||Significance|
|Overall cervical length (mm)||33.3 ± 6.2||35.0 ± 6.3||ns (P = 0.28)|
|Diameter (mm)||31.3 ± 5.7||31.9 ± 4.7||ns (P = 0.65)|
|Closed endocervical canal (mm)||21.2 ± 13.3||30.9 ± 10.3||P < 0.001 (t 3.89; df 118)|
|Internal os dilatation (mm)||13.7 ± 13.8||8.0 ± 9.9||P = 0.028*|
With gestational age at delivery as the dependent variable in multiple linear regression analysis, closed endocervical canal length was the only factor found to be of value. This parameter, prior to 20 weeks, had a coefficient of determination (R2) of 30.7% (F 34.99; d.f. 1,79 ; P < 0.001), and before 24 weeks, a coefficient of determination of 27.4% (F 36.99; d.f. 1,98 ; P < 0.001). Using this analysis, closed endocervical canal length of < 21 mm before 20 weeks of gestation is associated with delivery before 34 weeks in 95% of cases, and with delivery before 37 weeks in 95% of women where canal length is < 33 mm.
Using a logistic regression model with delivery before 34 weeks as the dependent variable, only the closed endocervical canal length was significant (P < 0.007; estimated odds 0.90; 95% CI 0.83–0.97) when controlling for possible confounding variables ( Table 4).
Table 4. Results of logistic regression analysis showing canal length to be the only significant factor with delivery < 34 weeks as the dependent variable
|Maternal age||0.991||0.319||0.952||0.86 to 1.05|
|History of:|| || || || |
| ≥ 3 × 1st trimester losses||0.231||0.631||0.701||0.16 to 2.98|
| 14–20 week pregnancy loss||0.956||0.328||1.919||0.52 to 7.09|
| Previous < 34 week delivery||2.603||0.107||3.126||0.78 to 12.48|
| Cervical surgery||0.036||0.851||0.789||0.07 to 9.28|
| Uterine anomaly||1.747||0.186||0.199||0.02 to 2.18|
|Intervention||0.709||0.400||0.554||0.14 to 2.19|
|Internal os dilatation||0.762||0.383||0.971||0.91 to 1.04|
|Closed endocervical canal length||7.346||0.007||0.900||0.83 to 0.97|
- Top of page
- MATERIALS and METHODS
Prediction and prevention of preterm delivery continue to be a vexatious problem in modern obstetric care. A number of methods have been investigated to determine predictive ability with varying degrees of success. Scoring systems based primarily on the woman’s history 4 have been developed but generally reported to have limited value 5. In this study, the woman’s history was the factor used to indicate ‘at-risk’. As the overall preterm delivery rate of 35% in this group was markedly higher than in the general population, this indicates that this is an appropriate starting point in determining which women may deliver preterm. In an attempt to narrow the focus, this study was designed to examine the predictive value of cervical parameters.
Cervical assessment has been used for many years to predict preterm delivery, initially using digital examination, which had limited success due to the inadequate information available. Vaginal speculum examination of the cervix is restricted to viewing the external os.
Abdominal ultrasound cervical assessment, though better than digital examination, also has a number of technical difficulties due to the full bladder causing compression and distortion of the pregnant cervix, together with problems in visualization6,7. Since transvaginal scanning was introduced and used for cervical examination 8, views of the cervix can be easily obtained and parameters measured 9. Some variations in cervical length measurements occur between different studies and these are probably due to measurement technique.
A number of studies have examined cervical status using transvaginal ultrasound in relation to preterm labor and delivery. They have all found associations between cervical shortening and preterm delivery 10–12 . A large population study of more than 3000 women found an increased relative risk of preterm delivery if cervical shortening was detected at either 24 or 28 weeks’ gestation 3. The current study comprised only ‘at-risk’ women and clearly demonstrates a relationship between an abnormal cervix and preterm delivery, particularly before 34 weeks. Further, it can be seen that the detection of an abnormal cervix by 20 weeks of gestation will detect about half of all those patients in this ‘at-risk’ group who will deliver before 34 weeks. Ideally, these characteristics should have been assessed in a blinded study to reduce the influence of interventions, as it is inevitable that referring clinicians would be more likely to use various treatments if cervical changes were detected. Such a study design was thought to be impractical as many referring clinicians would not have complied with a protocol in which they were blinded to ultrasound results.
Much information can be gained from vaginal ultrasound of the cervix and a variety of parameters can be measured. The question of which parameter is most predictive is important if this method is to be advocated for monitoring ‘at-risk’ pregnancies. In this study, the most obvious difference between women who deliver early compared to those who deliver at or near term is seen in the closed endocervical canal length. After logistic regression analysis this parameter is the only one which remains statistically significant. We would propose that endocervical canal length is the best indicator of increased risk of preterm delivery.
This work demonstrates a strong relationship between cervical status and pregnancy outcome, particularly the cervical findings before 20 and 24 weeks of gestation. Therefore, we recommend that careful surveillance is needed in women with a endocervical canal length of < 21 mm by 20 weeks of gestation. The regression analysis also shows that of the women whose endocervical canal length is > 21 mm by 20 weeks’ gestation, < 5% will deliver before 34 weeks’ gestation. Cervical ultrasound can therefore be used at these earlier gestations to target more appropriately those women who may benefit from interventions designed to reduce the risk of preterm delivery. In this study, the preterm delivery rate is still very high despite interventions, especially in those women with abnormal cervical findings. One can only speculate if the interventions did prolong the pregnancies in these women and whether the differences in gestational age at delivery would have been more marked had the study been blinded. A beneficial effect of this approach to ‘at-risk’ women is the reduction in unnecessary interventions in those with normal cervical findings, particularly in relation to cervical suture. In a large randomized trial of cervical cerclage, it was found that this intervention was likely to be of benefit in only a small subset (1 in 25) of women thought to be ‘at-risk’ because of previous history 13. Through the use of transvaginal ultrasound, it may be possible to target this intervention in the group of women who have cervical change before 20 weeks. A randomized trial design of transvaginal ultrasound followed by specific interventions would be needed to test this hypothesis.
This study has concentrated on a group of women ‘at-risk’ of preterm delivery based on a number of risk factors such as previous delivery before 34 weeks. The risk factors chosen were those in which it was thought that there was a reasonably high chance of cervical change being part of the mechanism of preterm labour. It is becoming increasingly obvious that the pathogenesis of preterm labor and delivery is multifactorial 14 and monitoring ‘at-risk’ women with transvaginal ultrasound of the cervix can only be effective in a subgroup of these women. However, as > 40% of women in this study demonstrated early cervical change, this would suggest that the potential for intervention is high. The challenge is to identify the most effective intervention for this group.