Among the 1,107 low-risk participants included in our study, 6.05% gave birth at < 37 weeks, which is lower than that reported in previous studies [19, 20].
As shown in Table 1, the mean CL was significantly shorter in the participants who delivered preterm than in those who delivered at term, and the mean UCA was significantly greater in those who delivered preterm. These results were in agreement with the findings of previous studies conducted in unselected singleton pregnant women; the UCA was significantly wider in singleton pregnant women with spontaneous PTB [21–25]. The combination of pressure applied by the surrounding pelvic organs, more importantly from the growing uterus during pregnancy, could result in the alteration of the internal ostium (os) and cervical function [26]. Some ultrasonographic parameters, such as the CL and UCA, have been used to assess cervical structure. In this regard, it can be said that an obtuse cervical angle is associated with a straight and direct force from the pregnant uterus, while an acute cervical angle is associated with a less direct force, maintaining the integrity of the cervix. In other words, the hypothesis is that the cervical angle acts as a barrier that affects the progression of labour when the angle is acute [12, 27, 28].
Our study revealed that the optimal UCA and CL cut-off values for predicting PTB at < 37 gestational weeks in low-risk participants were 98.96 degrees and 33.8 mm, respectively, which were similar to the results reported by Minoo Movahedi et al. [29], who recruited patients using the same criteria. In their study, the UCA and CL thresholds for PTB prediction were found to be 106° and 33 mm, respectively. These results were also consistent with the recent study by M Zhang et al. [30] on 275 singleton pregnancies in early and mid-pregnancy, that showed the optimal threshold of UCA and CL to predict PTB < 37 weeks gestation was 96° and 33.8 mm, respectively. Additionally, Sawaddisan et al. [16] reported that a UCA ≥ 110° measured later than 19.5 weeks of gestation corresponded to an 83.3% sensitivity for predicting spontaneous PTB. A study conducted by Paul Guerby et al. [31] revealed that a mid-trimester CL < 30 mm could detect 35% of all spontaneous PTBs before 35 weeks gestation at a false-positive rate of 5% in low-risk nulliparous women. It is important to note that their study included only women with no previous pregnancies, and ultrasound was conducted between 20 and 24 weeks gestation.
Our findings also demonstrated that at the cut-off points, a UCA ≥ 98.96°, a CL ≤ 33.8 mm, and the combination of a UCA ≥ 98.96° and a CL ≤ 33.8 mm were significantly associated with PTB at < 37 weeks gestation, with ORs (95% CIs) of 35.56 (14.88–84.96), 13.82 (6.52–29.29), and 24.28 (13.90-42.41), respectively.
Nonetheless, our study of low-risk pregnant women showed that CL alone did not have a significant impact on predicting PTB, unlike in the general population. A CL of 33.88 mm or less was a poor predictor for PTB, with a low sensitivity of 25%, a specificity of 66%, and an area under the curve (AUC) of only 0.240 (95% CI: 0.176–0.304). Jeanine van der Ven et al. [32] also reported that mid-trimester CL measurements had limited value for predicting spontaneous PTB in a large population of low-risk singleton pregnant women from 16+ 0-21+ 6 weeks gestation, with a poor AUC of 0.56 (95% CI 0.52–0.6) for PTB before 37 weeks gestation. This finding was consistent with a previous study by Iams et al. [33], which revealed that the positive predictive value of CL was poor (ranging from 6 to 44%), and the sensitivity was only 47% in a low-risk population.
In contrast, the UCA could be used as a predictive tool to identify women at risk of PTB in this low-risk population. The UCA alone had greater sensitivity and specificity than the CL, with a UCA of 98.96 degrees or more having a sensitivity and specificity of 91% and 75%, respectively. In addition, the AUC of the UCA was 0.899, which performed well in identifying women at risk of PTB in our study population. This finding contradicts the results reported by Sawaddisan et al. [16], who reported that the UCA in the second trimester is not a good predictor of PTB in low-risk pregnant women. The discrepancy in the findings may arise from the smaller sample size examined in their study compared to that in our study. In a recent systematic review, Goldstein et al. [21] analysed 15 studies that provided data on the test characteristics of the UCA and CL for predicting PTB. The review revealed that for the general population, the UCA alone had a significantly greater sensitivity (0.70; 95% CI: 0.66–0.73) than CL (0.46; 95% CI: 0.42–0.49). However, the UCA also had a significantly lower specificity (0.67; 95% CI: 0.66–0.68) than CL (0.90; 95% CI: 0.89–0.91) and a lower AUC (0.77) than CL (0.82).
Remarkably, our findings showed that while CL alone was not a reliable predictor of preterm birth in this low-risk population, combining CL with the UCA improved the predictive value compared to the UCA alone. The use of both a UCA ≥ 98.96° and a CL ≤ 33.8 mm increased the positive predictive value (PPV), positive likelihood ratio (LR+), and accuracy (ACC) for the prediction of PTB at < 37 weeks gestation to 36%, 8.87, and 91%, respectively, which agreed with the results reported in previous studies [14, 34]. Measuring the UCA at the same time as the CL for screening PTB according to recommendations (at a gestational age of 16–24 weeks) is convenient in clinical practice and may increase the effectiveness of PTB prediction when combining these two parameters.
The main strength of this study is that it is the first study conducted among low-risk singleton pregnant women in Vietnam, with a large sample size that demonstrated the ability of the UCA and CL to predict preterm birth. However, even when multivariate logistic regression was used to assess the associations among the UCA, CL, and preterm birth, we cannot deny that other factors may have influenced the outcome of preterm birth, such as smoking status and type of conception. Further multicentre studies are required to evaluate the consistency of these results and establish definitive conclusions.