Identification of risk factors for incident cervical insufficiency in nulliparous and parous women: a population-based case-control study

Cervical insufficiency refers to a condition in which patients present with a dilated cervix in the mid trimester of pregnancy, potentially with protruding membranes, in the absence of uterine contractions or vaginal bleeding [1]. It is one of the most important causes of late miscarriage and spontaneous preterm birth (sPTB, < 37 weeks’ gestation) [2]. Since women with a history of these conditions are more likely to be examined for cervical length in a subsequent pregnancy, cervical insufficiency is more easily detected in women with previous late miscarriages and in parous women. While “a painless cervical dilation, leading to recurrent second trimester births in the absence of other etiologies in otherwise normal pregnancy” is an accepted definition [3], there is a lack of consensus in published studies regarding both the definition and diagnosis [1, 3, 4] of cervical insufficiency, presenting a challenge for decisions regarding treatment. Cervical cerclage is accepted as an effective intervention for pregnancies where a history of previous pregnancy loss or extreme premature delivery indicates a risk of cervical insufficiency [1]. Such intervention requires women at risk to be identified before the cervix becomes too short or dilated. Several factors that could predict the risk for cervical insufficiency have been identified, with a history of late miscarriage or preterm birth being particularly important, and consequently included in some screening protocols [5]. Other reported predictors include a history of cervical procedures or prolonged second stage of labor [6, 7] and polycystic ovary syndrome (PCOS) [8, 9]. However, studies disagree on the importance of the factors investigated, and the definition and identification of cervical insufficiency remains elusive. Overcoming these difficulties is important in order to resolve the controversies concerning cerclage [10] and to realize the potential of this intervention and progesterone to reduce the risk of pregnancy loss or preterm birth [11].

To date, the independent associations of the various reported predictors of risk for cervical insufficiency have not been investigated in a population-based study, especially for incident cervical insufficiency (ICI), and differences between nulliparous and parous women have rarely been discussed in the literature. The objective of this study was to take advantage of a data linkage of Swedish national population health registers containing demographic information, reproductive history and surgical procedures, to identify independent correlates of ICI in nulliparous and parous women.

The comparison of the incident cases of cervical insufficiency to the random sample of unaffected pregnancies revealed some important differences in the presence and/or magnitude of associations with predictors in nulliparous and parous women. Increasing maternal age and shorter stature were associated with a reduced risk of ICI in both nulliparous and parous women, with ORs of 1.16 (1.05–1.28) and 1.05 (1.00–1.06) for each additional 5 years of age and 0.86 (0.90–0.95) and 0.95 (0.90–1.00) for a 5-cm difference in height. Other factors found to be independent positive correlates of risk in both nulliparous and parous women were region of birth, previous cervical procedures, and a history of miscarriage. The magnitude of the association with cervical procedures was similar in the two groups, whereas previous miscarriage had a much stronger association for nulliparous women, which was especially noted in multi-fetal gestations. In contrast, for parous women, where further adjustment was made for factors related to previous deliveries, the association of ICI with multi-fetal gestation was much weaker and the independent association with history of previous miscarriages was similar in singleton and multifetal pregnancies. For parous women, a prolonged second stage of labor in a previous delivery and previous delivery of a very large infant were also both inversely associated with risk of ICI in the current pregnancy.

We hypothesized that the negative association of ICI with taller stature may be related to a longer cervix. Weak evidence of an association between maternal height and cervical length was reported from a small study of 146 asymptomatic Turkish women [14], but a larger screening study of 5092 pregnant women in the USA found no such association [15]. BMI, which has previously been reported as a risk factor for sPTB [16, 17], has a weak independent association with ICI only for parous women in our study. There is no clear evidence in the literature regarding the importance of maternal age for the risk of cervical insufficiency. A weak negative association has been reported for older women [18], but the estimate was from a model that assumed common ORs for all other factors for nulliparous and parous women, whereas our study provides strong evidence against this assumption as we found many differences in the presence or magnitude of the risk factors for ICI in the two groups.

Multi-fetal pregnancies are known to be at higher risk of spontaneous preterm birth, and there have been reports of a lack of response to treatment, compared to singleton pregnancies [19, 20]. Recent guidelines [5] report that available data still indicate a risk of preterm birth following cerclage for short cervix (< 25mm) in twin pregnancies, but limited data providing evidence of an advantage for cervix < 15mm. Evidence from a recent clinical trial [18] supports the use of cerclage in twin pregnancies with a dilated cervix. However, further studies of multi-fetal pregnancies are required, as there is a lack of consensus regarding screening protocols or the risks and benefits of cerclage [21‐23] and it is notable that current RCOG guidelines [24] present key recommendations only for singleton pregnancies. This could be partly due to differences in parity as we found multi-fetal pregnancy to be a strong predictor of risk for nulliparous but not parous women. This might be explained by the myometrium and cervical tissues having been expanded and thus being more adaptable in a subsequent pregnancy. Evidence that would support this hypothesis comes from a meta-analysis of recurrent preterm birth [25] that reported a 10% rate of preterm deliveries for singleton pregnancies following a preterm twin delivery and only 1.3% following a term twin delivery. Another important difference we found with respect to parity concerns the importance of miscarriage history in singleton and multi-fetal gestation: in nulliparous women a history of 2 or more miscarriages was associated with a 24-fold higher risk of ICI in singleton pregnancies and 67-fold higher risk in multifetal pregnancies, whereas for parous women, the elevated risk for miscarriage history was similar in singleton and multifetal pregnancies. Although the parous women cannot be compared directly with the nulliparous women, as they provide more information (i.e., from their previous deliveries) and use a different model, the findings have value in identifying the important, and different, contribution of these risk factors for ICI in the two groups.

For parous women, we found several aspects of obstetric history to be strongly associated with ICI: a history of sPTB or PPROM were risk factors, in contrast to prolonged second stage of labor that showed an inverse association. As previously noted, a history of sPTB is widely recognized as a risk factor [5] and a small study reported a much higher rate of cervical insufficiency in the subsequent pregnancy of 102 women with PPROM than in 316 without PPROM: 14.7% vs. 1.0%; adjusted OR 3.8, 95% CI 1.2–11.6 [26]. The dose-response that we observed for the “weeks to term” of the delivery prior to the index is consistent with guidelines that recognize early premature delivery in risk evaluation protocols and consider cerclage for women with a history of extreme premature delivery [5]. Although previous prolonged second stage of labor was reported as a risk factor as early as 2006 [6], there is no consistent conclusion regarding the magnitude of the association. A prolonged second stage of labor has previously been investigated for its association with sPTB in the subsequent pregnancy, but these studies either found no association, or concluded that the risk is mediated by cesarean section following prolonged second stage of labor [7, 27, 28]. Prolonged second stage of labor is more commonly diagnosed in women with term birth and thus may rather be a sign of a strong cervix. In our study, a history of prolonged second stage of labor was an independent predictor of a lower risk of ICI. Although our data were insufficient for investigating cesarean section after prolonged second stage of labor (only 16 such cases), it is nonetheless interesting that we did not find any association between previous cesarean section and ICI.

A major strength of our study is that the large population-based data enabled the investigation of ICI in an un-selected national cohort of nulliparous and parous women separately, although the sample sizes were insufficient for further exploration of subgroups with specific reproductive or medical history. A further strength of the study is that pregnancy information was prospectively obtained at attendance for antenatal care and hospitalization information from the NPR, eliminating the potential for recall bias. Further, due to the rich data, we were able to study a broad spectrum of potential predictors for both nulliparous and parous women. Another strength of the study is that we focused on incident cervical insufficiency, while much of the literature is unclear regarding whether incident and/or recurrent cases are being considered [5].

Limitations of our study include the potential for misclassification of the cervical insufficiency diagnosis in both nulliparous and parous women, due to the partly unclear definition of the diagnosis and the inherent difficulty of discriminating between cervical insufficiency and late miscarriage, PPROM, or sPTB, which could occur in a continuum. Given the unclear definition of the diagnosis, the pregnancies captured by explicit ICD codes in the registers may nevertheless be more representative of true cases of cervical insufficiency, and the risk factors identified thus more specific. The diagnostic criteria commonly used are (i) cervical length less than 25 mm according to transvaginal ultrasound measurement before 24 gestational weeks, in a pregnant woman with no uterine contractions and at least one previous preterm birth or late miscarriage, or (ii) cervical length of 20mm or less in a woman with no history of preterm birth or late miscarriage. Since Sweden does not offer universal cervical length screening, the associations with previous miscarriages and previous sPTB in parous women may reflect diagnostic practice and/or detection bias from additional surveillance due to the already known correlation with cervical insufficiency [29]. However, in addition to considering history and number of sPTB, we found a positive association with the number of weeks’ preterm of the most recent delivery, and such dose-response seems unlikely to be the sole result of diagnostic practice or surveillance. These findings suggest that the full history of preterm deliveries prior to various cut-off times for gestational age and their potential interactions with multifetal gestation (such as we observed for miscarriage history) may provide information of value for interventions. Finally, while there was some evidence of changes in the rate of cervical insufficiency diagnosis over time, adjustment for calendar year had no influence on the associations of the predictors.

The differences we observed for nulliparous and parous women can have important consequences for screening and intervention for cervical insufficiency. The much stronger association with multi-fetal gestation in nulliparous women may contribute to the lack of consensus regarding screening protocols and cerclage for multi-fetal gestation [21, 22]. In addition, the difference in the importance of miscarriage history in singleton and multi-fetal gestation for nulliparous women but not in parous women highlights readily available information that could be used to inform screening protocols. Our work identified aspects of obstetric history that had strong (independent) associations with ICI: inclusion of these in studies of independent risk factors for parous women could help resolve some of the inconsistencies in the literature to date regarding the factors that are important for risk stratification. The identification of different risk factors of importance for nulliparous and parous women has clinical and scientific implications: these two groups of women can be better informed of their risks, and with additional scientific support, screened differently. Finally, stratification by parity in the reporting of risk factors for incident cervical insufficiency would facilitate meta-analyses that can produce more timely synthesized evidence for screening and eventually inform intervention efforts in both groups of mothers.