- Top of page
- The CERVO group
Cervical weakness and ascending infection have long been considered important causes of preterm birth. There may even be a synergy between these two factors: a weak cervix could facilitate infection and infection could cause contractions that weaken the cervix. The cervical mucus plug may prevent ascending infection by acting as an effective mechanical and immunological barrier, owing to its rich content of antimicrobial peptides, immunoglobulins and phagocytes.[2-4] A shortened cervix with a concomitant loss of the protective mucus plug could lead to subclinical ascending infection, amniotic fluid sludge and, consequently, preterm delivery.
Cerclage increases the resistance of a weakened cervix. The two commonly performed cervical cerclage procedures are both variations of the techniques described by Shirodkar and McDonald.[5, 6] A cervical cerclage may be inserted for prophylactic reasons before pregnancy or during the first trimester (primary cerclage). It may also be inserted later for therapeutic reasons when cervical shortening is detected without or with the membranes being exposed to the vagina (secondary and tertiary cerclage, respectively).
Cervical occlusion could theoretically be used for the treatment of cervical insufficiency and to protect the cervix against infection. A cervical cerclage could be used to increase the cervical resistance, and occlusion of the external os could be performed to retain the cervical mucus plug. Total cervical occlusion, described by Saling et al. in 1984, significantly increased the number of live births in women at risk of preterm birth.[9, 10] The procedure involved the surgical removal of the epithelial lining of the cervical canal in a similar fashion to conisation. A simpler, less traumatic technique has been developed, whereby the external cervical os is occluded using a continuous suture at the time that the cerclage is applied. In comparison with a historical cohort, we found that cerclage in which suture occlusion was applied was associated with a higher success rate (take-home baby rate). Two small descriptive studies of 53 and 26 women with cerclage and the less traumatic cervical occlusion recently showed take-home baby rates of 93 and 96%, respectively.[11, 12] As stated in the first of these reports, this take-home baby rate was superior to that reported in historical cohorts of abdominal cerclage, which gave an overall success rate of only 84%.
The aim of the present study was to evaluate the effect of cervical occlusion versus no cervical occlusion in women with prophylactic and therapeutic cerclages on the take-home baby rate, gestational age at delivery and the number of days spent by the neonate in the neonatal intensive care unit (NICU).
- Top of page
- The CERVO group
This was a hospital based, multicentre, stratified (prophylactic and therapeutic trial), randomised controlled trial with an allocation ratio of 1:1.
Women were allocated to the prophylactic trial on one of the following indications: (1) a history suggestive of cervical insufficiency (i.e. a prior delivery before 28 weeks of gestation); (2) a congenitally short cervix (secondary to maternal administration of diethylstilbestrol, DES) or traumatic/surgical damage (e.g. conisation); or (3) a successful or failed cerclage applied in a previous pregnancy. Women were placed in the therapeutic trial if they presented with a short cervix (less than 25 mm) with or without the membranes being exposed to the vagina. Furthermore, the gestational age at inclusion was between 12 and 27 completed weeks of gestation. Gestational age was estimated by ultrasound at 22 weeks of gestation or before, or by a reliable date for the last menstrual period. Women who were recruited had to be able to read and understand the national language and had to be 18 years or older (legally competent in all the participating countries). Consent was obtained according to the specifications of the local research ethics committee. Women were excluded if they had evidence of cervical infection, if complications were present in their current pregnancy, if they had a multiple pregnancy or if they had a previous history of abruptio placentae.
The study was hospital based and took place in 18 tertiary centres in nine countries (Australia, Denmark, India, Saudi Arabia, South Africa, Spain, Sweden, Switzerland and the UK). Eligible women were recruited from August 2006 to August 2011. A trial entry form was completed at randomisation and sent to the trial secretary. After delivery, a follow-up form was completed after the mother and child had been discharged, or 6 weeks after delivery if the mother or the child was still hospitalised. The follow-up form was sent to the trial secretary.
The cervical cerclage technique was left to the discretion of the treating doctor (McDonald or Shirodkar), and was performed with the materials used routinely in the participating departments. In addition, women randomised to cervical occlusion had the anterior cervical lip stitched to the posterior cervical lip to close the external os. The sutures were interrupted or continuous, placed approximately 1 cm deep on each lip and 0.5 cm apart, with a nylon 2–0/3–0 suture.
The prespecified primary end point was the take-home baby rate (the proportion of infants discharged alive from hospital from all pregnancies). The prespecified secondary end points were gestational age at delivery and preterm delivery before 34 weeks of gestation. The prespecified tertiary end point was the number of days the neonate spent in the NICU.
End points that were not prespecified were: preterm delivery before 28 weeks of gestation and admission to the NICU. Reports had shown a beneficial effect of cerclage in women with singleton gestation, previous preterm birth and a short cervix.[13, 14] All end points were therefore assessed in three exploratory strata: (1) secondary cerclage applied to a short cervix of less than 25 mm without membranes being exposed to the vagina, and with one prior birth between 16 and 34 weeks of gestation or (2) with two prior births between 16 and 34 weeks of gestation; and (3) tertiary cerclage applied to a short cervix of less than 25 mm with membranes exposed to the vagina.
With an expected extremely preterm/miscarriage rate with cerclage of 15% (R. Farquharson, personal communication) and an anticipated rate of 7.5% with cerclage and cervical occlusion, 604 women were required for the study to obtain 80% power and a significance level of 5%. We aimed for a total sample size of 650 women to take into account the risk of dropouts. We expected the duration of the study to be 2.5 years.
In accordance with prespecified guidelines in the protocol, the Data Monitoring and Safety Committee (DMC) recommended that the trial should be stopped in August 2011, because of an increasingly slow recruitment of women and an interim analysis blinded for allocation that showed no effect. The DMC was independent of the trial organisers.
The Perinatal Epidemiology Research Unit, Aarhus University Hospital, Denmark, provided a centralised telephone randomisation service with a computer system that was inaccessible to the investigators during enrolment and follow-up. The computer system generated a random allocation sequence and assigned the women after the doctor had entered all details by touch-tone buttons. Randomisation was stratified by centre and trial arm (prophylactic or therapeutic trial) in randomly varying block sizes of two, four or six. In the prophylactic trial, the women were further stratified into three strata: (1) cervical cerclage, with and without occlusion; (2) abdominal cerclage, with and without occlusion; and (3) abdominal cerclage versus cervical cerclage with occlusion. However, all women had cervical cerclage. Thus, none fell into the two last strata. Only 11 women entered the observational arm of the study, which was therefore not subjected to further analysis. The women and the doctors were aware of the allocated intervention at enrolment and follow-up, but the members of the DMC were blinded for the allocation during the interim analysis. The data analyst of the final analysis was not blinded.
Data were analysed using the statistical software package from stata/se 12.1 (StataCorp LP, College Station, TX, USA). Histograms and Q–Q plots were used to check the assumptions of normality. For binary outcome variables, the RR with 95% CI was calculated as a measure of association for the two intervention groups. For non-normally distributed continuous variables, a Wilcoxon rank sum test was used for comparison, and median differences with 95% CIs were obtained using the Hodges–Lehmann method. All outcomes were assessed for the three exploratory strata described above.
- Top of page
- The CERVO group
A total of 309 women were stratified into the prophylactic trial (n = 213) and the therapeutic trial (n = 96). They were randomised to cervical cerclage with or without occlusion (Figures 1 and 2). Twenty-five of 309 women (8%) were excluded in the final analysis. In total 284 women were analysed in the final intention-to-treat analysis. The intention-to-treat analyses were performed in accordance with the interventions to which the women had been allocated. Only nine of 284 women (3%) did not receive the allocated treatment.
Figure 1. Flow diagram for the prophylactic trial. N/A, not available. a Three patients lost record of assignment in prophylactic trial area.
Download figure to PowerPoint
As seen from Table 1, no differences in baseline characteristics were found for the prophylactic and the therapeutic trials. The prophylactic trial consisted of women whose mean age was 32 years, of which more than 95% were multigravida, around 40% had their last pregnancy end at between 16 + 0 and 24 + 6 weeks, approximately 75% had no prior cerclage and their mean cervical length was 30 mm. The therapeutic trial consisted of women whose mean age was 32 years, of which around 90% were multigravida, 40–46% had their last pregnancy end at between 16 + 0 and 24 + 6 weeks, more than 95% had no prior cerclage and their mean cervical length was 18 mm. There was no difference in terms of baseline characteristics for the three exploratory strata: secondary cerclage with one (n = 50) or two (n = 12) prior births between 16 and 34 weeks of gestation, and tertiary cerclage (n = 20).
Table 1. Baseline characteristics
| ||– occlusion ||+ occlusion|
| Prophylactic trial, n (%) || ( n = 97) || ( n = 100) |
|Maternal age (years), mean (SD)||31.9 (6.0)||31.5 (5.4)|
|Primigravida||3 (3)||4 (4)|
|Multigravida||94 (97)||96 (96)|
|Gestational age at entry (days), median (IQR)||100 (91 to 107)||98 (92 to 107)|
|Gestational age of last birth after 16+ weeks:|
|No prior births after 16 + weeks||4||5|
|16 + 0 to 24 + 6 weeks||36 (40)||38 (42)|
|25 + 0 to 28 + 6 weeks||12 (13)||5 (5)|
|29 + 0 to 35 + 6 weeks||15 (17)||20 (22)|
|36+ weeks||26 (29)||28 (31)|
|Congenital short cervix/trauma||10 (10)||11 (11)|
|No prior cerclagea||76 (78)||75 (75)|
|Successful||18 (19)||16 (16)|
|Failed||2 (2)||7 (7)|
|Both failed and successful||1 (1)||2 (2)|
|Cervical length at cerclage:|
|mm, mean (SD)||30.2 (10.7)||30.0 (9.8)|
|<15 mm||6 (7)||3 (4)|
|15–24 mm||15 (18)||17 (20)|
|25 + mm||61 (74)||65 (76)|
| Therapeutic trial, n (%) || ( n = 39) || ( n = 48) |
|Maternal age (years), mean (SD)||31.5 (5.3)||31.7 (5.4)|
|Primigravida||3 (8)||6 (13)|
|Multigravida||36 (92)||42 (88)|
|Gestational age at entry (days), median (IQR)||145 (128 to 157)||139 (127 to 154)|
|Gestational age of last birth after 16+ weeks:|
|No prior births after 16 + weeks||8||9|
|16 + 0 to 24 + 6 weeks||12 (40)||18 (46)|
|25 + 0 to 28 + 6 weeks||6 (20)||7 (18)|
|29 + 0 to 35 + 6 weeks||5 (17)||8 (21)|
|36+ weeks||7 (23)||6 (15)|
|Congenital short cervix/trauma||2 (5)||4 (8)|
|No prior cerclagea||38 (97)||46 (96)|
|Successful||1 (3)||0 (0)|
|Failed||0 (0)||2 (4)|
|Both failed and successful||0 (0)||0 (0)|
| Cervical length at cerclage:|
|mm, mean (SD)||18.2 (10.7)||17.1 (9.7)|
|<15 mm||14 (39)||20 (43)|
|15–24 mm||12 (33)||20 (43)|
|25 + mm||10 (28)||6 (13)|
The results are presented in Table 2. In the prophylactic trial, the take-home baby rate was 92% in women randomised to cerclage and occlusion, versus 90% in women randomised to routine cerclage (RR 1.03, 95% CI 0.94–1.12). In the therapeutic trial, the take-home baby rate was 81% in women randomised to cerclage and occlusion, versus 85% in women randomised to routine cerclage (RR 0.96, 95% CI 0.79–1.16). The results were similar when stratified by whether the cerclage was secondary or tertiary, and whether the women had a prior birth at 16–34 weeks of gestation.
Table 2. Results
| ||– occlusion||+ occlusion||RR or median difference (95% CI)|
| Prophylactic trial, n (%) and RR ||(n = 97)||(n = 100)|| |
|Take-home baby rate||87 (90)||92 (92)||1.03 (0.94 to 1.12)|
|Gestational age (days), median (IQR)||262 (241 to 276)||266 (251 to 273)||0 (–6 to 6)|
|Preterm <34 weeks of gestation||22 (23)||17 (17)||0.75 (0.42 to 1.32)|
|Preterm <28 weeks of gestation||10 (10)||8 (8)||0.78 (0.32 to 1.88)|
|Hospitalisation in NICU||19 (20)||21 (21)||1.07 (0.62 to 1.87)|
|Number of days in NICU, median (IQR)a||15 (2 to 42)||10 (3 to 21)||–1 (–25 to 6)|
| Therapeutic trial, n (%) and RR ||(n = 39)||(n = 48)|| |
|Take-home baby rate||33 (85)||39 (81)||0.96 (0.79 to 1.16)|
|Gestational age (days), median (IQR)||264 (222 to 274)||248 (205 to 275)||–8 (–23 to 7)|
|Preterm <34 weeks of gestation||13 (33)||17 (35)||1.06 (0.59 to 1.91)|
|Preterm <28 weeks of gestation||8 (21)||9 (19)||0.91 (0.39 to 2.15)|
|Hospitalisation in NICU||12 (31)||16 (33)||1.08 (0.58 to 2.01)|
|Number of days in NICU, median (IQR)b||32 (14 to 57)||33 (3 to 69)||–6.5 (–30 to 29)|
Figure 3 compares the two treatment groups within each of the five strata in terms of Kaplan–Meier plots of the gestational age at birth. No significant differences were found in gestational age at birth using a Wilcoxon rank sum test: prophylactic trial (P = 0.9); therapeutic trial (P = 0.3); secondary cerclage, with one (P = 0.4) and two (P = 0.4) prior births between 16 and 34 weeks of gestation; and tertiary cerclage (P = 1.0). Furthermore, no effect on delivery before 34 or 28 weeks of gestation was found.
Figure 3. Kaplan–Meier plots of gestational age at delivery for the prophylactic trial, the therapeutic trial and the three exploratory strata.
Download figure to PowerPoint
Using Wilcoxon rank sum test, we found no effect on admissions to the NICU within each of the five strata, and no difference in days in the NICU: prophylactic trial (P = 0.8); therapeutic trial (P = 0.6); secondary cerclage, with one (P = 0.5) and two (P = not available) prior births between 16 and 34 weeks of gestation; and tertiary cerclage (P = 0.5).
Logistic regression analysis after adjustment for treatment centre showed similar results for the prophylactic and the therapeutic trials in terms of take-home baby rates, delivery prior to 34 and 28 weeks, and admissions to the NICU. Cervical occlusion was associated with no harm.
- Top of page
- The CERVO group
This study is the first randomised study investigating the possible beneficial effect of cervical occlusion in women with cervical cerclage. We found that cervical occlusion along with routine cerclage had no significant effect on the take-home baby rate, gestational age, preterm delivery before 34 and 28 weeks of gestation, admissions to the NICU and days spent in the NICU in women with primary prophylactic cerclage (prophylactic trial), and also in women with secondary or tertiary therapeutic cerclage (therapeutic trial).
In a descriptive study of 26 women published in 2012, Zayyan et al. found that the addition of cervical occlusion to a McDonald cerclage yielded a take-home baby rate of 96% in women with at least two mid-trimester miscarriages, preterm deliveries or a prior failed cervical cerclage (McDonald or Shirodkar). Noori et al. found a take-home baby rate of 93% in a descriptive study of 53 women with at least one prior mid-trimester loss. We found similar results in our preliminary study. However, where former descriptive studies support the beneficial role of cervical occlusion, the present study showed no such benefit.
The major strengths of this trial were that it was randomised, that the DMC who performed the interim analyses was independent of the trial investigators, that almost all patients received the allocated treatment (97%) and that few were lost to follow-up or excluded in the final analysis (8%). The internal validity can accordingly be regarded as good.
The major weakness of this trial was that it stopped before it had achieved the desired number of study participants, which increases the risk of type–2 errors. An article discussing whether to trust trials that are stopped early argued that the results of randomised trials are, indeed, not biased if the trial is stopped purely for administrative reasons. The present study was stopped because of an increasingly slow enrolment of women (an administrative reason), and because an interim analysis showed no effect of occlusion (a data-driven reason). However, it is important to note that all interim analyses were performed by the DMC, and that the results were kept confidential to the study investigators. As the last interim analysis found no effect of cervical occlusion (as opposed to a significant P value), and because the decision to stop the trial was taken in accordance with the prespecified protocol guidelines, the derived estimates and confidence intervals were valid. Another weakness that could compromise the external validity of the study was the unknown number of eligible women who were not randomised.
A meta-analysis of four trials and a recent study showed beneficial effects of cerclage in women with a singleton gestation, a previous preterm birth and a short cervix.[13, 14] We therefore hypothesised that cervical occlusion might benefit women with one or two prior preterm births, with a short cervix and without membranes being exposed to the vagina (secondary cerclage), and also women with a short cervix and membranes that were exposed to the vagina (tertiary cerclage). However, no significant effect of cervical occlusion in these three exploratory strata was found.
Regarding the take-home baby rate in the prophylactic trial, the true effect of cervical occlusion (with 95% CI) was between a relative increase of 12% and a relative reduction of 6%. As the confidence interval was narrow, cervical occlusion had no clinically relevant effect on the take-home baby rate in women having undergone prophylactic cerclage. The confidence interval for preterm birth before 34 and 28 weeks of gestation was wide. It has therefore not been determined whether occlusion reduces or even increases the risk of preterm birth in women with a prophylactic cerclage.
Similarly, the confidence intervals for the take-home baby rate and preterm births in the therapeutic trial were wide. Whether or not occlusion is beneficial in women with a secondary and tertiary cerclage therefore remains to be settled.
Currently, it cannot be recommended to routinely perform cerclage with cervical occlusion in women with a history of cervical insufficiency or a short cervix, whether the membranes are exposed to the vagina or not. Another large randomised trial is needed to definitively confirm or reject the potentially positive effect of cervical occlusion. Considering the slow recruitment in the present study, we doubt whether such a study will ever be conducted. A positive effect of cervical occlusion, if any, seems small.
The CERVO group
- Top of page
- The CERVO group
Per Olofsson, Universitetssjukhuset MAS in Malmö, Sweden; Irene Hösli, Universitätshospital Basel, Switzerland; Hennie Lombaard, The Pretoria Academic Hospital Complex, South Africa; Lluis Cabero and Dr Maria Goya, Hospital Materno-Infantil Valle Hebron in Barcelona, Spain; Dr Caio Coelho Marques, Hospital São Lucas, PUCRS, Porto Alegre RS, Brazil; Jens Lyndrup, Roskilde Sygehus, Copenhagen, Denmark; Jens Langhoff-Roos, Rigshospitalet, Copenhagen, Denmark; Lone Hvidman, Aarhus University Hospital, Skejby, Denmark; Lone Krebs, Holbæk Sygehus, Denmark; Bente Sørensen and Marianne Christiansen, Horsens Sygehus, Denmark; Anni Holmskov, Regionshospitalet Viborg, Denmark; Nini Møller, Nordsjællands Hospital Hillerød, Copenhagen, Denmark; Jesper Legarth, Herlev Hospital, Copenhagen, Denmark; Maha I Tulbah, King Faisal Specialist Hospital & Research Centre, Makkah, Saudi Arabia.