Objective: To estimate the effectiveness of prophylactic and therapeutic cerclage by meta-analysis of randomized clinical trials.

Data Sources: We searched the Cochrane Pregnancy and Childbirth Group specialized register of clinical trials (May 2002). Congress proceedings of international society meetings of fetal-maternal and reproductive medicine were searched by hand.

Methods of Study Selection: Meta-analysis of randomized clinical trials comparing cervical cerclage with expectant management during pregnancy was performed. Further clarification was sought from trial authors when required.

Tabulation, Integration, and Results: Six trials describing a total of 2175 women were analyzed. Prophylactic cerclage was compared with no cerclage in four trials. Pooled results failed to show a statistically significant reduction in pregnancy loss and preterm delivery rates, although a small reduction in births less than 33 weeks' gestation was seen in the largest trial (relative risk [RR] 0.75; 95% confidence interval [CI] 0.58 to 0.98). Cervical cerclage was associated with mild pyrexia, increased use of tocolytic therapy, and hospital admission but no serious morbidity. Two trials examined the role of therapeutic cerclage when ultrasound examination revealed a short cervix. Pooled results failed to show a reduction in total pregnancy loss, early pregnancy loss, or preterm delivery before 28 and 34 weeks in women assigned to cervical cerclage.

Conclusion: The effectiveness of prophylactic cerclage in preventing preterm delivery in women at low or medium risk for second-trimester pregnancy loss has not been proven. The role of cerclage in women whose ultrasound reveals short cervix remains uncertain.

^aDepartment of Obstetrics and Gynecology, University of Liverpool, Liverpool, United Kingdom

Supported in part by the University of Liverpool, which supports the Cochrane Pregnancy and Childbirth Group. There was no external funding or conflict of interest.

The authors acknowledge the additional unpublished information provided by Drs. Sietske M. Althuisius and Orion A. Rust.

(Obstet Gynecol 2003:102:621-627. © 2003 by The American College of Obstetricians and Gynecologists.)

For some women, the cervix recurrently dilates in the second or early third trimester of pregnancy, leading to preterm birth. There is no consistent definition of cervical incompetence,¹ but one possibility is "the painless dilatation of the cervix resulting in ruptured membranes and midtrimester miscarriage (12-24 weeks) and the passage without resistance, of size 9 Hegar dilators (9 mm), through the cervix in the non-pregnant state."² Recently, transvaginal ultrasound measuring cervical length during pregnancy has been suggested as a way to identify women at increased risk of preterm labor.³ Cervical cerclage is proposed to offer some benefit in the prevention of preterm labor.⁴ Shirodkar⁵ reported the insertion of a cervical stitch at about 14 weeks of pregnancy. McDonald⁶ later described a simpler purse string stitch technique.

The main issues regarding the use of cervical cerclage include whether it should be offered at all; if so, to whom; and when and how the stitch should be inserted. We conducted a Cochrane systematic review to address these uncertainties and focused on the evidence provided from randomized clinical trials.

Sources

The main objectives of this study were to estimate the effectiveness and safety of cervical cerclage for prevention of preterm delivery. We specifically evaluated emergency cerclage inserted in women during the early stages of second-trimester miscarriage or extreme preterm labor and elective cerclage used before pregnancy by either the vaginal or abdominal route.⁷

Studies included women with confirmed or suspected cervical incompetence and desired future pregnancies and women who presented emergently with a diagnosis of preterm labor, presumed to be due to cervical incompetence. The diagnosis of cervical incompetence relied on researcher definitions such as clinical history, recurrent second-trimester losses, or cervical resistance studies.

We prespecified (decided) clinical outcomes of interest before data extraction and published them in the Cochrane Library (Appendix). ⁷ An additional comparison was included after completion of the protocol to examine the influence of ultrasound identification of a short cervix at stitch insertion.

This review is based on the search strategy developed for the Pregnancy and Childbirth Group of the Cochrane collaboration as a whole. The full list of journals and conference proceedings, as well as the search strategies for the electronic databases (which are searched by the Group on behalf of its reviewers), are described elsewhere.⁸ Briefly, the Cochrane Pregnancy and Childbirth Group's trials register is maintained by the Trials Search Co-coordinator and contains trials from quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL); monthly searches of Medline; hand-conducted searches of 30 journals and the proceedings of major conferences; and weekly current awareness searches of a further 37 journals. Details of the search strategies for CENTRAL and Medline, the list of hand-searched journals and conference proceedings, and the list of journals reviewed by the current awareness service can be found elsewhere.⁹ Congress proceedings of international society meetings of fetal-maternal and reproductive medicine were searched by hand. Randomized clinical trials comparing cervical cerclage with expectant management during pregnancy were included. Further clarification was sought from trial authors when required. The last search date was May 2002.

Two reviewers selected the randomized trials for inclusion in the review independently. We were not blinded to the authors and institutions of the trials under consideration. The third reviewer resolved any differences of opinion regarding trials for inclusion. The validity of each trial to be included was assessed according to the criteria in the Cochrane Reviewers' Handbook,⁸ with a grade allocated to each on the basis of allocation concealment, as follows: A (adequate), B (unclear), or C (clearly inadequate). Where the method of allocation concealment was unclear, authors were contacted to provide further details. Quasi-randomized studies in which allocation was transparent (eg, use of alternative allocation or medical record numbers) were excluded. We planned to exclude trials if it was not possible to present the data by intention to treat. Data for a given outcome were not considered if more than 20% of randomized participants were excluded.

Two reviewers who used previously prepared data extraction forms performed data extraction independently. Any discrepancy was resolved by discussion. If agreement was not reached, the data were excluded until further clarification was available from the authors. Statistical analysis was performed by RevMan 4.1.¹⁰ Results were reported as relative risk (RR), fixed, or random-effects model, as appropriate. Heterogeneity was tested automatically by RevMan 4.1. If significant heterogeneity was found by use of a fixed-effect model, and in the absence of obvious explanation for the heterogeneity, additional testing was performed with a more conservative random-effects model.

Study Selection

Six studies were included in the current review (Table 1). The United Kingdom-coordinated Medical Research College/Royal College of Obstetricians and Gynecologists (MRC/RCOG)¹¹ trial is the largest randomized controlled trial (RCT) in the review. This international multicenter study recruited women who were deemed at risk of a second-trimester pregnancy loss by history, but whose obstetrician was uncertain of the diagnosis of cervical weakness. A total of 1292 women were randomized to stitch versus no stitch. The trial by Rush et al¹² was set in a South African teaching hospital miscarriage clinic. It was designed to evaluate whether the policy of performing cerclage prolongs gestation in women with a history of late miscarriage. A total of 194 women with at least two previous preterm labors or pregnancy loss between 14 and 36 completed weeks were randomized to stitch or no stitch. The French study by Lazar and Guegan,¹³ set in four nearby hospitals, developed a scoring chart for suspected cervical incompetence and included women they deemed at moderate risk of preterm delivery. The scoring system included aspects of previous history, state of the cervix, and evolving signs of cervical change and vaginal bleeding. Five hundred and six women were included.

In studies from The Netherlands, Althuisius et al^14,15 initially screened a population of women with previous risk factors of preterm delivery and late miscarriage. The trials used two Amsterdam hospitals. The first RCT described prophylactic cerclage on the basis of history and reports on 70 women. In the second article, which described the same cohort, a second randomization was allowed for the initial no-stitch group if a woman's cervical length became less than 25 mm before 27 weeks' gestation. In this case, women were randomized either to therapeutic cerclage plus bed rest or else bed rest alone. All women were admitted to the hospital for 5 days. For the first 2 days, they had complete bed rest. On the third day, they were able to get up to use the bathroom. On the fourth day, they were allowed to walk around three times a day for a quarter of an hour each time. At home, they followed the same policy until 32 weeks' gestation. The women were initially randomized to receive a stitch or ultrasound surveillance in a 2:1 ratio, and analysis was by intention to treat for both articles. Although they took a pragmatic approach, the design of this study made it difficult for reviewers to enter data because initial randomization was complicated by a second randomization. We are grateful to Dr. Althuisius, who divided the trial data into two separate data sets for us: initial prophylactic and secondary therapeutic.

A US study, that of Rust et al,¹⁶ presented randomized data for 113 women, set in a single tertiary-care center. The study used a composite cervical score (Benham score) generated by dynamic imaging of the cervix and four measurements. The critical cutoff was 25 mm cervical length. All women in the Rust et al study underwent amniocentesis before randomization to exclude chorioamnionitis.

All six trials scored "A" for randomization allocation and were given equal weighting when computing relative risks and confidence intervals.

Results

Pooled results of four trials describing 2062 women showed no differences in total pregnancy loss and early pregnancy loss (less than 24 weeks) (RR 0.86; 95% confidence interval [95% CI] 0.59, 1.25). Two trials^12,14 reported on delivery at less than 28 weeks' gestation and three trials^12-14 on delivery at less than 32 weeks but failed to show beneficial effect of cerclage (RR 1.29; 95% CI 0.67, 2.49). There was also no difference in perinatal death (RR 0.8; 95% CI, 0.48 1.36) or the mean gestational age weighted mean difference, which was 0.15 (95% CI -0.35, 0.66) between the two groups. The MRC/RCOG trial used 33 weeks' gestation as an important milestone and appeared to suggest fewer deliveries in the cerclage arm (83 of 647, 12.8%, versus 110 of 645, 17.1%, RR 0.75; 95% CI 0.58, 0.98). This effect was magnified when considering a small subset of women who had three or more second-trimester losses. In this subgroup, which comprised 107 women, delivery rate before 33 weeks was 15% with cerclage compared with 32% in the control group (RR 0.46; 95% CI 0.22, 0.98). All four studies reported on preterm delivery at less than 37 weeks' gestation with no overall significant difference between the two groups (RR 1.04; 95% CI 0.99, 1.10).

There were no cases of maternal mortality in any of the reported trials. In the Althuisius et al trial, there was one uterine rupture in the cerclage arm and two major postpartum hemorrhages in the control arm. Compared with conservative therapy (Table 2), more women developed infection (defined as mild pyrexia by the trialists) after cervical stitch (6.7% versus 2.6%; RR 2.57; 95% CI 1.42, 4.64). Some studies^11,12 described infection as pyrexia greater than 38C, whereas the others were less specific. There was also a higher risk of minor maternal morbidity in the cerclage arm (RR 1.32; 95% CI 1.13, 1.55), which for most studies meant hospital admission and bed rest. There were more cesarean deliveries in the cervical suture group (14% versus 11.6%), but this did not reach statistical significance (RR 1.24; 95% CI 0.99, 1.55). No difference in the induction rate was observed (RR 1.13; 95% CI 0.92, 1.38). More tocolytic therapy was prescribed in the cerclage group (RR 1.29; 95% CI 1.14, 1.46). Only the Althuisius et al study reported on corticosteroid use for fetal pulmonary maturity, and no difference between the two groups was noted. The Althuisius et al study reported less preterm prelabor rupture of the membranes after cervical suture (8% versus 27%, RR 0.31; 95% CI 0.08, 1.28). There were no differences in antepartum hemorrhage rates (RR 0.77; 95% CI 0.16, 3.64), and neonatal birth weights were essentially the same (weighted mean difference 91; 95% CI -358, 540).

Two trials investigated ultrasound measurements of cervical length as the indicator for inserting a cervical stitch.^15,16 The Althuisius et al trial recruited women at risk of cervical incompetence in whom transvaginal ultrasound revealed a short cervix (less than 25 mm) before 27 weeks' gestational age. The Rust et al trial randomized women between 16 and 24 weeks who had demonstrable prolapse of the fetal membranes into the endocervical canal more than 25% of the total cervical length or a distal cervical length less than 25 mm according to transvaginal ultrasonography.

There was no difference in total pregnancy loss (RR 0.91; 95% CI 0.36, 2.27), early pregnancy loss (RR 0.17; 95% CI 0.01, 3.3), or preterm delivery before 28 weeks (RR 0.12; 95% CI 0.01, 2.19) and 34 weeks (RR 0.7; 95% CI 0.44, 1.12). The data for delivery at less than 34 weeks show significant heterogeneity (P = .03), and pooled RR by use of the random model is 0.31 (95% CI 0.02, 6.09). The Althuisius et al study reported delivery at less than 37 weeks' gestation, favoring cerclage (four of 19, 21.1%, versus ten of 16, 63%; RR 0.34; 95% CI 0.13, 0.87). The Althuisius et al study also showed a significantly longer gestation after cervical stitch (37.9 versus 33.1 weeks). However, when the data from both the Althuisius and Rust et al studies are combined, the prolongation of gestation was not significant (1.5 weeks; 95% CI -0.3, 3.3).

When evaluating maternal infection, the two studies conflict. The Althuisius et al trial showed the control group having more infection (one of 19, 5.3% cerclage, versus nine of 16, 56.3% control), but the Rust et al study shows the converse (11 of 55, 20% cerclage, versus six of 58, 10.3% control). The two studies provided prophylactic antibiotic therapy to both groups. The Althuisius et al study provided amoxicillin-clavulanic acid, and the Rust et al study provided clindamycin intravenously. The Rust et al study performed amniocentesis to exclude infection before inclusion. The Althuisius et al trial reported a lower incidence of preterm prelabor rupture of the membranes in the cerclage arm (0 of 19) compared with the no-stitch arm (eight of 16) (RR 0.03; 95% CI 0.0, 0.50). It is therefore debatable whether these results should be pooled together. The pooled RRs obtained by fixed effects were 0.78 (95% CI 0.39, 1.56) and 0.48 (95% CI 0.02, 10.52) by a random model. There were no observed differences in antepartum hemorrhage rate, use of corticosteroids, cesarean delivery, labor induction, or episodes of false labor. The Althuisius et al study reported higher birth weights in the cerclage arm compared with the control arm (3083 g versus 2224 g; weighted mean difference 859; 95% CI 287, 1430). There was heterogeneity between the two studies in terms of maternal infection, delivery at less than 34 weeks, and mean gestational age, which could be the result of the different selection criterion and management policies used.

Conclusion

We found no conclusive evidence from randomized studies that inserting a prophylactic cervical stitch in women perceived to be at risk of preterm birth or second-trimester pregnancy loss attributed to cervical factors reduces the risk of pregnancy loss, preterm delivery, or morbidity associated with preterm delivery. The largest study reported a significant reduction in preterm deliveries before 33 weeks' gestation, from 17% in the control group to 13% when cervical suture was inserted.¹¹ As far as maternal side effects are concerned, cervical cerclage is consistently associated with increased risk of maternal infection and pyrexia. Although undesirable and inconvenient, there is no evidence that maternal infection and pyrexia attributed to cervical cerclage causes long-term harm for mother or child. However, most authors used the term "infection" in a nonspecific manner, because pyrexia does not necessarily equate to infection. Chorioamnionitis was not included as a variable in any trial; however, the Rust et al trial excluded it by amniocentesis before recruitment. Exclusion of histologically proven, or at least clinical, chorioamnionitis would be preferable in future studies. No studies commented on possible interactions between cerclage and adjunct therapy such as antibiotics and nonsteroidal anti-inflammatory drugs.

The data on other associated maternal morbidity include increased use of tocolysis, possibly as a result of an increase in uterine irritability being triggered with a cervical stitch. A fear of allowing a woman to labor with a stitch, thus risking further damage to the cervix, or increased vigilance by the attending obstetrician might also be a contributing factor for increased use of tocolytics. Data from randomized trials provide no significant evidence that cerclage causes serious maternal or fetal morbidity.

We suggest that the source of heterogeneity for important clinical outcomes (maternal infection, preterm delivery at less than 34 weeks' gestation, and mean gestational age) is inconsistent clinical definitions used or different patient populations studied (Figure 1). For example, in describing the use of ultrasound, the Althuisius et al study included women before 27 weeks' gestation, 66% of whom had a previous preterm delivery at less than 28 weeks' gestation. In the Rust et al trial, the patients were included before 24 weeks' gestation, but the number of women with a previous preterm birth was not provided. This RCT also included some low-risk women in whom a short cervix was found incidentally. These two studies may illustrate a significant difference in the background risk of preterm labor.

Figure 1. Clinical outcomes for elective cerclage versus no cerclage. RR = relative risk; 95% CI = 95% confidence interval. Drakeley. Cervical Cerclage. Obstet Gynecol 2003.

The data from two studies where ultrasound was used to select patients at risk of preterm delivery show that the incidence of preterm prelabor rupture of the membranes was significantly reduced by stitch insertion, and the overall preterm delivery rate (less than 37 weeks) was lower. However, pregnancies were not significantly prolonged between 24 and 32 weeks, which we think is the most crucial time period associated with neonatal morbidity. It is noteworthy that in the Althuisius et al trial, a policy of strict bed rest without cerclage in women with a poor history and short cervixes resulted in a mean gestational age of 33 weeks and preterm delivery before 28 weeks' gestation of 15% (three of 16). This meta-analysis was not designed to determine risk. It is unrealistic to expect that data on fewer than 200 women could provide a clear picture on the effectiveness and safety of cervical cerclage.

No studies assessed the other comparisons, psychological effects of cervical stitch, whether in the short or long term, or associated interventions such as bed rest and cesarean delivery, on the patient or her family.

At present, there are inadequate data to support the use of cervical cerclage in women considered at low or medium risk of second-trimester miscarriage or extreme preterm labor. There may be a benefit to cervical cerclage for women considered at very high risk of second-trimester miscarriage as a result of a cervical factor (more than two second-trimester losses or progressive shortening of the cervix as revealed by ultrasound). However, predicting that miscarriage will occur in women with cervical factor remains unsatisfactory, and many women may be treated unnecessarily.

Ultrasound may help to reduce the number of unnecessary procedures, although our meta-analysis offers no firm evidence to support this view. Very promising data from Althuisius et al¹⁵ are confined to a highly selected group of women who had a combination of poor obstetric history and a progressively shortening cervix. All 19 women delivered after 34 weeks with a cervical stitch, whereas only seven of 16 without a stitch did so. Others need to confirm these results.

There remains uncertainty on how best to treat women who have experienced one or two second-trimester pregnancy losses. On the basis of the evidence, the skeptic would not offer cerclage at all. However, an enthusiast may offer cerclage in a woman with two previous losses and may even stretch the evidence to include one previous second-trimester loss on the basis of the low rate of serious morbidity. We think that with two previous losses due to cervical incompetence and no other likely diagnoses, it is reasonable to offer cerclage. We acknowledge that many women may be subjected to an unnecessary intervention. In fact, data from the MRC/RCOG trial suggest that cervical cerclage in women with a previous second-trimester loss (or previous preterm delivery) may help to prevent one delivery before 33 weeks for every 24 stitches inserted (95% CI 12, 295). In other words, 96% of operations would be performed on women who would not have labored prematurely.

As a result of the invasive nature of a cervical suture insertion and its dubious benefit, future evaluation of effectiveness and safety should only be performed within rigorous RCTs. One of the most important findings of our review is the lack of evidence for six prespecified clinically important clinical questions. This opinion is shared by a recently published expert view from the United States, published in Obstetrics & Gynecology.¹⁷ Although most authors of retrospective studies call for randomized trials, it appears that few researchers are prepared to conduct studies with the sufficient statistical power and external validity required to establish the appropriate indications for the operation.

Initial or secondary shortening of cervical length measured by ultrasound may or may not identify high-risk women who are going to miscarry, and it could be that these women may warrant further study. The obvious choice for the next randomized trial is a replication of the Althuisius et al therapeutic cerclage study.¹⁵ Assuming that the no-treatment group of women with poor obstetric history and short cervix will have a preterm delivery rate of 40%, on the basis of the data of the Althuisius et al study, we can expect a reduction in preterm delivery rate of at least 50%. Only 200 women would have to be randomized to prove beyond a reasonable doubt that cerclage is the treatment of choice for these women.

References

1. Berry CW, Brambati B, Eskes TKAB, Exalto N, Fox H, Geraedts JPM, et al. The Euro Team Early Pregnancy (ETEP) protocol for recurrent miscarriage. Hum Reprod 1995;10:1516-20.

2. Drakeley AJ, Quenby S, Farquahrson RG. Mid trimester loss—Appraisal of a screening protocol. Hum Reprod 1998;13:1975-80.

3. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad A, Das A, et al. The length of the cervix and the risk of spontaneous premature delivery. N Engl J Med 1996;334:567-72.

4. Matijevic R, Olujic B, Tumbri J, Kurjak A. Cervical incom- petence: The use of selective and emergency cerclage. J Perinat Med 2001;29:31-5.

5. Shirodkar VN. A new method of operative treatment for habitual abortions in the second trimester of pregnancy. Antiseptic 1955;52:299-300.

6. McDonald IA. Suture of the cervix for inevitable miscarriage. J Obstet Gynaecol Br Com 1957;64:346-53.

7. Drakeley AJ, Roberts D, Alfirevic Z. Cervical cerclage for prolonging pregnancy in women with cervical incompetence [protocol]. Cochrane Pregnancy and Childbirth Group. Cochrane Database Syst Rev 2002.

8. Clarke M, Oxman AD, eds. Cochrane reviewers' handbook 4.1 (updated June 2000). In: Review Manager 4.1. (RevMan) [computer program]. Oxford: The Cochrane Collaboration, 2000.

9. Editorial Team, Cochrane Pregnancy and Childbirth Group. In: The Cochrane Library, 2003. Oxford: Update Software.

10. Review Manager 4.1. (RevMan) [computer program]. Oxford: The Cochrane Collaboration, 2000.

11. MRC/RCOG Working Party on Cervical Cerclage. Final report of the Medical Research Council/Royal College of Obstetricians and Gynaecologists Multicentre Randomised trial of cervical cerclage. Br J Obstet Gynaecol 1993;100:516-23.

12. Rush RW, Isaacs S, McPherson K, Jones L, Chalmers I, Grant A. A randomised controlled trial of cervical cerclage in women at high risk of spontaneous preterm delivery. Br J Obstet Gynaecol 1984;91:724-30.

13. Lazar P, Guegan S. Multicentred controlled trial of cervical cerclage in women at moderate risk of preterm delivery. Br J Obstet Gynaecol 1984;91:731-5.

14. Althuisius SM, Dekker GA, van Geijn HP, Bekedam DJ, Hummel P. Cervical incompetence prevention randomised cerclage trial (CIPRACT): Study design and preliminary results. Am J Obstet Gynecol 2000;183:823-9.

15. Althuisius SM, Dekker GA, Hummel P, Bekedam DJ, van Geijn HP. Final results of the Cervical Incompetence Prevention Randomized Cerclage Trial (CIPRACT): Therapeutic cerclage with bed rest versus bed rest alone. Am J Obstet Gynecol 2001;185:1106-12.

16. Rust OA, Atlas RO, Reed J, van Gaalen, Balducci J. Revisiting the short cervix detected by transvaginal ultrasound in the second trimester: Why cerclage therapy may not help. Am J Obstet Gynecol 2001;185:1098-105

17. Harger JH. Cerclage and cervical insufficiency: An evidence-based analysis. Obstet Gynecol 2002;100:1313-27.

Address reprint requests to: Andrew J. Drakeley, MD, Department of Obstetrics and Gynecology, University of Liverpool, Liverpool Women's Hospital, Crown Street, Liverpool L8 7SS, UK; E-mail: adrakeley@yahoo.com

Received December 6, 2002.
Received in revised form May 20, 2003.
Accepted June 5, 2003.