- Email: [email protected]
Ultrasound cervical length measurement for prediction of delivery before 32 weeks in women with emergency cerclage for cervical insufficiency
International Journal of Gynecology and Obstetrics 110 (2010) 245–248
Contents lists available at ScienceDirect
International Journal of Gynecology and Obstetrics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j g o
CLINICAL ARTICLE
Ultrasound cervical length measurement for prediction of delivery before 32 weeks in women with emergency cerclage for cervical insufficiency Florent Fuchs a,b,c,⁎, Jean Bouyer b,c,d, Hervé Fernandez a,b,c, Amélie Gervaise a, René Frydman c,e,f, Marie-Victoire Senat a,b,c a
Department of Obstetrics and Gynecology, Hôpital Bicêtre, Assistance Publique Hôpitaux de Paris (APHP), Le Kremlin-Bicêtre, France Reproduction and Child Development, CESP Centre for Research in Epidemiology and Population Health, U1018, Inserm, Villejuif, France Université Paris Sud 11, UMRS 1018, Villejuif, France d INED, Paris, France e INSERM U782, Université Paris-Sud, Clamart, France f Department of Obstetrics, Gynecology and Reproductive Medicine, Hôpital Antoine Béclère, Assistance Publique Hôpitaux de Paris (APHP), Clamart, France b c
a r t i c l e
i n f o
Article history: Received 2 February 2010 Received in revised form 7 April 2010 Accepted 4 May 2010 Keywords: Bulging membranes Midtrimester cervical dilation Preterm birth Threatened second trimester loss Ultrasound measurement
a b s t r a c t Objective: To assess the accuracy of ultrasound cervical length for predicting delivery before 32 weeks among women with a threatened second trimester fetal loss treated by emergency cervical cerclage, and to compare it with prediction based on clinically-assessed cervical dilation. Methods: Retrospective study (1997–2006) of 70 women with singleton pregnancies who underwent emergency cervical cerclage and ultrasound cervical length measurement between 15 and 26 weeks. The associations between cervical length and delivery before 32 weeks were studied with univariate and multivariate regression. ROC curves were used to determine the most discriminating cut-off value. Results: Although ultrasound cervical length measurement was significantly associated with early preterm delivery, its predictive accuracy was moderate with an area under the ROC curve (AUC) of 0.68 (95% CI, 0.54–0.82), which was similar (P = 0.43) to the AUC of cervical dilation of 0.73 (95% CI, 0.61–0.85). The best cut-off value was 7 mm, corresponding to a doubled risk of delivery before 32 weeks. Its sensitivity was 52%, specificity 82%, PPV 62%, and NPV 76%. It was no more informative than cervical dilation of 3 cm. Conclusion: Ultrasound cervical length measurement does not predict early preterm birth better than clinically-assessed cervical dilation in women with an emergency cerclage. © 2010 International Federation of Gynecology and Obstetrics. Published by Elsevier Ireland Ltd. All rights reserved.
1. Introduction In recent years, efforts to prevent preterm delivery have focused on women at high risk for preterm and early preterm birth, such as those with cervical insufficiency. The literature describes various methods of managing known cervical insufficiency such as elective prophylactic cerclage [1] or ultrasound-indicated therapeutic cerclage. The decision to use therapeutic cerclage is based on the measurement of cervical length during transvaginal ultrasound performed at 14–24 weeks [2]. However, the accuracy of this measurement for predicting preterm birth has never been analyzed among women treated with emergency cervical cerclage. Published studies dealing with emergency cerclage have described only the clinical factors associated with cerclage failure, defined as preterm birth [3–9]. The objective of the present study was to assess the predictive accuracy of ultrasound cervical length measurement for predicting ⁎ Corresponding author. Hôpital Bicêtre - Service de Gynécologie-Obstétrique, 78 rue du Général Leclerc, 94270 Le Kremlin-Bicêtre, France. Tel.: +34 1 45 21 77 84; fax: +34 1 45 21 77 25. E-mail address: fl[email protected] (F. Fuchs).
early preterm birth (delivery before 32 weeks) among women undergoing emergency cerclage. More precisely, we intended to: (1) quantify the association between cervical length and early preterm birth; (2) determine whether ultrasound cervical length measurement is useful for predicting early preterm birth, and especially, which cervical length cut-off value discriminates best; and (3) compare the predictive accuracy of this measurement with that of clinical assessment of cervical dilation. 2. Materials and methods This retrospective study examined the records of women who underwent cerclage at Antoine Béclère Hospital, Clamart, France—a tertiary care center—from January 1997 through December 2006. We included all women who underwent emergency cerclage for a singleton pregnancy at 15–26 weeks of pregnancy. Demographic, pregnancy, and neonatal data were collected, as well as obstetric history. In this maternity ward, emergency cerclage was systematically performed for pregnant women with a threatened second trimester fetal loss, defined as a cervix spontaneously dilated 1 cm or more on physical examination and prolapse of an unopened amniotic
0020-7292/$ – see front matter © 2010 International Federation of Gynecology and Obstetrics. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijgo.2010.04.022
246
F. Fuchs et al. / International Journal of Gynecology and Obstetrics 110 (2010) 245–248
sac at the external cervical os of the uterus or bulging in the vagina during speculum examination, and with no signs of labor (progressive dilation with contractions). Emergency cerclage was always performed on the basis of the physical digital examination and never according to the results of the ultrasound examination. Exclusion criteria were fetal anomalies, preterm premature rupture of membranes (PPROM), and clinical signs of chorioamnionitis (maternal temperature N38 °C, uterine tenderness or contractions, fetal tachycardia, and positive microbiological findings after testing either vaginal swab or urine samples). Midwives with sterile gloves digitally evaluated cervical dilation during vaginal examination and routinely looked for bulging membranes, at or beyond external os, during sterile speculum examination. Experienced sonographers measured cervical length during all ultrasound examinations according to standard techniques after the bladder was emptied [10]. Fundal pressure was also applied as a provocative maneuver. The cervix was measured at least 3 times, and the shortest measurement was recorded. The protruding membranes meant that the closed canal usually measured was no longer present; therefore, the length that was measured was the remaining cervix between the internal and external os. Even when membranes were bulging in the vagina, sonographers could always measure the remaining cervical length on a sagittal view (Fig. 1). The goal of this transvaginal ultrasound was to obtain a standardized and objective measurement of cervical length, even for women with advanced cervical dilation. All the women included were treated according to the same protocol: admission to the high-risk pregnancy department for inhospital bed rest for at least 7 days; blood tests; cerclage according to the McDonald procedure with a single stitch applied in the operating room under general or epidural anesthesia; antibiotics (1 g of amoxicillin with clavulanic acid 3 times a day for 5 days); indomethacin suppositories (100 mg twice a day for 2 days); close clinical and laboratory monitoring; and administration of antenatal steroids after reaching 24 weeks of pregnany. When necessary, bulging membranes were treated with a balloon inflation device, as described by Tsatsaris et al. [11], or an inflated Foley catheter n°22. The study outcome was early preterm birth, i.e. delivery before 32 weeks of pregnancy. This gestational age cut-off was chosen according to findings from the French EPIPAGE cohort study [12], as it appears to be associated with a significant increase in neonatal mortality and morbidity. To describe the association between delivery before 32 weeks and cervical length, we used fractional polynomial (FP) modeling [13], designed for continuous variables such as cervical
Fig. 1. Sagittal view of cervix with membranes bulging beyond external os.
length, and produced a graph of the curve that best fitted the relation between cervical length and early preterm birth. Receiver operating characteristic (ROC) curves were used to identify the best ultrasound cervical length cut-off for predicting early preterm birth. To compare the predictive value of ultrasound cervical length measurements and clinically-measured cervical dilation, we compared the area under these ROC curves. The association between delivery before 32 weeks and cervical length was then analyzed with univariate and multivariate logistic regression to obtain odds ratios and with Poisson regression to obtain relative risks [14]. In these regressions, cervical length was modeled dichotomously, according to the cut-off found by the ROC curve. In the multivariate analysis, we adjusted for known confounders identified in previous studies [3–9] and for variables associated with delivery before 32 weeks in the univariate analysis with P b 0.2: gravidity, history of spontaneous abortion, history of second-trimester losses, prophylactic cerclage, gestational age at cerclage, and biological infection, defined as C-reactive protein N15 mg/L or white blood count ≥13 600 × 106/L [15–17]. Because our population included 15 women whose emergency cerclage was performed late, between 24 and 26 weeks, we verified that our results were similar after excluding those 15 women. Statistical analyses were performed with STATA software version 10 (Stata Corp, College Station, TX, USA). We considered results to be significant when P b 0.05. This study is a retrospective analysis of a medical database that included ultrasound cervical length measures. This database and its use for research have been approved by the French Data Protection Authority (CNIL, Commission Nationale de l'Informatique et des Libertes) under the notification number 1181076. 3. Results The mean age of the patients was 32.3 years (range, 21–43 years). Most were nulliparous (53%), and 57% were at high risk of preterm birth because they had one or more of the following risk factors (these percentages exceed 57% because some women had several risk factors): history of one or more second-trimester losses (17%); one or more preterm births (14%); one or more surgically induced abortions (21%); in utero exposure to diethylstilboestrol (10%); prophylactic cerclage for this pregnancy (14%); and surgery for a uterine malformation (10%). Median gestational age at cerclage was 22.3 weeks (range, 15–26 weeks) with 30 cerclages (43%) performed before 22 weeks, 25 (36%) between 22–24 weeks, and 15 (21%) after 24 weeks. Twenty-five women (36%) had membranes bulging in the vagina, and the other 45 (64%) had membranes at the external os. Digital clinical measurements of cervical dilation were distributed as follows: 40% with 1 cm dilation, 27% with 2 cm, 19% with 3 cm, and 14% with 4 cm or more. Ultrasound cervical length measurements were distributed as follows: 6% had a cervical length of 25 mm or more, 27% 15–24 mm, and 67% less than 15 mm. Median gestational age at delivery was 36.9 weeks (interquartile range, 27.1–38.4 weeks; range, 17–41.4 weeks), and 25 women (36%) gave birth before 32 weeks. Of 70 neonates, 57 (81%) survived to discharge from hospital, and 47 (82%) had no observable neonatal morbidity. Fig. 2 describes the probability of delivery before 32 weeks according to cervical length, as modeled by fractional polynomials. This modeling showed that the overall relationship was statistically significant (P = 0.036) and did not differ from a linear relationship (P = 0.13). The ROC analysis presented in Fig. 3 showed that, according to the standard classification of such curves [18], ultrasound cervical length alone was a poor to fair predictor of delivery before 32 weeks (area under curve [AUC] 0.68; 95% CI, 0.54–0.82). The best cut-off value was 7 mm, which was associated with a sensitivity of 52% (95% CI,
F. Fuchs et al. / International Journal of Gynecology and Obstetrics 110 (2010) 245–248
247
length shorter than 7 mm was associated with a more than doubled risk of early preterm birth (relative risk 2.4; 95% CI, 1.3–4.6). Exclusion of the 15 women whose emergency cerclage was perormed late, between 24 and 26 weeks, did not significantly change these results. The AUC of ultrasound cervical length (0.71; 95% CI, 0.56–0.87) and that of digital cervical dilation (0.75; 95% CI, 0.62– 0.88) were still not significantly different (P = 0.57). This finding confirmed the stability of our results. 4. Discussion
Fig. 2. Probability of early preterm birth as a function of cervical length, measured by transvaginal ultrasound and modeled by fractional polynomials [25].
31%–73%), a specificity of 82% (95% CI, 71%–94%), a positive predictive value of 62% (95% CI, 39%–85%), and a negative predictive value of 76% (95% CI, 63%–88%). Comparison of the ROC curves for ultrasound cervical length and for clinically-assessed cervical dilation showed similar results (AUC 0.68 [95% CI, 0.54–0.82] vs 0.73 [95% CI, 0.61–0.85]; P = 0.43). In particular, cervical dilation, assessed clinically at 3 cm, yielded a prediction very similar to that of a cervical length of 7 mm. The sensitivity of this 3-cm value for clinical dilation was 56% (95% CI, 35%–77%), its specificity 80% (95% CI, 68%–92%), its positive predictive value 61% (95% CI, 39%–82%), and its negative predictive value 77% (95% CI, 64%–89%). The strength of the association between cervical length and early preterm birth was shown by the univariate analysis, where we observed a substantial increase in the risk of delivery before 32 weeks (compared with delivery beyond that gestational age) for a cervix shorter than 7 mm, with a crude odds ratio (OR) of 5.0 (95% CI, 1.7–14.9). After adjustment for known risk factors in multivariate logistic regression analysis, this OR increased still further (adjusted OR 7.8; 95% CI, 1.9–31.2). Poisson regression showed that a cervical
Fig. 3. Receiver operating characteristic (ROC) curves of early preterm birth for ultrasound cervical length measurement and clinical assessment of cervical dilation.
The study showed that in emergency situations such as threatened second-trimester fetal loss, very short cervical length measured by ultrasound before cerclage is significantly associated with early preterm birth. However, the predictive accuracy of this measurement for early preterm birth was no better than that of clinically measured cervical dilation. Numerous publications report the widespread use of transvaginal ultrasound measurement of cervical length to predict preterm birth, and it is well known to be an important prognostic tool [19,20]. Transvaginal ultrasound yields precise and reproducible values whereas digital examination of the cervix is subjective [21]. In the general population, a cervical length of less than 25 mm at 24 weeks is associated with an increased risk of preterm delivery, but because the prevalence of preterm birth is fairly low in the general population, its positive predictive value is only 18% [19]. Similarly, a cervical length of less than 15 mm at 24 weeks has a positive predictive value of 50% [22]. In contrast, for asymptomatic women at high risk of preterm birth because of their obstetric history, the positive predictive value of a cervical length of less than 25 mm is 55% [23]. As our population was also at very high risk of early preterm birth, with 36% of the women giving birth before 32 weeks, we considered it important to check the utility of ultrasound cervical length in this situation of threatened second-trimester fetal loss and to determine the appropriate cut-off. The association between cervical length and early preterm birth appears clearly in Fig. 2 and is confirmed in univariate and multivariate analyses. Whether a short cervical length, less than 7 mm, was considered before or after adjustment for known risk factors, the odds ratio was statistically significant and very high. However, because the prevalence of early preterm birth was high (36%) in our population, the odds ratios must not be interpreted as multiplicative factors. We thus estimated rate ratios with Poisson regression and found in the multivariate analysis that cervical shortening to less than 7 mm was associated with a more than doubled increase in the risk of early preterm birth (adjusted RR 2.4; 95% CI, 1.3–4.6). That measurement, although extremely short, seems logical as the cervix is dilated with bulging membranes in the cervix or in the vagina. It corresponds to the remaining cervical length should surgeons try to perform emergency cerclage. Despite this strong statistical association, ultrasound cervical length measurement was not a good predictor of early preterm birth. The ROC curve in Fig. 3 shows a poor to fair predictive ability, with an area under the curve of 0.68 (95% CI, 0.54–0.82), close to the 0.5 that corresponds to a predictive ability no better than random allocation. Comparing the areas under the ROC curves showed that the clinical and ultrasound predictions were very similar. Similarly, the sensitivity and specificity of ultrasound cervical length less than 7 mm and cervical dilation of 3 cm appeared very close. These predictive factors are associated with one other, as the remaining cervical length rarely exceeds 7 mm when the cervix is dilated by more than 3 cm. In this situation of a threatened pregnancy loss, we believe that ultrasound examination is relevant only when speculum examination shows cervical dilation greater than 3 cm, to reduce the potential induction of infection by digital examination [24]. Some characteristics of our study must be mentioned. We did not choose to exclude women who had undergone a prophylactic
248
F. Fuchs et al. / International Journal of Gynecology and Obstetrics 110 (2010) 245–248
cerclage, although their risk of preterm delivery might be higher than that of other women, because we sought to include every woman who met the criteria for emergency cerclage (defined as cervix dilation and membrane protrusion). Our study was intended to be a pragmatic one. It is also relevant that our hospital is a tertiary care center with experienced operators familiar with emergency cerclage practice. This factor may limit the extrapolation of our results to the general population. Another possible limitation of our study may be our inclusion of women up to a gestational age of 26 weeks. In recent years, emergency cerclage after 24 weeks has been rare, as indeed our study shows: most of the cerclages at 24–26 weeks were performed before 2002. However, as we have shown, excluding these women did not change our results. In view of the increased success in neonatal resuscitation for very preterm neonates, we believe that the practice of emergency cervical cerclage above 24 weeks should be considered and debated, but not routinely performed. Each woman was included based on clinical factors and not on ultrasound indicators. This mode of selection may explain why ultrasound and clinical digital examination appear equivalent. Indeed, it may have led to a bias in favor of the predictive value of clinical examination over ultrasound, because women with only an ultrasound-measured "short" cervix did not meet the inclusion criteria and would have been omitted from the study because they had no clinical changes of the cervix. Other studies should attempt to confirm our results about the predictive value of transvaginal ultrasound measurement of cervical length for outcome after emergency cerclage. Our results indicate that there is currently no evidence to recommend the use of such measurement for women presenting with a dilated cervix and visible membranes during speculum examination. Conflict of interest The authors have no conflicts of interest to disclose. References [1] Final report of the Medical Research Council/Royal College of Obstetricians and Gynaecologists multicentre randomised trial of cervical cerclage. MRC/RCOG Working Party on Cervical Cerclage. Br J Obstet Gynaecol 1993;100(6):516–23. [2] Berghella V, Odibo AO, To MS, Rust OA, Althuisius SM. Cerclage for short cervix on ultrasonography: meta-analysis of trials using individual patient-level data. Obstet Gynecol 2005;106(1):181–9. [3] Althuisius SM, Dekker GA, Hummel P, van Geijn HP. Cervical incompetence prevention randomized cerclage trial: emergency cerclage with bed rest versus bed rest alone. Am J Obstet Gynecol 2003;189(4):907–10. [4] Daskalakis G, Papantoniou N, Mesogitis S, Antsaklis A. Management of cervical insufficiency and bulging fetal membranes. Obstet Gynecol 2006;107(2 Pt 1): 221–6.
[5] Debby A, Sadan O, Glezerman M, Golan A. Favorable outcome following emergency second trimester cerclage. Int J Gynecol Obstet 2007;96(1):16–9. [6] Novy MJ, Gupta A, Wothe DD, Gupta S, Kennedy KA, Gravett MG. Cervical cerclage in the second trimester of pregnancy: a historical cohort study. Am J Obstet Gynecol 2001;184(7):1447–54 discussion 1454-6. [7] Olatunbosun OA, al-Nuaim L, Turnell RW. Emergency cerclage compared with bed rest for advanced cervical dilatation in pregnancy. Int Surg 1995;80(2):170–4. [8] Pereira L, Cotter A, Gomez R, Berghella V, Prasertcharoensuk W, Rasanen J, et al. Expectant management compared with physical examination-indicated cerclage (EM-PEC) in selected women with a dilated cervix at 14(0/7)-25(6/7) weeks: results from the EM-PEC international cohort study. Am J Obstet Gynecol 2007;197(5):483 e1-8. [9] Stupin JH, David M, Siedentopf JP, Dudenhausen JW. Emergency cerclage versus bed rest for amniotic sac prolapse before 27 gestational weeks A retrospective, comparative study of 161 women. Eur J Obstet Gynecol Reprod Biol 2008;139: 32–7. [10] Nzeh DA, Adetoro OO. Sonographic assessment of the incompetent cervix in pregnancy. Int J Gynecol Obstet 1992;37(3):179–84. [11] Tsatsaris V, Senat MV, Gervaise A, Fernandez H. Balloon replacement of fetal membranes to facilitate emergency cervical cerclage. Obstet Gynecol 2001;98(2): 243–6. [12] Marret S, Ancel PY, Marpeau L, Marchand L, Pierrat V, Larroque B, et al. Neonatal and 5-year outcomes after birth at 30-34 weeks of gestation. Obstet Gynecol 2007;110(1):72–80. [13] Royston P, Ambler G, Sauerbrei W. The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol 1999;28(5):964–74. [14] McNutt LA, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol 2003;157(10):940–3. [15] Hvilsom GB, Thorsen P, Jeune B, Bakketeig LS. C-reactive protein: a serological marker for preterm delivery? Acta Obstet Gynecol Scand 2002;81(5):424–9. [16] Subtil D. Prognostic and therapeutic value of biologic signs of infection in the management of preterm labor (amniocentesis excepted). J Gynecol Obstet Biol Reprod 2002;31(7 Suppl):5S43-51. [17] Watts DH, Krohn MA, Wener MH, Eschenbach DA. C-reactive protein in normal pregnancy. Obstet Gynecol 1991;77(2):176–80. [18] Swets JA. Measuring the accuracy of diagnostic systems. Science 1988;240(4857): 1285–93. [19] 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. National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996;334(9):567–72. [20] Rozenberg P, Goffinet F, Malagrida L, Giudicelli Y, Perdu M, Houssin I, et al. Evaluating the risk of preterm delivery: a comparison of fetal fibronectin and transvaginal ultrasonographic measurement of cervical length. Am J Obstet Gynecol 1997;176(1 Pt 1):196–9. [21] Gomez R, Galasso M, Romero R, Mazor M, Sorokin Y, Goncalves L, et al. Ultrasonographic examination of the uterine cervix is better than cervical digital examination as a predictor of the likelihood of premature delivery in patients with preterm labor and intact membranes. Am J Obstet Gynecol 1994;171(4):956–64. [22] Hassan SS, Romero R, Berry SM, Dang K, Blackwell SC, Treadwell MC, et al. Patients with an ultrasonographic cervical length b or = 15 mm have nearly a 50% risk of early spontaneous preterm delivery. Am J Obstet Gynecol 2000;182(6):1458–67. [23] Owen J, Yost N, Berghella V, Thom E, Swain M, Dildy III GA, et al. Mid-trimester endovaginal sonography in women at high risk for spontaneous preterm birth. Jama 2001;286(11):1340–8. [24] Krebs-Jimenez J, Neubert AG. The microbiological effects of endovaginal sonographic assessment of cervical length. J Ultrasound Med 2002;21(7):727–9. [25] Royston P, Sauerbrei W. Multivariable model-building. A pragmatic approach to regression analysis based on fractional polynomials for modelling continuous variables. Wiley series in probability and statistics. Chichester: John Wiley & Sons; 2008.
Contents lists available at ScienceDirect
International Journal of Gynecology and Obstetrics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j g o
CLINICAL ARTICLE
Ultrasound cervical length measurement for prediction of delivery before 32 weeks in women with emergency cerclage for cervical insufficiency Florent Fuchs a,b,c,⁎, Jean Bouyer b,c,d, Hervé Fernandez a,b,c, Amélie Gervaise a, René Frydman c,e,f, Marie-Victoire Senat a,b,c a
Department of Obstetrics and Gynecology, Hôpital Bicêtre, Assistance Publique Hôpitaux de Paris (APHP), Le Kremlin-Bicêtre, France Reproduction and Child Development, CESP Centre for Research in Epidemiology and Population Health, U1018, Inserm, Villejuif, France Université Paris Sud 11, UMRS 1018, Villejuif, France d INED, Paris, France e INSERM U782, Université Paris-Sud, Clamart, France f Department of Obstetrics, Gynecology and Reproductive Medicine, Hôpital Antoine Béclère, Assistance Publique Hôpitaux de Paris (APHP), Clamart, France b c
a r t i c l e
i n f o
Article history: Received 2 February 2010 Received in revised form 7 April 2010 Accepted 4 May 2010 Keywords: Bulging membranes Midtrimester cervical dilation Preterm birth Threatened second trimester loss Ultrasound measurement
a b s t r a c t Objective: To assess the accuracy of ultrasound cervical length for predicting delivery before 32 weeks among women with a threatened second trimester fetal loss treated by emergency cervical cerclage, and to compare it with prediction based on clinically-assessed cervical dilation. Methods: Retrospective study (1997–2006) of 70 women with singleton pregnancies who underwent emergency cervical cerclage and ultrasound cervical length measurement between 15 and 26 weeks. The associations between cervical length and delivery before 32 weeks were studied with univariate and multivariate regression. ROC curves were used to determine the most discriminating cut-off value. Results: Although ultrasound cervical length measurement was significantly associated with early preterm delivery, its predictive accuracy was moderate with an area under the ROC curve (AUC) of 0.68 (95% CI, 0.54–0.82), which was similar (P = 0.43) to the AUC of cervical dilation of 0.73 (95% CI, 0.61–0.85). The best cut-off value was 7 mm, corresponding to a doubled risk of delivery before 32 weeks. Its sensitivity was 52%, specificity 82%, PPV 62%, and NPV 76%. It was no more informative than cervical dilation of 3 cm. Conclusion: Ultrasound cervical length measurement does not predict early preterm birth better than clinically-assessed cervical dilation in women with an emergency cerclage. © 2010 International Federation of Gynecology and Obstetrics. Published by Elsevier Ireland Ltd. All rights reserved.
1. Introduction In recent years, efforts to prevent preterm delivery have focused on women at high risk for preterm and early preterm birth, such as those with cervical insufficiency. The literature describes various methods of managing known cervical insufficiency such as elective prophylactic cerclage [1] or ultrasound-indicated therapeutic cerclage. The decision to use therapeutic cerclage is based on the measurement of cervical length during transvaginal ultrasound performed at 14–24 weeks [2]. However, the accuracy of this measurement for predicting preterm birth has never been analyzed among women treated with emergency cervical cerclage. Published studies dealing with emergency cerclage have described only the clinical factors associated with cerclage failure, defined as preterm birth [3–9]. The objective of the present study was to assess the predictive accuracy of ultrasound cervical length measurement for predicting ⁎ Corresponding author. Hôpital Bicêtre - Service de Gynécologie-Obstétrique, 78 rue du Général Leclerc, 94270 Le Kremlin-Bicêtre, France. Tel.: +34 1 45 21 77 84; fax: +34 1 45 21 77 25. E-mail address: fl[email protected] (F. Fuchs).
early preterm birth (delivery before 32 weeks) among women undergoing emergency cerclage. More precisely, we intended to: (1) quantify the association between cervical length and early preterm birth; (2) determine whether ultrasound cervical length measurement is useful for predicting early preterm birth, and especially, which cervical length cut-off value discriminates best; and (3) compare the predictive accuracy of this measurement with that of clinical assessment of cervical dilation. 2. Materials and methods This retrospective study examined the records of women who underwent cerclage at Antoine Béclère Hospital, Clamart, France—a tertiary care center—from January 1997 through December 2006. We included all women who underwent emergency cerclage for a singleton pregnancy at 15–26 weeks of pregnancy. Demographic, pregnancy, and neonatal data were collected, as well as obstetric history. In this maternity ward, emergency cerclage was systematically performed for pregnant women with a threatened second trimester fetal loss, defined as a cervix spontaneously dilated 1 cm or more on physical examination and prolapse of an unopened amniotic
0020-7292/$ – see front matter © 2010 International Federation of Gynecology and Obstetrics. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijgo.2010.04.022
246
F. Fuchs et al. / International Journal of Gynecology and Obstetrics 110 (2010) 245–248
sac at the external cervical os of the uterus or bulging in the vagina during speculum examination, and with no signs of labor (progressive dilation with contractions). Emergency cerclage was always performed on the basis of the physical digital examination and never according to the results of the ultrasound examination. Exclusion criteria were fetal anomalies, preterm premature rupture of membranes (PPROM), and clinical signs of chorioamnionitis (maternal temperature N38 °C, uterine tenderness or contractions, fetal tachycardia, and positive microbiological findings after testing either vaginal swab or urine samples). Midwives with sterile gloves digitally evaluated cervical dilation during vaginal examination and routinely looked for bulging membranes, at or beyond external os, during sterile speculum examination. Experienced sonographers measured cervical length during all ultrasound examinations according to standard techniques after the bladder was emptied [10]. Fundal pressure was also applied as a provocative maneuver. The cervix was measured at least 3 times, and the shortest measurement was recorded. The protruding membranes meant that the closed canal usually measured was no longer present; therefore, the length that was measured was the remaining cervix between the internal and external os. Even when membranes were bulging in the vagina, sonographers could always measure the remaining cervical length on a sagittal view (Fig. 1). The goal of this transvaginal ultrasound was to obtain a standardized and objective measurement of cervical length, even for women with advanced cervical dilation. All the women included were treated according to the same protocol: admission to the high-risk pregnancy department for inhospital bed rest for at least 7 days; blood tests; cerclage according to the McDonald procedure with a single stitch applied in the operating room under general or epidural anesthesia; antibiotics (1 g of amoxicillin with clavulanic acid 3 times a day for 5 days); indomethacin suppositories (100 mg twice a day for 2 days); close clinical and laboratory monitoring; and administration of antenatal steroids after reaching 24 weeks of pregnany. When necessary, bulging membranes were treated with a balloon inflation device, as described by Tsatsaris et al. [11], or an inflated Foley catheter n°22. The study outcome was early preterm birth, i.e. delivery before 32 weeks of pregnancy. This gestational age cut-off was chosen according to findings from the French EPIPAGE cohort study [12], as it appears to be associated with a significant increase in neonatal mortality and morbidity. To describe the association between delivery before 32 weeks and cervical length, we used fractional polynomial (FP) modeling [13], designed for continuous variables such as cervical
Fig. 1. Sagittal view of cervix with membranes bulging beyond external os.
length, and produced a graph of the curve that best fitted the relation between cervical length and early preterm birth. Receiver operating characteristic (ROC) curves were used to identify the best ultrasound cervical length cut-off for predicting early preterm birth. To compare the predictive value of ultrasound cervical length measurements and clinically-measured cervical dilation, we compared the area under these ROC curves. The association between delivery before 32 weeks and cervical length was then analyzed with univariate and multivariate logistic regression to obtain odds ratios and with Poisson regression to obtain relative risks [14]. In these regressions, cervical length was modeled dichotomously, according to the cut-off found by the ROC curve. In the multivariate analysis, we adjusted for known confounders identified in previous studies [3–9] and for variables associated with delivery before 32 weeks in the univariate analysis with P b 0.2: gravidity, history of spontaneous abortion, history of second-trimester losses, prophylactic cerclage, gestational age at cerclage, and biological infection, defined as C-reactive protein N15 mg/L or white blood count ≥13 600 × 106/L [15–17]. Because our population included 15 women whose emergency cerclage was performed late, between 24 and 26 weeks, we verified that our results were similar after excluding those 15 women. Statistical analyses were performed with STATA software version 10 (Stata Corp, College Station, TX, USA). We considered results to be significant when P b 0.05. This study is a retrospective analysis of a medical database that included ultrasound cervical length measures. This database and its use for research have been approved by the French Data Protection Authority (CNIL, Commission Nationale de l'Informatique et des Libertes) under the notification number 1181076. 3. Results The mean age of the patients was 32.3 years (range, 21–43 years). Most were nulliparous (53%), and 57% were at high risk of preterm birth because they had one or more of the following risk factors (these percentages exceed 57% because some women had several risk factors): history of one or more second-trimester losses (17%); one or more preterm births (14%); one or more surgically induced abortions (21%); in utero exposure to diethylstilboestrol (10%); prophylactic cerclage for this pregnancy (14%); and surgery for a uterine malformation (10%). Median gestational age at cerclage was 22.3 weeks (range, 15–26 weeks) with 30 cerclages (43%) performed before 22 weeks, 25 (36%) between 22–24 weeks, and 15 (21%) after 24 weeks. Twenty-five women (36%) had membranes bulging in the vagina, and the other 45 (64%) had membranes at the external os. Digital clinical measurements of cervical dilation were distributed as follows: 40% with 1 cm dilation, 27% with 2 cm, 19% with 3 cm, and 14% with 4 cm or more. Ultrasound cervical length measurements were distributed as follows: 6% had a cervical length of 25 mm or more, 27% 15–24 mm, and 67% less than 15 mm. Median gestational age at delivery was 36.9 weeks (interquartile range, 27.1–38.4 weeks; range, 17–41.4 weeks), and 25 women (36%) gave birth before 32 weeks. Of 70 neonates, 57 (81%) survived to discharge from hospital, and 47 (82%) had no observable neonatal morbidity. Fig. 2 describes the probability of delivery before 32 weeks according to cervical length, as modeled by fractional polynomials. This modeling showed that the overall relationship was statistically significant (P = 0.036) and did not differ from a linear relationship (P = 0.13). The ROC analysis presented in Fig. 3 showed that, according to the standard classification of such curves [18], ultrasound cervical length alone was a poor to fair predictor of delivery before 32 weeks (area under curve [AUC] 0.68; 95% CI, 0.54–0.82). The best cut-off value was 7 mm, which was associated with a sensitivity of 52% (95% CI,
F. Fuchs et al. / International Journal of Gynecology and Obstetrics 110 (2010) 245–248
247
length shorter than 7 mm was associated with a more than doubled risk of early preterm birth (relative risk 2.4; 95% CI, 1.3–4.6). Exclusion of the 15 women whose emergency cerclage was perormed late, between 24 and 26 weeks, did not significantly change these results. The AUC of ultrasound cervical length (0.71; 95% CI, 0.56–0.87) and that of digital cervical dilation (0.75; 95% CI, 0.62– 0.88) were still not significantly different (P = 0.57). This finding confirmed the stability of our results. 4. Discussion
Fig. 2. Probability of early preterm birth as a function of cervical length, measured by transvaginal ultrasound and modeled by fractional polynomials [25].
31%–73%), a specificity of 82% (95% CI, 71%–94%), a positive predictive value of 62% (95% CI, 39%–85%), and a negative predictive value of 76% (95% CI, 63%–88%). Comparison of the ROC curves for ultrasound cervical length and for clinically-assessed cervical dilation showed similar results (AUC 0.68 [95% CI, 0.54–0.82] vs 0.73 [95% CI, 0.61–0.85]; P = 0.43). In particular, cervical dilation, assessed clinically at 3 cm, yielded a prediction very similar to that of a cervical length of 7 mm. The sensitivity of this 3-cm value for clinical dilation was 56% (95% CI, 35%–77%), its specificity 80% (95% CI, 68%–92%), its positive predictive value 61% (95% CI, 39%–82%), and its negative predictive value 77% (95% CI, 64%–89%). The strength of the association between cervical length and early preterm birth was shown by the univariate analysis, where we observed a substantial increase in the risk of delivery before 32 weeks (compared with delivery beyond that gestational age) for a cervix shorter than 7 mm, with a crude odds ratio (OR) of 5.0 (95% CI, 1.7–14.9). After adjustment for known risk factors in multivariate logistic regression analysis, this OR increased still further (adjusted OR 7.8; 95% CI, 1.9–31.2). Poisson regression showed that a cervical
Fig. 3. Receiver operating characteristic (ROC) curves of early preterm birth for ultrasound cervical length measurement and clinical assessment of cervical dilation.
The study showed that in emergency situations such as threatened second-trimester fetal loss, very short cervical length measured by ultrasound before cerclage is significantly associated with early preterm birth. However, the predictive accuracy of this measurement for early preterm birth was no better than that of clinically measured cervical dilation. Numerous publications report the widespread use of transvaginal ultrasound measurement of cervical length to predict preterm birth, and it is well known to be an important prognostic tool [19,20]. Transvaginal ultrasound yields precise and reproducible values whereas digital examination of the cervix is subjective [21]. In the general population, a cervical length of less than 25 mm at 24 weeks is associated with an increased risk of preterm delivery, but because the prevalence of preterm birth is fairly low in the general population, its positive predictive value is only 18% [19]. Similarly, a cervical length of less than 15 mm at 24 weeks has a positive predictive value of 50% [22]. In contrast, for asymptomatic women at high risk of preterm birth because of their obstetric history, the positive predictive value of a cervical length of less than 25 mm is 55% [23]. As our population was also at very high risk of early preterm birth, with 36% of the women giving birth before 32 weeks, we considered it important to check the utility of ultrasound cervical length in this situation of threatened second-trimester fetal loss and to determine the appropriate cut-off. The association between cervical length and early preterm birth appears clearly in Fig. 2 and is confirmed in univariate and multivariate analyses. Whether a short cervical length, less than 7 mm, was considered before or after adjustment for known risk factors, the odds ratio was statistically significant and very high. However, because the prevalence of early preterm birth was high (36%) in our population, the odds ratios must not be interpreted as multiplicative factors. We thus estimated rate ratios with Poisson regression and found in the multivariate analysis that cervical shortening to less than 7 mm was associated with a more than doubled increase in the risk of early preterm birth (adjusted RR 2.4; 95% CI, 1.3–4.6). That measurement, although extremely short, seems logical as the cervix is dilated with bulging membranes in the cervix or in the vagina. It corresponds to the remaining cervical length should surgeons try to perform emergency cerclage. Despite this strong statistical association, ultrasound cervical length measurement was not a good predictor of early preterm birth. The ROC curve in Fig. 3 shows a poor to fair predictive ability, with an area under the curve of 0.68 (95% CI, 0.54–0.82), close to the 0.5 that corresponds to a predictive ability no better than random allocation. Comparing the areas under the ROC curves showed that the clinical and ultrasound predictions were very similar. Similarly, the sensitivity and specificity of ultrasound cervical length less than 7 mm and cervical dilation of 3 cm appeared very close. These predictive factors are associated with one other, as the remaining cervical length rarely exceeds 7 mm when the cervix is dilated by more than 3 cm. In this situation of a threatened pregnancy loss, we believe that ultrasound examination is relevant only when speculum examination shows cervical dilation greater than 3 cm, to reduce the potential induction of infection by digital examination [24]. Some characteristics of our study must be mentioned. We did not choose to exclude women who had undergone a prophylactic
248
F. Fuchs et al. / International Journal of Gynecology and Obstetrics 110 (2010) 245–248
cerclage, although their risk of preterm delivery might be higher than that of other women, because we sought to include every woman who met the criteria for emergency cerclage (defined as cervix dilation and membrane protrusion). Our study was intended to be a pragmatic one. It is also relevant that our hospital is a tertiary care center with experienced operators familiar with emergency cerclage practice. This factor may limit the extrapolation of our results to the general population. Another possible limitation of our study may be our inclusion of women up to a gestational age of 26 weeks. In recent years, emergency cerclage after 24 weeks has been rare, as indeed our study shows: most of the cerclages at 24–26 weeks were performed before 2002. However, as we have shown, excluding these women did not change our results. In view of the increased success in neonatal resuscitation for very preterm neonates, we believe that the practice of emergency cervical cerclage above 24 weeks should be considered and debated, but not routinely performed. Each woman was included based on clinical factors and not on ultrasound indicators. This mode of selection may explain why ultrasound and clinical digital examination appear equivalent. Indeed, it may have led to a bias in favor of the predictive value of clinical examination over ultrasound, because women with only an ultrasound-measured "short" cervix did not meet the inclusion criteria and would have been omitted from the study because they had no clinical changes of the cervix. Other studies should attempt to confirm our results about the predictive value of transvaginal ultrasound measurement of cervical length for outcome after emergency cerclage. Our results indicate that there is currently no evidence to recommend the use of such measurement for women presenting with a dilated cervix and visible membranes during speculum examination. Conflict of interest The authors have no conflicts of interest to disclose. References [1] Final report of the Medical Research Council/Royal College of Obstetricians and Gynaecologists multicentre randomised trial of cervical cerclage. MRC/RCOG Working Party on Cervical Cerclage. Br J Obstet Gynaecol 1993;100(6):516–23. [2] Berghella V, Odibo AO, To MS, Rust OA, Althuisius SM. Cerclage for short cervix on ultrasonography: meta-analysis of trials using individual patient-level data. Obstet Gynecol 2005;106(1):181–9. [3] Althuisius SM, Dekker GA, Hummel P, van Geijn HP. Cervical incompetence prevention randomized cerclage trial: emergency cerclage with bed rest versus bed rest alone. Am J Obstet Gynecol 2003;189(4):907–10. [4] Daskalakis G, Papantoniou N, Mesogitis S, Antsaklis A. Management of cervical insufficiency and bulging fetal membranes. Obstet Gynecol 2006;107(2 Pt 1): 221–6.
[5] Debby A, Sadan O, Glezerman M, Golan A. Favorable outcome following emergency second trimester cerclage. Int J Gynecol Obstet 2007;96(1):16–9. [6] Novy MJ, Gupta A, Wothe DD, Gupta S, Kennedy KA, Gravett MG. Cervical cerclage in the second trimester of pregnancy: a historical cohort study. Am J Obstet Gynecol 2001;184(7):1447–54 discussion 1454-6. [7] Olatunbosun OA, al-Nuaim L, Turnell RW. Emergency cerclage compared with bed rest for advanced cervical dilatation in pregnancy. Int Surg 1995;80(2):170–4. [8] Pereira L, Cotter A, Gomez R, Berghella V, Prasertcharoensuk W, Rasanen J, et al. Expectant management compared with physical examination-indicated cerclage (EM-PEC) in selected women with a dilated cervix at 14(0/7)-25(6/7) weeks: results from the EM-PEC international cohort study. Am J Obstet Gynecol 2007;197(5):483 e1-8. [9] Stupin JH, David M, Siedentopf JP, Dudenhausen JW. Emergency cerclage versus bed rest for amniotic sac prolapse before 27 gestational weeks A retrospective, comparative study of 161 women. Eur J Obstet Gynecol Reprod Biol 2008;139: 32–7. [10] Nzeh DA, Adetoro OO. Sonographic assessment of the incompetent cervix in pregnancy. Int J Gynecol Obstet 1992;37(3):179–84. [11] Tsatsaris V, Senat MV, Gervaise A, Fernandez H. Balloon replacement of fetal membranes to facilitate emergency cervical cerclage. Obstet Gynecol 2001;98(2): 243–6. [12] Marret S, Ancel PY, Marpeau L, Marchand L, Pierrat V, Larroque B, et al. Neonatal and 5-year outcomes after birth at 30-34 weeks of gestation. Obstet Gynecol 2007;110(1):72–80. [13] Royston P, Ambler G, Sauerbrei W. The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol 1999;28(5):964–74. [14] McNutt LA, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol 2003;157(10):940–3. [15] Hvilsom GB, Thorsen P, Jeune B, Bakketeig LS. C-reactive protein: a serological marker for preterm delivery? Acta Obstet Gynecol Scand 2002;81(5):424–9. [16] Subtil D. Prognostic and therapeutic value of biologic signs of infection in the management of preterm labor (amniocentesis excepted). J Gynecol Obstet Biol Reprod 2002;31(7 Suppl):5S43-51. [17] Watts DH, Krohn MA, Wener MH, Eschenbach DA. C-reactive protein in normal pregnancy. Obstet Gynecol 1991;77(2):176–80. [18] Swets JA. Measuring the accuracy of diagnostic systems. Science 1988;240(4857): 1285–93. [19] 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. National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996;334(9):567–72. [20] Rozenberg P, Goffinet F, Malagrida L, Giudicelli Y, Perdu M, Houssin I, et al. Evaluating the risk of preterm delivery: a comparison of fetal fibronectin and transvaginal ultrasonographic measurement of cervical length. Am J Obstet Gynecol 1997;176(1 Pt 1):196–9. [21] Gomez R, Galasso M, Romero R, Mazor M, Sorokin Y, Goncalves L, et al. Ultrasonographic examination of the uterine cervix is better than cervical digital examination as a predictor of the likelihood of premature delivery in patients with preterm labor and intact membranes. Am J Obstet Gynecol 1994;171(4):956–64. [22] Hassan SS, Romero R, Berry SM, Dang K, Blackwell SC, Treadwell MC, et al. Patients with an ultrasonographic cervical length b or = 15 mm have nearly a 50% risk of early spontaneous preterm delivery. Am J Obstet Gynecol 2000;182(6):1458–67. [23] Owen J, Yost N, Berghella V, Thom E, Swain M, Dildy III GA, et al. Mid-trimester endovaginal sonography in women at high risk for spontaneous preterm birth. Jama 2001;286(11):1340–8. [24] Krebs-Jimenez J, Neubert AG. The microbiological effects of endovaginal sonographic assessment of cervical length. J Ultrasound Med 2002;21(7):727–9. [25] Royston P, Sauerbrei W. Multivariable model-building. A pragmatic approach to regression analysis based on fractional polynomials for modelling continuous variables. Wiley series in probability and statistics. Chichester: John Wiley & Sons; 2008.