Increasing the effectiveness of external cephalic version

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Velzel, J.

Publication date2018Document VersionFinal published versionLicenseOther

Link to publication

Citation for published version (APA):Velzel, J. (2018). Increasing the effectiveness of external cephalic version.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an opencontent license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, pleaselet the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the materialinaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letterto: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. Youwill be contacted as soon as possible.

Download date:10 Sep 2022

Increasing the effectiveness of External Cephalic Version

Joost Velzel

The work described in this thesis was performed at the Amsterdam UMC, location AMC, The Netherlands.

Cover design and layout: © evelienjagtman.comPrinted by: Gilde Print

ISBN: 978-94-6323-399-6

Financial support for printing this thesis was kindly provided by: Amsterdam UMC, location AMC, and OLVG, Amsterdam

Increasing the effectiveness of External Cephalic Version

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctoraan de Universiteit van Amsterdamop gezag van de Rector Magnificus

prof. dr. ir. K.I.J. Maex ten overstaan van een door het College voor Promoties ingestelde commissie,

in het openbaar te verdedigen in de Agnietenkapelop woensdag 12 december 2018, te 12.00 uur

Joost Velzel

geboren te Purmerend

Promotiecommissie

Promotores: Prof. dr. J.A.M. van der Post AMC – Universiteit van AmsterdamProf. dr. B.W.J. Mol AMC – Universiteit van Amsterdam

Co-promotores: Dr. M. Kok AMC – Universiteit van AmsterdamDr. J.F.M. Molkenboer Spaarne Gasthuis

Overige leden: Prof. dr. P.M.M. Bossuyt AMC – Universiteit van AmsterdamDr. I.M. de Graaf AMC – Universiteit van AmsterdamProf. dr. C.J.M de Groot Vrije Universiteit AmsterdamProf. dr. E. Pajkrt AMC – Universiteit van AmsterdamProf. dr. V.J.M. Pop Tilburg University Prof. dr. F. Scheele Vrije Universiteit Amsterdam Faculteit der Geneeskunde

Voor mijn vader – trost zou hij zijn geweest

Contents

Chapter 1 General introduction 9

Chapter 2 A randomized controlled trial comparing atosiban and feno-terol as a uterine relaxant for external cephalic version (NTR 1877)

Chapter 3 A systematic review and network meta-analysis comparing tocolytics for external cephalic version

Chapter 4 Effectiveness of moxibustion in women with breech pre-sentation in combination with external cephalic version: a systematic review and meta-analysis

Chapter 5 Beta-mimetics for external cephalic version, a secondary analysis of the external cephalic version with uterine relax-ation trial to determine who benefits most

Chapter 6 Prediction models for successful external cephalic version: a systematic review

Chapter 7 Development and internal validation of a clinical prediction model for external cephalic version

Chapter 8 Mode of delivery after successful external cephalic version: a systematic review and meta-analysis

Chapter 9 Summary, general discussion and future perspectives 163

Appendices Nederlandse samenvattingList of co-authorsList of publications PhD PortfolioDankwoordAbout the author

179187189191193197

Chapter 1General introduction

General introduction | 11

1General introduction

The ultimate goal in obstetrical care is to achieve that a mother delivers a healthy baby at the end of her pregnancy and remains in optimal physical and mental condition. Women, when problems arise or are anticipated during pregnancy and delivery, require appropri-ate counseling in order to come to informed decision making to obtain the optimal fetal and maternal health goals. This thesis provides novel insights to improve individualized informed decision making for the prevention of breech presentation at term delivery. Breech presentation at term compared to vertex position with headfirst has potentially worse outcome for both mother and child.

Clinical management of breech presentation at term: vaginal birth or caesarean delivery?At term, in three to four per cent of all singleton pregnancies the fetus presents itself in breech position.1 In only 15% of the women with a breech presentation, an identifiable factor exists such as uterine malformation, placenta praevia, polyhydramnios, (congenital) abnormalities of the fetus like anencephaly and disorders accompanied by motor impair-ment that are reported to be associated with breech presentation.2 Compared to birth in cephalic position, vaginal breech delivery of healthy fetuses is associated with increased risk of neonatal birth trauma, low Apgar scores and neonatal mortality.3,4 The results of the Term Breech Trial, a randomized controlled trial (RCT) comparing the mode of delivery in term breech presentations, showed a clear reduction in overall risk of perinatal and neo-natal mortality and morbidity in favor of elective caesarean delivery compared to planned vaginal breech delivery (respectively 0.3% v 1.3%, RR 0.23, 95% CI 0.07 to 0.81 and 1.4% v 3.8%, RR 0.33, 95% CI 0.19 to 0.56). Publication had an unprecedented large impact on daily clinical practice even though results correspond to a number needed to treat of 100, which means performing 100 elective caesarean deliveries to prevent one perinatal or neonatal death.5 So gain in newborn health was judged superior to the accompanied increased risk for the mother because of the associated abdominal surgery. In addition, the 2-year follow up of this study demonstrated no difference in ‘neonatal death or neu-rodevelopmental delay’ (RR 1.09, 95% CI 0.52 to 2.3).6 The authors concluded that a policy of elective caesarean delivery is substantially better, especially in Western countries and in clinical practice this policy was adopted quickly. Although the results of this study seemed to confirm the presumption that an elective caesarean delivery could reduce neonatal morbidity and mortality, criticism on the trial followed.7,8 This critique mostly regarded methodological issues such as case selection, incomplete follow-up and questions on clini-cal competence of personnel responsible for breech delivery in the participating hospitals. The PREMODA study group conducted a prospective cohort study on breech delivery in 174 hospitals where breech delivery was still common practice.9 This study demonstrated

12 | Chapter 1

no significant differences in perinatal and neonatal mortality between both modes of delivery (OR 0.64, 95% CI 0.13 to 3.1). The authors concluded that a planned vaginal breech delivery remains a safe option and can be offered to women in hospitals when strict criteria are met before and during labor. However, a recent published meta-analysis, including cohort studies and RCTs, confirmed the findings of the Term Breech Trial.10 This study showed that in absolute risks, perinatal mortality is approximately 2.0/1000 with planned vaginal breech birth and 0.5/1000 with caesarean delivery. The authors however also pointed out the relatively low absolute risks of vaginal breech delivery leading to a high number needed to treat (approximately 67) with caesarean delivery having negative effects on mothers’ health and outcome of future pregnancies. They stated that their study underscored the importance of the practice of individualized decision making on the mode of delivery in a term breech presentation.

Even though there is an ongoing discussion in literature about the preferred mode of delivery, clinical practice indicates that clinicians’ and patients’ choice is unambiguous in breech presentation at term since after publication of the Term Breech Trial, caesarean delivery rates for breech delivery increased significantly worldwide. In the Netherlands, the rate of elective caesarean delivery for breech presentation in the years before this publication varied between 20 and 25%, and from 2001 onwards, it is more than 60%.11,12 In Sweden, the UK and Australia, this percentage shifted up to 80-95%.13–15 Nowadays, in Western countries, breech presentation is the third most common indication for caesar-ean delivery and is responsible of 15% of all caesarean deliveries.16

The ongoing debate on this topic is understandable, considering the maternal down-sides that come with caesarean delivery being the proposed safest route of delivery for the fetus. Women undergoing caesarean delivery are at higher risk of mortality and morbidity compared with women after vaginal delivery.17 A large population-based study from Canada found a threefold risk of short-term combined severe maternal morbidities for caesarean delivery as compared to vaginal delivery (2.7% v 0.9%).18 Serious mater-nal morbidity was found like hemorrhage requiring hysterectomy or transfusion, uterine rupture, anesthetic complications, assisted ventilation, venous thromboembolic event and major infection. Also, there were concerns regarding future pregnancies in the light of the increased incidence of abnormal placentation, which can lead to hemorrhage requiring hysterectomy, and rupture of the uterine scar, which is highly associated with perinatal death (6% to 10% of all uterine ruptures).19–23 Therefore, family planning is an important topic in counseling women with breech presentation near term as a study by Vlemmix et al (unpublished observation) demonstrates that elective caesarean delivery for term pregnancy leads to a significant increase in maternal morbidity, and adverse neonatal outcome in subsequent pregnancies.

1Since currently one in three women give birth by caesarean delivery in the United States, the American College of Obstetricians and Gynecologist published jointly with the Society of Maternal-Fetal Medicine an obstetric care consensus recommending prevention of the elective caesarean delivery.24 They conclude on an existence of global overuse of caesar-ean delivery because of the rapid increase of caesarean births in the last decade without the clear evidence of decreases in neonatal morbidity or mortality. Prevention of breech at term would solve the dilemma discussed in this paragraph.

Prevention of breech presentation at term: overviewAs fetal breech presentation is the reason for a caesarean delivery in approximately 15% of all cases16, prevention is an important issue. Though theoretically vaginal breech deliv-ery could reduce the number of caesarean deliveries and has to be discussed as an option with women, logically reducing the number of breech babies at term is a strategy to invest. Several methods to achieve this are described in literature and are based on different mechanisms. A preventive measure to reduce the incidence of breech presen-tation at term is postural management from 32 weeks of gestation onwards. This includes knee-chest position, and a supine position with the pelvis elevated with a wedge-shaped cushion for several minutes to one hour per day. However, a Cochrane review from 2012 concluded insufficient evidence to support the use of postural management as standard treatment.25 Moxibustion therapy is believed to increase fetal movement and could result in a spontaneous version to cephalic presentation from 32 weeks of gestation onwards. External cephalic version (ECV) is an obstetrical procedure in which a baby is manipulated by applying pressure externally through the gravid abdomen into a cephalic position. It requires training and manual skills. ECV is proven to be effective and reduces non-cephalic birth (pooled RR 0.42, 95% CI 0.29 to 0.61) and caesarean delivery (pooled RR 0.57, 95% CI 0.40 to 0.80).17 Sources from the time of Aristotle suggest that it has been practiced since the classical antiquity. In post medieval literature, ECV is first described in 1807. It became routine practice for breech presentation, and after a period of declined interests between 1960 and 1980, due to publications reporting on high rates of reversions and fetal complications, the revival of the use of ECV came in the early 1980’s.17 ECV is considered a safe procedure with few contraindications and can prevent breech delivery in 40-60% of cases.26 The complication rate is low, with reported serious perina-tal complications of 2.0/1000, including still-birth and placental abruption.27–30 In case of still-birth, there does not seem to be a relationship with ECV, as it equals the reported still-birth rate in the general population.31 It is important to compare perinatal risk associated with ECV with adverse outcome associated with the alternatives to ECV, namely perina-tal mortality in planned vaginal breech birth (2.0/1000) and elective caesarean delivery (0.5/1000).32 As the ultimate goal of ECV is to achieve vaginal cephalic birth, there remains,

14 | Chapter 1

however, controversy on the question whether the risk of caesarean delivery is increased for women with a fetus in cephalic position after a successful version compared with women with a spontaneous cephalic presentation. A review by Chan et al in 2004 found a two times increased risk for caesarean delivery in women after a successful external cephalic version.33 Since 2004 many studies on ECV were published, therefore an update on mode of delivery after successful ECV is needed and could shed more light on this issue.

At present, ECV is recommended to all women with an uncomplicated breech pregnancy near term.24 Furthermore, Tan et al demonstrated ECV to be cost-effective when the probability of successful ECV was greater than 32%.34 Then, the question remains: is there room for improvement?

There is still room for improvement of ECV both in the way it is executed and imple-mentedTo enhance success rates of ECV procedures, several strategies have been proposed. These methods include the use of tocolytic medication or epidural or spinal analgesia to increase uterine relaxation and amnioinfusion to increase fetal room to maneuver and methods not based on principles used in Western regular medicine like moxibustion therapy.35

Several studies have shown that uterine relaxation with tocolytic medication is effective.17 Types of tocolytics described in literature are nitric oxide, beta-mimetics, oxytocin recep-tor blockers and calcium channel blockers. The majority of studies that evaluated the effectiveness of tocolytics have used beta-mimetics and meta-analysis of randomized controlled trials demonstrated that the use of beta-mimetics can enhance the ECV success rate compared with placebo (nine studies, pooled RR 1.6, 95% CI 1.2 to 2.0) for cephalic presentation after ECV attempt.35 However, beta-mimetics have known adverse maternal cardiovascular side effects in terms of flushing, chest pain and palpitations, and as a result the implementation of routine uterine relaxation is low.36 Patients’ willingness to undergo ECV was evaluated in a vignette study by Vlemmix et al.37 This study showed that the use of intravenous uterus relaxants with side effects decreased the chance of opting for ECV. Therefore, evaluating alternative tocolytic medication could be useful. A possible tocolytic agent that has not been evaluated for ECV is an oxytocin receptor antagonist. This type of drug could be considered for ECV as its been used as uterine relaxant to delay or prevent preterm birth and has fewer maternal side effects compared to other tocolytics.38,39 To our knowledge, there are no randomized controlled trials assessing the effectiveness of an oxytocin receptor antagonist in ECV until now.

1In addition to planning this type of trials and in the absence of trials that compare all pharmacological interventions, a network meta-analysis could be helpful. This analysis allows to rank treatments by effectiveness in combining direct (available from standard pairwise meta-analysis) and indirect evidence. Even when the results of direct evidence are conclusive, combining them with the results of indirect evidence may contribute to a more refined and precise estimate of the outcome since it maximizes existing information within the network of treatment comparisons.40,41 Furthermore, since routine administration of tocolytic medication always carries a risk for unwanted side-effects, careful weighing who benefits most is needed in providing the best a priori chance of success.

Another used method to prevent breech presentation is the use of moxibustion, part of Complementary and Alternative Medicine (CAM). CAM is defined as non-mainstream prac-tices or products used as primary treatment or as a complement to mainstream health care.42 In healthy pregnant women, CAM treatments have been reported to help lower stress, improve sleep, and reduce pain, and some care providers regularly refer pregnant patients to yoga classes, massage therapists, and other CAM practitioners, though this is found only in low impact literature.43 Notwithstanding this lack of solid evidence, a survey among pregnant women and physicians revealed that up to 49% of women consult a CAM practitioner, of which only half discloses this to their physician.44 In addition, breech presentation is one of the most common topics in pregnancy to consult a CAM practi-tioner for treatment with moxibustion therapy alone or in combination with acupuncture to establish cephalic presentation.

Moxibustion therapy is a traditional Chinese medical intervention that uses the heat gener-ated by burning herbal preparations containing Artemisia vulgaris (mugwort) to stimulate the acupoint BL67 (beside the outer corner of the fifth toenail). The exact mechanism cannot be understood with Western physiological principles, however, it is thought that this method increases fetal activity and can be applied to correct breech presentation from 32 weeks of gestation onwards.45 A recent published Cochrane review in 2012 found limited evidence to support the beneficial effect of moxibustion, applied alone or in combi-nation with acupuncture or postural measures, compared to observation alone or postural measures.46 However, this review did not distinguish between studies offering ECV or not. As moxibustion in practice is used from 32 weeks which is prior to a possible ECV attempt, executed from 36 weeks of gestation, potentially both could be recommended provided a beneficial effect is present. We decided that there is a need to determine the effectiveness of moxibustion as a complementary treatment for women eligible for an ECV attempt in order to be able to answer questions from clients in daily practice.

16 | Chapter 1

Implementation of ECV in clinical practice: barriers and facilitatorsConsidering the safety and effectiveness of the ECV procedure, all eligible women should be offered this treatment option. However, a study on the implementation of ECV in the Neth-erlands in 2014 showed that the implementation rate varied between 8% and 84%, with an average of 72%.47 In the United States, the implementation of ECV is even lower. A nationwide cohort study showed that only 5% of the women with breech presentation had a successful version. Assuming an average ECV success rate of 40% would imply that only 13% received an ECV attempt.48 These low implementation rates are both doctor and patient related. Among professionals the main barriers are a lack of knowledge and skills to inform and counsel patients.49 For patients the success rate and expected pain are the most important factors influencing the willingness to opt for ECV.37 Higher success rates of vaginal delivery after suc-cessful ECV increased women’s willingness (OR 3.4, 95% CI 2.0 to 5.7). As the success rate of ECV varies from approximately 35% up to 86% in the literature with an average of 50–60%17, an a priori chance of success estimate can be helpful to counsel women for an ECV.

Previous studies have shown that clinical and ultrasound characteristics are associated with success or failure of an ECV procedure.50,51 The clinical factors multiparity, non engage-ment of the breech, a relaxed uterus and a palpable fetal head were associated with successful ECV. Ultrasound factors predictive for successful ECV were posterior placenta localization, complete breech position and an amniotic fluid index >10 cm. In literature, there are several prediction models that enable individualized prediction of the outcome of an ECV attempt. For implementing in clinical practice, it is important to assess their quality by determining which predictors they used and determining their performance. Accurate and individualized prediction of the outcome of an ECV attempt is helpful for optimal and accurate shared decision making.

Objective and outline of the thesisThe aims of the presented studies in this thesis address the following research questions:• What is the effectiveness of atosiban as a uterine relaxant for ECV compared to

fenoterol?• Which tocolytic agents are reported to be the most effective for ECV?• Is moxibustion alone or in combination with acupuncture to correct breech presen-

tation in combination with ECV effective? • Can we determine who benefits most of tocolysis with beta-mimetics for ECV?• Which prediction model performs best in predicting successful ECV?• Can we develop a clinical prediction model for ECV that performs better in predicting

successful version and beneficial neonatal outcome compared to existing models? • Is there an increased risk for caesarean section after external cephalic version result-

ing in cephalic presentation compared to spontaneous cephalic presentation?

1In chapter 2 we report on the results of a randomized controlled trial assessing the effectiveness of the oxytocin receptor antagonist atosiban as a uterine relaxant for ECV compared to the beta-mimetic fenoterol.

In chapter 3 we describe the results of a network meta-analysis that was performed to identify the most effective tocolytic agent for uterine relaxation for ECV.

In chapter 4 we systematically review the medical literature reporting on the use of mox-ibustion alone or in combination with acupuncture to correct breech presentation in combination with ECV.

In chapter 5 we report on a secondary analysis of a randomized controlled trial to deter-mine who benefits most of tocolysis with beta-mimetics for ECV.

In chapter 6 we demonstrate the results of a systematic review of medical literature on prediction models for successful ECV.

In chapter 7 we present a model for the prediction of successful ECV based on clinical variables determined alongside a randomized controlled trial.

In chapter 8 we demonstrate the results of a systematic review of medical literature on the mode of delivery after ECV.

In chapter 9 we discuss the results of the studies presented in this thesis and give clinical implications and implications for future research in this field.

18 | Chapter 1

Reference

1. Hickok, D. E., Gordon, D. C., Milberg, J. A., Williams, M. A. & Daling, J. R. The frequency of breech presentation by gestational age at birth: a large population-based study. Am. J. Obstet. Gynecol. 166, 851–2 (1992).

2. Rayl, J., Gibson, J. & Hickok, D. A popu-lation-based case-control study of risk factors for breech presentation. Am. J. Obstet. Gynecol. 174, 28–32 (1996).

3. Rietberg, C. C. T. et al. Term breech pre-sentation in The Netherlands from 1995 to 1999: Mortality and morbidity in relation to the mode of delivery of 33,824 infants. Br. J. Obstet. Gynaecol. 110, 604–609 (2003).

4. Bjellmo, S. et al. Is vaginal breech delivery associated with higher risk for perinatal death and cerebral palsy compared with vaginal cephalic birth? Registry-based cohort study in Norway. BMJ Open 7, (2017).

5. Hannah, M. E. et al. Planned caesarean section versus planned vaginal birth for breech presentation at term: a randomised multicentre trial. Term Breech Trial Collab-orative Group. Lancet 356, 1375–83 (2000).

6. Whyte, H. et al. Outcomes of children at 2 years after planned cesarean birth versus planned vaginal birth for breech presenta-tion at term: The international randomized Term Breech Trial. Am. J. Obstet. Gynecol. 191, 864–871 (2004).

7. Glezerman, M. Five years to the term breech trial: The rise and fall of a random-ized controlled trial. Am. J. Obstet. Gynecol. 194, 20–25 (2006).

8. Goffinet, F., Azria, E., Kayem, G., Schmitz, T. & Deneux-Tharaux, C. Re: The risks of planned vaginal breech delivery versus planned caesarean section for term breech birth: A meta-analysis including observational studies: Let’s avoid simplis-tic radicalism when reality is complex. Br. J. Obstet. Gynaecol. 123, 145–147 (2016).

9. Goffinet, F. et al. Is planned vaginal deliv-ery for breech presentation at term still an option? Results of an observational pro-spective survey in France and Belgium. Am. J. Obstet. Gynecol. 194, 1002–1011 (2006).

10. Berhan, Y. & Haileamlak, A. The risks of planned vaginal breech delivery versus planned caesarean section for term breech birth: A meta-analysis including observa-tional studies. Br. J. Obstet. Gynaecol. 123, 49–57 (2016).

11. Molkenboer, J. F. M., Bouckaert, P. X. J. M. & Roumen, F. J. M. E. Recent trends in breech delivery in the Netherlands. Br. J. Obstet. Gynaecol. 110, 948–51 (2003).

12. Rietberg, C. C. T., Elferink-Stinkens, P. M. & Visser, G. H. A. The effect of the Term Breech Trial on medical intervention behaviour and neonatal outcome in The Netherlands: an analysis of 35,453 term breech infants. Br. J. Obstet. Gynaecol. 112, 205–9 (2005).

13. Alexandersson, O., Bixo, M. & Högberg, U. Evidence-based changes in term breech delivery practice in Sweden. Acta Obstet. Gynecol. Scand. 84, 584–587 (2005).

14. Bragg, F. et al. Variation in rates of cae-sarean section among English NHS trusts after accounting for maternal and clinical risk: cross sectional study. BMJ 341, c5065 (2010).

15. Sullivan, E. A., Moran, K. & Chapman, M. Term breech singletons and caesarean section: A population study, Australia 1991-2005: Original Article. Aust. New Zeal. J. Obstet. Gynaecol. 49, 456–460 (2009).

16. Barber, E. L. et al. Indications contributing to the increasing cesarean delivery rate. Obstet. Gynecol. 118, 29–38 (2011).

17. Hofmeyr, G. J., Kulier, R. & West, H. M. External cephalic version for breech pre-sentation at term. Cochrane Database Syst. Rev. 4, (2015).

118. Liu, S. et al. Maternal mortality and severe

morbidity associated with low-risk planned cesarean delivery versus planned vaginal delivery at term. Can. Med. Assoc. J. 176, 455–460 (2007).

19. Silver, R. M. et al. Maternal morbidity associated with multiple repeat cesarean deliveries. Obstet. Gynecol. 107, 1226–1232 (2006).

20. Marshall, N. E., Fu, R. & Guise, J. M. Impact of multiple cesarean deliveries on mater-nal morbidity: A systematic review. Am. J. Obstet. Gynecol. 205, (2011).

21. Solheim, K. N. et al. The effect of cesarean delivery rates on the future incidence of placenta previa, placenta accreta, and maternal mortality. J. Matern. Neonatal Med. 24, 1341–1346 (2011).

22. Guise, J. M. et al. Vaginal birth after Cesar-ean. Obstet. Gynecol. 115, 1267–78 (2010).

23. Gregory, K. D., Jackson, S., Korst, L. & Frid-man, M. Cesarean versus vaginal delivery: Whose risks? Whose benefits? Am. J. Peri-natol. 29, 7–18 (2012).

24. ACOG & SMFM. Safe Prevention of the Pri-mary Cesarean Delivery. Obstet. Gynecol. 123, 693–711 (2014).

25. Hofmeyr, G. J. & Kulier, R. Cephalic version by postural management for breech presenta-tion. Cochrane Database Syst. Rev. (2012).

26. Rosman, A. N. et al. Contraindications for external cephalic version in breech position at term: A systematic review. Acta Obstet. Gynecol. Scand. 92, 137–142 (2013).

27. Collaris, R. J. & Oei, S. G. External cephalic version: a safe procedure? A systematic review of version-related risks. Acta Obstet. Gynecol. Scand. 83, 511–8 (2004).

28. Nassar, N. et al. Systematic review of adverse outcomes of external cephalic version and persisting breech presenta-tion at term. Paediatr. Perinat. Epidemiol. 20, 163–71 (2006).

29. Collins, S., Ellaway, P., Harrington, D., Pandit, M. & Impey, L. W. M. The complications of external cephalic version: results from 805

consecutive attempts. Br. J. Obstet. Gynae-col. 114, 636–8 (2007).

30. Rijnders, M. et al. A retrospective study of the success, safety and effectiveness of external cephalic version without tocoly-sis in a specialised midwifery centre in the Netherlands. Midwifery 24, 38–45 (2008).

31. Vilchez, G. et al. Risk of Expectant Man-agement and Optimal Timing of Delivery in Low-Risk Term Pregnancies: A Popu-lation-Based Study. Am. J. Perinatol. 35, 262–270 (2018).

32. No, G. G. Management of Breech Pre-sentation: Green-top Guideline No. 20b Management of Breech Presentation : Green-top Guideline No. 20b. Br. J. Obstet. Gynaecol. 151–177 (2017).

33. Chan, L. Y. et al. Intrapartum cesarean delivery after successful external cephalic version: A meta-analysis. Obstet. Gynecol. 104, 155–160 (2004).

34. Tan, J. M., Macario, A., Carvalho, B., Druzin, M. L. & El-Sayed, Y. Y. Cost-effectiveness of external cephalic version for term breech presentation. BMC Pregnancy Childbirth 10, 3 (2010).

35. Cluver, C., Gyte, G. M. L., Sinclair, M., Dow-swell, T. & Hofmeyr, G. J. Interventions for helping to turn term breech babies to head first presentation when using external cephalic version. Cochrane database Syst. Rev. 2, CD000184 (2015).

36. Feitsma A.H., Middeldorp J.M., O. D. De uitwendige versie bij de a terme stuit. Nederlands tijdschrift voor Obstetrie & Gynaecologie (2010).

37. Vlemmix, F. et al. Patient’s willingness to opt for external cephalic version. J. Psychosom. Obstet. Gynaecol. 34, 15–21 (2013).

38. Flenady, V., Reinebrant, H. E., Liley, H. G., Tambimuttu, E. G. & Papatsonis, D. N. M. Oxytocin receptor antagonists for inhibit-ing preterm labour. Cochrane Database Syst. Rev. 6 (2014).

39. Van Vliet, E. O. G. et al. Nifedipine versus atosiban for threatened preterm birth

20 | Chapter 1

(APOSTEL III): A multicentre, randomised controlled trial. Lancet 387, 2117–2124 (2016).

40. Jansen, J. P. et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: Report of the ISPOR task force on indirect treatment comparisons good research practices: Part 1. Value Heal. 14, 417–428 (2011).

41. Hoaglin, D. C. et al. Conducting indi-rect - t reatment -compar ison and network-meta-analysis studies: Report of the ISPOR task force on indirect treatment comparisons good research practices: Part 2. Value Heal. 14, 429–437 (2011).

42. National Center for Complementary and Integrative Health. Complementary, Alter-native, or Integrative Health: What’s In a Name? NCCIH 1–6 (2015).

43. Wang, S. M. et al. Complementary and alternative medicine for low-back pain in pregnancy: a cross-sectional survey. J. Altern. Complement. Med. 11, 459–464 (2005).

44. Steel, A. et al. Utilisation of complementary and alternative medicine (CAM) practi-tioners within maternity care provision: results from a nationally representative cohort study of 1,835 pregnant women. BMC Pregnancy Childbirth 12, 146 (2012).

45. Coyle, M. E., Smith, C. a & Peat, B. Cephalic

version by moxibustion for breech pre-sentation. Cochrane Database Syst. Rev. (2005).

46. Vas, J. et al. Using moxibustion in primary healthcare to correct non-vertex presenta-tion: a multicentre randomised controlled trial. Acupunct. Med. 31, 31–8 (2013).

47. Vlemmix, F. et al. Implementation of exter-nal cephalic version in the Netherlands: a retrospective cohort study. Birth 41, 323–9 (2014).

48. Weiniger, C. F. et al. Maternal outcomes of term breech presentation delivery: Impact of successful external cephalic version in a nationwide sample of delivery admissions in the United States. BMC Pregnancy Child-birth 16, 1–7 (2016).

49. Rosman, A. N. et al. Facilitators and barri-ers to external cephalic version for breech presentation at term among health care providers in the Netherlands: A quantita-tive analysis. Midwifery 30, e145-50 (2014).

50. Kok, M. et al. Clinical factors to predict the outcome of external cephalic version: a metaanalysis. Am. J. Obstet. Gynecol. 199, 630.e1-7; discussion e1-5 (2008).

51. Kok, M., Cnossen, J., Gravendeel, L., Van Der Post, J. A. & Mol, B. W. Ultrasound factors to predict the outcome of external cephalic version: a meta-analysis. Ultrasound Obstet. Gynecol. 33, 76–84 (2009).

Chapter 2A randomized controlled trial comparing atosiban and fenoterol as a uterine relaxant for external cephalic version (NTR 1877)

Joost Velzel, Floortje VlemmixBrent C. OpmeerJan F. M. MolkenboerCorine J. VerhoevenMariëlle G. van PampusDimitri N. M. PapatsonisJoke M. J. BaisKarlijn C. VollebregtLiesbeth van der EschJoris A. M. van der PostBen Willem MolMarjolein Kok

British Medical Journal, January 2017; 26;356:i6773.

24 | Chapter 2

Abstract

Objective: We compared the effectiveness of atosiban and fenoterol as a uterine relaxant in women undergoing ECV for breech presentation.

Design: A multicenter, open label, randomized controlled trial.

Setting: Eight hospitals in the Netherlands.

Participants: 830 women with a singleton fetus in breech presentation and a gestational age beyond 34 weeks were randomly allocated in a 1:1 ratio to either 6.75 mg atosiban (n=416) or 40 μg fenoterol (n=414) for uterine relaxation.

Main outcome measures: The primary outcome measures were a fetus in cephalic posi-tion 30 minutes after the procedure and cephalic presentation at delivery. Secondary outcome measures were mode of delivery, neonatal outcome and adverse events during ECV. All analyses were done on an intention-to-treat basis.

Results: Cephalic position 30 minutes after ECV occurred significantly less in the atosiban group than in the fenoterol group (34% v 40%, RR 0.73, 95% CI 0.55 to 0.93). Presenta-tion at birth was cephalic in 35% (n=139) of the atosiban group and 40% (n=166) of the fenoterol group (RR 0.86, 95% CI 0.72 to 1.03), and caesarean delivery was performed in 60% (n=240) of women in the atosiban group and 55% (n=218) in the fenoterol group (RR 1.09, 95% CI 0.96 to 1.20). No significant differences were found in neonatal outcomes or drug related adverse events.

Conclusions: In women undergoing ECV for breech presentation, uterus relaxation with fenoterol increases the cephalic rate immediately after the procedure. No statistically significant difference was found for cephalic presentation at delivery.

Trial registration: Dutch Trial Register, NTR 1877.

RCT comparing atosiban and fenoterol as uterine relaxant for ECV | 25

Introduction

Breech presentation occurs in up to 4% of pregnancies at term. Since publication of the Term Breech Trial in 2000, elective caesarean delivery has been the dominant mode of delivery in most countries.1–3 In 2015 the World Health Organization reported that globally caesarean delivery was overused.4,5 Breech presentation is the third most common indi-cation for elective caesarean delivery.6 External cephalic version (ECV) is a safe obstetrical procedure that reduces non-cephalic birth and caesarean delivery by approximately 50%.7 In 2014 the American College of Obstetricians and Gynaecologists and the Society for Maternal-Fetal Medicine published a joint consensus statement on the safe prevention of elective caesarean delivery, in which ECV is highly recommended.6 This is especially relevant in low and middle income countries. recommendation is specifically relevant in low and middle income countries where the impact of caesarean delivery on morbidity and mortality is more severe.8

Several methods have been proposed to enhance the outcome of ECV, including uterine relaxation with tocolytic medication, epidural or spinal analgesia, and amnio-infusion, as well as complementary methods such as vibro-acoustic stimulation, acupuncture and moxibustion.9 Drug induced uterine relaxation using nitric oxide, beta mimetics, calcium channel antagonists, or oxytocin antagonists, can increase the success rate of ECV.6,10 Most studies evaluating the effectiveness of tocolysis have used beta mimetics (nine studies of beta mimetics versus placebo, pooled RR 1.6 (95% CI 1.2 to 2.0) for cephalic presentation after ECV attempt).9 However, beta-mimetics have known adverse maternal cardiovascular side effects such as flushing and palpitations.9

Randomized trial data show that nifedipine, a calcium channel blocker, is not effective for increasing cephalic presentation after ECV compared with no drugs (RR 1.1, 95% CI 0.85 to 1.5). 11 Atosiban, an oxytocin receptor antagonist, has no cardiovascular side effects and is becoming more widely used for ECV, yet no randomized controlled trial has assessed its effectiveness. We compared the effectiveness of atosiban with the beta mimetic fenoterol for achieving a cephalic presentation 30 minutes after ECV and cephalic presentation at delivery.

26 | Chapter 2

Methods

Study Design and participantsWe performed a multicenter, open label, randomized controlled trial in one academic and seven teaching hospitals in The Netherlands. Together, these hospitals are responsible for 16 000 increased risk deliveries a year, with annual hospital delivery rates ranging from 1500 to 3000. We followed the CONSORT guidelines to report this study.

Women with a singleton fetus in breech position who were scheduled for ECV were eligible for the study. Exclusion criteria were maternal age less than 18 years , gestational age less than 32 weeks, any contraindication to vaginal birth (such as placenta praevia), any contra-indication for ECV according to the Guideline of the Dutch Association for Obstetrics and Gynecology (scarred uterus other than transverse in the lower segment, known uterine anomalies, history or signs of placental abruption, severe pre-eclampsia or HELLP syn-drome, bleeding less than seven days before ECV attempt, and ruptured membranes),12 any known contraindication to one of the two study drugs, suspected intrauterine growth restriction (defined as estimated fetal weight less than the fifth centile for gestational age assessed by ultrasonography), severe oligohydramnios (deepest pool <2 cm), fetal anom-alies, or non-reassuring fetal heart rate. Midwives, residents, or gynecologists identified eligible women. After counselling and reading the patient information form, patients were asked for written informed consent.

RandomizationThe women were randomly allocated (1:1) to receive atosiban (intervention group) or fenoterol (control group). An independent data manager assigned the women to groups based on a computer generated random sequence, stratified by hospital and parity. EVC was scheduled within seven days after randomization. Participants and investigators were aware of allocation; blinding was not possible owing to obvious maternal side effects that commonly occur with fenoterol, such as tachycardia, dizziness, and flushing.

InterventionsIn each hospital, a team of experienced obstetricians and midwives performed the ECV. A trained sonographer, obstetrician, or midwife carried out ultrasonography to assess the position of the fetus, including type of breech (frank, complete, or footling), location of the placenta, estimation of fetal weight, and estimation of amniotic fluid volume, including measurement of the largest pocket depth. Fetal wellbeing was established with electronic fetal heart rate monitoring for at least 30 minutes before and after ECV. Fifteen minutes before ECV a doctor gave the mother an intravenous bolus of atosiban (6.75 mg in 0.9 mL (7.5 mg/mL)) or fenoterol (40 μg in 0.8 mL (0.5 mg/10 mL)). ECV could comprise a forward

and a backward roll. Fetal heart rate was monitored intermittently with ultrasonography. The duration of any fetal bradycardia was registered. Women with non-sensitized Rh neg-ative blood received anti-D immunoglobulin (1000 IU intramuscularly) after ECV.

OutcomesThe primary outcome measures were cephalic presentation 30 minutes after the proce-dure, confirmed by ultrasound, and cephalic presentation at delivery. Secondary outcomes were mode of delivery, and complications of ECV events due to atosiban or fenoterol. Complications of ECV were persistent non-reassuring fetal CTG after ECV, occult or overt umbilical cord prolapse, placental abruption and emergency delivery. Adverse events due to atosiban or fenoterol were defined as chest pain, nausea, vomiting, headaches, flushing, dizziness, hypotension (associated with fainting or CTG abnormalities), tachycardia result-ing in palpitations, local reaction of the skin on injection of the medication, anaphylactic shock, cessation of treatment due to side effects.

Post hoc outcomes were gestational age at delivery, time to delivery, admission to neona-tal intensive care for >24 hours, Apgar <7 at 5 minutes, birth weight, blood loss, women requiring blood transfusion, maternal admission in hospital postpartum in days, maternal complication postpartum defined as puerperal fever, (suspected) endometritis, mastitis, operation for placental rest, pulmonary embolism. Appendix table 1 lists all outcomes.

The data were collected on web based electronic case record forms, and uploaded cases were stored in a database. Paper delivery forms were supplied to participants who did not deliver in a participating hospital and their primary caretakers, such as midwives in pri-mary-care settings, were asked to ensure that these forms were returned after the study.

Sample sizeThe trial was designed to detect a 10% improvement in the primary outcome of cephalic presentation 30 minutes after ECV, assuming a 50% success rate in the fenoterol group (b error 20%, a error 5%, two sided test). We calculated that 806 women needed to be randomized to show an improvement from 50% to 60% with atosiban.

AnalysisTo ensure intention-to-treat analysis, we carried out five imputations on the primary out-come using baseline covariates. The primary analysis was subsequently performed on imputed datasets, and we pooled estimates from these datasets using Rubin’s rule. Com-plete case analysis was used to analyze secondary outcomes. Baseline data were analyzed descriptively for the women and their pregnancies. The c2 test was used to compare the rates of the primary and secondary outcomes between the two groups. We considered a

28 | Chapter 2

P value of less than 0.05 to indicate statistical significance. For each outcome we calculated relative risks and appropriate 95% confidence intervals (CI).

We performed three post hoc sensitivity analyses for the primary outcome of cephalic presentation 30 minutes after ECV retrospectively. The first was a complete case analysis. The second sensitivity analysis was performed because of an imbalance in baseline charac-teristics, despite randomization. We adjusted the imputed data for age, parity, gestational age at ECV, ethnicity, body mass index, estimated fetal weight, type of breech presentation, location of the placenta, and estimated amniotic fluid index. A third post analysis was performed to evaluate the robustness of the treatment effect on cephalic presentation 30 minutes after ECV between centers. The results were pooled into relative risks, and heterogeneity was explored in a random effects meta-analysis model. An interim analysis was planned half way through inclusion to evaluate safety. Serious adverse events were reported to an independent data safety monitoring board. They noted no conditions to stop the trial.

Patient involvementNo patients were involved in setting the research question or the outcome measures, nor were they involved in developing plans for design or implementation of the study. No patients were asked to advise on interpretation or writing up of results. Information will be used in a guideline and will be added to patient information brochures.

Results

Between August 2009 and May 2014, 830 women were enrolled and randomized; 416 to atosiban (intervention group) and 414 to fenoterol (control group). The mean gestational age at which ECV was performed was 36 weeks, and 62% of the women were nulliparous. Baseline characteristics were comparable between the two groups (table 1). For the com-plete case analysis, primary outcome data were available for 410 women in the atosiban group and 408 in the fenoterol group (Figure 1).

Assessed for eligibility (n=910)

Allocated to atosiban (n=416)

Received allocated intervention (n=416)

Did not receive allocated intervention (n=0)

Lost to follow up (n=6)

Randomized (n=830)

Excluded (n=80)

Not meeting inclusion criteria (n=18)

Declined to participate (n=58)

Declined ECV (n=4)

Allocated to fenoterol (n=414)

Intention to treat analysis (n=416)

Complete-case analysis (n=410)

Intention to treat analysis (n=414)

Complete-case analysis (n=408)

Figure 1. Randomization, treatment and follow-up of participants

The primary outcome of cephalic presentation 30 minutes after ECV occurred in 34% (n=140) of women in the atosiban group compared with 40% (n=166) in the fenoterol group (RR 0.73, 95% CI 0.55 to 0.93). At delivery, 35% (n=139) of fetuses in the atosiban group and 40% (n=160) in the fenoterol group were in the cephalic position (RR 0.86, 95% CI 0.72 to 1.03). Caesarean delivery was performed in 60% (n=240) of women in the ato-siban group and 55% (n=218) in the fenoterol group (RR 1.09, 95% CI 0.96 to 1.22; Table 2).

30 | Chapter 2

Table 1. Baseline characteristics of study participants by intervention group for complete case analysis. Values are numbers (percentages) unless stated otherwise

Atosiban (n=410) Fenoterol (n=408)

Mean (SD) age (years) 32.1 (4.0) 32.4 (4.3)

Multiparous women 154 (38) 153 (27)

Mean (SD) gestational age at ECV (weeks) 35.8 (0.9) 35.9 (1.0)

White ethnicity 350 (85.4) 355 (87.0)

Mean (SD) body mass index 24.1 (4.3) 24.2 (4.8)

Mean (SD) estimated fetal weight 1 2622 (391.0) 2572 (457.5)

Frank breech presentation 2 300 (76) 273 (69.8)

Anterior placental 3 127 (33) 144 (37.3)

Mean (SD) estimated amniotic fluid index 4 13.8 (4.9) 13.7 (4.7)

ECV = external cephalic version.1 Missing data: n=234 for atosiban, n=250 for fenoterol. 2 Missing data: n=392 for atosiban, n=391 for fenoterol. 3 Missing data: n=386 for atosiban, n=386 for fenoterol. 4 Missing data: n=263 for atosiban, n=281 for fenoterol.

Table 2. Results for primary and secondary outcomes in intention-to-treat analysis

Atosiban (n=416)

Fenoterol (n=414)

Relative risk (95% CI)

Cephalic presentation 30 minutes after ECV (No; %) 1 140 (34) 166 (40) 0.73 (0.55 to 0.93)

Cephalic presentation at delivery 2 (No; %) 139 (35) 160 (40) 0.86 (0.72 to 1.03)

Mode of delivery 3

Vaginal delivery (No; %) 163 (40) 180 (45) 0.89 (0.76 to 1.05)

Spontaneous 146 167

Instrumental 17 13

Caesarean delivery (No; %) 240 (60) 218 (55) 1.09 (0.96 to 1.22)

Elective 199 158

No progress of labor 27 26

Suspected fetal distress 4 21

Other 10 13

ECV = external cephalic version.1 Imputation for primary outcome. 2 Missing data: n=402 for atosiban, n=397 for fenoterol. 3 Missing data: n=403 for atosiban, n=398 for fenoterol.

In the atosiban group, of the 139 women with a fetus in the cephalic position at delivery, 109 (81%) had a spontaneous vaginal delivery, 15 (11%) an instrumental delivery, and 9 (6%) an intrapartum caesarean delivery. In the fenoterol group, of the 160 women with a fetus in the cephalic position at delivery, 125 (78%) had a spontaneous vaginal delivery, 13 (8%) an instrumental delivery, and 19 (12%) an intrapartum caesarean delivery. Emer-gency caesarean delivery for suspected fetal distress was more common in the fenoterol group: two cases immediately after ECV and 19 during labor. In the atosiban group, four of the 204 women who had planned a vaginal birth had emergency caesarean delivery for suspected intrapartum fetal distress compared with 19 of 218 participants in the fenoterol group (RR 0.75, 95% CI 0.24 to 1.29).

The frequency of poor fetal and maternal complications did not differ between the two groups (table 3). The average time between ECV and delivery was 22 days in both groups. Labor was induced only for additional indications such as hypertension. No perinatal or neo-natal mortality occurred in the atosiban group, whereas two children died in the fenoterol group. One neonate, born five weeks after ECV, died of an autosomal recessive congenital disorder. The other baby died two hours after a vacuum assisted delivery for prolonged second stage of labor. ECV had been performed five weeks before labor. The baby had normal Apgar scores (9 at five minutes), and autopsy did not reveal cause of death.

The frequency of fetal and maternal complications after ECV did not differ significantly between groups (table 4). No woman required emergency delivery in the atosiban group but two did in the fenoterol group. In one woman, the fetal heart rate was non-reassuring after an ECV attempt, and in the other woman, ultrasonography showed the umbilical cord to be lying before the fetal head as presenting part after successful ECV. No drug related adverse events were reported in the atosiban group. The procedure was stopped in one woman in the fenoterol group because she experienced severe hypotension. A significant difference in side effects was found between the groups, with 15 women (5%) in the atosiban group compared with 209 women (71%) in the fenoterol group experiencing palpitations, (RR 0.07, 95% CI 0.04 to 0.12).

Complete case analysis showed a significant difference in favor of fenoterol for cephalic presentation 30 minutes after ECV (relative risk 0.82, 95% CI 0.69 to 0.98). After adjusting for baseline characteristics the odds ratio for successful ECV was 0.85 (95% CI 0.72 to 1.02). The pooled result of the post hoc analysis indicated a consistent treatment effect and no heterogeneity was identified among centers (RR 0.82, 95% CI 0.69 to 0.98), I2 0%) (see supplementary appendix 1).

32 | Chapter 2

Table 3. Neonatal and maternal outcomes in complete case analysis. Values are numbers (per-centages) unless stated otherwise

Atosiban (n=410)

Fenoterol (n=408)

P value

Mean (SD) gestational age at delivery (weeks

38.9 (1.3) 38.9 (1.9) - 0.70

Mean (SD) time to delivery in days (mean; SD)

22 (9.1) 22 (9.6) - 0.23

Fetal mortality 0 (0.0) 0 (0.0) - -

Neonatal mortality 0 (0.0) 2 (0.1) - 0.15

Admission to neonatal intensive care for >24 hours (days) 1

16 (0.4) 17 (0.4) 0.94 (0.48 to 1.8) -

Apgar < 7 at 5 minutes 2 6 (1) 13 (3) 0.45 (0.17 to 1.20) -

Mean (SD) birth weight (g) 3 3356 (460) 3364 (523) - 0.81

Female 204 (50) 204 (50) 1.00 (0.87 to 1.14) -

Mean (SD) blood loss (ml) 4 474 (473.2) 470 (447.3) - 0.92

Women requiring blood transfusions 5 6 (1) 7 (2) 0.86 (0.29 to 2.5) -

Mean (SD) maternal admission in hospital postpartum (days) 6

290 (3.1) 287 (3.1) - 0.94

Maternal postpartum complications 7 15 (4) 16 (4) 0.94 (0.74 to 1.9) -1 Missing data: n=399 for atosiban, n=397 for fenoterol. 2 Missing data: n=399 for atosiban, n=397 for fenoterol. 3 Missing data: n=394 for atosiban, n=396 for fenoterol. 4 Missing data: n=390 for atosiban, n=389 for fenoterol. 5 Missing data: n=400 for atosiban, n=401 for fenoterol. 6 Missing data: n=398 for atosiban, n=395 for fenoterol.7 Missing data: n=385 for atosiban, n=386 for feno-terol. Postpartum complication: puerperal fever, (suspected) endometritis, mastitis, operation for placental rest, pulmonary embolism.

Table 4. Complications of external cephalic version (ECV) and drug related adverse events in com-plete case analysis. Values are numbers (percentages) unless stated otherwise

Atosiban (n=410)

Fenoterol (n=408)

Non-reassuring CTG registration after ECV attempt resulting in emergency delivery 1

0 1 (0.2) -

Placental abruption 2 1 0 -

Emergency delivery 3 0 2 (0.4) -

Adverse events due to medication 4 0 1 (0.2) -

Cessation of treatment due to side effects 5 0 1 (0.2) -

Minor side effects 6 86 (30.0) 224 (75.7) 0.40 (0.33 to 0.48)

Palpitations 7 15 (5) 209 (71) 0.07 (0.04 to 0.12)

Dizziness 8 25 (9) 55 (19) 0.47 (0.30 to 0.73)

Flushes 9 17 (6) 99 (34) 0.18 (0.11 to 0.29)

Minor side effects are defined as one or a combination of the following complaints: palpitations, nausea, vomiting, headaches, flushing, dizziness.1 Missing data: n=402 for atosiban, n=399 for fenoterol. 2 Missing data: n=392 for atosiban, n=390 for fenoterol. 4 Missing data: n=391 for atosiban n=391 for fenoterol. 5 Missing data: n=399 for ato-siban, n=396 for fenoterol. 6 Missing data: n=399 for atosiban, n=396 for fenoterol. 7 Missing data: n=287 for atosiban, n=296 for fenoterol. 8 Missing data: n=287 for atosiban, n=296 for fenoterol. 9 Missing data: n=285 for atosiban, n=292 for fenoterol.

34 | Chapter 2

Discussion

In this randomized controlled trial of 818 women with singleton pregnancies of more than 34 weeks’ gestation, uterine relaxation with atosiban for ECV resulted in a lower rate of fetuses in the cephalic position after 30 minutes compared with fenoterol. This led to a higher risk of caesarean deliveries after atosiban treatment than after fenoterol treatment. Although the difference was not statistically significant, it is highly plausible given the lower rate of successful ECV with atosiban.

Strengths and limitations of this studyA beta-mimetic rather than placebo was used for the control because at the start of the trial it was the best treatment in a clinical setting, with a 10% absolute increase in cephalic presentations at delivery.9 To allow the drugs to reach a full therapeutic blood level and to overcome unintended protocol violations, the protocol for both drug administrations were equalized; the ECV attempt was started 15 minutes after administration of one of either drug. The mean time for both drugs to reach their maximum concentrations in serum is 30 minutes, with a half life of 1.4 hours for atosiban and 2 hours for fenoterol.13–17

A limitation of our study is that doctors and patients could not be blinded to treatment owing to the obvious and common side effects of beta mimetics. However, side effects were never a reason for stopping the intervention, and bias due to non-blinding was virtually absent.

Our success rate for cephalic presentations at 30 minutes after ECV was lower than anticipated,9 which might be due to the referral of some women after an initial unsuccessful attempt by independent midwives in an out of hospital setting. A cohort study of ECV attempts by midwives outside the hospital setting in the Netherlands reported a higher success (41%) rate than ours.18 The success rate in our study corresponds to that in other cohort studies and randomized controlled trials comparing tocolytic agents.10,11,19,20 Our sample size calculation was still adequate to detect the anticipated increase in cephalic presentations at 30 minutes after ECV, as lower or higher baseline rates have more sta-tistical power to detect similar changes of 10% absolute risk difference.

Even though there was a difference in effect size between the imputed analysis and complete case analysis, the results are still likely to be generalizable. Missing data were balanced between groups and assumed to be random. We included a wide range of both positive and negative predictors for successful ECV in our imputation model. Correction for confounding from potential baseline imbalance showed a strong correlation between baseline characteristics and successful ECV. Uterine relaxation might not be the strongest effector influencing the success rate of ECV.

After adjustment for potential baseline imbalance, the relative risk increased from 0.73 to 0.85, indicating a smaller magnitude of effect. However, the majority of the confidence intervals still favored fenoterol.

Comparison with other studiesA Cochrane review on uterine relaxants for ECV concluded that, compared with placebo, beta mimetics were the most effective drugs for increasing cephalic presentation in labor and thereby reducing the caesarean section rate (six studies, 742 participants, RR 0.77, 95% CI 0.67 to 0.88).9 Our results indicate that beta mimetics are the most effective in achieving cephalic presentation after ECV. Moreover fenoterol is much cheaper than ato-siban (€0.90 (£0.80; $1.00) versus € 31.80 in the Netherlands).

Our study clearly shows that fenoterol has more side effects than atosiban (71% v 5.2%). Discomfort due to palpitations was the most reported side effect, and severe hypotension occurred in one participant after taking fenoterol. Women must be adequately coun-selled about the side effects in relation to the effectiveness of beta mimetics, as evidence shows that women are willing to undergo treatment if the gain in success outweighs the possible adverse side effects.21 In 2011 one in three deliveries in the United States was a caesarean delivery, making caesarean delivery the most common major surgical proce-dure for women in the country.22,23 Without clear evidence of an accompanying decrease in maternal and neonatal morbidity or mortality, caesarean delivery might be overused. Therefore, prevention of planned caesarean section is one of the main ways to improve maternal and neonatal outcome in current obstetric practice.6,24 As breech presentation is the third most common indication for primary caesarean delivery in the United States, and ECV is proven to be a safe procedure, implementation of beta-mimetics as routine uterine relaxant in ECV should be high priorty.25–28

Conclusions

Fenoterol is more effective than atosiban for uterine relaxation before ECV. Fenoterol use is recommended to increase the success rate of ECV and decrease the caesarean delivery rate.

36 | Chapter 2

References

4. Human Reproduction Programme. WHO Statement on Caesarean Section Rates. 1–8 (2015).

5. Gibbons, L. et al. Inequities in the use of cesarean section deliveries in the world. Am. J. Obstet. Gynecol. 206, 331.e1-331.e19 (2012).

7. Hofmeyr, G. J., Kulier, R. & West, H. M. External cephalic version for breech pre-sentation at term. Cochrane Database Syst. Rev. 4 (2015).

8. Mishra, M. & Sinha, P. Does caesarean sec-tion provide the best outcome for mother and baby in breech presentation? A per-spective from the developing world. J. of Obstet. Gynaecol. 31, 495–498 (2011).

9. Cluver, C., Gyte, G. M. L., Sinclair, M., Dow-swell, T. & Hofmeyr, G. J. Interventions for helping to turn term breech babies to head first presentation when using external cephalic version. Cochrane Database Syst. Rev. 2 (2015).

10. Hofmeyr, G. J. Interventions to help external cephalic version for breech presentation at term. Cochrane Database Syst. Rev. (2004).

11. Kok, M. et al. Nifedipine as a uterine relaxant for external cephalic version: a randomized controlled trial. Obstet. Gyne-col. 112, 271–6 (2008).

12. Obstetrische Werkgroep. Stuitligging. Guidel. Dutch Coll. Obstet. Gynaecol. (2008).

13. Lundin, S., Broeders, A. & Melin, P. Pharma-cokinetic properties of the tocolytic agent [Mpa1, D-Tyr(Et)2, Thr4, Orn8]-oxytocin (antocin) in healthy volunteers. Clin. Endo-crinol. (Oxf). 39, 369–74 (1993).

14. Akerlund, M., Hauksson, A., Lundin, S., Melin, P. & Trojnar, J. Vasotocin analogues which competitively inhibit vasopres-sin stimulated uterine activity in healthy women. Br. J. Obstet. Gynaecol. 93, 22–7 (1986).

15. Lippert, T. H., Peters, F. D. & Kidess, E. Dose-response study of fenoterol on uterine activity in labor at term. Gynecol. Obstet. Invest. 11, 219–24 (1980).

16. Rasmussen, B. B., Larsen, L. S. & Sendero-vitz, T. Pharmacokinetic interaction studies of atosiban with labetalol or betametha-sone in healthy female volunteers. Br. J. Obstet. Gynaecol. 112, 1492–9 (2005).

17. Reinheimer, T. M. et al. Barusiban, a new highly potent and long-acting oxytocin antagonist: pharmacokinetic and pharma-codynamic comparison with atosiban in a cynomolgus monkey model of preterm labor. J. Clin. Endocrinol. Metab. 90, 2275–81 (2005).

18. Beuckens, a et al. An observational study of the success and complications of 2546 external cephalic versions in low-risk pregnant women performed by trained midwives. Br. J. Obstet. Gynaecol. (2015).

19. Collins, S., Ellaway, P., Harrington, D., Pandit, M. & Impey, L. W. M. The complications of external cephalic version: results from 805 consecutive attempts. Br. J. Obstet. Gynae-col. 114, 636–8 (2007).

20. Vlemmix, F. et al. Implementation of client versus care-provider strategies to improve external cephalic version rates: a cluster randomized controlled trial. Acta Obstet. Gynecol. Scand. 94, 518–26 (2015).

22. Hamilton, B. E., Hoyert, D. L., Martin, J. A., Strobino, D. M. & Guyer, B. Annual summary of vital statistics: 2010-2011. Pediatrics 131, 548–58 (2013).

23. MacDorman, M., Declercq, E. & Menacker, F. Recent Trends and Patterns in Cesarean and Vaginal Birth After Cesarean (VBAC)

Deliveries in the United States. Clin. Peri-natol. 38, 179–192 (2011).

24. Gregory, K. D., Jackson, S., Korst, L. & Frid-man, M. Cesarean versus vaginal delivery: Whose risks? whose benefits? Am. J. Peri-natol. 29, 7–18 (2012).

25. ACOG. ACOG Committee Opinion No. 340. Mode of term singleton breech delivery. ACOG Guidel. No 340 108, 235–7 (2006).

26. Barber, E. L. et al. Contributing Indica-tions to the Rising Cesarean Delivery Rate. Obstet. Gynecol. 118, 29–38 (2011).

28. Grootscholten, K., Kok, M., Oei, S. G., Mol, B. W. J. & van der Post, J. A. External cephalic version-related risks: a meta-analysis. Obstet. Gynecol. 112, 1143–51 (2008).

38 | Chapter 2

Supplementary material

Appendix table 1. All outcomes

Primary:

cephalic presentation 30 minutes after the procedure

cephalic presentation at delivery

Secondary:

mode of delivery

complications of ECV

adverse events due to atosiban or fenoterol

Post hoc:

gestational age at delivery

time to delivery

admission to neonatal intensive care for >24 hours

Apgar score <7 at 5 minutes

birth weight

blood loss

women requiring blood transfusion

maternal admission in hospital postpartum in days

maternal complication postpartum

Chapter 3A systematic review and network meta-analysis comparing tocolytics for external cephalic version

Joost VelzelFloortje VlemmixDanielle D. EgbertsJacqueline C.E.M. LimpensJan F.M. MolkenboerJoris A. M. van der PostBen Willem MolMarjolein KokEwoud Schuit

Submitted

42 | Chapter 3

Abstract

Objective: To determine the most effective tocolytic agent to enhance the effectiveness of external cephalic version (ECV).

Data source: We searched MEDLINE, ClinicalTrials.gov, Embase, Central, Cochrane Dare and NSHEED and PubMed Publisher from inception until January 2018 for randomized controlled trials of tocolytic therapy for ECV in women with singleton pregnancies in breech position after 34 weeks of gestation.

Methods of study selection: Two reviewers independently selected studies, assessed risk of bias and extracted data. The primary outcome measure was a fetus in cephalic position after ECV. Secondary outcomes were caesarean delivery and vaginal cephalic delivery. A random effects network meta-analysis was performed.

Tabulation, integration, and results: We identified 505 titles, of which 18 randomized controlled trials assessed uterine relaxants for ECV. We found the following comparisons: beta-mimetics versus placebo (nine studies, n=894), nitric oxide versus placebo (three studies, n=282), calcium channel blockers versus placebo (one study, n=310), beta-mimet-ics versus nitric oxide (two studies, n=133), beta-mimetics versus calcium channel blockers (two studies, n=176), and beta-mimetics versus oxytocin receptor blockers (one study, n=830). Compared with placebo, beta-mimetics was the only tocolytic agent to increase the chance of a fetus in cephalic position after the procedure (RR 1.6, 95% CI 1.28 to 2.0) and increase in vaginal cephalic delivery (RR 1.6, 95% CI 1.15 to 2.3). Compared to placebo, beta-mimetics and oxytocin receptor blockers showed to decrease caesarean delivery (respectively RR 0.78, 95% CI 0.70 to 0.86 and RR 0.85, 95% CI 0.72 to 0.99). Beta-mimetics had the highest probability of being ranked the best treatment option for ECV success (71%), caesarean delivery (70%) and cephalic vaginal delivery (54%).

Conclusion: Beta-mimetics had the highest probability to enhance the effectiveness of ECV as indicated by an increased ECV success rate and a reduced caesarean delivery rate.

Systematic review registration: PROSPERO CDR42014008720

Systematic review and network meta-analysis comparing tocolytics for ECV | 43

Introduction

Breech presentation occurs in 3 to 4% of all term singleton pregnancies. Since publication of the Term Breech Trial, elective caesarean delivery is the dominant mode of delivery for breech presentation and makes breech presentation the third most common indication for elective caesarean delivery in the United States.1,2 External cephalic version (ECV) aims to reduce the number of breech presentations at term and a recent meta-analysis showed ECV can be considered as a safe obstetrical procedure that reduces both non-cephalic birth and caesarean delivery by approximately 40-60%.3

Several methods have been proposed to enhance the outcome of ECV, including uterine relaxation, epidural or spinal analgesia and amnioinfusion as well as complementary meth-ods such as vibro-acoustic stimulation, acupuncture and moxibustion.4 Of these methods, drug induced uterine relaxation is the most promising to increase the success of ECV.4 The types of drugs used are nitric oxide, beta-mimetics, calcium channel antagonists and oxytocin antagonists. Each tocolytic has unique pathways and mechanisms, side effects, and different ways to administer.3

In randomized controlled trials (RCTs), individual tocolytic agents have been compared with placebo or other tocolytic agents. Although a recent Cochrane review demonstrated that beta-mimetics, compared to placebo, are the most effective in reducing non-cephalic birth (pooled RR 1.7, 95% CI 1.14 to 2.5) and caesarean delivery (pooled RR 0.77, 95% CI 0.67 to 0.88)4, this meta-analysis did not assess the comparative effectiveness of all available treatment options.

In the absence of robustly designed trials that compare all interventions of interest, an indirect comparison can provide useful evidence of the comparative effectiveness of com-peting interventions, especially those that have not been compared directly in randomized trials. A so-called network meta-analysis allows us to combine both direct and indirect evidence, and to rank treatments by comparative effectiveness. Even when the results of the direct evidence are conclusive, combining them with the results of indirect evidence in a network meta-analysis may contribute to a more refined and precise estimate of the outcome because it links and maximizes existing information within the network of treatment comparisons.5,6

We have systematically reviewed and analyzed RCTs on tocolytics for ECV and carried out a network meta-analysis to determine the most effective agent for enhancing the effectiveness of ECV.

44 | Chapter 3

Methods

Our study protocol was registered in PROSPERO (CDR42014008720). Our meta-analysis is reported in accordance with the PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions and the extension of that statement for network meta-analysis.7,8

SourcesA medical librarian conducted a systematic search in OVID MEDLINE, OVID Embase, the Cochrane Central Register of Controlled Trials (CENTRAL, DARE, NSHEED and the clinical trial register ClinicalTrials.gov) from inception to January 2018 to identify RCTs on the tocolytic treatment of ECV. To this end we searched for controlled terms (i.e. MESH) and free text terms for (1) breech or fetal version and (2) tocolyse or tocolytic agents, includ-ing specific tocolytic drugs. In MEDLINE and EMBASE this search was combined with a search-filter adapted from the Cochrane to identify RCTs. No language or date restrictions were applied. We cross-checked the reference lists and the citing articles of the identified relevant papers and adapted the search in case of additional relevant studies. The bib-liographic records retrieved were imported and de-duplicated in ENDNOTE. The complete search strategies are presented in appendix 1.

Study selectionWe included randomized trials comparing tocolytics to another tocolytic, placebo or no treatment in women with singleton pregnancies with a fetus in breech position from 34 weeks of gestation onwards. Two independent reviewers (JV and DE) first assessed eligibility of identified studies by title and abstract, and then by full text assessment. Dis-agreements was resolved by consensus and, if necessary, by consulting a third reviewer (MK).

Risk of bias was assessed by two authors (JV and DE) using the risk of bias tool devel-oped by the Cochrane Collaboration, which contains specific items that assess random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias. Each domain was scored as low, high or unclear.9 A third author (MK) reviewed studies in which there was any disagreement about the risk of bias assess-ment.

Both reviewers used a standardized data extraction form to extract data on baseline characteristics (maternal age, maternal BMI, parity, gestational age, type of breech pre-sentation, mean estimated fetal weight, mean amniotic fluid index, placental location),

maternal adverse events as major side effects due to medication (e.g. hypotension, aller-gic reaction, anaphylactic shock, cessation of treatment), neonatal adverse events due to ECV (e.g. persisting non-reassuring fetal heart rate monitoring which necessitated an emergency caesarean delivery, placental abruption and fetal death), and the primary and secondary outcomes described below. When data or results were unclear or missing for the paper, the corresponding author of the trial was contacted for further information. Again, disagreements were resolved by consensus and, if necessary, by a third reviewer (MK).

OutcomesThe primary effectiveness outcome was successful ECV defined as a fetus in cephalic presentation immediately after ECV attempt. Secondary outcomes were mode of delivery (caesarean delivery and vaginal cephalic delivery) and incidence of fetal and maternal complications, and study drug related adverse events. Fetal complications were defined as persistent as non-reassuring fetal heart rate on cardiotocography (CTG) leading to emergency caesarean delivery, placental abruption, fetal and neonatal death, admission to neonatal intensive care, Apgar score less than 7 at five minutes. Maternal complica-tions were defined as the need of cessation of treatment because of side effects due to medication.

InterventionsWe distinguished five intervention classes: placebo, beta-mimetics (ritodrine, terbutaline, nylidrin, salbutamol, fenoterol, hexoprenaline, isoxsuprine), calcium channel blockers (nifedipine, nicardipine), nitric oxide (nitroglycerine, nitric oxide) and oxytocin receptor blockers (atosiban, barusiban).

Statistical analysisAll analyses were conducted within a frequentist framework and according to the inten-tion-to-treat principle. We conducted a network meta-analysis using a multivariate random effects meta-analysis to simultaneously compare the different tocolytic therapies for each outcome. Where head-to-head data were available we also performed random effects pair-wise meta-analyses, which were compared with the network meta-analysis estimates to assess consistency. An important assumption underpinning network meta-analysis is that there is consistency between the direct and indirect evidence10. This may not be the case if there is an imbalance in effect-modifiers across treatment comparisons. We there-fore pre-specified potential key effect modifiers i.e. maternal age, maternal BMI, parity, gestational age, type of breech presentation, mean estimated fetal weight, mean amniotic fluid index, placental location. We assessed whether they were similarly distributed across trials, through cross-tabulations and if necessary assessed the impact of these on the level

46 | Chapter 3

of the trial by adding these as covariates to a meta-regression model.11 Consistency was explored statistically assuming comparison-specific heterogeneity estimates. If the result-ing relative odds ratio (ROR) were higher than 2 inconsistency would be considered to be present. Relative treatment effects were reported as a risk ratios (RR) and 95% Confidence Intervals (CI) and Prediction Intervals (PrI).

We calculated the probability of each treatment being the most or least effective for each outcome. As this can be an unstable metric, we also reported the rank for each treatment, with 1 being the best and 5 being the worst treatment. Statistical analyses were performed using STATA/SE 13.1 for Windows.

Results

Of the 505 titles identified, 35 potentially eligible studies were assessed full text and 18 met the inclusion criteria (Figure 1). The characteristics of the included studies are outlined in Table 1 and all baseline characteristics of the included studies are presented in a separate table in the supplementary materials.

Records identified in

MEDLINE (n=222)

EMBASE (n=454)

CENTRAL (n=53)Sc

Records after duplicates removed(n=505)

Records excluded(n=470)

Full-text articles assessed for eligibility(n=35)

Studies included in systematic review(n=18)

Full-text articles excluded, with reasons (n = 17)

3 systematic reviews; 5 prospective cohort studies; 3 trials comparing ECV to no ECV; 2 abstracts for conference; 3 retrospective cohorts; 1 prospective

cohort study compared to historic cohort

Cochrane CDSR, Dare DARE &

NHSEED (n=5)

ClinicalTrials.gov (n=24)

Figure 1. PRISMA flow chart of search results and article selection

48 | Chapter 3

Table 1. Baseline characteristics of included studies

1st author Publication year Country Study design Sample size Overall Quality Index treatment Control treatment

Robertson20 1987 USA RCT 58 Low Ritodrine hydrochloride infusion 200 μg/minute for 20 mins

Tan21 1989 Singapore RCT 90 Low Salbutamol 4 mg orally 3 times a day for at least 1 day.

Stock22 1993 Hong Kong RCT 63 Low Ritodrine 0.3 mg per min infusion for 30 mins and during the procedure, and placebo bolus injection. Or placebo infusion and hexoprenaline 10 micrograms bolus injection.

Placebo infusion and bolus injection

Chung23 1996 Hong Kong RCT (doctor blinded)

50 High Ritodrine 0.4 mg/ml in 5% dextrose at 1.5 ml/min via an infusion pump,for 15 minutes before and during ECV attempt.

Placebo in matching infusion of 5% dextrose via in infusion pump, 15 minutes before and during ECV attempt

Marquette24 1996 Canada RCT (double-blinded)

283 Low Ritodrine infusion 1.5 ml vials with a ritodrine concentration of 10 mg/ml (111 μg/min) in 20 minutes before attempt.

Placebo in matching infusion and in same rate.

Fernandez25 1997 USA RCT (doctor blinded)

103 High Terbutaline 0.25 mg subcutaneously in unlabeled insulin syringe 15-30 minutes before ECV attempt

Placebo equal volume of unlabeled insulin syringe 15-30 minutes before ECV attempt

Yanny26 2000 UK RCT (double-blinded)

57 Low Glyceryl trinitrate sublingual spray 800 μg

Placebo spray

Bujold27 2003 Canada RCT 99 High Nitroglycerine 2 sublingual sprays of 400 mg 3 minutes before ECV attempt.

Placebo 2 sublingual sprays of placebo 3 minutes before ECV attempt.

Bujold28 2003 Canada RCT (double-blinded)

74 High Intravenous placebo for 20 minutes with sublingual nitroglycerine 0.8 mg 3 minutes before ECV attempt.

Ritodrine intravenously (111 μg/min) for 20 minutes with sublingual placebo 3 minutes before ECV attempt.

El-Sayed29 2004 USA RCT 59 Low Nitroglycerin intravenous 100 μg immediately before and after beginning ECV

Terbutaline subcutaneous injection 0.25 mg 5 minutes before ECV

Impey30 2005 UK RCT (double-blinded)

124 High Ritodrine hydrochloride (Yutopar) infusion of 50 mg (10 mg/ml) added to 12 ml dextrose saline via a syringe pump, starting at 1 ml/hour and increasing at 10-minute intervals to a maximum of 5 ml /hour.

Nor Azlin31 2005 Malaysia RCT (double-blinded)

60 High Ritodrine inusion (0.4 mg/ml in 5% dextrose) via a syringe pump at the rate of 1.5 ml/min, beginning 15 minutes before and continuing throughout the procedure

Collaris32 2008 Malaysia RCT (double-blinded)

90 High Nifedipine 10 mg tablet and intravenous placebo 0.9% sodium chloride and 1 ml syringe

Terbutaline sulphate (500 micrograms/ml) together and 1 ml syringe and a placebo tablet

Table 1. Baseline characteristics of included studies

Robertson20 1987 USA RCT 58 Low Ritodrine hydrochloride infusion 200 μg/minute for 20 mins

Tan21 1989 Singapore RCT 90 Low Salbutamol 4 mg orally 3 times a day for at least 1 day.

Stock22 1993 Hong Kong RCT 63 Low Ritodrine 0.3 mg per min infusion for 30 mins and during the procedure, and placebo bolus injection. Or placebo infusion and hexoprenaline 10 micrograms bolus injection.

Placebo infusion and bolus injection

Chung23 1996 Hong Kong RCT (doctor blinded)

50 High Ritodrine 0.4 mg/ml in 5% dextrose at 1.5 ml/min via an infusion pump,for 15 minutes before and during ECV attempt.

Placebo in matching infusion of 5% dextrose via in infusion pump, 15 minutes before and during ECV attempt

Marquette24 1996 Canada RCT (double-blinded)

283 Low Ritodrine infusion 1.5 ml vials with a ritodrine concentration of 10 mg/ml (111 μg/min) in 20 minutes before attempt.

Fernandez25 1997 USA RCT (doctor blinded)

103 High Terbutaline 0.25 mg subcutaneously in unlabeled insulin syringe 15-30 minutes before ECV attempt

Placebo equal volume of unlabeled insulin syringe 15-30 minutes before ECV attempt

Yanny26 2000 UK RCT (double-blinded)

57 Low Glyceryl trinitrate sublingual spray 800 μg

Placebo spray

Bujold27 2003 Canada RCT 99 High Nitroglycerine 2 sublingual sprays of 400 mg 3 minutes before ECV attempt.

Placebo 2 sublingual sprays of placebo 3 minutes before ECV attempt.

Bujold28 2003 Canada RCT (double-blinded)

74 High Intravenous placebo for 20 minutes with sublingual nitroglycerine 0.8 mg 3 minutes before ECV attempt.

Ritodrine intravenously (111 μg/min) for 20 minutes with sublingual placebo 3 minutes before ECV attempt.

El-Sayed29 2004 USA RCT 59 Low Nitroglycerin intravenous 100 μg immediately before and after beginning ECV

Terbutaline subcutaneous injection 0.25 mg 5 minutes before ECV

Impey30 2005 UK RCT (double-blinded)

124 High Ritodrine hydrochloride (Yutopar) infusion of 50 mg (10 mg/ml) added to 12 ml dextrose saline via a syringe pump, starting at 1 ml/hour and increasing at 10-minute intervals to a maximum of 5 ml /hour.

Nor Azlin31 2005 Malaysia RCT (double-blinded)

60 High Ritodrine inusion (0.4 mg/ml in 5% dextrose) via a syringe pump at the rate of 1.5 ml/min, beginning 15 minutes before and continuing throughout the procedure

Collaris32 2008 Malaysia RCT (double-blinded)

90 High Nifedipine 10 mg tablet and intravenous placebo 0.9% sodium chloride and 1 ml syringe

Terbutaline sulphate (500 micrograms/ml) together and 1 ml syringe and a placebo tablet

50 | Chapter 3

The treatment drug under study was beta-mimetics in nine studies, calcium channel block-ers in three studies, five studies used nitric oxide and one study used oxytocin receptor blockers. Thirteen trials were placebo controlled and the other five studies used beta-mi-metics as a control group. Figure 2 presents the evidence network of the trials that had data available on ECV success rate (2A; 18 trials), caesarean delivery (2B; 12 trials), and vaginal cephalic delivery (2C; 12 trials).

No evidence network was constructed for the outcomes of poor fetal and severe maternal complications due to the low incidence of these outcomes. An overall estimate of the inci-dence of these outcomes was determined by dividing the number of events by the total number of women. No formal meta-analysis was performed for these outcomes. The risk of bias of the included trials is demonstrated in Supplementary Figure 1.

Successful ECVEighteen trials (n = 2 613) were included in the network meta-analysis for the outcome suc-cessful ECV (Figure 2A). A total of 1009 women (39%) were assigned to beta-mimetics, 241 (9%) were assigned to calcium channel blockers, 212 (8%) were assigned to nitric oxide, 410 (16%) were assigned to oxytocin receptor blockers, and 741 (28%) were assigned to placebo.

Figure 3 shows the pooled estimates of the network meta-analysis for each comparison. Beta-mimetics was the only tocolytic agent that was found to increase ECV success rate in comparison to placebo (RR 1.6, 95% CI 1.28 to 2.0). Oxytocin receptor blockers showed a non-significant increase in ECV success rate (RR 1.31, 95% CI 0.78 to 2.21). Both calcium

Table 1. Continued

Kok33 2008 The Netherlands

Multicenter RCT (double-

blind)

310 High Nifedipine 10 mg 30 and 15 minutes before ECV attempt.

Placebo 30 and 15 minutes before ECV attempt.

Nor Azlin34 2008 Malaysia RCT (doctor blinded)

86 High Nifedipine 20 mg tablet 20 minutes before ECV attempt.

Terbutaline 50 μg intravenously slow bolus 20 minutes before ECV attempt.

Hilton35 2009 Canada Multicenter RCT (double-

blinded)

126 High Nitroglycerin intravenously 10 ml of 100 μg/ml

Placebo in matching infusion

Vani36 2009 Malaysia RCT (open-label)

114 Low Salbutamol 0.1 mg intravenously every 5 minutes till maternal pulse was > 100 bpm.

Velzel16 2017 The Netherlands

Multicenter RCT

830 High intravenous bolus of atosiban of 6.75 mg in 0.9 ml (7.5 mg/ml)

fenoterol bolus of 40 μg in 0.8 ml (0.5 mg/10 ml)

channel blockers and nitric oxide performed almost equal in comparison with placebo (respectively RR 1.09, 95% CI 0.76 to 1.6 and RR 0.98, 95% CI 0.68 to 1.4). RORs in the two closed loops in the network were 1.7 (95% CI 1.0 to 4.2) for the placebo – beta-mimetics – nitric oxide loop, and 1.0 (95% CI 1.0 to 1.8) for the placebo – beta-mimetics – calcium channel blockers loop, which provides support for the assumption of consistency required for network meta-analysis. Results of the pairwise meta-analyses are presented in Supple-mentary Material, and showed to be comparable to the risk ratios found in the network meta-analysis.

Table 2 shows the rank probability for each intervention for successful ECV. Beta-mimetics had the highest probability (77.5%) of being ranked the best treatment option and had a mean rank of 1.2. Oxytocin receptor blockers had the second highest probability (20.6%) of being the best treatment method, with a mean rank of 2.4.

Caesarean deliveryCaesarean delivery was reported in 12 trials (n = 2231). A total of 895 women (40%) were assigned to beta-mimetics, 241 (11%) were assigned to calcium channel blockers, 95 (4%) were assigned to nitric oxide, 410 (18%) were assigned to oxytocin receptor blockers, and 590 (26%) were assigned to placebo (Figure 2B).

Table 1. Continued

Kok33 2008 The Netherlands

Multicenter RCT (double-

blind)

310 High Nifedipine 10 mg 30 and 15 minutes before ECV attempt.

Placebo 30 and 15 minutes before ECV attempt.

Nor Azlin34 2008 Malaysia RCT (doctor blinded)

86 High Nifedipine 20 mg tablet 20 minutes before ECV attempt.

Terbutaline 50 μg intravenously slow bolus 20 minutes before ECV attempt.

Hilton35 2009 Canada Multicenter RCT (double-

blinded)

126 High Nitroglycerin intravenously 10 ml of 100 μg/ml

Placebo in matching infusion

Vani36 2009 Malaysia RCT (open-label)

114 Low Salbutamol 0.1 mg intravenously every 5 minutes till maternal pulse was > 100 bpm.

Velzel16 2017 The Netherlands

Multicenter RCT

830 High intravenous bolus of atosiban of 6.75 mg in 0.9 ml (7.5 mg/ml)

fenoterol bolus of 40 μg in 0.8 ml (0.5 mg/10 ml)

52 | Chapter 3

Beta-mimetic

Oxytocin receptorblockers

Placebo

Nitricoxide

Calciumchannelblockers

9 / 894

2 / 133

3 / 2821 / 310

1 / 830

2 / 176

A: network for the outcome successful ECV

Beta-mimetic

Placebo

Nitricoxide

5 / 512

1 / 74

3 / 2821 / 310

1 / 830

2 / 176

B: network for the outcome caesarean delivery

Figure 2. Graphic representation of tocolytic trials retrieved for network meta-analysis. Lines rep-resent trials comparing two classes of drug. Numbers on lines represent number of trials and total number of participants in those trials.

Beta-mimetic

Placebo

Nitricoxide

5 / 459

2 / 133

3 / 282

1 / 830

2 / 176

C: network for the outcome vaginal cephalic delivery

Figure 2. Continued.

Figure 3 shows the pooled estimates of the network meta-analysis for each comparison for caesarean delivery. Compared to placebo, both beta-mimetics and oxytocin receptor blockers showed to decrease the caesarean delivery rate significantly (respectively RR 0.78, 95% CI 0.70 to 0.86 and RR 0.85, 95% CI 0.72 to 0.99), and nitric oxide and calcium channel blockers are less likely to decrease the caesarean delivery rate (respectively RR 0.84, 95% CI 0.70 to 1.01 and RR 1.09, 95% CI 0.92 to 1.29). RORs in the two closed loops in the net-work were 1.0 (95% CI 1.0 to 1.8) for the placebo – beta-mimetics – nitric oxide loop, and 1.0 (95% CI 1.0 to 1.5) for the placebo – beta-mimetics – calcium channel blockers loop, which provides support for the assumption of consistency required for network meta-anal-ysis. Results of the pairwise meta-analyses are presented in Supplementary Material, and showed to be comparable to the risk ratios found in the network meta-analysis.

54 | Chapter 3

RR 95% CI 95% PrI

1.60 (1.28 to 1.98) (0.93 to 2.75)0.98 (0.68 to 1.42) (0.52 to 1.85)1.09 (0.76 to 1.56) (0.58 to 2.05)1.31 (0.78 to 2.21) (0.62 to 2.79)

0.62 (0.42 to 0.89) (0.33 to 1.16)0.68 (0.48 to 0.97) (0.36 to 1.27)0.82 (0.51 to 1.32) (0.40 to 1.68)

1.11 (0.68 to 1.81) (0.53 to 2.30)1.33 (0.73 to 2.44) (0.58 to 3.05)

1.21 (0.67 to 2.18) (0.53 to 2.73)

Treatment Effect

Beta-mimetic vs Placebo

Nitric oxide vs Placebo

Calcium channel blocker vs Placebo

Oxytocin receptor blocker vs Placebo

Nitric oxide vs Beta-mimetic

Calcium channel blocker vs Beta-mimetic

Oxytocin receptor blocker vs Beta-mimetic

Calcium channel blocker vs Nitric oxide

Oxytocin receptor blocker vs Nitric oxide

Oxytocin receptor blocker vs Calcium channel blocker

0.2 0.5 2 51

Relative Risk active vs. control

Figure 3. Pooled estimates of treatment effects for each comparison for successful ECV. Black line represent risk ratios (RR) and 95% Confidence Intervals (CI) and the outer line represent prediction Intervals (PrI).

0.5 21

RR 95% CI 95% PrI

0.78 (0.70 to 0.86) (0.68 to 0.89)

0.84 (0.70 to 1.01) (0.67 to 1.06)

1.09 (0.92 to 1.29) (0. 98 to 1.35)

0.85 (0.72 to 0.99) (0.69 to 1.04)

1.08 (0.88 to 1.32) (0.84 to 1.39)

1.40 (1.19 to 1.66) (1.14 to 1.73)

1.09 (0.96 to 1.23) (0.93 to 1.27)

1.30 (1.02 to 1.66) (0.96 to 1.76)

1.01 (0.80 to 1.27) (0.75 to 1.35)

0.78 (0.63 to 0.95) (0.60 to 1.01)

Treatment Effect

Beta-mimetics had the highest probability (70%) of being ranked the best treatment option with a mean rank of 1.3. Nitric oxide had the second highest probability (22%) of being the best treatment method with a mean rank of 2.3 (Table 2).

Vaginal cephalic deliveryVaginal cephalic delivery was reported in 12 trials (n = 1811). A total of 795 women (44%) were assigned to beta-mimetics, 87 (5%) were assigned to calcium channel blockers, 181 (10%) were assigned to nitric oxide, 410 (23%) were assigned to oxytocin receptor blockers, and 338 (19%) were assigned to placebo (Figure 2C).

Figure 4 shows the pooled estimates of the network meta-analysis for each comparison for vaginal cephalic delivery. Compared to placebo, only beta-mimetics showed a significant increase in vaginal cephalic deliveries (RR 1.6, 95% CI 1.15 to 2.3). Oxytocin receptor block-ers showed non-statistically significant increase in vaginal cephalic deliveries (RR 1.45, 95% CI 0.71 to 3.0). Both nitric oxide and calcium channel blockers performed almost equal in comparison to placebo (respectively RR 1.13, 95% CI 0.71 to 1.8 and RR 1.04, 95% CI 0.51 to 2.1). RORs in the one closed loop in the network was 1.0 (95% CI 1.0 to 3.3) for the placebo – beta-mimetics – nitric oxide loop, which provides support for the assumption of consistency required for network meta-analysis. Results of the pairwise meta-analyses are presented in Supplementary Material, and showed to be comparable to the risk ratios found in the network meta-analysis.

RR 95% CI 95% PrI

1.64 (1.15 to 2.34) (0.71 to 3.77)

1.13 (0.71 to 1.82) (0.46 to 2.83)

1.04 (0.51 to 2.14) (0.34 to 3.21)

1.45 (0.71 to 2.96) (0.47 to 4.45)

0.69 (0.42 to 1.13) (0.27 to 1.75)

0.64 (0.34 to 1.19) (0.23 to 1.80)

0.89 (0.48 to 1.65) (0.31 to 2.49)

0.92 (0.42 to 2.03) (0.28 to 3.03)

1.28 (0.58 to 2.82) (0.39 to 4.21)

1.39 (0.58 to 3.35) (0.39 to 5.00)

Treatment Effect

210.2 0.5 5

Beta-mimetics had the highest probability (54%) of being ranked the best treatment option with a mean rank of 1.5. Oxytocin receptor blockers had the second highest probability (34%) of being the best treatment method with a mean rank of 2.3 (Table 2).

56 | Chapter 3

Table 2: The rank probabilities for each intervention for successful ECV, caesarean delivery rate and vaginal cephalic delivery

Tocolytic agent Probability of being the best treatment (%) based on estimated probabilities

Mean Rank

Probability of being the best treatment (%) based on predictive probabilities

Mean Rank

Successful ECV

Beta-mimetic 77.5 1.2 64.3 1.4

Calcium channel blockers 1.4 3.4 6.3 3.3

Nitric oxide 0.5 4.0 3.0 3.9

Oxytocin receptor blockers 20.6 2.4 25.0 2.5

Placebo 1.4 3.4 1.4 3.8

Caesarean delivery

Beta-mimetic 70.1 1.3 70.3 1.3

Nitric oxide 22.1 2.3 21.9 2.3

Placebo 0.0 4.1 0.0 4.1

Vaginal cephalic delivery

Beta-mimetic 54.7 1.5 45.6 1.8

Nitric oxide 5.3 3.7 9.4 3.3

Placebo 0.1 4.1 3.0 3.9

Effect modificationThe availability of effect modifiers in the included trials was limited (Table 3). To prevent too much loss of information and to maintain a substantial network of evidence we therefore only assessed effect modification for the three potential effect modifiers that were most frequently available, i.e. maternal age (17/18 trials for ECV success, and 12/12 trials for caesarean delivery and vaginal cephalic delivery), parity (14/18 trials for ECV success, 10/12 trials for caesarean delivery, and 9/12 trials for vaginal cephalic delivery), and gestational age at ECV (15/18 trial for ECV success, 11/12 trials for caesarean delivery and vaginal cephalic delivery). Meta-regression indicated no modification of treatment effects across all comparisons and investigated outcomes.

Table 3. Availability of effector modifiers per study

Study Outcomes Potential effect modifier

Successful ECV

Caesarean delivery

Cephalic vaginal delivery

Mean age

Parity

Maternal w

Body Mass Index

Gestational age at ECV

Estimated fetal w

Ethnicity

Frank breech

Placental location

Amniotic Fluid Index

Robertson ● ● ● ● ● ● ● ●

Tan ● ● ●

Stock ● ● ● ●

Chung ● ● ●

Marquette ● ● ● ● ● ● ●

Fernandez ● ● ● ● ● ● ●

Yanny ● ● ● ● ●

Bujold ● ● ● ● ● ● ● ● ● ●

Bujold ● ● ● ● ● ● ● ● ●

El-Sayed ● ● ● ● ● ● ● ● ●

Impey ● ● ● ● ● ● ● ●

Nor Azlin ● ● ● ● ● ● ●

Collaris ● ● ● ● ● ● ● ● ● ●

Kok ● ● ● ● ● ● ● ● ● ● ● ● ●

Nor Azlin ● ● ● ● ● ● ● ●

Hilton ● ● ● ● ● ●

Vani ● ● ● ● ● ● ● ●

Velzel ● ● ● ● ● ● ● ● ● ● ● ● ●

Fetal and maternal complicationsIn Table 4, we demonstrate the frequencies of poor fetal and severe maternal complica-tions among the included studies. Fetal and neonatal death occurred in 12 of 2156 women (0.19%). The risk for a non-reassuring CTG leading to an emergency caesarean delivery was 0.47%. Placental abruption occurred in 1 of 1238 women (0.08%). The ECV attempt was stopped in 2 women out of 2006 (0.10%) due to severe hypotension. The distribution of fetal and severe maternal was similar between the intervention and control groups.

58 | Chapter 3

Table 4: Complications of ECV and drug related adverse events in the included studies

Outcome Number of studies

Total number of patients

Result N (%)

Adverse Fetal events after ECV 13 2319 11 (0.47)*

Fetal/neonatal death 12 2156 4 (0.19)†

Placental abruption 4 1238 1 (0.08)

NICU admission 5 1363 52 (3.8)

AS < 7 after 5 minutes 6 1521 22 (1.4)

Adverse Maternal events during ECV 9 2006 2 (0.10)‡

* All cases non-reassuring CTG leading to emergency caesarean † Two babies died after delivery of severe congenital disorder, one intra-uterine fetal death 4 weeks after ECV and one baby died two hours after a vacuum assisted delivery for prolonged second stage of labor‡ Both women had severe hypotension with the need to stop the procedure. One woman received nitric oxide, the other woman a beta-mimetic

Discussion

In this network meta-analysis we assessed the comparative effectiveness of four different classes of tocolytic agents and placebo to enhance ECV success rate. We demonstrated that beta-mimetics was the most effective tocolytic to increase ECV success rate and the number of vaginal cephalic deliveries. It also decreased caesarean delivery rates.

This is the first network meta-analysis comparing tocolytic agents used for ECV. Our net-work meta-analysis considered all available knowledge on tocolytic agents used for ECV obtained from randomized clinical trials and as such is the best available evidence on the comparative effectiveness of these agents. Using network meta-analysis, we were able to provide the comparative effectiveness of treatments that have not been compared before in head-to-head trials. Therefore, we were able to rank these agents based on the probabilities of the best available treatment option.

A potential limitation of this study is that we were not able to report on all relevant neo-natal and maternal outcomes. Unfortunately, these outcomes were poorly reported in the included studies, and if reported their occurrence was rare. Therefore, we were not able to perform an accurate analysis for these outcomes. We found that the distribution of fetal and severe maternal outcomes was similar between the intervention and control groups, and therefore it seems that these outcomes occur independent of the use of tocolytics. The reported rates of serious complications such as still-birth and placental abruption in our study correspond to that in other systematic reviews of cohort studies (in our study 0.27% and in systematic reviews 0.24%).12–14 It is important to weigh this risk of complications with an ECV in light of the mode of delivery for term breech presenta-tion as alternative, in which the risk of perinatal mortality is approximately 2.0/1000 with planned vaginal breech birth and 0.5/1000 with caesarean section and in future preg-nancies 1.0/1000.15

Our results are in line with a previous pairwise meta-analysis. A recent Cochrane review found that women using beta-mimetics for uterine relaxation for ECV had a higher rate of cephalic birth (pooled RR 1.7, 95% CI 1.14 to 2.5) and a lower chance on caesarean delivery (pooled RR 0.77, 95% CI 0.67 to 0.88), compared to placebo.4 There was no signif-icant difference with regards to caesarean delivery for calcium channel blockers and nitric oxide compared to placebo. Our study complements this knowledge with the finding that beta-mimetics are not only preferable to placebo but also to the other tocolytic agents.Women who are eligible for an ECV attempt should be adequately counselled about the effectiveness of beta-mimetics and the potential side effects. Side effects occur frequently (71%)16, but are limited to transient feelings of tachycardia, while severe hypotension

60 | Chapter 3

occurred in only in one out of 408 patients (0.2%) after the use of fenoterol.16 In our opin-ion ECV with fenoterol is the optimal treatment for women with a breech presentation. Despite the truly unpleasant though transient side effects of tachycardia and flushes, it outweighs the risk of a caesarean section for breech presentation and associated com-plication risks. The results of a patient’s preference study confirm this: as long the gain in success rate outweighs possible side effects like tachycardia, most women are willing to undergo treatment with beta-mimetics.17

We want to emphasize the importance of core clinical outcome sets for clinical studies. The first systematic review on tocolytic agents for ECV was published in 200418 and the first systematic review on safety of ECV was published 2 years later in 2006.19 Now, another 12 years later we still need to conclude that treatment safety is poorly reported in randomized clinical trials reporting on ECV. The fact that earlier studies did not show frequent occurring major side effects does not relieve researchers from the obligation to report on important safety outcomes. Especially with rare events, only large (combined) data sets can truly inform us on the potential burdens of treatments for mothers, fetus and future offspring. An interesting topic of study is whether all women benefit from tocolytic therapy in an equal matter, or if women with more favorable baseline characteristics (which correspond to an increased success rate such as parity, placenta localization, frank breech etc.) ben-efit more or whether there are women that would benefit less. Withholding unnecessary treatment is a noble endeavor and an obligation for doctors searching for best treatment options for individual patients.

In conclusion, beta-mimetics are the most effective tocolytic agent with respect to enhance the effectiveness of ECV as indicated by an increased ECV success rate and cephalic vaginal delivery rate, and a reduced caesarean delivery rate. Oxytocin receptor blockers were less effective than beta-mimetics, while calcium channel blockers and nitric oxide are not effective in ECV. We recommend beta-mimetics as the tocolytic of first choice in women undergoing ECV.

References

1. Molkenboer, J. F. M., Bouckaert, P. X. J. M. & Roumen, F. J. M. E. Recent trends in breech delivery in the Netherlands. Br. J. Obstet. Gynaecol. 110, 948–51 (2003).

5. Jansen, J. P. et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: Report of the ISPOR task force on indirect treatment comparisons good research practices: Part 1. Value Heal. 14, 417–428 (2011).

6. Hoaglin, D. C. et al. Conducting indi-rect - t reatment -compar ison and network-meta-analysis studies: Report of the ISPOR task force on indirect treatment comparisons good research practices: Part 2. Value Heal. 14, 429–437 (2011).

7. Moher, D., Liberati, A., Tetzlaff, J. & Altman, D. G. Preferred reporting items for sys-tematic reviews and meta-analyses: the PRISMA statement. BMJ 339, b2535 (2009).

8. Hutton, B. et al. The PRISMA extension statement for reporting of systematic reviews incorporating network meta-anal-yses of health care interventions: Checklist and explanations. Ann. Intern. Med. 162, 777–784 (2015).

9. Higgins, J. P. T. et al. The Cochrane Collab-oration’s tool for assessing risk of bias in randomized trials. (2011).

10. Lu, G. & Ades, A. E. Assessing evidence inconsistency in mixed treatment com-parisons. J. Am. Stat. Assoc. 101, 447–459 (2006).

11. Jansen, J. P. Network meta-analysis of indi-vidual and aggregate level data. Res. Synth. Methods 3, 177–190 (2012).

16. Velzel, J. et al. Atosiban versus fenoterol as a uterine relaxant for external cephalic version: A randomized controlled trial. BMJ 356, (2017).

18. Hofmeyr, G. J. Interventions to help external cephalic version for breech presentation at term. Cochrane Database Syst. Rev. (2004).

62 | Chapter 3

20. Robertson, A. W. et al. External cephalic version at term: is a tocolytic necessary? Obstet. Gynecol. 70, 896–9 (1987).

21. Tan, G. W., Jen, S. W., Tan, S. L. & Salmon, Y. M. A prospective randomised controlled trial of external cephalic version compar-ing two methods of uterine tocolysis with a non-tocolysis group. Singapore Med. J. 30, 155–8 (1989).

22. Stock, A., Chung, T., Rogers, M. & Ming, W. W. Randomized, double blind, placebo controlled comparison of ritodrine and hexoprenaline for tocolysis prior to exter-nal cephalic version at term. Aust. New Zeal. J. Obstet. Gynaecol. 33, 265–8 (1993).

23. Chung, T., Neale, E., Lau, T. K. & Rogers, M. A randomized, double blind, controlled trial of tocolysis to assist external cephalic version in late pregnancy. Acta Obstet. Gynecol. Scand. 75, 720–4 (1996).

24. Marquette, G. P., Boucher, M., Thériault, D. & Rinfret, D. Does the use of a tocolytic agent affect the success rate of external cephalic version? Am. J. Obstet. Gynecol. 175, 859–61 (1996).

25. Fernandez, C. O., Bloom, S. L., Smulian, J. C., Ananth, C. V & Wendel, G. D. A Ran-domized Placebo-Controlled Evaluation of Terbutaline for External Cephalic Version. Obstet. Gynecol. 90, 775–9 (1997).

26. Yanny, H. et al. Double-blind randomised controlled trial of glyceryl trinitrate spray for external cephalic version. Br. J. Obstet. Gynaecol. 107, 562–4 (2000).

27. Bujold, E. et al. Sublingual nitroglycerine as a tocolytic in external cephalic version: a comparative study. J. Obstet. Gynaecol. Can. 25, 203–7 (2003).

28. Bujold, E., Marquette, G. P., Ferreira, E., Gauthier, R. J. & Boucher, M. Sublingual nitroglycerin versus intravenous ritodrine as tocolytic for external cephalic version: A double-blinded randomized trial. Am. J. Obstet. Gynecol. 188, 1454–1459 (2003).

29. El-Sayed, Y. Y. et al. Randomized com-parison of intravenous nitroglycerin and subcutaneous terbutaline for external cephalic version under tocolysis. Am. J. Obstet. Gynecol. 191, 2051–5 (2004).

30. Impey, L. & Pandit, M. Tocolysis for repeat external cephalic version in breech presentation at term: a randomised, dou-ble-blinded, placebo-controlled trial. Br. J. Obstet. Gynaecol. 112, 627–31 (2005).

31. Nor Azlin, M. I. et al. Tocolysis in term breech external cephalic version. Int. J. Gynaecol. Obstet. 88, 5–8 (2005).

32. Collaris, R. & Tan, P. C. Oral nifepidine versus subcutaneous terbutaline tocolysis for external cephalic version: a double-blind randomised trial. Br. J. Obstet. Gynaecol. 116, 74-80–1 (2009).

33. Kok, M. et al. Nifedipine as a uterine relaxant for external cephalic version: a randomized controlled trial. Obstet. Gyne-col. 112, 271–6 (2008).

34. Mohamed Ismail, N. A. et al. Nifedipine versus terbutaline for tocolysis in external cephalic version. Int. J. Gynaecol. Obstet. 102, 263–6 (2008).

35. J. Hilton, B. Allen, C. Swaby, R. W. et al. Intra-venous Nitroglycerin for External. Obstet. Gynecol. 114, 560–567 (2009).

36. Vani, S., Lau, S. Y., Lim, B. K., Omar, S. Z. & Tan, P. C. Intravenous salbutamol for external cephalic version. Int. J. Gynaecol. Obstet. 104, 28–31 (2009).

Suplemantary material:

Table with all baseline characteristics of all included studies

Study Baseline characteristics IncidenceIndex group

IncidenceControl group

N or mean

SD N or mean

Robertson Total 28 30

Mean age 24,1 0,4 22,4 0,3

Nulliparous 16 17

Mean maternal weight (lbs) 158 5,1 161 4,4

Mean gestational age (wks) 38,9 0,2 38,9 0,2

Mean estimated fetal weight (g) 3180 60 3043 51

Tan Total 30 30

Nulliparous 17 17

Gestation < 36 weeks 9 9

Gestation > = 36 weeks 21 21

Stock Total 21 21

Mean age 26,9 4,1 29,2 4,7

Nulliparous 9 9

Mean maternal weight (kg) 58,8 5,4 64,5 5,9

Chung Total 25 25

Mean age 30,1 24-38 (range)

28,8 18-43 (range)

Nulliparous 15 15

Mean gestational age (wks) 36,6 36-38 (range)

37,1 36-38 (range)

Marquette Total 138 145

Mean age 28,5 0,45 29,3 0,41

BMI 27,7 3,1 28,1 3,4

AFI 11,8 0,34 12,2 0,36

Nulliparous 58 49

Frank Breech 59 43

Fernandez Total 52 51

Mean age 23,4 4,9 25,7 5,4

AFI 13,3 4,8 13,2 4,3

Mean estimated fetal weight (g) 3142 434 3162 484

64 | Chapter 3

N or mean

SD N or mean

Yanny Total 31 26

Mean age 28 4,1 28,1 5

Frank Breech 14 12

Bujold Total 50 50

Mean age 31,5 21-41 (range)

31,7 21-44 (range)

Mean maternal weight (kg) 75 58-102 (range)

73 46-102 (range)

AFI 11,2 5,6-29,3 (range)

13,8 5,0-25,0 (range)

Nulliparous 0 0

Frank Breech 16 25

Mean estimated fetal weight (g) 3063 2218-4305

(range)

2979 2010-4031

(range)

Anterior Placenta 21 21

Bujold Total 38 36

Mean age 30 19-42 (range)

29 19-38 (range)

Mean maternal weight (kg) 72 59-113 (range)

74 37-117 (range)

AFI 11,8 7,9-20,8 (range)

12,6 6,3-22,4 (range)

Mean gestational age (wks) 37,4 36,1-39,3 (range)

37,4 36,3-38,7 (range)

Frank Breech 20 19

Mean estimated fetal weight (g) 3014 2458-4544

(range)

3050 2490-3880

(range)

El-Sayed Total

Maternal age (y) 31,1 5,6 31,7 4,8

Gestational age at ECV (wk) 38,4 0,7 38,4 0,8

Multiparity 13 11

Estimated fetal weight (g) 3266 336 3227 398

Body mass index 28,6 4 30,2 7

Anterior placenta 21 24

N or mean

SD N or mean

Impey Total

Maternal age 30,6 4,5 30,9 5,5

Nulliparous 44 45

Caucasian race 61 61

Frank breech 38 42

Liquor pool depth (mm) 43,3 47,7

Gestation at ECV (weeks) 37,5 0,81 37,5 0,85

Nor Azlin Total 30 30

Mean age 29,13 4,49 27,5 4,28

AFI 15,2 4,18 14,6 3,67

Nulliparous 22 23

Frank Breech 18 14

Collaris Total 44 46

Mean age 30 5 30 5

Nulliparous 23 25

Frank Breech 21 25

Gestation at ECV (weeks) 38 1 38 1

AFI 9,5 2,6 10,6 3,3

Anterior placenta 43,2 32,6

Estimated fetal weight (kg) 2,8 0,4 2,8 0,3

Kok Total 154 156

Mean age 33,6 4,2 34,1 4,5

Nulliparous 78 83

Gestation at ECV (days) 258,8 5,9 259,3 6,2

Caucasian race 126 133

BMI 24,7 4,6 24,8 4,3

Estimated fetal weight (kg) 2732 396 2672 376

Frank Breech 119 114

AFI <10 19 23

66 | Chapter 3

N or mean

SD N or mean

Nor Azlin Total 43 43

Mean age 28,5 4,06 29,9 5,15

Nulliparous 18 21

Gestation at ECV (weeks) 37,8 0,8 37,5 0,4

Frank Breech 24 24

AFI 12,9 2,7 12,3 2,8

Hilton Total 65 61

Maternal age 30 5 (range) 29 4 (range)

Gestation at ECV (weeks) 37 5 (range) 37

Vani Total 57 57

Mean age 28,2 4,8 (range)

28,7 4,3 (range)

Nulliparous 31 27

Gestation at ECV (weeks) 38 0,6 (range)

38 0,6 (range)

Frank Breech 29 28

AFI <10 28 24

Velzel Total 410 408

Age in years (mean; SD) 32.1 (4.0) 32.4 (4.3)

Multiparous women (%) 154 (37.6%) 153 (37.5%)

Gestational age at ECV (mean; SD)

35.8 (0.9) 35.9 (1.0)

Caucasian ethnicity (%) 350 (49.6%) 355 (50.4%)

BMI (mean; SD) 24.1 (4.3) 24.2 (4.8)

Estimated fetal weight (mean; SD)

2622 (391.0) 2572 (457.5)

Frank breech (%) 300 (76.5) 273 (69.8)

Anterior placental location (%) 127 (32.9) 144 (37.3)

Risk of bias:

Random sequence generation

(selection bias)

Allocation concealment

(selection bias)

Blinding of participants and personnel(perform

ance bias)

Blinding of outcome assessm

ent(detection bias)

Incomplete outcom

e data(attrition bias)

Selective reporting(reporting bias)

Other bias

Robertson ● ● ● ● ● ● ●

Tan ● ● ● ● ● ● ●

Stock ● ● ● ● ● ● ●

Chung ● ● ● ● ● ● ●

Marquette ● ● ● ● ● ● ●

Fernandez ● ● ● ● ● ● ●

Yanny ● ● ● ● ● ● ●

Bujold ● ● ● ● ● ● ●

El-Sayed ● ● ● ● ● ● ●

Impey ● ● ● ● ● ● ●

Nor Azlin ● ● ● ● ● ● ●

Collaris ● ● ● ● ● ● ●

Kok ● ● ● ● ● ● ●

Nor Azlin ● ● ● ● ● ● ●

Hilton ● ● ● ● ● ● ●

Vani ● ● ● ● ● ● ●

Velzel ● ● ● ● ● ● ●

68 | Chapter 3

Supplementary table pairwise meta-analysis of interventions in random effects model.

Outcome Intervention Control Number of studies RR 95% CI I2

Successful ECV Beta-mimetics Placebo 9 1.55 1.22 to 2.0 52.3

Nitric oxide Placebo 3 1.30 0.63 to 2.7 72.0

Calcium channel blocker Placebo 1 1.12 0.85 to 1.47 0.0

Nitric oxide Beta-mimetics 2 0.49 0.30 to 0.80 0.0

Calcium channel blocker Beta-mimetics 2 0.67 0.48 to 0.93 0.0

Oxytocin receptor blocker Beta-mimetics 1 0.82 0.69 to 0.98 0.0

Caesarean delivery Beta-mimetics Placebo 6 0.77 0.67 to 0.88 24.6

Cephalic vaginal delivery Beta-mimetics Placebo 5 1.7 1.10 to 2.5 52.3

Supplementary table pairwise meta-analysis of interventions in random effects model.

Outcome Intervention Control Number of studies RR 95% CI I2

Successful ECV Beta-mimetics Placebo 9 1.55 1.22 to 2.0 52.3

Caesarean delivery Beta-mimetics Placebo 6 0.77 0.67 to 0.88 24.6

Cephalic vaginal delivery Beta-mimetics Placebo 5 1.7 1.10 to 2.5 52.3

Chapter 4Effectiveness of moxibustion in women with breech presentation in combination with external cephalic version: a systematic review and meta-analysis

Joost VelzelRui WangGabriëlle R.M. KoningJan F.M. MolkenboerJoris A. M. van der PostBen Willem MolMarjolein Kok

Submitted

72 | Chapter 4

Abstract

Background: Breech presentation is mostly managed with elective caesarean delivery. The objective of this study is to evaluate the effectiveness of ante partum moxibustion on acupoint BL67 as intervention followed by a possible ECV attempt.

Methods: To identify relevant studies, we searched the literature for articles published from inception to January 2017, using Medline, EMBASE and Cochrane databases. We also hand-searched the bibliographies of retrieved articles to identify additional related studies. We included randomized controlled trials reporting on moxibustion to correct breech presentation. Study quality was assessed by using the Cochrane Handbook for Systematic Reviews of Interventions. The primary outcome was cephalic presentation after treatment of moxibustion (at 35 weeks of gestation and before possible ECV). Outcomes were reported as a pooled risk ratio with 95% confidence interval (95% CI).

Results: A total of 13 studies were analyzed, including 2,555 patients. Cephalic presen-tation after moxibustion treatment occurred more in the moxibustion group compared to routine care (RR 2.8, 95% CI 1.6 to 4.7, I2 74%). In studies offering ECV to all women in case of persisting breech presentation after moxibustion, neither cephalic presentation at delivery nor caesarean delivery differed significantly (RR 1.11, 95% CI 0.96 to 1.28, I2 10%), and, RR 1.01, 95% CI 0.82 to 1.20, I2 21%).

Conclusions: In women with breech presentation, the use of moxibustion may increase the number of babies presenting in cephalic version. However, moxibustion followed by a possible ECV attempt does not reduce the caesarean delivery rate.

Moxibustion for breech presentation in combination with ECV: a systematic review and meta-analysis | 73

Introduction

Breech presentation occurs in 3 to 4% of term pregnancies1, and elective caesarean delivery seems beneficial for short-term neonatal outcome as established in the Term Breech Trial.2 After publication of this trial, caesarean delivery rates for this indication increased significantly worldwide. Nowadays, in Western countries, breech presenta-tion is the third most indication of elective caesarean delivery.3 However, caesarean delivery is associated with short and long-term consequences for maternal and neona-tal health.4,5 In 2015, the World Health Organization reported that globally caesarean delivery was overused.6

External cephalic version (ECV) is a safe obstetrical procedure that causes a statistically significant and clinically meaningful reduction in non-cephalic (pooled RR 0.42, 95% CI 0.29 to 0.61) and caesarean delivery (pooled RR 0.57, 95% CI 0.40 to 0.82).7 ECV is recommended in national guidelines and all women with an uncomplicated breech presentation should be counselled to undergo ECV.8–10 Besides, it is considered a safe procedure with few contraindications.11 The reported rates of serious complications are 0.24%, including still-birth and placental abruption, and the pooled risk on fetal death is 0.19%.12–15 However, implementation of ECV is limited to only two-third of all patients.16 For professionals, the main barriers to counsel women for ECV are a lack of knowledge to fully inform and counsel patients and the inability to counsel women who preferred a primary caesarean delivery.17 Most important factors influencing patients’ willingness to undergo ECV are expected pain, adverse side effects of tocolytic drugs and success rate of ECV.18

Worldwide an interest of complementary and alternative medicine (CAM) exists, despite lack of evidence for effect.12,13 In obstetrics, surveys among patients and physicians revealed that up to 49% of women consult a CAM practitioner, of which only half dis-closes this to their physician. Therefore, there is a need for provision of valid information concerning these alternative methods.14–16

Moxibustion is a traditional Chinese medicine therapy that uses the heat generated by burning herbal preparations containing Artemisia vulgaris (mugwort) to stimulate the acu-point BL67 (beside the outer corner of the fifth toenail). It is thought that this method increased fetal activity and can be applied to correct breech presentation.17 Moxibustion to correct breech presentation has been evaluated in multiple randomized controlled trials. A recent published Cochrane review in 2012 revealed a beneficial effect of moxibustion, applied alone or in combination with acupuncture or postural measures, compared to observation alone or postural measures.17 Since then three studies were published18–20,

74 | Chapter 4

of which only one confirms these findings. Alongside an update of this review, we want to focus on ECV as previous reviews did not distinguish between studies who offer ECV and studies that do not. Furthermore, moxibustion therapy is applied from 32 weeks of gestation onwards and ECV is performed near term and therefore, both could be rec-ommended if a beneficial effect is present. Therefore, the aim of the current systematic review and meta-analysis is to evaluate the effectiveness of moxibustion to correct breech presentation followed by a possible ECV attempt.

Methods

Our study protocol was designed a priori and was registered in PROSPERO: the inter-national prospective register of systematic reviews (CRD42014013524). This work was performed in accordance with the PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions.21

SearchWe conducted a computerized search in Medline, Embase, and the current Cochrane data-bases from inception until January 2018. We identified published randomized controlled trials reporting on the use of moxibustion to correct breech presentation.

To ensure the inclusion of all relevant articles, references from identified publications were manually searched for additional relevant articles. Language restrictions were applied for common Western languages (English, French, German, Spanish e.g.) and Chinese. The complete electronic search is available from the first author. This study had no funding source.

Study selectionTwo independent reviewers (JV and GK) screened the electronic search for relevant articles. Based on title and abstract, full text articles were obtained if they seemed eligible. Both reviewers used the same standardized data extraction form. If disagreements occurred, they were resolved by consensus and, if necessary, by a third reviewer (MK). Extracted data on study design, number of patients, characteristics of the patients, inclusion and exclusion criteria, intervention and outcome measures were reported. Studies retrieved from references in Chinese, data were extracted by a Chinese native speaker (RW). Inclu-sion criteria were full text article availability, women with singleton breech presentation, comparing the effect of moxibustion to a control group, providing information on our pre-specified outcomes of interest.

Quality AssessmentThe same reviewers evaluated the methodological quality of the articles, using the crite-ria stated in the Cochrane Handbook for Systematic Reviews of Interventions.22 The risk of bias was assessed at study level, taking into account random sequence generation, concealment of allocation, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and other sources of bias. For each study, the risk of bias was separately scored in each domain as low risk, high risk or unclear. When at least five out of the seven domains had a low risk score, with at least random sequence generation, we considered the overall study quality to be high.

76 | Chapter 4

Interventions and outcome measuresThe experimental group consists of moxibustion alone or in combination with acupunc-ture. We described moxibustion as stimulation of acupoint BL67 by moxibustion rolls and acupuncture as puncture of acupoint BL67. The control group consists of routine care or placebo (stimulation of another acupoint than BL67 with laser or acupuncture). Routine care was described as care as usual or in combination with frequent knee to chest posi-tioning management.

The primary outcome measure was cephalic presentation after treatment of moxibustion (at 35 weeks of gestation). Secondary outcomes were cephalic presentation at delivery, caesarean delivery rate, total number of ECV attempts and ECV success rate defined as fetus in cephalic presentation after ECV attempt. We have categorized all studies into two different strategies; one strategy in which ECV is offered in both intervention and control group if persisting breech presentations exists, and the other strategy includes studies that did not offer ECV either in the treatment or control group. This applied for the out-comes caesarean delivery and cephalic presentation at delivery.

Statistical analysisOutcome data from studies were pooled into relative risks (RR) for all dichotomous out-comes. Heterogeneity was explored for all the meta-analyses. Analyses were undertaken using a fixed effects model. In case of substantial heterogeneity (I2 ≥50%) the random effects model of DerSimonian and Laird was used to obtain the pooled risk ratio estimate. Outcome measures were reported as RR with 95% confidence interval (95% CI). P value < 0.05 was considered statistically significant. Review Manager 5.3 was used to carry out the statistical analyses.

Results

We identified 85 citations and reviewed 46 articles for inclusion. A total of 13 studies (2,555 patients) met the inclusion criteria and were eligible.18–20,23–32 A flow diagram of the electronic search and selection process of the articles through the different stages of the review is shown in Figure 1.

MEDLINE (n=62)

EMBASE (n=79)Sc

Records after duplicates removed(n=85)

Full-text articles assessed for eligibility(n=46)

Full-text articles excluded, with reasons (n=33)

20 reviews; 2 case series; 1 non controlled study; 2 reported other outcome; 2 Cochrane review;

2 studies including cephalic presentation; 4 used other technique

CENTRAL (n=2)

Records after cross referencing (n=5)

Records excluded(n=44)

Figure 1. PRISMA flow chart of search results and article selection

78 | Chapter 4

Table 1. Characteristics of the included studies

Study Country Sample size (n) Inclusion criteria GA atrandom-ization (wk)

Treatment method

Control group Study primary outcome

ECV offered

Huang 1990 China 387 (193/194) Breech presentation (Not reported whether confirmed by ultrasound scan)

28 - 32 Moxibustion Routine care Cephalic presentation after treatment

Cardini 1998 China/Italy

260 (130/130) Nulliparous, breech presentation diagnosed by US, normal foetal biometry

321 - 337 Moxibustion Routine care Cephalic presentation at 35th week

Yes. For both groups subjects with persistent breech presentation after 2 weeks of treatment could undergo external cephalic version anytime between 35 weeks’ gestation and delivery.

Lin 2002 China 122 (63/59) Singleton breech presentation confirmed by physical examination and ultrasound scan.

Chen 2004 China 143 (73/69) Singleton breech presentation confirmed by ultrasound scan.

Neri 2004 Italy 240 (120/120) Caucasian, weeks GA, breech presentation confirmed by US

33 - 35 Moxibustion + acupuncture

Routine care Presentation at birth No

Cardini 2005 Italy 123 (65/58) Healthy, non-Chinese nulliparous women, breech presentation confirmed by US, normal foetal biometry

Yes. For both groups participants with persistent breech presentation could request ECV after the 37th week.

Yang 2006 China 206 (103/103) Breech presentation confirmed by palpation and ultrasound scan.

Neri 2007 Italy 41 (31/10) Maternal age 20-40 yrs, frank position, 33-36 weeks gestational age

33 - 36 Moxibustion or moxibustion + acupuncture

Routine care CTG characteristics No

Millereau 2009 France 68 (33/35) Singleton breech presentation, confirmed by US at 32 weeks

34 Moxibustion Routine care Cephalic presentation at birth

Guittier 2009 Switzerland 212 (106/106) Women with single foetus in breech presentation confirmed by US

34 - 36 Moxibustion Routine care Cephalic presentation at delivery or before ECV if attempted

Yes. For both groups, participants with persistent breech presentation could request ECV after the 37th week.

Do 2011 Australia 20 (10/10) 18 years or older, singleton breech presentation confirmed by US, normal foetal biometry

34 - 365 Moxibustion Routine care Cephalic presentation at birth

Vas 2013 Spain 406 (136/270) Breech presentation confirmed by US, at least 18 years old, normal foetal biometry, no prior treatment with moxibustion to achieve version

33 - 35 Moxibustion Routine care, knee-chest postural management

Cephalic presentation at birth

Coulon 2014 France 328 (164/164) Breech presentation confirmed by US, at least 18 years old

Placebo (inactivated laser)

Breech presentation at 372 weeks of GA

Yes, ECV was offered between 37+2 and 37+5 weeks of gestation for both groups.

Table 1. Characteristics of the included studies

Study Country Sample size (n) Inclusion criteria GA atrandom-ization (wk)

Treatment method

Control group Study primary outcome

ECV offered

Huang 1990 China 387 (193/194) Breech presentation (Not reported whether confirmed by ultrasound scan)

Cardini 1998 China/Italy

260 (130/130) Nulliparous, breech presentation diagnosed by US, normal foetal biometry

Yes. For both groups subjects with persistent breech presentation after 2 weeks of treatment could undergo external cephalic version anytime between 35 weeks’ gestation and delivery.

Lin 2002 China 122 (63/59) Singleton breech presentation confirmed by physical examination and ultrasound scan.

Chen 2004 China 143 (73/69) Singleton breech presentation confirmed by ultrasound scan.

Neri 2004 Italy 240 (120/120) Caucasian, weeks GA, breech presentation confirmed by US

Routine care Presentation at birth No

Cardini 2005 Italy 123 (65/58) Healthy, non-Chinese nulliparous women, breech presentation confirmed by US, normal foetal biometry

Yes. For both groups participants with persistent breech presentation could request ECV after the 37th week.

Yang 2006 China 206 (103/103) Breech presentation confirmed by palpation and ultrasound scan.

Neri 2007 Italy 41 (31/10) Maternal age 20-40 yrs, frank position, 33-36 weeks gestational age

33 - 36 Moxibustion or moxibustion + acupuncture

Routine care CTG characteristics No

Millereau 2009 France 68 (33/35) Singleton breech presentation, confirmed by US at 32 weeks

34 Moxibustion Routine care Cephalic presentation at birth

Guittier 2009 Switzerland 212 (106/106) Women with single foetus in breech presentation confirmed by US

34 - 36 Moxibustion Routine care Cephalic presentation at delivery or before ECV if attempted

Do 2011 Australia 20 (10/10) 18 years or older, singleton breech presentation confirmed by US, normal foetal biometry

34 - 365 Moxibustion Routine care Cephalic presentation at birth

Vas 2013 Spain 406 (136/270) Breech presentation confirmed by US, at least 18 years old, normal foetal biometry, no prior treatment with moxibustion to achieve version

33 - 35 Moxibustion Routine care, knee-chest postural management

Cephalic presentation at birth

Coulon 2014 France 328 (164/164) Breech presentation confirmed by US, at least 18 years old

Placebo (inactivated laser)

Breech presentation at 372 weeks of GA

Yes, ECV was offered between 37+2 and 37+5 weeks of gestation for both groups.

80 | Chapter 4

The characteristics of the included studies are outlined in Table 1. All studies enrolled only women with a singleton breech presentation. The gestational age ranged from 28 to almost 35 weeks. In five studies, ECV was recommended in case of persisting breech presentation in both the intervention group as the routine care group.18,20,24,28,32 In those studies, ECV was offered from 37 weeks of gestation. In four studies the number of ECV attempts and the success rate in each randomization group was registered.18,20,24,32 A total of 10 studies compared moxibustion to routine care18,19,23–26,28,29,31,32, two studies compared moxibustion together with acupuncture to routine care.27,30 One study compared mox-ibustion with acupuncture to an inactive laser.20 Study quality of the included studies is demonstrated in Figure 2. We considered the study quality of six studies high.19,20,24,27,28,32 All studies that offered ECV were of high quality. All studies in Chinese were of low quality.

Figure 2. Methodological quality of the included studies

Figure 3 demonstrates forest plots for all three comparisons of the outcome cephalic presentation after treatment. Cephalic presentation after treatment occurred more in the moxibustion group compared to routine care (eight studies, pooled RR 2.8, 95% CI 1.6 to 4.7, I2 74%) (Figure 3a). Also for the comparison moxibustion with acupuncture to routine care (one study, RR 1.91, 95% CI 1.12 to 3.2) (Figure 3b). There was no significant difference for this outcome in the comparison moxibustion with acupuncture to placebo (one study, RR 1.48, 95% CI 0.93 to 2.4) (Figure 3c). All studies measured cephalic presen-tation at 35 weeks of gestation.

Figure 3a. Cephalic presentation after treatment in comparison moxibustion to routine care

Figure 3b. Cephalic presentation after treatment in comparison moxibustion with acupuncture to routine care

Figure 3c. Cephalic presentation after treatment in comparison moxibustion with acupuncture to placebo

82 | Chapter 4

Table 2. Outcome of studies after categorizing studies into two strategies; one strategy in which ECV is offered in both intervention and control group if persisting breech presentations exists, and the other strategy includes studies that did not offer ECV either in the treatment or control group.

Studies offering ECV Studies not offering ECV

moxibustion versus routine care moxibustion versus routine care

moxibustion and acupuncture versus placebo moxibustion and acupuncture versus routine care

No data available

Table 2. Outcome of studies after categorizing studies into two strategies; one strategy in which ECV is offered in both intervention and control group if persisting breech presentations exists, and the other strategy includes studies that did not offer ECV either in the treatment or control group.

Studies offering ECV Studies not offering ECV

No data available

84 | Chapter 4

In table 2, we have divided studies that offer ECV in both intervention and control group if persisting breech presentations occurs, and studies that did not offer ECV either in the treatment or control group. In studies that offered ECV for persisting breech presentation, in which moxibustion was evaluated as an additional intervention versus routine care (i.e. moxibustion in advance of ECV versus ECV alone), cephalic presentation at delivery and caesarean delivery rate did not differ significantly (respectively, four studies, pooled RR 1.11, 95% CI 0.96 to 1.28, I2 10% and three studies, pooled RR 1.01 95% CI 0.82 to 1.20, I2 21%)).

Further, table 2 also demonstrates studies that did not offer ECV either in the treatment or control group. Caesarean delivery did not differ significantly for the comparison mox-ibustion versus routine care (two studies, pooled RR 0.87 (95% CI 0.71 to 1.06, I2 0%)). One study compared moxibustion with acupuncture to routine care and demonstrated a significant difference in favor of moxibustion with acupuncture for caesarean delivery (RR 0.79, 95% CI 0.58 to 0.95).

Four studies reported on outcome measures of ECV, all comparing moxibustion to routine care. There were no significant differences for the outcome need for ECV (pooled RR 0.79, 95% CI 0.47 to 1.32, I2 79%) (Figure 4) and ECV success rate (pooled RR 1.09, 95% CI 0.78 to 1.54, I2 32%) (Figure 5).

Figure 4. Need for ECV in the moxibustion and control group

Figure 5. ECV success rate in the moxibustion and control group

Discussion

This meta-analysis of 13 randomized controlled trials shows that the use of moxibustion may increase the rate of cephalic presentation after treatment. If ECV is offered as addi-tional intervention for breech presentation, caesarean section rates are similar between the groups.

To our knowledge, this is the first meta-analysis that was performed to determine the effect of moxibustion in combination with ECV. Given the earlier positive results in favor of moxibustion and the increased acceptance and use of CAM and therefore the associated questions to discuss with patients in shared decision making, makes this an interesting topic to discuss.24

The strength of our review is based on compliance with strict criteria for performing a systematic review. We analyzed the studies with and without the use of ECV, gaining new insight in the efficacy of the combination of moxibustion and ECV and the efficacy of those treatments used separately. Also, we included studies performed in China and a native speaker extracted data.

Not all included studies were methodologically well executed. The randomization was not performed adequately in every study. Only seven out of the 13 studies used an intention to treat analysis. As a result of our quality assessment, we found that all the studies that also offered ECV were of high quality. Studies with the most positive outcome in favor of moxibustion, had the lowest quality. Furthermore, we could only derive one outcome from these studies and we were not able to compare mode of delivery.

Although the present study was performed according to established methodology, there are limitations to our findings. Most studies used different study protocols. There was a large variation on gestational age at the start of the treatment, amount of treatment, and performance of treatment by a health professional or by the patient self. Moreover, we found a possible beneficial effect of moxibustion on acupoint BL67 if ECV is not offered, these results should be interpreted with caution due to clinical heterogeneity (including differences in interventions, sample size, and study populations), and varying levels of statistical heterogeneity. Furthermore, publication bias for this topic could exist as RCTs with a non-significant treatment effect may not be published.

We chose the need for ECV as an outcome of interest. Even though there was no significant difference between the groups for the need for ECV, the numbers of ECV that were per-formed varied widely between studies and within groups. This might be through possible

86 | Chapter 4

treatment bias. For example, in the study of Caridini24, there were no participants with a fetus in non cephalic position after moxibustion that underwent ECV in the treatment group, in contrast to 19% of the participants in the routine care group. Implementation of ECV is low in the included trials, and therefore, ECV does not appear to be standard treatment. Even though ECV is considered as a safe obstetric intervention with a low risk of complications33,34, concerns about safety are an important reason for patients not to opt for ECV. For moxibustion, contraindications and adverse events are not described, nor to expect from hypothetical effect. For now, it remains unclear if moxibustion followed by an ECV attempt is beneficial in terms of a reduced need for ECV as implementation is limited and therefore could result in a lower caesarean delivery rate. Future research should include women with breech presentation at 32 weeks of gestation and ECV should be recommended in case of persisting breech presentation in both treatment and control group. This study should be of sufficient power to answer this research question.

The positive effect of moxibustion treatment is thought to be the increase of fetal move-ment, which is demonstrated by two trials24,27, included in this review. Several hypotheses are suggested to explain this increase of fetal movements. It is for example thought that moxibustion therapy might stimulate the production of placental estrogens and maternal prostaglandin due to hyperthermia.35 It also might increase free cortisol plasma levels in women.27 Therefore, future trials evaluating moxibustion therapy for breech presentation should also focus on its pathophysiology, and measure for example estrogen and cortisol levels before and after treatment.

In conclusion, prevention of breech presentation at term is an important topic in cur-rent obstetric practice considering the high caesarean delivery rate for this indication.3,4 For clinical practice, if breech presentation exists at 32 weeks gestation, moxibustion on acupoint BL67 may increase cephalic version and could be discussed in shared decision making. If persisting breech presentation occurs, ECV should be recommended from 36 weeks of gestation onwards.

References

1. Hickok DE, Gordon DC, Milberg JA, Williams MA, Daling JR. The frequency of breech presentation by gestational age at birth: a large population-based study. Am J Obstet Gynecol. 1992;166(3):851–2.

2. Hannah ME, Hannah WJ, Hewson SA, Hodnett ED, Saigal S, Willan AR. Planned caesarean section versus planned vagi-nal birth for breech presentation at term: a randomised multicentre trial. Term Breech Trial Collaborative Group. Lancet. 2000;356(9239):1375–83.

3. Molkenboer JFM, Bouckaert PXJM, Roumen FJME. Recent trends in breech delivery in the Netherlands. Br. J. Obstet. Gynaecol. 2003;110(10):948–51.

4. Rietberg CCT, Elferink-Stinkens PM, Visser GHA. The effect of the Term Breech Trial on medical intervention behaviour and neonatal outcome in The Netherlands: an analysis of 35,453 term breech infants. Br. J. Obstet. Gynaecol. 2005;112(2):205–9.

5. Villar J, Valladares E, Wojdyla D, et al. Caesarean delivery rates and pregnancy outcomes: the 2005 WHO global survey on maternal and perinatal health in Latin America. Lancet. 2006;367(9525):1819–29.

6. Lydon-Rochelle M, Holt VL, Easterling TR, Martin DP. Risk of uterine rupture during labor among women with a prior cesarean delivery. New Engl J. 2001;345(1):3–8.

7. Hofmeyr GJ, Kulier R, West HM. External cephalic version for breech presentation at term. Cochrane database Syst Rev. 2015;4

8. Grootscholten K, Kok M, Oei SG, Mol BWJ, van der Post JA. External cephalic ver-sion-related risks: a meta-analysis. Obstet Gynecol. 2008;112(5):1143–51.

9. Rosman AN, Vlemmix F, Beuckens A, et al. Facilitators and barriers to external cephalic version for breech presentation at term among health care providers in the Netherlands: A quantitative analysis.

Midwifery. 2014;30(3):e145–50.10. Rosman ANA, Vlemmix FF, Fleuren MAHM,

et al. Patients’ and professionals’ barriers and facilitators to external cephalic version for breech presentation at term, a qualita-tive analysis in the Netherlands. Midwifery. 2014;30(3):324–30.

11. Vlemmix F, Kuitert M, Bais J, et al. Patient’s willingness to opt for external cephalic version. J Psychosom Obstet Gynaecol. 2013;34(1):15–21.

12. Eisenberg DM, Davis RB, Ettner SL, et al. Trends in alternative medicine use in the United States, 1990-1997: results of a follow-up national survey. JAMA. 1998;280(18):1569–75.

13. Su D, Li L. Trends in the use of comple-mentary and alternative medicine in the United States: 2002-2007. J Health Care Poor Underserved. 2011;22(1):296–310.

14. Harrigan JT. Patient disclosure of the use of complementary and alternative medicine to their obstetrician/gynaecologist. J Obstet Gynaecol. 2011;31(1):59–61.

15. Steel A, Adams J, Sibbritt D, Broom A, Gallois C, Frawley J. Utilisation of com-plementary and alternative medicine (CAM) practitioners within maternity care provision: results from a nationally repre-sentative cohort study of 1,835 pregnant women. BMC Pregnancy Childbirth. 2012;12:146.

16. Münstedt K, Maisch M, Tinneberg HR, Hübner J. Complementary and alternative medicine (CAM) in obstetrics and gynaecol-ogy: a survey of office-based obstetricians and gynaecologists regarding attitudes towards CAM, its provision and coopera-tion with other CAM providers in the state of Hesse, Germany. Arch Gynecol Obstet. 2014;290(6):1133–9.

17. Coyle ME, Smith C a, Peat B. Cephalic version by moxibustion for breech pre-

88 | Chapter 4

sentation. Cochrane Database Syst Rev. 2005;(2).

18. Do CK, Smith C a, Dahlen H, Bisits A, Schmied V. Moxibustion for cephalic ver-sion: a feasibility randomised controlled trial. BMC Complement Altern Med. 2011;11(1):81.

19. Vas J, Aranda-Regules JM, Modesto M, et al. Using moxibustion in primary health-care to correct non-vertex presentation: a multicentre randomised controlled trial. Acupunct Med. 2013;31(1):31–8.

20. Coulon C, Poleszczuk M, Paty-Montaigne M-H, et al. Version of breech fetuses by moxibustion with acupuncture: a ran-domized controlled trial. Obstet Gynecol. 2014;124(1):32–9.

21. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535.

22. Higgins JPT, Altman D, Gøtzsche P, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomized trials. 2011.

23. Huang W. Comparison study on mox-ibustion and knee- chest position for conversion of fetal presentation. Zhongguo Zhongxiyi Jiehe Zazhi. 1990;10:105–106.

24. Cardini F, Weixin H. Moxibustion for correc-tion of breech presentation: a randomized controlled trial. JAMA. 1998;280(18):1580–1584. doi:10.1001/jama.280.18.1580.

25. Lin Y, Zhang D, Hao Y, Duan X. Combination of moxibustion at point Zhiyin and knee-chest position for correction of breech pregnancy in 63 cases. Chinese Acupunct Moxibustion. 2002;22(12):811–812.

26. Chen Y, Yang LW. Moxibustion on Zhiyin plus raising buttocks in a lateral position for correction fetal presentation in 73 cases.

Clin J Tradit Chinese Med. 2004;16(4):333.27. Neri I, Airola G, Contu G, Allais G, Facchinetti

F, Benedetto C. Acupuncture plus moxi-bustion to resolve breech presentation: a randomized controlled study. J Matern Fetal Neonatal Med. 2004;15(4):247–252.

28. Cardini F, Lombardo P, Regalia AL, et al. A randomised controlled trial of moxibus-tion for breech presentation. Br. J. Obstet. Gynaecol. 2005;112(6):743–747.

29. Yang F. Comparison of knee-chest position plus moxibustion on Zhiyin with knee-chest position for breech presentation. J Sichuan Tradit Chinese Med. 2006;24:106–107.

30. Neri I, De Pace V, Venturini P, Facchinetti F. Effects of three different stimulations (acu-puncture, moxibustion, acupuncture plus moxibustion) of BL.67 acupoint at small toe on fetal behavior of breech presentation. Am J Chin Med. 2007;35(1):27–33.

31. Millereau M, Branger B, Darcel F. Fetal ver-sion by acupuncture (moxibustion) versus control group. J Gynecol Obstet Biol Reprod (Paris). 2009;38(6):481–7.

32. Guittier M-J, Pichon M, Dong H, Irion O, Boul-vain M. Moxibustion for Breech Version. J Obstet Gynaecol. 2009;114(5):1034–1040.

34. Grootscholten, K., Kok, M., Oei, S. G., Mol, B. W. J. & van der Post, J. A. External cephalic version-related risks: a meta-analysis. Obstet. Gynecol. 112, 1143–51 (2008).

35. Vas J, Aranda JM, Nishishinya B, et al. Correction of nonvertex presentation with moxibustion: a systematic review and metaanalysis. Am J Obstet Gynecol. 2009;201(3):241–259.

Chapter 5Beta-mimetics for external cephalic version, a secondary analysis of the external cephalic version with uterine relaxation trial to determine who benefits most

Joost VelzelEwoud SchuitFloortje VlemmixJoris A. M. van der PostBen Willem MolMarjolein KokMohammad H. Zafarmand

Submitted

92 | Chapter 5

Abstract

Objective: To investigate patient characteristics that influence the chance for a successful external cephalic version (ECV) with tocolytics.

Study design: This is a secondary analysis of a multicenter, open-label randomized con-trolled trial that assessed the effectiveness of atosiban (n=416) compared to fenoterol (n=414) as uterine relaxant during ECV in women with a singleton fetus in breech presen-tation with a gestational age of 34 weeks or more. The primary outcome measures were a fetus in cephalic position 30 minutes after the procedure and cephalic presentation at delivery. Using logistic regression modelling, we investigated associations between out-come and 14 different maternal and fetal characteristics, all collected at study entry.

Results: Among the studied characteristics nine of them showed a statistically significant association with successful ECV. Multiparity, higher gestational age, higher estimated fetal weight, higher Amniotic Fluid Index and presence of relaxation of uterus were associated with a higher chance of ECV success; while Frank breech, lateral left position of the fetal spine and engaged breech were associated with a lower chance of ECV success. However, none of the markers were associated with higher chances of a successful ECV following administration of atosiban compared to fenoterol (P-value for interaction >0.10).

Conclusions: In pregnancies with a singleton fetus in breech presentation and a gesta-tional age of more than 34 weeks, the investigated prognostic markers for successful EVC are unlikely to be helpful in identifying women who have a high chance to benefit more from administration of atosiban or fenoterol as uterine relaxant during ECV.

Beta-mimetics for ECV, a secondary analysis of the RCT to determine who benefits most | 93

Introduction

External cephalic version (ECV) is a safe obstetrical procedure that reduces non-cephalic birth and caesarean delivery by approximately 50%.1 Improving the effectiveness of ECV can highly contribute to the goal to reduce elective caesarean deliveries.

To enhance the success rate of ECV, drug induced uterine relaxation is the most promising to increase the success of ECV.2 In the past, beta-mimetics have shown to be the most effective as uterine relaxation for this matter.2 As this type of drug gives substantial side effects in terms of palpations, flushes and headaches, routine implementation could be low. The oxytocin receptor blocker, atosiban, which is used in the prevention of preterm birth3, was recently studied as uterine relaxant for ECV by our group in an RCT. We found that uterine relaxation with atosiban for ECV resulted in a lower rate of fetuses in the cephalic position after 30 minutes compared with fenoterol (34% v 40%, RR 0.73, 95% CI 0.55 to 0.93).4 This led to a higher risk of caesarean deliveries after atosiban treatment than after fenoterol treatment (RR 1.09, 95% CI 0.96 to 1.20).4

A matter of concern is whether this finding is generalizable to all women who opt for ECV. For successful ECV, multiple predictors are known, and beta-mimetics give side effects in up to 75% of all women.4 To examine who is likely to benefit most from tocolytic medica-tion, we performed a post hoc analysis of data from the “ECV with uterine relaxation” trial. We evaluated the known markers for successful ECV with atosiban compared to fenoterol for uterine relaxation during the procedure.

94 | Chapter 5

Methods

Study design and patientsThe ECV with uterine relaxation trial was a multicenter, open-label randomized controlled trial in one academic and seven teaching hospitals in the Netherlands. The background, methods, baseline characteristics of the randomized patients, and results has previously been reported elsewhere.4

In brief, the trial included 830 women with a singleton fetus in breech position who were scheduled for ECV. The study was approved by the research ethics committee of the Academic Medical Center in Amsterdam (reference number MEC 08-364) and by the board of directors of each of the participating hospitals and was registered in the Dutch Trial register (NTR 1877). All women presenting between August 2009 and May 2014 with a singleton fetus in breech position who were scheduled for ECV were eligible for the study. Exclusion criteria were maternal age less than 18 years old, gestational age below 34 weeks, any contra-indication to vaginal birth (e.g. placenta praevia), any con-traindication for ECV (scarred uterus other than transverse in the lower segment, known uterine anomalies, placental abruption in history or signs of placental abruption, severe preeclampsia or HELLP syndrome, third trimester blood loss or seven days prior to ECV, ruptured membranes), any known contra-indication to one of the two drugs, suspected intrauterine growth restriction (defined as estimated fetal weight <P5 for gestational age assessed by ultrasonography), severe oligohydramnios (deepest pool < 2 cm), fetal anom-alies or non-reassuring fetal heart rate monitoring. Women were randomly assigned (1:1 ratio) to receive atosiban (intervention group, n=416) or fenoterol (control group, n=414). Participating woman received an intravenous bolus of atosiban (6.75 mg in 0.9 mL (7.5 mg/mL)) or fenoterol (40 μg in 0.8 mL (0.5 mg/10 mL)) administered by a physician fifteen minutes before starting ECV. Experienced obstetricians and midwives performed the ECV procedure, and the procedure was successful if the fetus was still in cephalic position 30 minutes after the ECV attempt. For the ECV procedure a forward and backward roll was allowed. The fetal heart rate was monitored intermittently by ultrasound scanning during the procedure, followed by electronic fetal heart rate monitoring after the procedure. If fetal bradycardia would occur, the duration was registered. Non-sensitized Rhesus neg-ative women received anti-D immunoglobulin (1,000 international units intramuscularly) after the ECV procedure.

The primary outcome measures were cephalic presentation 30 minutes after the proce-dure, confirmed by ultrasound, and cephalic presentation at delivery. Secondary outcomes were mode of delivery, and complications of ECV events due to atosiban or fenoterol. Based on literature5,6 and clinical reasoning, we selected 14 potential markers which were

reportedly prognostic and therefore could have potentially been informative for treatment selection. All markers are available in clinical practice and were recorded before the ECV attempt. The markers we selected were parity (multiparitiy vs. nulliparitiy), maternal age (years), body mass index (BMI, kg/m2), ethnicity (non-Caucasian vs. Caucasian), gestational age (weeks), placental location (non-anterior vs, anterior), Type of breech (other types vs. Frank breech), estimated fetal weight (EFW, gr), amniotic fluid index in cm (AFI), position of the fetal spine (others vs. lateral left), breech engagement (any engagement vs. non-en-gaged), symphysis fundal height (weeks), possibility of palpation of the fetal head (difficult vs. easy), and relaxation of the uterus (no vs. yes). Ethnicity initially consisted of six differ-ent groups, but it was recorded as Caucasian or non-Caucasian in this study, as a review showed a negative effect of Caucasian ethnicity on successful ECV.5 EFW was calculated with the Hadlock formula.7

Statistical analysisWe performed an intention-to-treat analysis at the randomization time. To increase the statistical power and reduce bias from a complete cases analysis, we used Multivariate Imputations by Chained Equations (MICE) for missing values.8 Ten datasets were created to handle missing values. For each marker we developed a univariable logistic regression model to predict the chance of successful ECV by marker (independent variable), treat-ment (administration of atosiban vs. fenoterol), and marker-by-treatment interaction. The latter term expresses to what extent the treatment effect is associated with the marker value. In this context, we define that markers could be potentially useful for treatment selection if they show a biologically plausible qualitative interaction with treatment.9 A marker has a qualitative interaction with treatment if it is risk factor (odds ratio >1) in one arm of trial and protective factor in the other arm of trial (odds ratio <1). The presence of a qualitative interaction would imply that for a subgroup of pregnant women uterine relaxation with atosiban is beneficial, and for the other subgroup administration of feno-terol is beneficial. We selected markers that showed marker-by-treatment interaction effect P-values below 0.10 for multivariable model building.10 To investigate the possibility of non-linear associations for continuous markers, we repeated model building using natural splines up to four degrees of freedom and tested if using splines would improve model fit in terms of Akaike Information Criterion (AIC). Given the exploratory nature of our analysis, we focused on the magnitude of the interaction and the associated precision, not on statistical hypothesis testing. All analyses were done using R for Windows (Version 3.4.3; R Foundation for Statistical Computing, Vienna, Austria).

96 | Chapter 5

Results

Between August 2009 and May 2014, 830 women were enrolled and randomized (Figure 1); 416 to atosiban (intervention group) and 414 to fenoterol (control group). Baseline characteristics of the women and fetuses who participated in the initial trial are summa-rized in Table 1. Cephalic presentation 30 minutes after ECV occurred in 34% of women in the atosiban group compared with 40% in the fenoterol group (RR 0.73, 95% CI 0.55 to 0.93). Caesarean delivery was performed in 60% (n=240) of women in the atosiban group and 55% (n=218) in the fenoterol group (RR 1.09, 95% CI 0.96 to 1.22).

Assessed for eligibility (n=910)

Allocated to atosiban (n=416)

Randomized (n=830)

Excluded (n=80)

Not meeting inclusion criteria (n=18)

Declined to participate (n=58)

Declined ECV (n=4)

Allocated to fenoterol (n=414)

Intention to treat analysis (n=416) Intention to treat analysis (n=414)

Figure 1. Randomization, treatment and follow-up of participants

Table 2 summarizes the results of our analysis of the associations between markers and successful ECV separately for atosiban and fenoterol. Among the studied markers nine of them showed a statistically significant association with successful ECV (p-value main effect < 0.05). Multiparity, gestational age, estimated fetal weight, Amniotic Fluid Index and presence of relaxation of uterus showed a positive association with ECV success; while Frank breech, lateral left position of the fetal spine and engaged breech showed a

negative association with ECV success. However, the strength of associations were similar in women who received atosiban or fenoterol and none of the markers were associated with higher chances of a successful ECV following administration of atosiban compared to fenoterol (P-value for interaction >0.10).

Table 1. Maternal and fetal characteristics of the participants in initial trial at study entry

Baseline characteristics Atosiban (n=416)

% missing Fenoterol (n=414)

% missing

Nulliparity (%) 62.2 0 62.3 0

Mean maternal age (years) 32.2 0 32.4 0

Mean body mass index (kg/m2) 24.1 0 24.2 0

Caucasian ethnicity (%) 85.3 4 86.7 3

Gestational age (weeks) 35.8 1 35.9 0

Estimated fetal weight (grams) 2614 25 2598 24

Amniotic Fluid Index (cm) 13.8 37 13.7 32

Frank breech presentation (%) 71.2 7 65.7 6

Anterior placental location (%) 29.6 8 34.5 7

Lateral left position of spine (%) 40.6 17 42.8 12

Mean fundal Height (weeks) 36.2 14 36.3 14

Non-engaged breech (%) 25.6 10 26.3 8

Easy palpation of head (%) 72.1 18 73.9 17

Relaxed uterus (%) 63.0 18 67.9 17

98 | Chapter 5

Table 2. The relationship between characteristics and successful ECV is shown separately in women who received atosiban and in women who received fenoterol. For each biomarker, odds ratios (OR) show the relative change in the risk per unit increase in the marker. The interaction OR presents the ratio of the two odds ratios.

Potential treatment selection factors Atosiban (n=416) Fenoterol (n=414)

Subgroup(n)

Successful ECV n (%)

Odds Ratio(95% CI)

Subgroup(n)

Odds Ratio(95% CI)

p-value main effect c

p-valuefor interaction

Maternal age, years a 1.04 (0.99-1.1)

1.03 (0.98-1.07)

0.059 0.675

<25 15 5 (33.3) 18 5 (27.8)

25-29 102 25 (24.5) 98 39 (39.8)

30-34 205 74 (36.1) 188 79 (42.0)

≥35 94 34 (36.2) 110 45 (40.9)

Body Mass Index preconceptionally, kg/m2 a 1.01 (0.97-1.06)

1.04 (0.99-1.08)

0.122 0.459

≤20 49 16 (32.7) 66 25 (37.9)

20-25 225 75 (33.3) 216 83 (38.4)

25-30 97 31 (32.0) 84 36 (42.9)

≥30 45 16 (35.6) 48 24 (50.0)

Parity 3.12 (2.04-4.78)

4.76 (3.1-7.29)

<0.001 0.171

Nulliparous b 259 61 (23.6) 258 69 (26.7)

Multiparous 157 77 (49.0) 156 99 (63.5)

Gestational age at ECV, weeks a 1.11 (0.89-1.39)

1.25 (1.04-1.52)

0.014 0.424

≤36 144 50 (34.7) 145 60 (41.4)

36-36.9 208 61 (29.3) 194 68 (35.1)

≥37 64 27 (42.2) 75 40 (53.3)

Female ethnicity 0.72 (0.37-1.41)

0.82 (0.43-1.57)

0.261 0.790

Caucasian b 368 125 (34.0) 370 152 (41.1)

Non-Caucasian 48 13 (27.1) 44 16 (36.4)

Estimated Fetal Weight, kg a 1.85 (0.99-3.47)

2.47 (1.35-4.51)

0.001 0.517

<2.5 143 43 (30.1) 169 52 (30.8)

2.5-2.9 229 75 (32.8) 198 89 (44.9)

≥3.0 44 20 (45.5) 47 27 (57.4)

Amniotic Fluid Index, cm a 1.09 (1.03-1.14)

1.05 (1.01-1.1)

<0.001 0.373

<10 49 7 (14.3) 40 10 (25.0)

10-14.9 140 40 (28.6) 162 54 (33.3)

≥ 15 227 91 (40.1) 212 104 (49.1)

Table 2. The relationship between characteristics and successful ECV is shown separately in women who received atosiban and in women who received fenoterol. For each biomarker, odds ratios (OR) show the relative change in the risk per unit increase in the marker. The interaction OR presents the ratio of the two odds ratios.

Subgroup(n)

Odds Ratio(95% CI)

Subgroup(n)

Odds Ratio(95% CI)

Maternal age, years a 1.04 (0.99-1.1)

1.03 (0.98-1.07)

0.059 0.675

<25 15 5 (33.3) 18 5 (27.8)

25-29 102 25 (24.5) 98 39 (39.8)

30-34 205 74 (36.1) 188 79 (42.0)

≥35 94 34 (36.2) 110 45 (40.9)

Body Mass Index preconceptionally, kg/m2 a 1.01 (0.97-1.06)

1.04 (0.99-1.08)

0.122 0.459

≤20 49 16 (32.7) 66 25 (37.9)

20-25 225 75 (33.3) 216 83 (38.4)

25-30 97 31 (32.0) 84 36 (42.9)

≥30 45 16 (35.6) 48 24 (50.0)

Parity 3.12 (2.04-4.78)

4.76 (3.1-7.29)

<0.001 0.171

Nulliparous b 259 61 (23.6) 258 69 (26.7)

Multiparous 157 77 (49.0) 156 99 (63.5)

Gestational age at ECV, weeks a 1.11 (0.89-1.39)

1.25 (1.04-1.52)

0.014 0.424

≤36 144 50 (34.7) 145 60 (41.4)

36-36.9 208 61 (29.3) 194 68 (35.1)

≥37 64 27 (42.2) 75 40 (53.3)

Female ethnicity 0.72 (0.37-1.41)

0.82 (0.43-1.57)

0.261 0.790

Caucasian b 368 125 (34.0) 370 152 (41.1)

Non-Caucasian 48 13 (27.1) 44 16 (36.4)

Estimated Fetal Weight, kg a 1.85 (0.99-3.47)

2.47 (1.35-4.51)

0.001 0.517

<2.5 143 43 (30.1) 169 52 (30.8)

2.5-2.9 229 75 (32.8) 198 89 (44.9)

≥3.0 44 20 (45.5) 47 27 (57.4)

Amniotic Fluid Index, cm a 1.09 (1.03-1.14)

1.05 (1.01-1.1)

<0.001 0.373

<10 49 7 (14.3) 40 10 (25.0)

10-14.9 140 40 (28.6) 162 54 (33.3)

≥ 15 227 91 (40.1) 212 104 (49.1)

100 | Chapter 5

Table 2. Continued.

Subgroup(n)

Odds Ratio(95% CI)

Subgroup(n)

Odds Ratio(95% CI)

Type of breech 1.73 (1.09-2.76)

2.61 (1.71-3.99)

<0.001 0.204

Frank breech b 318 96 (30.2) 283 94 (33.2)

Other types 98 42 (42.9) 131 74 (56.5)

Placental location 1.37 (0.87-2.14)

1.29 (0.86-1.94)

0.091 0.852

Anterior b 133 38 (28.6) 155 57 (36.8)

Non-anterior 283 100 (35.3) 259 111 (42.9)

Position of the fetal spine 1.16 (0.77-1.74)

1.58 (1.07-2.35)

0.037 0.282

Lateral left b 197 62 (31.5) 203 71 (35.0)

Other positions 219 76 (34.7) 211 97 (46.0)

Breech engagement 0.25 (0.16-0.39)

0.35 (0.22-0.54)

<0.001 0.298

Non-engaged b 120 67 (55.8) 119 70 (58.8)

Engaged (descent into pelvis or fixed in pelvis)

296 71 (24.0) 295 98 (33.2)

Fundal height, weeks a 1.16 (0.98-1.38)

1.13 (0.99-1.29)

0.012 0.788

<36 63 17 (27.0) 74 29 (39.2)

36-36.9 229 73 (31.9) 210 80 (38.1)

≥37 124 48 (38.7) 130 59 (45.4)

Palpation of head possible 0.71 (0.37-1.37)

0.56 (0.29-1.1)

0.051 0.616

Easy b 364 124 (34.1) 369 155 (42.0)

Difficult 52 14 (26.9) 45 13 (28.9)

Relaxation of uterus 0.46 (0.27-0.78)

0.41 (0.24-0.72)

<0.001 0.785

Yes b 317 117 (36.9) 337 149 (44.2)

No 99 21 (21.2) 77 19 (24.7)a The odds ratios are calculated using the variables as a continuous factor and not a categorical factor. The categories presented here are just made for presentation purposes.b Reference category.c p-value main effect is the p-value of the association between the marker and successful ECV, irrespective of the treatment women received.

Table 2. Continued.

Subgroup(n)

Odds Ratio(95% CI)

Subgroup(n)

Odds Ratio(95% CI)

Type of breech 1.73 (1.09-2.76)

2.61 (1.71-3.99)

<0.001 0.204

Frank breech b 318 96 (30.2) 283 94 (33.2)

Other types 98 42 (42.9) 131 74 (56.5)

Placental location 1.37 (0.87-2.14)

1.29 (0.86-1.94)

0.091 0.852

Anterior b 133 38 (28.6) 155 57 (36.8)

Non-anterior 283 100 (35.3) 259 111 (42.9)

Position of the fetal spine 1.16 (0.77-1.74)

1.58 (1.07-2.35)

0.037 0.282

Lateral left b 197 62 (31.5) 203 71 (35.0)

Other positions 219 76 (34.7) 211 97 (46.0)

Breech engagement 0.25 (0.16-0.39)

0.35 (0.22-0.54)

<0.001 0.298

Non-engaged b 120 67 (55.8) 119 70 (58.8)

Engaged (descent into pelvis or fixed in pelvis)

296 71 (24.0) 295 98 (33.2)

Fundal height, weeks a 1.16 (0.98-1.38)

1.13 (0.99-1.29)

0.012 0.788

<36 63 17 (27.0) 74 29 (39.2)

36-36.9 229 73 (31.9) 210 80 (38.1)

≥37 124 48 (38.7) 130 59 (45.4)

Palpation of head possible 0.71 (0.37-1.37)

0.56 (0.29-1.1)

0.051 0.616

Easy b 364 124 (34.1) 369 155 (42.0)

Difficult 52 14 (26.9) 45 13 (28.9)

Relaxation of uterus 0.46 (0.27-0.78)

0.41 (0.24-0.72)

<0.001 0.785

Yes b 317 117 (36.9) 337 149 (44.2)

No 99 21 (21.2) 77 19 (24.7)a The odds ratios are calculated using the variables as a continuous factor and not a categorical factor. The categories presented here are just made for presentation purposes.b Reference category.c p-value main effect is the p-value of the association between the marker and successful ECV, irrespective of the treatment women received.

102 | Chapter 5

Discussion

In this study of women undergoing ECV with tocolytic medication, we investigated the 14 well-known clinical predictors of successful ECV for identification of women who might benefit most from the administration of atosiban compared to fenoterol. The treatment interaction tests could not identify any potential treatment selection marker that could indicate a successful ECV with administration of atosiban compared to fenoterol (P-values for interaction >0.10).

Our analysis was based on data of a nationwide randomized controlled trial of 830 women who underwent ECV and were randomly allocated to atosiban or fenoterol. None of the evaluated baseline clinical markers had affected the choice of tocolytic agent. This gave us the opportunity to study the association between treatment selection markers and tocolytic agents without the risk of selection bias. We incorporated all relevant markers identified by previous systematic reviews on both clinical and ultrasound variables which are easy available and inexpensive to measure in daily clinical practice and found them to be informative for the chance of success.

One should keep in mind that the real chance of success could be higher as our data come from a RCT in which women already decided on an ECV attempt compared to women who are considering this intervention. Nevertheless, our success rate corresponds to a large cluster RCT in the Netherlands in which all women counselled for ECV were included.11 Furthermore, it might be that women with an anticipated high chance of successful ECV might not have been included by caregivers in the RCT about uterine relaxants. On the other hand, our success rate in the RCT could be higher compared to daily practice as ECV attempts were undertaken in centers that perform ECV frequently and it is strongly believed that experience of the performing caretaker and in which setting the ECV attempt is performed, effects the success rate of ECV positivily.12,13 Unfortunately, information on who supervised the attempt and in which setting an ECV took place (special ECV clinic or in outpatient setting) was not available in our study, meaning that we were unable to determine treatment selection in different types of ECV management.

The initial trail was conducted to evaluate the oxytocin receptor blocker atosiban, to feno-terol, a beta-mimetic. For this trial, we initially chose a beta-mimetic as a recent Cochrane review found that women using beta-mimetics for uterine relaxation for ECV had a higher rate of cephalic birth (pooled RR 1.7, 95% CI 1.14 to 2.5) and a lower chance on caesarean delivery (pooled RR 0.77, 95% CI 0.67 to 0.88), compared to placebo.2 After the trail, we concluded that atosiban is inferior to fenoterol to increase successful ECV, and this was supported by the development of the prediction model of data of this trial, in which the

administration of fenoterol is a predictor for successful ECV.4,14 In counseling women for an ECV attempt, it is important to inform that beta-mimetics give side effects, though, this is mostly limited to transient feelings of tachycardia (71%). Occurrence of unwanted side effects with the need to stop the ECV attempt, such as severe hypotension, is minimal (less than 1%).4 Also, fetal adverse events due to beta-mimetics, hypothesized as transient hypoglycemia, is not observed.15 In addition, even though the complication rate of ECV is very low (pooled risk on fetal death is 0.19%16), it seems to be independently to the use of tocolytics.16 Therefore, we believe that beta-mimetics are safe to use for tocolysis for ECV.

Even though we could not find an association between subgroups and the benefit of administration of a tocolytic agent, the findings of our RCT and prediction model clearly demonstrate that fenoterol increases the chance of successful ECV and reducing caesar-ean delivery for this matter.4,14 Supported by these conclusions, beta-mimetics for ECV need to be a discussed in shared decision making. Especially for caretakers, as we know that implementation of beta-mimetics to improve ECV outcome is limited due to mater-nal side effects17, it is important to stress that a patient’s preference study found that women are willing to undergo ECV as the gain in success rate outweigh the unpleasant side effects.18 Implementation of ECV varies widely. Therefore, future research could focus on whether administration of beta-mimetics for ECV in selected subgroups compared to routine administration would be equally successful but more acceptable in daily practice. A pragmatic RCT in which women are randomized to an attempt with a beta-mimetic, or to a first attempt without beta-mimetics and if unsuccessful, a repeat ECV with beta-mimetics, could potentially provide evidence to define these subgroups.

Conclusion

In pregnancies with a singleton fetus in breech presentation and a gestational age of more than 34 weeks, most of the known prognostic markers for successful ECV seem unlikely to be helpful in identifying a subgroup who could benefit from administration of atosiban or fenoterol as uterine relaxant during ECV. For clinical practice, counseling about tocolysis with fenoterol must be performed before a first attempt.

104 | Chapter 5

References

3. Van Vliet, E. O. G. et al. Nifedipine versus atosiban for threatened preterm birth (APOSTEL III): A multicentre, randomised controlled trial. Lancet 387, 2117–2124 (2016).

5. Kok, M. et al. Clinical factors to predict the outcome of external cephalic version: a metaanalysis. Am. J. Obstet. Gynecol. 199, 630.e1-7; e1-5 (2008).

7. Hadlock, F. P., Harrist, R. B., Sharman, R. S., Deter, R. L. & Park, S. K. Estimation of fetal weight with the use of head, body, and femur measurements- a prospective study. Am. J. Obstet. Gynecol. 151, 333–337 (1985).

8. Buuren, S. van & Groothuis-Oudshoorn, K. mice : Multivariate Imputation by Chained Equations in R. J. Stat. Softw. (2011).

9. Polley, M.-Y. C. et al. Statistical and Practi-cal Considerations for Clinical Evaluation of Predictive Biomarkers. J. Natl. Cancer Inst. (2013).

10. Steyerberg, E. W., Eijkemans, M. J. & Habbema, J. D. Stepwise selection in small data sets: a simulation study of bias in logistic regression analysis. J. Clin. Epide-miol. (1999).

12. Kuppens, S. M. I., Francois, A. M. H., Hasaart, T. H. M., van der Donk, M. W. P. & Pop, V. J. M. [Fewer breech deliveries after imple-mentation of a modified cephalic version protocol]. Ned. Tijdschr. Geneeskd. 154, A63 (2010).

13. Kim, S. Y. et al. Evaluation of the learning curve for external cephalic version using cumulative sum analysis. Obstet. Gynecol. Sci. 60, 343–349 (2017).

14. Velzel, J. et al. Development and internal validation of a clinical prediction model for external cephalic version. Eur. J. Obstet. Gynecol. Reprod. Biol. 228, 137–142 (2018).

15. Neilson, J. P., West, H. M. & Dowswell, T. Betamimetics for inhibiting preterm labour. Cochrane Database Syst. Rev. 2 (2014).

Chapter 6Prediction models for successful external cephalic version: a systematic review

Joost VelzelMarcella de HundtFrederique M. MulderJan F. M. MolkenboerJoris A. M. van der PostBen Willem MolMarjolein Kok

European Journal of Obstetrics & Gynecology and Reproductive Biology, December 2015; 195:160-167

108 | Chapter 6

Abstract

Objective: To provide an overview of existing prediction models for successful ECV, and to assess their quality, development and performance.

Study design: We searched MEDLINE, EMBASE and the Cochrane Library to identify all articles reporting on prediction models for successful ECV published from inception to January 2015. We extracted information on study design, sample size, model-building strategies and validation. We evaluated the phases of model development and summa-rized their performance in terms of discrimination, calibration and clinical usefulness. We collected different predictor variables together with their defined significance, in order to identify important predictor variables for successful ECV.

Results: We identified eight articles reporting on seven prediction models. All models were subjected to internal validation. Only one model was also validated in an external cohort. Two prediction models had a low overall risk of bias, of which only one showed promising predictive performance at internal validation. This model also completed the phase of external validation. For none of the models their impact on clinical practice was evaluated. The most important predictor variables for successful ECV described in the selected articles were parity, placental location, breech engagement and the fetal head being palpable. One model was assessed using discrimination and calibration using inter-nal (AUC 0.71) and external validation (AUC 0.64), while two other models were assessed with discrimination and calibration, respectively.

Conclusions: We found one prediction model for breech presentation that was validated in an external cohort and had acceptable predictive performance. This model should be used to council women considering ECV.

Prediction models for successful ECV: a systematic review | 109

Introduction

Breech presentation occurs in 3 to 4% of all term pregnancies.1,2 Vaginal delivery of a fetus in breech position is associated with higher rates of neonatal morbidity and mor-tality compared to a planned caesarean delivery.3 Consequently, worldwide the rate of term vaginal breech deliveries has declined substantially.4,5 However, caesarean delivery is associated with short and long-term consequences for maternal and neonatal health.6,7 External cephalic version (ECV) is a safe and effective procedure to reduce the number of breech presentations at term and consequently the caesarean delivery rate for this indication.8–12 Considering the worldwide general rise in caesarean delivery rate in the last decade13–15 and the urgency to put a hold to this rise6,7, ECV is an important intervention that can contribute to this goal.

Even though procedure related complications rarely occur, concerns about safety of the procedure can be a reason for women not to accept an ECV attempt.16,17 Additionally, knowledge about the effectiveness of ECV also influences acceptance of ECV.16,17 The success rate of ECV varies from approximately 35% up to 86% in the literature with an average of 50-60%.4,18 Thus, a reliable more precise and individualized prediction of suc-cessful ECV could be useful to counsel women for an ECV attempt. Previous studies have shown that clinical and ultrasound characteristics are associated with success or failure of an ECV procedure.19,20 There are several prediction models that enable individualized prediction of the outcome of an ECV attempt. However, since the use of poor-quality prediction models could have a negative effect on decision-making21, careful evaluation is needed before these models can be implemented in clinical practice.

Therefore, the aim of this review is to give an overview of existing prediction models for successful ECV, to evaluate their quality, development and performance and to identify important predictor variables described in the selected articles.

110 | Chapter 6

Methods

Study identificationWe performed a computerized MEDLINE, EMBASE and Cochrane databases search to identify all studies reporting on prediction models for successful ECV published from inception to April 2015. Language restrictions were not applied. Together with a clinical librarian we developed a search strategy including all known synonyms for the term ‘exter-nal cephalic version’ and we performed a search filter for prediction models. References from selected publications were manually searched for additional relevant articles not identified by the computerized search.

Study selectionThis review focused on articles that reported on a prediction model to predict the out-come of an ECV attempt. A prediction model was defined as a multivariable model that expressed the chance of successful ECV as a function of at least two predictor variables. In order to be eligible and selected, the articles had to report on a prediction model sufficient to make predictions for individual cases. Two independently working reviewers (JV and FM) selected articles by assessing titles and abstracts. When there were any doubts about the eligibility after reading title and abstract, the article was included for full text reading to make sure no potential eligible article was missed. Any disagreements were resolved by consensus and, if necessary, by a third reviewer (MK).

Study quality assessmentA framework was developed based on the recommended guideline of Hayden et al22 in combination with recommendations derived from the Quality In Prognosis Studies tool23, in order to adequately operationalize items for assessing bias and study quality of the selected articles. The framework was divided into four sections: study participation, pre-dictor variables/prognostic factor measurement, outcome measurement and analysis.

As recommended by Hayden et al, we assessed our four potential bias domains in two steps. During the first step, we assessed the fully operationalized relevant quality items which were scored with ‘yes’, ‘no’, ‘partly’ or ‘unclear’. During the second step, we used these item responses to judge each of the four potential bias domains. Each of these domains was rated as having high, moderate or low risk of bias. This approach to qual-ity appraisal follows a method that has been described by Wortman24 as ‘mixed-criteria’ quality assessment.

Additionally, we assessed the overall risk of bias and therefore the overall study quality for each study by combining the ratings of the four bias domains of our framework. We took into account the importance of each individual bias domain, with ‘analysis’ and ‘predictor variables’ being the most important domains, which were determined a priori.

Predictor variablesWe collected all predictor variables used in the selected articles together with their defined significance, in order to identify the most important predictor variables for successful ECV. Predictor variables are the potential predictors, which were tested both during model development and in the final model. The most important predictor variables were those described as statistically significant input variables in both univariable and multivariable regression analysis in three or more studies. When a predictor variable was only tested in two studies, and not in the other included studies, both studies had to describe this variable as statistically significant to label it as one of the most important predictors.

Model development assessmentWe assessed the development of the prediction models with a published evaluation scheme25, which distinguishes three phases of model development: model derivation, model validation and impact analysis. In the model derivation phase, predictors are iden-tified, based on prior knowledge, and the weight of each predictor (regression coefficient) is calculated. Model validation, the second phase, consists of an internal and external validation phase. With internal validation, the ability of the prediction model is evaluated in the group of patients in which the model was developed. Since internal validation consis-tently gives a too optimistic impression about the predictive accuracy of a model, external validation is a crucial next step in assessing the performance of a model. With external validation, ‘generalizability’ or ‘transportability’ of the prediction model is evaluated. This means that the ability of the model to predict the outcome in populations other than the population in which the model was developed, is evaluated.

The third and final phase of model development is impact analysis. During this phase, prediction models are tested for their ability to change and improve clinicians’ decisions and patient outcomes.

All prediction models identified in this review are classified into the different phases of model development. We tried to contact all authors of the selected articles by email to investigate if there were new results of model validation or impact analysis, which were not published yet.

112 | Chapter 6

Model performanceFor the models that were evaluated we quantified the performance of the prediction models by assessing calibration, discrimination and clinical usefulness. Calibration describes the agreement between observed outcomes and predictions of the outcomes. Calibration usually is reported as a calibration plot in which predicted probability is plot-ted against the actual outcome fraction. Usually reported in a graphical illustration of the Hosmer-Lemeshow goodness-of-fit test.

Overall statistical performance measures incorporate both calibration and discrimination aspects. In this case, discrimination relates to how well a prediction model can discriminate those with the outcome from those without the outcome. Discrimination is commonly reported as the concordance (c) statistic, which indicates the discriminative ability of gen-eralized linear regression models. For a binary outcome, c is identical to the area under the receiver operating characteristic (ROC) curve (AUC). An AUC of 1 implies perfect dis-crimination, whereas an AUC of 0.5 means that the test does not discriminate at all. For this review, a model is considered to have poor performance if the AUC lies between 0.50 and 0.70. An AUC between 0.70 and 0.80 represents fair performance, and an AUC between 0.80 and 0.90 represents good performance. Another way to evaluate model performance is by assessing its clinical usefulness. Both clinical usefulness and discrimi-nation describe the ability of a model in distinguishing between women with a poor and women with a good chance of successful ECV, which is influenced by the cut-off point used for this classification. Once the cut-off point is chosen, clinical usefulness measures such as accuracy, sensitivity or specificity, positive or negative predictive values or likelihood ratios can be defined.

Results

Study identification and selectionWe identified 373 potential eligible articles with the computerized search. After screening titles and abstracts we selected 21 articles for further reading. One paper was retrieved from cross-references. One article was not obtainable in full text version and was therefore excluded since it could not be assessed for study selection. Fourteen articles used uni-variable and/or multivariable regression analysis to evaluate the significance of predictor variables but did not develop a prediction model in order to express the individual chance of successful ECV. Finally, eight articles met our inclusion criteria and were included in our review (Figure 1).26–33

tification

Screen

Eligibility

Includ

Full-text articles assessed for eligibility (n=24)

Records excluded on base of title/abstract (n=349)

Records identified MEDLINE, EMBASE and Cochrane databases (n=373)

One study retrieved from cross-references (n=1)

Full-text articles excluded(n=15)

Figure 1. Flowchart of study selection process

114 | Chapter 6

Study characteristicsStudy characteristics are shown in Table 1. We identified eight articles reporting on seven prediction models. One study described the development and internal validation of one prediction model in two separate articles.27,28 One study29 described the development and internal validation in a first article and the external validation of the prediction model in a separate article.33 One study30 described the development of two separate prediction models in one article.

Table 1. Study characteristics of included articles

Study (year) Patients Exclusion criteria Outcome Individuals (n) Overall success rate (%)

Study design

Burgos et al. (2011/ 2012)27,28

Women with a singleton pregnancy at ≥37 weeks, a live foetus in a breech position, and no contraindications to vaginal birth

Placenta praevia, placental abruption, oligohydramnios, foetal compromise, foetal death, severe foetal malformation, multiple gestation, Rhesus sensitization, uterine abnormalities and clotting disorders

Successful ECV 1000(phase 1 = 500; phase 2 = 500)

phase 1 = 52,2%;phase 2 = 51,2% Overall = 51,7%

Prospective two-phase study

Kok et al. (2011)29

Women with a singleton pregnancy in breech presentation at ≥36 weeks of gestation

- Successful ECV 310 121 (39%) RCT

Aisenbrey et al. (1999)26

Women with breech presentation at ≥37 weeks of gestation

Placenta praevia, history of vaginal bleeding, AFI of at most 5 cm, foetal anomalies, foetal growth restriction, previous uterine scar other than low-transverse, non-reassuring non-stress test, EFW exceeding 4 kg, suspicion of nuchal cord or uterine malformation.

Successful ECV 128 78 (64%) Prospective cohort study

Wong et al. (2000)34

Intrauterine growth retardation, hypertension, placenta praevia, oligohydramnios, previous caesarean section and multiple pregnancy

Successful ECV 140 (phase 1 = 53; phase 2 = 87)

phase 1 = 64,2%; phase 2 =61,4%

Lau et al. (1997)30

Described by Stock et al37, full text version not available

Successful ECV 243 169 (69.5%) Prospective cohort study

Newman et al. (1993)31

Women with singleton gestations Mentioned contraindications for ECV: abruptio placentae, placenta praevia, premature rupture of membranes, foetal compromise during preliminary foetal heart rate monitoring

Successful ECV 394 (phase 1 = 108; phase 2 = 286 (20 for transverse lie))

166 (62.4%) in phase 2 (success rate transverse lie 18 (90%))

Retrospective and prospective two-phase study

Patient selection and inclusion criteria were not the same in all articles. Three articles included women with breech presentation with a gestational age of 36 weeks or more in their study29,30,34 whereas two studies included women with a gestational age of 37 weeks or more.26–28 One study did not report their inclusion criteria concerning gestational age.35 Two studies did not mention contraindications to perform ECV.29,30

Table 1. Study characteristics of included articles

Study (year) Patients Exclusion criteria Outcome Individuals (n) Overall success rate (%)

Study design

Burgos et al. (2011/ 2012)27,28

Women with a singleton pregnancy at ≥37 weeks, a live foetus in a breech position, and no contraindications to vaginal birth

Placenta praevia, placental abruption, oligohydramnios, foetal compromise, foetal death, severe foetal malformation, multiple gestation, Rhesus sensitization, uterine abnormalities and clotting disorders

Successful ECV 1000(phase 1 = 500; phase 2 = 500)

phase 1 = 52,2%;phase 2 = 51,2% Overall = 51,7%

Kok et al. (2011)29

Women with a singleton pregnancy in breech presentation at ≥36 weeks of gestation

- Successful ECV 310 121 (39%) RCT

Placenta praevia, history of vaginal bleeding, AFI of at most 5 cm, foetal anomalies, foetal growth restriction, previous uterine scar other than low-transverse, non-reassuring non-stress test, EFW exceeding 4 kg, suspicion of nuchal cord or uterine malformation.

Successful ECV 128 78 (64%) Prospective cohort study

Wong et al. (2000)34

Intrauterine growth retardation, hypertension, placenta praevia, oligohydramnios, previous caesarean section and multiple pregnancy

Successful ECV 140 (phase 1 = 53; phase 2 = 87)

phase 1 = 64,2%; phase 2 =61,4%

Lau et al. (1997)30

Described by Stock et al37, full text version not available

Successful ECV 243 169 (69.5%) Prospective cohort study

Newman et al. (1993)31

Women with singleton gestations Mentioned contraindications for ECV: abruptio placentae, placenta praevia, premature rupture of membranes, foetal compromise during preliminary foetal heart rate monitoring

Successful ECV 394 (phase 1 = 108; phase 2 = 286 (20 for transverse lie))

166 (62.4%) in phase 2 (success rate transverse lie 18 (90%))

Retrospective and prospective two-phase study

116 | Chapter 6

Study qualityThe results of the study quality assessment are shown in Table 2. Three studies27–29 were rated of high study quality. The four other studies26,30,34,35 were rated as having an overall moderate risk of bias and therefore moderate study quality. After critically appraising these four articles by comparing the ratings of each prompting item and bias domain, one study35 was considered to have the lowest overall study quality. This was due to the fact that, in addition to the ratings of moderate risk of bias for ‘analysis’ and ‘predictor variables’, this study did not completely describe inclusion criteria and patient characteristics.

Table 2. Quality of included studies

Burgos et al. (2011/2012)

Kok et al. (2011)

Wong et al. (2000)

Aisenbrey et al. (1999)

Lau et al. (1997)

man et al. (1993)

Study participation:

Description of setting and study period Yes Yes Yes Yes Yes Yes

Description of inclusion and exclusion criteria Yes Partly Yes Yes Partly Partly

Consecutive patient selection Yes Yes Yes Yes Yes Yes

Number of patients reported Yes Yes Yes Yes Yes Yes

Success rate of external cephalic version reported

Yes Yes Yes Yes Yes Yes

Clear description of patient characteristics Yes Yes Yes Yes Yes Partly

Predictor variables:

Clear definition of all predictor variables evaluated

Yes Yes Yes Yes Yes Yes

Description of proportion of participants with complete/missing data

No Yes No No No No

Outcome measures:

Clear definition of outcome Yes Yes Yes Yes Yes Yes

Analysis:

Description of evaluation measures Yes Yes Partly Partly Partly Partly

Description of model-building strategies Yes Yes Partly Partly Partly Partly

Overall quality*: + + +/- +/- +/- +/-

* This only concerns study quality based on operationalized items as mentioned in this table. Other quality aspects will be discussed in the discussion. + Low risk of bias; high study quality; +/- Moderate risk of bias; moderate study quality

Predictor variablesTable 3 shows a total of 23 different possible predictor variables, which were reported in the seven selected studies. We divided these predictor variables into three groups: patient characteristics, ultrasonographic findings and findings at clinical examination. The variables most commonly described, in four or more articles, were parity, gestational age, type of breech, placental location, estimated fetal weight (EFW) and amniotic fluid index. The most important predictor variables, described as statistically significant input variables in both univariable and multivariable analysis in three or more studies, were parity, pla-cental location, breech engagement and amniotic fluid index. Parity and placental location were described by all studies included in this review and were found to be statistically significant in both univariable and multivariable analysis in four studies and were incorpo-rated in their final model. Breech engagement was described and found to be significant in three studies and was incorporated in three final prediction models.30,34,35 Amniotic fluid index was described in all selected articles and was found to be statistically significant in three studies.27–29,34 However, only two studies included amniotic fluid index in their final model.27–29 Palpation of the fetal head was described and found to be significant in two studies, leading to incorporation of this predictor variable in two final models.30,34

Phases of model developmentAll of the included models had completed the phase of model derivation. With the excep-tion of one study26 all studies passed the phase of internal validation.26–29,31–33 One study30 performed an internal validation by cross-validation, by randomly dividing their dataset in two parts, developing two regression models and testing both models on the other subgroup. Three models were tested by temporal validation, where their model was tested on a group of subsequent patients in the same center.27,28,34,35 One study29 used bootstrapping for internally validating their model and this was also the only study that reached the phase of external validation.33 After contacting the authors we received one response from the study-group of Burgos et al, who claimed that their model is currently not being externally validated.

Performance of the prediction modelsAn overview and specification of the performance of the models described in the selected articles is presented in Table 4. Discrimination was described for two models27–29, and calibration was described for two models26,29,27. Only the performance of one model29 was evaluated by calibration and discrimination in both internal and external validation. Aisenbrey et al26 did not perform an internal validation of their model, nevertheless they assessed calibration with the Hosmer-Lemeshow test, which had a value of 0.92.

118 | Chapter 6

Table 3. Predictor variables evaluated and used in prediction models

Burgos et al. (2011/2012)

Kok et al. (2011)

Aisenbrey et al. (1999)

Wong et al. (2000)

Lau et al. (1997)

man et al. (1993)

Patient characteristics

Maternal age 3 2 3

Ethnicity 3

Parity 1 1 2 3 1 1

Gestational age 3 3 2 3 3

Previous Caesarean Section 3

Ultrasonographic characteristics

Type of breech 1 2 2 3

Placental location 1 1 1 3 3(+) 1

Position of foetal spine 2 3(+)

Foetal position 2

EFW 1 3 2 1

Amniotic fluid (index) 1 1 3 2 2 3

Foetal growth parameters* 3 2 (AC only)

Clinical examination characteristics

Maternal weight 3 3 3

Maternal height and weight 3 2 (height only)

BMI 3 3 3

Uterine tone 1 1 2

Palpation of head possible 1 1

Cervical dilation 1

Breech location 1

Breech engagement 1 1 1

Symphysis fundal height 1 3

* These contain biparietal diameter(cm), femur length(cm) and abdominal circumference (AC) (cm); EFW: Estimated Foetal Weight1 Statistically significant in both univariate and multivariate analysis and therefore included in prediction model2 Statistically significant in univariate analysis and not included in prediction model3 Not statistically significant and not included in prediction model+ Not statistically significant variables in univariate analysis but included in multivariate analysis based on prior research

In the study by Burgos et al27,28 there are three different AUC described; the discriminative ability or their model was of fair performance in model derivation, 0.74 (95% confidence interval 0.70 – 0.78) of poor performance in internal validation, (AUC of 0.67 (95% confi-dence interval 0.62 – 0.72)), and in total of fair performance 0.70 (95% confidence interval 0.67 – 0.74). The model by Kok et al29,33 had a fair performance in internal (AUC of 0.71 (95% confidence interval 0.66 – 0.77)) and a poor performance in external validation (0.66 (95% confidence interval 0.60-0.72)). Calibration of the model was visualized in calibration plots. Calibration was good at internal validation. At external validation, overall a slight underestimation of ECV success was seen, with a maximum difference of 14% between observed and predicted probability. However there was no overlap in the confidence intervals between women with a poor prognosis (less than 20% chance of successful ECV) and a good prognosis (more than 50% chance of successful ECV). The Hosmer-Lemeshow test in both internal (p = 0.66) and external validation (p = 0.30) was not significant.

Three studies30,34,35, described the performance of their models at internal validation by clinical usefulness, in terms of likelihood ratios or accuracy. Overall, the performance of these three models was fair.

120 | Chapter 6

Table 4a. Evaluation of model development and model performance

Study (year) Specification of model development

Type of model & presentation (clinical decision rule) Development phase Internal validation (method) External validation

Burgos et al. (2011/ 2012)27,28 Multivariable logistic regression analysis (maximum modelling strategy) Score Chart Rule and index score

Internal validation completed Prospective Group of subsequent patients in same centre n=500

Kok et al. (2011)29 Multivariable logistic regression analysis & stepwise backward selection procedure Regression formula

Internal and external validation completed

Bootstrapping (200 times) De Hundt et al., 201233 n = 320

Multivariable logistic regression analysis with backward selection Presentation of regression coefficients and constant; No model presentation

Model derivation & calibration measure completed, no internal validation

Wong et al. (2000)34 Univariate analysis (χ2 or Student t test) Score Chart Rule combined with likelihood ratios

Internal validation completed In the same study, group of subsequent patients in same centre n= 87

Lau et al. (1997)30 Multivariable logistic regression modelling with stepwise forward selection 2 sets of regression coefficients and constants

Internal validation completed Split-sample; Two patient subgroups in same centre;Group 1 n=129Group 2 n=114

Newman et al. (1993)31 Stepwise linear regression analysis & discriminate analysis & χ2 and Student t test Score Chart Rule

Internal validation completed In the same study, group of subsequent patients in same centre n=266