Method of Delivery of the Nonvertex SecondTwin: A Community Hospital ExperienceStephen J. Smith, MD,* Joseph Zebrowitz, MD, and Richard A. Latta, MD

Abington Memorial Hospital, Abington, Pennsylvania

Abstract The purpose of this study is to examine the incidence of cesarean section and fetal distress complicatingthe delivery of the second twin in vertex–nonvertex twin gestations in which the second twin underwent either breechextraction or external version. The intrapartum courses of 510 twin gestations delivered at a community hospital over a10-year period were retrospectively analyzed. All vertex–nonvertex twin gestations were identified in which the secondtwin underwent attempted breech extraction or external version. Exclusion criteria included birthweight#1,500 g, fetalanomaly, intrauterine demise, and monoamniotic twins. Of the 76 twin sets that met inclusion criteria, 33 underwentexternal version and 43 underwent primary breech extraction. The two groups had similar demographic characteristics.External version compared to breech extraction was associated with a significantly greater incidence of cesarean section(8/33 vs. 1/43, P 5 .008) and fetal distress (8/33 vs. 1/43, P 5 .008). There was no difference between groups in neonataloutcome for the second twin as measured by length of stay, 5-minute Apgar ,7, intensive care unit admissions, hyalinemembrane disease, intraventricular hemorrhage, and traumatic birth injury. In conclusion, the increased incidence ofcesarean section and fetal distress in patients undergoing attempted external version suggests that breech extraction maybe the preferable route of delivery for the nonvertex second twin weighing more than 1,500 g. J. Matern.–Fetal Med.6:146–150, 1997. r 1997 Wiley-Liss, Inc.

Key Words: twin; external version; breech extraction; perinatal outcome

INTRODUCTIONThe optimal method of delivery of vertex–nonvertex

twins is controversial. Options include cesarean section,breech extraction of the second twin following vaginaldelivery of the first twin, or external version of the secondtwin following vaginal delivery of the first. Cesarean sectionwas advocated in this setting as recently as 1986 [1].Subsequent retrospective studies have shown that theperinatal morbidity and mortality of the nonvertex secondtwin weighing more than 1,500 g is not significantlyinfluenced by route of delivery [2–5]. The AmericanCollege of Obstetricians and Gynecologists Technical Bulle-tin 131 describes all three modes of delivery as reasonablemanagement options [6].The experience with external version and breech extrac-

tion of the nonvertex second twin has been variable. Since1989, three studies have compared breech extraction toexternal version. Each study showed a significantly higherincidence of abdominal delivery and fetal distress associatedwith external version [7–9]. Alternatively, Kaplan reportedno complications in 96 nonvertex second twins undergoingthe procedure [10].We sought to clarify this issue at our institution, a

community hospital, for two reasons. First, prior studies

provide conflicting information regarding the safety ofexternal version of the second twin. Second, since thesestudies were performed at university centers, it is unknownwhether the results are applicable to a community hospitalsetting. The objective of this study was to examine theincidence of cesarean section and fetal distress complicatingthe delivery of the second twin in vertex–nonvertex twingestations in which, after vaginal delivery of the first twin,the second twin underwent either breech extraction orexternal version. The secondary objective was to comparematernal and neonatal outcome using the two deliverymethods.

SUBJECTS AND METHODSThe maternal and neonatal records of all twin deliveries

occurring at Abington Memorial Hospital between January1, 1986 and December 31, 1995 were reviewed. Patientswith a twin gestation were included if the presentation wasvertex–nonvertex, the first fetus delivered vaginally, and

Received 1 August 1996; Revised 12 November 1996; Accepted3 December 1996.*Correspondence to: Stephen J. Smith, M.D., Suite 119, Levy Medical

Plaza, 1235 Old York Road, Abington, PA 19001.

The Journal of Maternal-Fetal Medicine 6:146–150 (1997)

r 1997 Wiley-Liss, Inc.

the nonvertex second twin underwent an attempt at eitherexternal version or breech extraction. Exclusion criteriaincluded 1) intrauterine death of at least one fetus, 2)known fetal anomaly in either twin, 3) prenatally diagnosedmonoamniotic twins, and 4) birthweight of the second twin#1,500 g.All patients were cared for and delivered by private

obstetricians, maternal-fetal medicine subspecialists, or resi-dents supervised by attending obstetricians. The decision toperform external version versus breech extraction of twin Bwas made at the discretion of the attending physician.External version was performed as described by Cherve-

nak et al. [11]. Most patients received either epiduralanesthesia and/or a uterine relaxant (terbutaline 0.25 mgintravenously or subcutaneously) prior to attempt at ver-sion. Immediate access to cesarean section was available.After delivery of the first twin, real-time ultrasound wasused to determine the presentation of the second twin.Initially, gentle pressure with the ultrasound transducer wasused in an attempt to guide the fetus into a vertexpresentation. If unsuccessful, a forward or backward roll wasattempted. The fetal heart rate was monitored throughoutthe procedure. If the version was successful, membraneswere ruptured and oxytocin augmentation was used ifnecessary. If the version was unsuccessful, cesarean sectionor breech extraction was performed.Breech extractions were classified as total or assisted.

Total breech extraction, with or without internal podalicversion, was performed by grasping the fetal feet through theintact membranes. The feet were pulled gently and continu-ously toward the birth canal, while the other hand was usedto dislodge and elevate the head toward the uterine fundus.Membranes were ruptured with the twin in longitudinal lieand breech presentation. In assisted breech extraction, thefetus was delivered spontaneously as far as the umbilicus butthe remainder of the body was extracted [12,13].Maternal and neonatal length of stay were calculated by

counting the number of days the patient remained in thehospital after the day of delivery. A patient was considerednulliparous if she had never delivered a fetus weighing morethan 500 g. Fetal distress was defined as a nonreassuringfetal heart rate pattern felt to require emergent delivery bythe obstetrician. Birth trauma was defined as neurologicinjury or bone fracture in the neonate that resulted fromdelivery. An obstetrician’s years in practice was calculatedby counting the number of years since the completion ofhis/her residency.Descriptive statistical methods were used to characterize

all patients. Between-group comparisons of continuousvariables were performed using parametric (Student t-test)and non-parametric (Kruskal-Wallis one-way analysis ofvariance) methods. Categorical variables were evaluatedusing the chi-square statistic; the Fisher’s exact test was usedwhen any cell was ,5. A value of P , .025 was used to

determine statistical significance because of a Bonferroniadjustment to allow for multiple comparisons.

RESULTSThere were 510 live-born twin deliveries out of 31,391

total deliveries during the study period. One-hundredeighty-two (35.7%) presented with a vertex–nonvertexpresentation. Cesarean section was performed for medical orobstetric indications before or during the first stage of laborin 79 (43.4%). In 17 cases (9.3%), the second twinspontaneously converted to vertex after delivery of the firsttwin. There were ten cases (5.5%) in which, after vaginaldelivery of the first twin but before attempted externalversion or breech extraction, the second fetus was deliveredby cesarean section for obstetric indications. The remaining76 sets of vertex–nonvertex twins (41.8%) were eligible forinclusion into the study.Sixty-six percent of the eligible twins were delivered by

private obstetricians, 18% by maternal-fetal medicine sub-specialists, and 16% by residents supervised by attendingobstetricians. Of these 76 twin sets, 43 underwent breechextraction and 33 underwent external version. The demo-graphic characteristics, the gestational age distribution, andthe mean birthweight of Baby A and Baby B were similarbetween the external version and breech extraction groups(Table 1). The incidence of breech (17/33 vs. 29/43,P 5 .24), transverse (14/33 vs. 13/43, P 5 .39), and obliquepresentation (2/33 vs. 1/43, P 5 1.0) of the second twinafter delivery of the first was also similar between theexternal version and breech extraction groups.External version was successful in 17 of 33 patients

(51.5%). Four patients required abdominal delivery after

TABLE 1. Maternal Demographics, Neonatal Birthweight,and the Incidence of Cesarean Section and Fetal Distress inPatients Undergoing External Version or Breech Extraction

Externalversion(n 5 33)

Breechextraction(n 5 43)

Pvalue*

Maternal age (yr.) 30.6 6 5.3 29.9 6 5.4 NSRaceWhite 31 38 NSBlack 2 5 NS

Nulliparous patients 14 14 NSGestational age (wk) 36.1 6 2.5 36.3 6 2.4 NSGestational age distribution300–320 wk 3 3 NS321–340 wk 6 6 NS$341 wk 24 34 NS

Birthweight A (g) 2,494 6 545 2,591 6 455 NSBirthweight B (g) 2,495 6 496 2,454 6 477 NSCesarean section 8 1 .008Fetal distress 8 1 .008

*NS, not significant.

DELIVERY OF NONVERTEX SECOND TWIN 147

successful external version due to fetal distress, includingtwo cases of umbilical cord prolapse. Of the 16 patients inwhom version was unsuccessful, 12 subsequently weredelivered by uncomplicated breech extraction. The remain-ing four patients required abdominal delivery because of thepersistent nonvertex presentation and retraction of thecervix, not allowing safe secondary breech extraction.Forty-two of the 43 patients in the primary breech

extraction group delivered by uncomplicated breech extrac-tion. One patient required abdominal delivery due to fetaldistress (bradycardia).Patients who underwent external version had a signifi-

cantly higher incidence of abdominal delivery than patientswho underwent primary breech extraction (Table 1). Theexternal version group also had a higher incidence of fetaldistress requiring cesarean section (n 5 4) or operativevaginal delivery (n 5 4). In this group, fetal distress wascharacterized by bradycardia in six cases and repetitive latedecelerations in two cases.Obstetricians choosing to perform primary breech extrac-

tion were in practice a mean of 15.1 years compared to 9.8years for obstetricians performing external version(P 5 .009). The maternal length of stay was similar be-tween the breech extraction and external version groups(2.7 6 2.2 vs. 3.2 6 1.1, P 5 .10). Postpartum endometritiswas diagnosed in three patients undergoing primary breechextraction compared to four patients undergoing externalversion. This difference was not statistically significant(P 5 .46). In two patients undergoing attempted primarybreech extraction, complications of postpartum endometri-tis (septic pelvic thrombophlebitis in one patient, septicshock in one patient) resulted in a 10-day hospitalization.The neonatal outcomes of the second twins were similar

in patients who underwent breech extraction compared tothose in whom external version was attempted (Table 2).There were no neonatal deaths. There was no difference inthe length of stay, the number of intensive care unitadmissions, or the incidence of morbidities, including5-minute Apgar,7, hyaline membrane disease, grade 3 or 4intraventricular hemorrhage, and traumatic delivery. Onesecond twin delivered by uncomplicated primary breechextraction suffered an Erb’s palsy. At 6 months of age, thepalsy had resolved.In comparing patients who underwent a successful exter-

nal version to those in whom the version was unsuccessful,there was no difference in maternal demographic character-istics, the mean birthweight of either twin, or the meandifference in birthweight between the first and secondtwins. There was also no difference in the incidence ofbreech, transverse, or oblique presentations of Baby B afterthe delivery of Baby A. Two patients, both of whom had anunsuccessful version, did not receive either epidural anesthe-sia or a uterine relaxant prior to version (Table 3).

DISCUSSIONThe significant finding of this study is that delivery of the

nonvertex second twin by attempted external version moreoften results in the need for cesarean section when com-pared to primary breech extraction. Attempted externalversion was associated with a 24% cesarean section rateregardless of whether the version was successful, while a 2%rate was achieved in patients undergoing primary breechextraction. Fetal distress was a common complication ofexternal version and a common indication for abdominaldelivery.Three prior studies, all retrospective, have compared

breech extraction to external version for delivery of thenonvertex second twin [7–9]. In accord with our findings,

TABLE 2. Neonatal Outcome of the Second Twin in PatientsUndergoing External Version or Breech Extraction

Externalversion(n 5 33)

Breechextraction(n 5 43)

Pvalue*

Neonatal death 0 0 NSLength of stay (days) 8.4 6 9.6 7.5 6 7.5 NS5-min Apgar score ,7 3 1 NSIntensive care unit admission 21 28 NSHyaline membrane disease 5 3 NSIntraventricular hemorrhagea 0 0 NSTraumatic delivery 0 1b NSaGrade 3 or 4.bErb’s palsy.*NS, not significant.

TABLE 3. Maternal Demographics, Neonatal Birthweights,Presentation of the Second Twin, and the Use of Epiduralor Uterine Relaxants in Patients With Successful and Failed

External Version

Successfulexternalversion(n 5 17)

Failedexternalversion(n 5 16)

Pvalue*

Maternal age (yr.) 31.5 6 6.6 29.6 6 3.5 NSNulliparous patients 7 7 NSGestational age (wk) 36.5 6 2.6 35.6 6 2.4 NSBirthweight A (g) 2,538 6 594 2,448 6 502 NSBirthweight B (g) 2,526 6 472 2,461 6 534 NSBirthweight difference (g)

between A & B 332 6 264 248 6 198 NSPresentation of BBreech 9 8 NSTransverse 6 8 NSOblique 2 0 NS

Patients with epiduraland/or uterinerelaxant 17 14 NS

*NS, not significant.

SMITH ET AL.148

each study showed a significantly increased incidence ofabdominal delivery and fetal distress after external versionwhile neonatal outcome was unaffected by mode of delivery(Table 4). Our study is unique because it was performed atan institution and in a patient population markedly differ-ent from the previous studies, i.e., a community hospitalwith a primarily white suburban private population. Theprior studies were performed at university centers onprimarily indigent populations. Since ours is the first studyto address this question in this population type, the findingsare especially important to the obstetrician practicing in asimilar setting.Since recent studies have questioned its safety, it is

helpful to review the entire experience with externalversion of the nonvertex second twin. Combining theresults of eight prior studies with this report, a total of 300evaluable patients have undergone attempted externalversion of the second twin [2,4,7–11,14]. The number ofpatients in each individual study ranges from 15 to 96(mean 33). The mean success rate is 67% (range 47–90%),the mean cesarean section rate is 19% (0–48%), and themean incidence of fetal distress is 8% (0–24%). Althoughresults are variable between studies and the sample sizes aresmall, there appears to be a significant potential forcomplications associated with external version. A notewor-thy exception is the largest of these studies, in which Kaplanretrospectively described his experience with external ver-sion in 96 patients [10]. During the 5-year study period, allnonvertex second twins underwent attempted externalversion after vaginal delivery of the first twin. Breechextraction was reserved for unsuccessful versions. Thesuccess rate was 75%; the cesarean section rate 2%; and nocase was complicated by fetal distress. Because detailsregarding their patient population and operator experiencewere not provided, the reason for the success is not readilyidentified. It is possible that the use of a well-definedprotocol that utilizes external version as the single methodof delivery may result in the acquisition of the expertisenecessary to achieve a success rate comparable to primarybreech extraction. However, apart from this isolated experi-ence, the remaining combined data reveal a cesareansection rate for the second twin that is approximately 10

times greater after attempted external version compared toprimary breech extraction.A prospective randomized trial comparing external ver-

sion to breech extraction would be required to conclusivelyascertain the optimal mode of delivery of the nonvertexsecond twin. The need for a large number of patients makesthis a difficult task. In order to achieve appropriate powerassuming a reduction in the cesarean section rate from 15%using external version to 5% using breech extraction, a totalsample size of 318 patients with nonvertex second twinswould be required (a 5 0.05, b 5 0.80). Another impedi-ment to a randomized trial is the variable expertise amongoperators. Obstetricians at our institution consistently favorone method of delivery and become considerably moreexperienced in that method. Participation in a randomizedtrial would require random assignment of delivery method, arequirement that many obstetricians would find unsatisfac-tory and which may, given the relative inexperience of theoperator with one method or the other, be unethical to thepatient. These potential problems suggest the need for amulticenter study with participating operators skilled inboth external version and breech extraction. Until such astudy is performed, management must be predicated on theavailable studies. Although these studies are limited by theirretrospective design and small sample sizes, they do provideuseful information to guide management decisions andcounsel patients.In this study, the obstetricians’ number of years in

practice was the major determinant in selecting the route ofdelivery. Recently trained obstetricians at our institutionrely on external version as the procedure of choice while themore senior obstetricians practice primary breech extrac-tion. Since external version of the nonvertex second twinwas advocated by Chervenak, many residency programshave taught and even emphasized the use of this technique[11]. This training and the decline in the frequency ofsingleton vaginal breech delivery, has markedly limited theresident experience in intrapartum management of thevaginal breech.In our study, 12 patients underwent uncomplicated

secondary breech extraction after an initial unsuccessfulversion. The initial guidelines for the performance of

TABLE 4. Review of the Studies Examining the Use of External VersionVersus Breech Extraction

Author

Cesarean section Fetal distress

Externalversion

Breechextraction

Pvalue

Externalversion

Breechextraction

Pvalue

Gocke et al., 1989 [7] 16/41 2/55 ,.001 5/41 0/55 .01Wells et al., 1991 [8] 11/23 1/43 ,.001 4/17 0/43 .01Chauhan et al., 1995 [9] 10/21 1/23 .001 4/21 0/23 .04Smith, 1996 8/33 1/43 .008 8/33 1/43 .008

Total 45/118 5/164 ,.001 21/118 1/164 ,.001(38%) (3%) (18%) (0.6%)

DELIVERY OF NONVERTEX SECOND TWIN 149

external version of the second twin included the recommen-dation for breech extraction if the version was unsuccessful[11]. Other authors have criticized this recommendation fortwo reasons. First, external version may provoke fetal heartrate abnormalities and, second, attempted external versionmay convert the breech to a more unstable lie makingbreech extraction more difficult [9]. These concerns, how-ever, are not supported by the limited data available.Combining this study with the eight previous reports onexternal version of the nonvertex second twin, a total of 51secondary breech extractions was performed after unsuccess-ful external version [2,4,7–11,14]. No maternal or neonatalcomplications were reported. Secondary breech extractionof the fetus weighing more than 1,500 g seems to be areasonable option if version is attempted and is unsuccess-ful. This experience is compiled from several small retrospec-tive studies, however, thus limiting its applicability.In conclusion, the increased incidence of cesarean sec-

tion and fetal distress in patients undergoing attemptedexternal version suggests that, at a community hospital withexperienced obstetricians, breech extraction may be thepreferable route of delivery for the nonvertex second twinweighing more than 1,500 g. The primary determinant ofthe selection of delivery mode is an obstetrician’s number ofyears in practice, with recently trained obstetricians relyingon external version. These findings suggest that additionaltraining of obstetrical residents in vaginal breech delivery ofthe second twin may be necessary.

REFERENCES1. Cetrulo CL: The controversy of mode of delivery in twins: intrapartum

management of twin gestation (Part 1). Semin Perinatol 10(1):39–43,1986.

2. Adam C, Allen AC, Baskett TF: Twin delivery: influence of thepresentation and the method of delivery on the second twin. Am JObstet Gynecol 165:23–27, 1991.

3. Fishman A, Grubb DK, Kovacs BW: Vaginal delivery of the nonvertexsecond twin. Am J Obstet Gynecol 168:861–864, 1993.

4. Greig PC, Veille J, Morgan T, Henderson L: The effect of presentationand mode of delivery on neonatal outcome in the second twin. Am JObstet Gynecol 167:901–906, 1992.

5. Chervenak FA, Johnson RE, Berkowitz RL, Grannum P, Hobbins JC:Is routine cesarean section necessary for vertex-breech and vertex-transverse twin gestations? Am J Obstet Gynecol 148:1–5, 1984.

6. American College of Obstetricians and Gynecologists: ‘‘MultipleGestation.’’ Washington: American College of Obstetricians andGynecologists, 1989; ACOG Committee Opinion no. 131.

7. Gocke SE, Nageotte MP, Garite T, Towers CV, Dorcester W: Manage-ment of the nonvertex second twin: primary cesarean section, externalversion or primary breech extraction. Am J Obstet Gynecol 161:111–114, 1989.

8. Wells SR, Thorp JM, Bowes WA: Management of the nonvertexsecond twin. Surgery 172:383–385, 1991.

9. Chauhan SP, Roberts WE, McLaren RA, Roach H, Morrison JC,Martin JN: Delivery of the nonvertex second twin: breech extractionversus external cephalic version. Am J Obstet Gynecol 173:1015–1020, 1995.

10. Kaplan B, Peled Y, Rabinerson D, Goldman GA, Nitzan Z, Neri A:Successful external version of B-twin after birth of A-twin forvertex-nonvertex twins. Eur J Ostet Gynecol Reprod Biol 58:157–160,1995.

11. Chervenak FA, Johnson RE, Berkowitz RL, Hobbins JC: Intrapartumexternal version of the second twin. Obstet Gynecol 62:160–165,1983.

12. Cunningham FG, MacDonald PC, Gant NF, Leveno KJ, Gilstrap LCIII: ‘‘Williams Obstetrics,’’ 19th ed. Norwalk, CT: Appleton & Lange,1993.

13. Rabinovici J, Barkai G, Reichman B, Serr DM, Mashiach S: Internalpodalic version with unruptured membranes for the second twin intransverse lie. Obstet Gynecol 71:428–430, 1988.

14. Tchabo J, Tomai T: Selected intrapartum external cephalic version ofthe second twin. Obstet Gynecol 79:421–423, 1992.

SMITH ET AL.150