Original Article

487

Optimal Timing for Term Delivery of Twin Pregnancies: A Population-Based Study Gustavo A. Vilchez, MD1 Jing Dai, PhD2 Luis R. Hoyos, MD1 Ray Bahado-Singh, MD3 Robert J. Sokol, MD2,3

Detroit Medical Center, Detroit, Michigan 2 C.S. Mott Center for Human Growth and Development, Wayne State University, Detroit, Michigan 3 Department of Obstetrics and Gynecology, Wayne State University, Detroit, Michigan

Address for correspondence Gustavo A. Vilchez, MD, Department of Obstetrics and Gynecology, Wayne State University/Detroit Medical Center, 3990 John R, 7-Brush N, Box 165, Detroit, MI 48201 (e-mail: [email protected]).

Am J Perinatol 2015;32:487–496.

Abstract

Keywords

► ► ► ►

twin pregnancy neonatal morbidity NICU admission gestational age

Objective The objective of this study was to examine the risk of adverse neonatal outcomes after twin delivery according to gestational age. Materials and Methods The U.S. Natality Database from 2007 to 2010 was reviewed. Inclusion criteria were twin deliveries and gestational age of 37 to 42 weeks. Exclusion criteria were congenital anomalies and missing/incomplete data. Cases were subdivided by gestational age into early term, term, and late term. Singleton pregnancies matched by delivery time and location were selected as controls. Outcome variables included were low Apgar score, assisted ventilation, neonatal intensive care unit admission, surfactant/antibiotic use, seizures, and birth injury. Logistic regression analysis was used to calculate adjusted odds ratios according to gestational age and plurality, using singleton term as reference. Results A total of 220,169 twin and 270,540 singleton deliveries were identified. The risk of adverse neonatal outcomes for twins was higher than for singletons. For twins, the distribution of the risks of the composite of adverse neonatal outcomes was linear, being the lowest at early term and the highest at late term, whereas the distribution for singletons was U-shaped being lowest at term compared with early and late term. Conclusions Twins are at higher risk of suboptimal neonatal outcomes than singletons, but do better when delivered at early term rather than term or late term.

Throughout the last few decades, there has been a substantial increase in twin pregnancies with an estimated rise of 76% since 1980. According to National Vital Statistics, currently, the birth rate for twins is 33.2 per 1,000 total births and approximately 41.2% are born after 37 completed weeks of gestation.1 In the practice of obstetrics, this is one of the most commonly encountered conditions associated with increased risk and a substantial proportion achieve term gestation. Management of twin pregnancy at term is problematic due to lack of more solid data rather than expert opinion consen-

sus regarding optimal timing for delivery, whether to intervene at a specific gestational age or await spontaneous labor. Recently, considerable attention to this matter has risen because several studies have raised the concern of an increased risk of perinatal death with prolonged gestation after 36 to 38 weeks.2–7 However, unlike term singleton pregnancies in which current recommendations and a recent metaanalysis suggest induction of labor between 41 and 42 weeks of gestation rather than expectant management without spontaneous labor,8–10 the question regarding the optimal

received June 16, 2014 accepted after revision October 2, 2014 published online December 29, 2014

Copyright © 2015 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA. Tel: +1(212) 584-4662.

DOI http://dx.doi.org/ 10.1055/s-0034-1396693. ISSN 0735-1631.

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

1 Department of Obstetrics and Gynecology, Wayne State University/

Anushka Chelliah, MD1

Optimal Timing for Term Delivery of Twin Pregnancies timing for delivery of twins at term remains controversial. This is mainly because a gestational age in which fetal and neonatal risks and benefits are balanced still needs to be determined. Despite published guidelines recommending delivery of dichorionic and monochorionic twins at 37 and 36 weeks of gestation, respectively11; as well as delivery at 38 weeks for dichorionic and between 34 and 37 weeks for monochorionic gestations,12 there is limited evidence concerning neonatal morbidity that might be entailed by implementing such recommendations.11 Evidence about neonatal outcomes for twin deliveries at different gestational ages at term is sparse, despite the frequency of this high-risk situation.3,4,13–19 To provide more information on neonatal morbidity following delivery of twins at term, we examined the risk of adverse neonatal outcomes according to gestational age at delivery compared with singletons using a composite score of neonatal complications in a population study in the context of the current literature.

Materials and Methods Study approval was obtained from the Institutional Review Board of the Wayne State University/Detroit Medical Center in Detroit, Michigan. The design of the study was populationbased retrospective cohort. The birth data files from the Vital Statistics Data-Center of the Disease Control and Prevention of the United States from the year 2007 to 2010 were reviewed. This database contains all deliveries in the United States and represents all births registered in the 50 states of the United States, the District of Columbia, and New York City. To identify cases, the inclusion criteria used for this study were twin deliveries with gestational age at delivery from 37 to 42 completed weeks. The exclusion criteria used were cases with any reported congenital anomaly and cases with missing or incomplete data. Among twin deliveries, information about chorionicity was not available within the database, hence making impossible its identification. A looping algorithm identifying patterns of “1221” or “121” (being 1 ¼ singleton delivery and 2 ¼ twin delivery) was used to select two singleton deliveries matched by proximity and location immediately before and after each selected twin delivery, to be used as controls (match ratio between 2:1 and 1:1). The selected singleton and twin pregnancies were subdivided into three subgroups according to gestational age—early term (delivery occurred between 37 and 38 weeks), term (39–40 weeks), and late term (41–42 weeks), as recently recommended.20 Variables related to adverse neonatal outcomes, Apgar score less than 7 at 5 minutes, assisted ventilation, assisted ventilation for more than 6 hours, neonatal intensive care unit (NICU) admission, use of surfactant, use of antibiotics, neonatal seizures, and birth injury (including skeletal fracture[s], peripheral nerve injury, and/or soft tissue/solid organ hemorrhage that required intervention) were selected for analysis. These eight variables were used to generate a composite adverse outcome score, ranging from 0 to 8, as well as individually. Frequencies for adverse neonatal outcomes American Journal of Perinatology

Vol. 32

No. 5/2015

Vilchez et al. were calculated for both twin and singleton pregnancies. Forward logistic regression analysis was used to calculate the adjusted odds ratios (ORs) for dichotomous outcome variables according to gestational age and plurality, using singleton term deliveries as the reference group and standard cofounders in the literature (maternal age, live birth order, maternal race, newborn gender, year of delivery) as covariates throughout. To understand the main contributor(s) to the increased risk of the composite of adverse neonatal outcomes, a subsequent analysis was performed in twin cases only. Frequencies of adverse neonatal outcomes were calculated according to timing of delivery. R2 changes were calculated from a stepwise linear regression using the composite score (0–8 scale) for the neonatal complications as the response variable and the eight individual neonatal complication variables as the predictor variables. Further, to identify preceding factors with increased risk associated with the main contributor(s) to adverse neonatal outcomes, a forward logistic regression analysis was performed in cases with positive status of these main contributor(s). Variables considered for this subanalysis were maternal (race, age, live birth order, history of hypertensive or diabetic-related disorders, premature rupture of membranes); delivery mode (induction of labor, cesarean, operative, or vaginal delivery, year of delivery), adverse neonatal outcomes (as previously identified), and adverse labor outcomes (chorioamnionitis, fetal distress, meconium staining). Information regarding assisted reproductive techniques was available, but its determination was not the scope of this article specifically, but will be used for future analysis. Despite the fact that information for birth weight was available, neither estimated fetal weight nor discordancy was available, even though growth restriction may increase as term is approached which is more common among twins. Data were analyzed using Statistical Package for Social Sciences (SPSS) version 21.0 (IBM Corp, Armonk, NY, 2012). Finally, efforts were made to address the potentially underestimated variability within the twin pregnancies, related to including two offspring for a single pregnancy. Because of lack of pairing information in the database, all twin pregnancies were analyzed independently even though twin pairs are not truly independent. Matching the twin pregnancies case by case would have been computationally difficult or impossible and, in any case, grossly inaccurate given the lack of key information, such as date of birth or name of hospital, as well as the large sample size. Instead, the original sample was compared with a subsample containing all the singleton pregnancies and a random half of the twin pregnancies for the coefficients of variation of the outcome variables and a few pregnancy-related variables, such as hypertension, diabetes, and preeclampsia. Adjusted ORs estimated from the previously mentioned logistic regression models were compared for the original and the reduced samples for neonatal and pregnancy variables. The pregnancy-related variables, while not of particular interest per se, served as a valuable reference in this comparison because the potential dependence within the twin pregnancies should theoretically have more

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

488

Optimal Timing for Term Delivery of Twin Pregnancies

Vilchez et al.

489

Table 1 Frequencies of adverse neonatal outcomes in twin and singleton deliveries (N ¼ 490,709) Early term

Term

Late term

Singleton

Twin

Singleton

Twin

Singleton

Twin

N ¼ 85,394

N ¼ 169,292

N ¼ 148,517

N ¼ 41,402

N ¼ 35,795

N ¼ 9,475

Composite score  1

4,808 (5.61)

16,246 (9.60)

7,049 (4.73)

4,943 (11.94)

1,859 (5.18)

1,223 (12.91)

Apgar score at 5 min < 7

919 (1.08)

2,352 (1.40)

1,542 (1.04)

793 (1.93)

441 (1.23)

201 (2.13)

Assisted ventilation

1,911 (3.22)

5,482 (5.01)

3,102 (3.02)

1,473 (5.58)

819 (3.40)

347 (6.07)

Assisted ventilation > 6 h

255 (0.43)

969 (0.88)

299 (0.29)

250 (0.95)

92 (0.38)

101 (1.77)

Admission to NICU

2,601 (4.39)

11,026 (10.07)

3,085 (3.00)

3,573 (13.54)

817 (3.39)

935 (16.34)

Surfactant use

56 (0.09)

200 (0.18)

47 (0.05)

88 (0.33)

14 (0.06)

43 (0.75)

Antibiotics use

749 (1.26)

1,907 (1.74)

1,086 (1.06)

498 (1.89)

303 (1.26)

183 (3.20)

Neonatal seizures

16 (0.03)

18 (0.02)

34 (0.03)

4 (0.02)

7 (0.03)

1 (0.02)

Birth injury

34 (0.06)

44 (0.04)

73 (0.07)

5 (0.02)

22 (0.09)

10 (0.17)

Abbreviation: NICU, neonatal intensive care unit.

impact on pregnancy-related variables than on neonatal outcomes.

Results A total of 220,537 twin deliveries were identified, with 220,169 twin deliveries being included for analysis. As summarized in ►Table 1, of these, 76.89% were delivered at early term, 18.80% at term, and 4.30% at late term. Of the 270,540 controls, 31.56% were delivered at early term, 54.90% at term, and 13.23% at late term (►Table 1). Frequencies of adverse neonatal outcomes by timing of delivery and plurality are also shown in ►Table 1 and ►Fig. 1. For twins, the frequency of the composite of adverse neonatal outcomes for twins was lower at early term (9.60%) compared with term and late term though always higher than for singletons. Similarly, for individual adverse neonatal outcomes the frequencies were lowest at early term, rising at term and late term. For singletons, the distribution of the frequencies appeared U-shaped, appearing lower at term than early and late term for both the composite (4.73%) and individual adverse neonatal outcome. The adjusted ORs of adverse neonatal outcomes with singleton term as the reference group are shown in ►Table 2 and ►Fig. 2. For twins, the distribution of the risks of the composite of adverse neonatal outcomes was linear, lower at early term (OR [95% confidence interval, CI] ¼ 2.55 [2.48–2.630]) and higher at late term (OR [95% CI] ¼ 4.37 [4.11–4.64]) compared with singletons at term. For singletons, the risk of the composite of neonatal complications was U-shaped, lower at term compared with early and late term. There was no difference in the risk of neonatal seizures among the twin gestations across the gestational age categories and there was no difference in the risk of birth injury between the early-term and term birth twins (►Table 2 and ►Fig. 2). To identify the main contributors to the increased risk of the composite of adverse neonatal outcomes, a subanalysis in

twins only was performed. Frequencies and adjusted ORs (with early term twins as the group of reference) are shown in ►Table 3 and ►Fig. 3. The distribution of frequencies of all adverse neonatal outcomes is linear, lower at early term and higher at late term, that is, the risk of all adverse neonatal outcomes increases as gestation progresses. The highest

Fig. 1 Adverse neonatal outcomes after singleton and twin deliveries. NICU, neonatal intensive care unit. American Journal of Perinatology

Vol. 32

No. 5/2015

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Number of cases (frequency)

Optimal Timing for Term Delivery of Twin Pregnancies

Vilchez et al.

Table 2 Adjusted odds ratios of adverse neonatal outcomes in twins compared with singleton deliveries (N ¼ 220,169) Adjusted odds ratios (95% CI)a Composite score  1 Apgar score < 7

Early term

Term

Late term

Singleton

1.30 (1.25–1.35)

Reference

1.16 (1.10–1.22)

Twins

2.55 (2.48–2.63)

3.26 (3.13–3.38)

4.37 (4.11–4.64)

Singleton

1.03 (0.95–1.12)

Reference

1.18 (1.06–1.31)

Twins

1.43 (1.34–1.53)

1.96 (1.80–2.14)

2.18 (1.88–2.53)

Assisted ventilation

Singleton

1.08 (1.02–1.14)

Reference

1.11 (1.03–1.20)

Twins

1.77 (1.69–1.85)

1.97 (1.85–2.10)

2.15 (1.91–2.41)

Assisted ventilation > 6 h

Singleton

1.48 (1.26–1.76)

Reference

1.28 (1.01–1.62)

Twins

3.32 (2.91–3.79)

3.51 (2.96–4.16)

6.57 (5.23–8.26)

NICU admission

Singleton

1.49 (1.41–1.57)

Reference

1.13 (1.04–1.22)

Twins

3.78 (3.63–3.94)

5.30 (5.03–5.57)

6.59 (6.08–7.13)

Surfactant use

Singleton

2.01 (1.36–2.98)

Reference

1.24 (0.68–2.25)

Twins

4.30 (3.12–5.94)

7.72 (5.40–11.04)

17.49 (11.52–26.54)

Antibiotics use Neonatal seizures Birth injury

Singleton

1.20 (1.10–1.32)

Reference

1.16 (1.02–1.32)

Twins

1.80 (1.66–1.94)

1.93 (1.73–2.15)

3.30 (2.81–3.87)

Singleton

0.85 (0.47–1.54)

Reference

0.87 (0.39–1.97)

Twins

0.55 (0.31–0.99)

0.50 (0.18–1.42)

0.57 (0.08–4.14)

Singleton

0.80 (0.53–1.21)

Reference

1.28 (0.80–2.07)

Twins

0.56 (0.39–0.82)

0.21 (0.08–0.58)

2.44 (1.26–4.75)

Abbreviations: CI, confidence interval; NICU, neonatal intensive care unit. a Adjusted for maternal age, live birth order, maternal race, newborn gender, and year of delivery.

frequencies at early term, term, and late term were observed in NICU admission, at 10.07, 13.54, and 16.34%, respectively. Ninety-five percent of the explained variance in the composite of adverse neonatal outcomes score was contributed by NICU admission. The risk of NICU admission compared with early term was significantly higher at term (OR [95% CI] ¼ 1.40 [1.34–1.45]) and at late term (OR [95% CI] ¼ 1.73 [1.61–1.86]). Once NICU admission had been identified as the main indicator of increased composite risk for adverse neonatal outcome, a subanalysis of cases with NICU admission was performed. The frequencies and relative risks (RRs) of factors that preceded NICU admission are shown in ►Table 4. The highest frequencies were observed in African Americans (17.1%), primiparas (25.1%), delivered by cesarean delivery (77.5%), complicated with preeclampsia (10.3%), with newborns of male gender (52.3%), who required antibiotic use (13.4%), and assisted ventilation (19.0%). The highest RRs were identified in cases with antibiotic use (RR [95% CI] ¼ 37.73 [34.20–41.63]), assisted ventilation > 6 hours (RR [95% CI] ¼ 50.71 [43.50–59.11]), neonatal seizures (RR [95% CI] ¼ 85.33 [20.01–363.95]), and surfactant use (RR [95% CI] ¼ 74.70 [52.12–7.07]). However, note the limited sample in some cells. The adjusted ORs of factors preceding NICU admission are shown in ►Table 5. The highest risks were identified for low Apgar score (OR [95% CI] ¼ 6.05 [5.57–6.58]), chorioamnioAmerican Journal of Perinatology

Vol. 32

No. 5/2015

nitis (OR [95% CI] ¼ 3.10 [2.61–3.69]) and premature rupture of membranes (OR [95% CI] ¼ 3.39 [3.12–3.67]). Finally, for the subsample that included only half of the twin pregnancies, thus presumably representing half of the potential dependence within the twin pregnancies, coefficients of variation increased by 1.9% on average among neonatal outcomes and by 4.7% among pregnancy-related variables. Adjusted ORs remained largely unchanged (3% maximum) for the reduced sample (data not shown), compared with the full sample. For example, the adjusted OR for early-term twin and late-term twin estimated from the reduced sample were 2.52 (2.43–2.60) and 4.50 (4.15– 4.88), respectively, compared with 2.55 (2.48–2.63) and 4.37 (4.11–4.14) for the neonatal complication composite score.

Discussion There are limited high-quality data on which to base a recommendation for timing of delivery in twin pregnancies21; the optimal gestational age is still based on expert opinion data rather than randomized controlled trials.21–23 We present population-based data of every twin delivery in the United States over a course of 4 years which strikingly demonstrates the increasing neonatal risk following delivery beyond 38 weeks of gestation. Some previous studies have been underpowered and findings have been unclear. This

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

490

Vilchez et al.

Fig. 2 Adjusted odds ratios of the composite of neonatal complications in twins using term singleton deliveries as reference (N ¼ 220,169). NICU, neonatal intensive care unit.

study adds information not only regarding the optimal gestational age for delivery but also regarding associated medical/obstetrical risks and short-term neonatal morbidity. The key findings of this population-based study are that for term twins, earlier delivery is associated with significantly lower risk for the neonate than later delivery, despite twin risk being uniformly higher than risk for singletons. We found a significant increase in a composite score of neonatal complications, which had been constructed a priori, for twin deliveries from the early- to the late-term periods, related primarily to risks leading to NICU admission. According to our findings, term twin neonates requiring NICU admission followed a profile of males delivered by cesarean to African American primiparous mothers with medical complications of pregnancy, such as preeclampsia, chronic hypertension, or diabetes, with evidence of infection and meconium during labor. Major warning signs for adverse perinatal outcome were premature

rupture of membranes (PROM), chorioamnionitis and low 5-minute APGAR score. We also found an increase in the use of surfactant with advanced gestational age at term. This use of surfactant is likely unrelated to respiratory distress syndrome (RDS), since RDS has not been reported in term newborns.16,18 Therefore, it was most likely related to other causes of respiratory distress at term such as meconium aspiration syndrome or pneumonia. Focusing on the optimal timing for delivery in twins, as shown in ►Table 6, there are two randomized controlled studies,15,24 one systematic review,22 and one prospective cohort study21 in the literature. It took a long time to recruit enough cases for the trials and yielded inconsistent results most likely due to the small samples and consequently low power.25–27 Different groups have previously attempted to shed light on the relationship between neonatal morbidity and gestational age at delivery in term twin pregnancies. American Journal of Perinatology

Vol. 32

No. 5/2015

491

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Optimal Timing for Term Delivery of Twin Pregnancies

Optimal Timing for Term Delivery of Twin Pregnancies

Vilchez et al.

Table 3 Adverse neonatal outcomes after delivery of twin pregnancies (N ¼ 220,169) Early term

Term

Late term

N ¼ 169,292

N ¼ 41,402

N ¼ 9,475

2,352 (1.40)

793 (1.93)

201 (2.13)

Number of cases (percentage) Apgar score at 5 min < 7 Assisted ventilation

5,482 (5.01)

1,473 (5.58)

347 (6.07)

Assisted ventilation > 6 h

969 (0.88)

250 (0.95)

101 (1.77)

NICU admission

11,026 (10.07)

3,573 (13.54)

935 (16.34)

Surfactant use

200 (0.18)

88 (0.33)

43 (0.75)

Antibiotics use

1,907 (1.74)

498 (1.89)

183 (3.20)

18 (0.02)

4 (0.02)

1 (0.02)

Reference

Neonatal seizures Adjusted odds ratios (95% CI)

a

Apgar score at 5 min < 7

1.36 (1.26–1.48)

1.51 (1.30–1.74)

Assisted ventilation

1.11 (1.05–1.18)

1.21 (1.08–1.35)

Assisted ventilation > 6 h

1.05 (0.92–1.21)

1.97 (1.60–2.43)

NICU admission

1.40 (1.34–1.45)

1.73 (1.61–1.86)

Surfactant use

1.79 (1.39–2.31)

4.06 (2.91–5.65)

Antibiotics use

1.07 (0.97–1.18)

1.81 (1.55–2.12)

Neonatal seizures

0.87 (0.30–2.58)

0.96 (0.13–7.24)

Birth injury

0.37 (0.13–1.04)

4.35 (2.18–8.65)

Apgar score at 5 min < 7

1.36 (1.26–1.48)

1.51 (1.30–1.74)

Assisted ventilation

1.11 (1.05–1.18)

1.21 (1.08–1.35)

Abbreviations: CI, confidence interval; NICU, neonatal intensive care unit. a Adjusted for maternal age, live birth order, maternal race, newborn gender, and year of delivery.

In 2012, Dodd et al22 presented the results of a multicenter randomized controlled trial that compared elective birth at 37 weeks of gestation versus birth after 38 weeks using a composite of serious adverse outcomes for the infant. They found that neonates in an elective birth group at 37 weeks of gestation were at significantly lower risk of morbidity when compared with neonates not electively induced at 37 weeks. Although the methodology and results of the study have been questioned,25,27 other authors have highlighted the conclusions and the contribution of that article.23 The current study adds to our knowledge base concerning term twin birth. The limited power of the published trials gives population-based studies such as this one a real role in the evaluation of outcomes related to this particular relatively rare scenario. It is very difficult to imagine powering another clinical trial to discern such small differences in a statistically meaningful way22; given the already available trials, the additional findings of a population-based study approach like ours is likely to be the best evidence that will be available. Our study strongly suggests that while term delivery is desirable, earlier is better than later. Perhaps, given reasonable dating, twins should be delivered before 39 weeks, that is, by 38 completed weeks of gestation. Such a conclusion is also supported by population-based studies. For example, a decade ago, Dodd et al3 also performed a population-based study focused mainly on stillbirth and American Journal of Perinatology

Vol. 32

No. 5/2015

mortality risks. They also analyzed composite neonatal morbidity outcomes and showed at each gestational week infants of twin pregnancies had a greater frequency of adverse outcomes than infants of singleton pregnancies. The lowest occurrence of adverse outcomes in twins was found to be at 37 weeks of gestation only to increase with advancing gestation. These results, which coincide with ours, support the assertion that the lowest risk of neonatal morbidity in term twin pregnancies occurs earlier rather than later and that with prolonged gestation there is a greater risk of adverse neonatal outcomes. However, not all studies in this area support our results. Bakr and Karkour13 performed a prospective cohort study of twin deliveries in an attempt to evaluate perinatal outcomes including neonatal morbidity. They found the need for NICU admission as well as a lower birth weight was significantly more frequent in those delivered at term before 39 weeks, when compared with those delivered after 39 weeks. Unlike their findings, based on outcomes of essentially all-term twins born in the country over 4 years, we report that in late-term pregnancy, the assisted ventilation and prolonged assisted ventilation (> 6 hours) incidence was increased. Luke et al16 found similar results, with the incidence of ventilator support increasing after 38 weeks, but other groups such as Soucie et al17 have reported opposite findings. Dividing the outcomes for twin A and twin B, the latter group reported that

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

492

Vilchez et al.

Fig. 3 Frequencies and adjusted ORs of neonatal outcomes by timing of delivery in twins (N ¼ 220,169). NICU, neonatal intensive care unit; ORs, odds ratios.

there was a decreased risk of assisted ventilation from 38 to 39 weeks in twin A and from 39 to  40 weeks for twin B when compared with 37 weeks of gestation. One reason these differences may exist can be related to the construction of the comparison groups, as well as the number of infants in each group. Although we divided our groups into 37 to 38, 39 to 40, and 41 to 42 weeks of gestation with 169,292, 41,402, and 9,475 infants in each group, respectively, Bakr and Karkour13 only reported 16 patients in the >39 weeks of gestation group. Soucie et al17 as well as Luke et al16 studied all infants born  40 weeks of gestation in a single group and the others by independent gestational weeks; we, on the other hand, specifically grouped gestational ages as previously mentioned in independent categories.20 The strengths of this study are intrinsic to its design since by using a population base, sampling bias is avoided, external validity is increased and greater sample sizes are available, allowing better statistical power. However, our study has limitations which must be acknowledged. One unavoidable weakness is the lack of information about chorionicity which, unfortunately, is unavailable in the database. However, our statistics may fairly represent those for dichorionic pregnancies, since monochorionic pregnancies only occur in approximately 25% of twin pregnancies. Besides, the risk of complications in monochorionic is greater than in dichorionic twins which might carry an increased risk of preterm delivery either because of intrinsic or iatrogenic reasons. In a historical cohort, delivery before 37 weeks of gestation occurred more

often in monochorionic than in dichorionic twin pregnancies.28 In addition, even in uncomplicated monochorionic twin pregnancies guidelines and expert opinion recommend delivery between 34 and 37 weeks.11,12,29,30 The aforementioned highlights the fact that the results here presented, given that it only included twins at term, likely represent primarily those of dichorionic pregnancies. However, we believe that these results could also be applied to monochorionic twins at term since to our knowledge there is no other study which evaluates complications among this subgroup by gestational age after term and according to our results, regardless of chorionicity, twins at term tend to do better when delivered early. Furthermore, there are recent data that do not support the increased risk of morbidity/mortality in uncomplicated monochorionic twins before 37 weeks.31 An additional issue is the inability to sort down to one twin per pregnancy, but detailed statistical analysis showed little effect of this limitation. Although pairing information was unavailable in this database, the large sample size clearly provided sufficient stability and robustness in the statistical analyses. The potential deviations from the true population parameters as a result of discounting the dependence within twin pregnancies were minimal and should not affect the credibility of our estimates. Finally, other limitations that are recognized include possible errors in data entry and that, from the perspective of changing clinical practice, this is not a randomized controlled trial. Errors in data entry are always a possible limitation when public health vital statistics are American Journal of Perinatology

Vol. 32

No. 5/2015

493

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Optimal Timing for Term Delivery of Twin Pregnancies

Optimal Timing for Term Delivery of Twin Pregnancies

Vilchez et al.

Table 4 Characteristics of cases with newborns admitted to NICU after twin deliveries (N ¼ 141,606) No NICU admission

NICU admission

(N ¼ 126,072)

(N ¼ 15,534)

p Value

N (percentage) Chronic hypertension

1,535 (1.2)

301 (1.9)

0.000

Diabetes

7,691 (6.1)

1,239 (8.0)

0.000

Eclampsia

274 (0.2)

88 (0.6)

0.000

Hypertension-related disorder

7,524 (6)

1,602 (10.3)

0.000

Apgar score < 7

1,471 (1.2)

1,066 (6.9)

0.000

Antibiotic use

513 (0.4)

2,075 (13.4)

0.000

Assisted ventilation

4,348 (3.4)

2,954 (19.0)

0.000

Assisted ventilation > 6 h

194 (0.2)

1,126 (7.2)

0.000

Birth injury

40 (0)

19 (0.1)

0.000

Neonatal seizures

2 (0)

21 (0.1)

0.000

Surfactant use

33 (0)

298 (1.9)

0.000

Male newborn gender

62,236 (49.4)

8,125 (52.3)

0.000

Chorioamnionitis

512 (0.4)

205 (1.3)

0.000

Fetal intolerance

4,362 (3.5)

820 (5.3)

0.000

Induction of labor

19,328 (15.4)

1,861 (12)

0.000

Vaginal delivery

33,641 (26.7)

3,495 (22.5)

0.000

Cesarean delivery

92,390 (73.3)

12,035 (77.5)

0.000

Operative delivery

3,024 (2.4)

341 (2.2)

0.116

Rupture of membranes

2,211 (1.8)

931 (6.0)

0.000

Meconium staining

1,558 (1.2)

291 (1.9)

0.000

Black maternal race

17,941 (14.2)

2,655 (17.1)

0.000

Primiparous

29,174 (23.3)

3,872 (25.1)

0.000

Abbreviation: NICU, neonatal intensive care unit.

Table 5 Adjusted ORs for NICU admission (N ¼ 141,606) Variable

Adjusted ORs (95% CI)

Newborn gender

1.13 (1.09–1.16)

Black maternal race

1.17 (1.12–1.22)

Primiparous

1.06 (1.02–1.10)

Cesarean delivery

1.18 (1.12–1.22)

Chronic hypertension

1.61 (1.42–1.82)

Diabetes

1.24 (1.17–1.32)

Preeclampsia

1.77 (1.67–1.87)

Apgar score at 5 min < 7

6.05 (5.57–6.58)

Chorioamnionitis

3.10 (2.61–3.69)

Fetal intolerance

1.43 (1.32–1.55)

Premature rupture of membranes

3.39 (3.12–3.67)

Meconium staining

1.35 (1.19–1.54)

Induction of labor

0.75 (0.71–0.80)

Abbreviations: CI, confidence interval; NICU, neonatal intensive care unit; ORs, odds ratios. American Journal of Perinatology

Vol. 32

No. 5/2015

used; this limitation could be significant for the results and applicability of the present study if errors in the assignment of gestational age at delivery were made; nevertheless, high to moderate exact agreement for gestational age at delivery has been reported among birth certificate data and hospital medical records.32 Despite these caveats, based on the balance of evidence in the literature and this population-based study, in terms of clinical implications, there is strong evidence of the optimal timing of delivery of twin pregnancies by 38 weeks of gestation. The profile of cases with an increased risk for NICU admission should enable obstetricians and neonatologists to prepare accordingly for potential complications in these pregnancies and newborns. Recognition of this high-risk profile may allow for timely assessment and early intervention to improve outcomes for these high-risk perinates. Given the practical limits of even multicenter trials for testing interventions for relatively rare complications in relatively rare clinical situations; however, it is not clear that better, higher level evidence will be forthcoming. We believe that the results presented here support delivery of twins in the early-term period. Delivery between 37 and 38 weeks of

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

494

38 weeks

Conflict of Interest The authors report no conflict of interest.

Natl Vital Stat Rep 2011;60(1):1–70 2 Cheung YB, Yip P, Karlberg J. Mortality of twins and singletons by

3

4

1,486 twin pregnancies Retrospective cohort study

Population-based Vilchez 2014, United States

220,169 twin pregnancies

5,594 twin pregnancies Retrospective cohort study

Luke 2005, United States

Dodd 2003, Australia

113 twin pregnancies Retrospective cohort study Suzuki 2010, Japan

273 twin sets Prospective cohort study Bakr 2006, Egypt

2 randomized trials

1,028 twin pregnancies Prospective cohort study

Randomized controlled trial

Systematic review

Breathnach 2012, Ireland

Dodd 2014, Australia

235 twin pregnancies

Randomized controlled trial

Dodd 2012, Australia

Participants

2,804 twin pregnancies

Design Study, setting

Barrett 2013, Canada

6

7 8 9

10

11

12

13 14

15

–

16

3

18

13

21

22

15

24

17

gestational age: a varying-coefficient approach. Am J Epidemiol 2000;152(12):1107–1116 Dodd JM, Robinson JS, Crowther CA, Chan A. Stillbirth and neonatal outcomes in South Australia, 1991-2000. Am J Obstet Gynecol 2003;189(6):1731–1736 Hartley RS, Emanuel I, Hitti J. Perinatal mortality and neonatal morbidity rates among twin pairs at different gestational ages: optimal delivery timing at 37 to 38 weeks’ gestation. Am J Obstet Gynecol 2001;184(3):451–458 Kahn B, Lumey LH, Zybert PA, et al. Prospective risk of fetal death in singleton, twin, and triplet gestations: implications for practice. Obstet Gynecol 2003;102(4):685–692 Minakami H, Kimura H, Honma Y, Tamada T, Sato I. When is the optimal time for delivery?—Purely from the fetuses’ perspective Gynecol Obstet Invest 1995;40(3):174–178 Minakami H, Sato I. Reestimating date of delivery in multifetal pregnancies. JAMA 1996;275(18):1432–1434 National Collaborating Centre for Women’s and Children’s Health. Induction of Labour. London, UK: NICE Clinical Guideline; 2008 Delaney M, Roggensack A, Leduc DC, et al; Clinical Practice Obstetrics Committee; Maternal Fetal Medicine Committee. Guidelines for the management of pregnancy at 41þ0 to 42þ0 weeks. J Obstet Gynaecol Can 2008;30(9):800–823 Gülmezoglu AM, Crowther CA, Middleton P, Heatley E. Induction of labour for improving birth outcomes for women at or beyond term. Cochrane Database Syst Rev 2012;6:CD004945 National Collaborating Centre for Women’s and Children’s Health. Multiple Pregnancy: the Management of Twin and Triplet Pregnancies in the Antenatal Period. London, UK: NICE Clinical Guideline; 2011 ACOG Practice Bulletin No. ACOG Practice Bulletin No. 144: Multifetal gestations: twin, triplet, and higher-order multifetal pregnancies. Obstet Gynecol 2014;123(5):1118–1132 Bakr AF, Karkour T. What is the optimal gestational age for twin delivery. BMC Pregnancy Childbirth 2006;6:3 Dodd JM, Crowther CA. Elective delivery of women with a twin pregnancy from 37 weeks’ gestation. Cochrane Database Syst Rev 2003;(1):CD003582 Dodd JM, Crowther CA, Haslam RR, Robinson JS; Twins Timing of Birth Trial Group. Elective birth at 37 weeks of gestation versus standard care for women with an uncomplicated twin pregnancy at term: the Twins Timing of Birth Randomised Trial. BJOG 2012; 119(8):964–973 Luke B, Brown MB, Alexandre PK, et al. The cost of twin pregnancy: maternal and neonatal factors. Am J Obstet Gynecol 2005;192(3): 909–915 Soucie JE, Yang Q, Wen SW, Fung Kee Fung K, Walker M. Neonatal mortality and morbidity rates in term twins with advancing gestational age. Am J Obstet Gynecol 2006;195(1):172–177 American Journal of Perinatology

Vol. 32

No. 5/2015

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

1 Martin JA, Hamilton BE, Ventura SJ, et al. Births: final data for 2009.

16

Reference

495

gestation provides the lowest risk of neonatal morbidity. Delivery after this period appears to carry no additional benefit, but on the contrary, imparts unnecessary risk to the well-being of the fetus–neonate. Further studies to clearly determine the optimal gestational age of delivery, specifically late preterm versus early term, among apparently uncomplicated monochorionic twins are needed.

5

Table 6 Studies of timing of delivery for term twin pregnancies

Vilchez et al.

References

Neonatal morbidity

37 weeks

37–38 weeks Maternal and neonatal morbidity

Stillbirth and neonatal outcomes

37 weeks or later Perinatal morbidity and mortality

38 weeks or later Maternal and neonatal morbidity

38 weeks

37 weeks Maternal/neonatal morbidity and mortality

Perinatal morbidity and mortality

37 weeks Composite of infant outcomes

Optimal timing of delivery Outcomes

Perinatal morbidity and mortality

N/A

Optimal Timing for Term Delivery of Twin Pregnancies

Optimal Timing for Term Delivery of Twin Pregnancies

Vilchez et al.

18 Suzuki S, Inde Y, Miyake H. Comparison of short-term outcomes of

19

20

21

22

23 24

25

late pre-term singletons and dichorionic twins and optimal timing of delivery. J Obstet Gynaecol 2010;30(6):574–577 Suzuki S, Otsubo Y, Sawa R, Yoneyama Y, Araki T. Clinical trial of induction of labor versus expectant management in twin pregnancy. Gynecol Obstet Invest 2000;49(1):24–27 Spong CY. Defining “term” pregnancy: recommendations from the Defining “Term” Pregnancy Workgroup. JAMA 2013;309(23): 2445–2446 Breathnach FM, McAuliffe FM, Geary M, et al; Perinatal Ireland Research Consortium. Optimum timing for planned delivery of uncomplicated monochorionic and dichorionic twin pregnancies. Obstet Gynecol 2012;119(1):50–59 Dodd JM, Deussen AR, Grivell RM, Crowther CA. Elective birth at 37 weeks’ gestation for women with an uncomplicated twin pregnancy. Cochrane Database Syst Rev 2014;2:CD003582 Jackson M. What is new in multiple gestations? Best articles from the past year. Obstet Gynecol 2014;123(2 Pt 1):359–360 Barrett JF, Hannah ME, Hutton EK, et al; Twin Birth Study Collaborative Group. A randomized trial of planned cesarean or vaginal delivery for twin pregnancy. N Engl J Med 2013;369(14): 1295–1305 Dias T, Thilaganathan B. Elective birth at 37 weeks of gestation versus standard care for women with an uncomplicated twin

American Journal of Perinatology

Vol. 32

No. 5/2015

26

27

28

29

30 31

32

pregnancy at term: the Twins Timing of Birth Randomised Trial. BJOG 2012;119(13):1676, author reply 1676–1677 Scholl J, Russell M. Optimum timing for planned delivery of uncomplicated monochorionic and dichorionic twin pregnancies. Obstet Gynecol 2012;119(6):1276, author reply 1276 Welsh A, Clements S, Henry A, Bisits A. Elective birth at 37 weeks of gestation versus standard care for women with an uncomplicated twin pregnancy at term: the Twins Timing of Birth Randomised Trial. BJOG 2012;119(13):1675–1676, author reply 1676–1677 Hack KE, Derks JB, de Visser VL, Elias SG, Visser GH. The natural course of monochorionic and dichorionic twin pregnancies: a historical cohort. Twin Res Hum Genet 2006;9(3):450–455 American College of Obstetricians and Gynecologists. ACOG committee opinion no. 560: medically indicated late-preterm and early-term deliveries. Obstet Gynecol 2013;121(4):908–910 Cleary-Goldman J, D’Alton ME. Uncomplicated monochorionic diamniotic twins and the timing of delivery. PLoS Med 2005;2(6):e180 Burgess JL, Unal ER, Nietert PJ, Newman RB. Risk of late-preterm stillbirth and neonatal morbidity for monochorionic and dichorionic twins. Am J Obstet Gynecol 2014;210(6):e1–e9 Martin JA, Wilson EC, Osterman MJ, Saadi EW, Sutton SR, Hamilton BE. Assessing the quality of medical and health data from the 2003 birth certificate revision: results from two states. Natl Vital Stat Rep 2013;62(2):1–19

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

496

Copyright of American Journal of Perinatology is the property of Thieme Medical Publishing Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.