SEARCH

SEARCH BY CITATION

Keywords:

  • chorionicity;
  • fetal loss;
  • perinatal outcome;
  • twin pregnancy

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Objective

To assess outcome in twin pregnancies according to chorionicity.

Methods

A cohort was retrieved from local ultrasound databases at 14 obstetric departments in Denmark, comprising all twin pregnancies with two live fetuses scanned between weeks 11 and 14 in the period 1 January 2004 to 31 December 2006. Outcome data were retrieved from the National Board of Health.

Results

Among 2038 twin pregnancies, 1757 (86.2%) were dichorionic (DC) and 281 (13.8%) were monochorionic diamniotic (MC). In MC pregnancies, the rate of spontaneous fetal loss in both second and third trimesters was more than threefold higher than the comparable rate in DC pregnancies: 6.0% vs. 1.9% for at least one fetus in the second trimester (P < 0.001) and 2.1% vs. 0.7% in the third trimester (P = 0.03). In 98.4% of DC pregnancies and in 91.1% of MC pregnancies, at least one infant was liveborn. Amongst pregnancies with two live fetuses at 24 weeks, the proportion with two live infants at 28 days after delivery was 97.5% and 95.1%, respectively.

Conclusions

The increased incidence of fetal loss in MC pregnancies compared with DC pregnancies predominantly occurs before 24 weeks' gestation. After this stage, although the risk of intrauterine fetal death is still higher in MC than in DC pregnancies, if both fetuses are alive at 24 weeks, the chance of a woman having two live infants 1 month after delivery is similar in MC and DC pregnancies. Copyright © 2011 ISUOG. Published by John Wiley & Sons, Ltd.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

In Denmark, twin pregnancies account for more than 2% of all pregnancies, and the proportion has been rising as a result of the increased use of assisted reproductive technologies (ARTs) and a higher mean maternal age. Twin pregnancies have higher rates of perinatal morbidity and mortality compared with singleton pregnancies1. The increased perinatal mortality for twins is mostly the result of a high incidence of preterm birth, which also often has severe consequences in the neonatal period and in later life2.

Twin pregnancies are classified according to either zygosity or chorionicity, and chorionicity rather than zygosity determines the outcome3, 4. During pregnancy, twins are classified as monochorionic (MC) or dichorionic (DC) according to placentation, and MC twin pregnancies are further differentiated as monoamniotic or diamniotic. MC twins have a higher mortality and morbidity than DC twins, primarily because of the development of twin-to-twin transfusion syndrome (TTTS), which is diagnosed in 8–10% of MC pregnancies5, which also contributes to a higher rate of preterm birth6. Hack et al.7 thus found that median gestational age at delivery was 1 week older in DC twins than in MC twins and consequently found a higher mean birth weight in DC twins.

Prenatal ultrasound has allowed us to accurately determine chorionicity before 15 weeks of gestation based on the presence or absence of the lambda sign8, 9. All pregnant women in Denmark are offered a risk assessment for Down syndrome, including measurement of the fetal nuchal translucency (NT) thickness, between gestational weeks 11 and 14, and the majority of women choose this test10. Thus, most twin pregnancies are recognized in the first trimester, and chorionicity and amnionicity are determined before 15 weeks of gestation. To detect possible complications, MC pregnancies are often more intensively monitored with a routine scan every fortnight11. TTTS can, depending on the Quintero staging12, be effectively treated by selective fetoscopic laser coagulation of anastomosing vessels or selective cord occlusion13. A crown–rump length (CRL) discrepancy at the first-trimester scan may be an early sign of TTTS in MC pregnancies and can also be associated with the risk of adverse short-term outcome in DC pregnancies14–17.

The aim of our study was to evaluate pregnancy outcome in a large unselected cohort of twin pregnancies, with the main focus on describing mortality rates according to chorionicity from the first trimester until the neonatal period.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This cohort study is based on data entered into a fetal medicine software program (Astraia, Munich, Germany), which was taken into use by all Departments of Obstetrics and Gynecology in Denmark in the period 1 January 2003 to 1 July 2006. The study population was identified by retrieving data on twin pregnancies with two live fetuses at the time of the NT scan in the period between 1 January 2004 and 31 December 2006 from local Astraia servers in 14 of the 21 Departments of Obstetrics and Gynecology in Denmark, which provide care for more than 75% of the pregnant population.

Outcome data were collected from the Medical Birth Register (MBR) at the National Board of Health to which all pregnancies and deliveries in Denmark are reported consecutively. In all cases of fetal loss and where the outcome was unclear or missing in the MBR, we retrieved data from medical records. Outcome data were cross-linked with the ultrasound data from Astraia using the unique personal identification number held by every citizen in Denmark. All data had been recorded prospectively and were analyzed retrospectively.

Our inclusion criteria were twin pregnancies with a chorionicity determination and two live fetuses identified with a CRL of 45–84 mm at the time of the NT scan. The earliest assessment of chorionicity and CRL was used, and chorionicity was assessed in all departments based on the presence or absence of the lambda sign and/or two separate placentae. Exclusion criteria were pregnancies with unknown chorionicity, monoamniotic pregnancies, more than two fetuses and pregnancies with a known reduction from a higher number of multiples.

To describe baseline characteristics for the population, we collected information on maternal age at the time of the NT scan, conception method, smoking status, body mass index (BMI) and parity. The most reliable data concerning ART were on in-vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI). Hence, ART in this study does not include ovulation induction and intrauterine insemination. Information about ART was obtained from the Danish IVF Register at the National Board of Health. Smoking status was categorized as smokers and non-smokers. Prepregnancy BMI was defined according to the World Health Organization (WHO) criteria: underweight, < 18.5 kg/m2; normal weight, 18.5–24.9 kg/m2; overweight, 25.0–29.9 kg/m2; and obese > 29.9 kg/m2. Parity was categorized as nulliparous vs. parous.

The primary outcomes of interest were fetal loss and neonatal death. These outcomes were evaluated with regard to chorionicity for all diamniotic pregnancies in the population. Our hypothesis was that mortality rate early in pregnancy is higher in MC twin pregnancies compared with DC pregnancies. We further wanted to assess whether mode of conception affected mortality rate.

We dichotomized spontaneous fetal loss into miscarriage before 23 + 6 weeks of gestation and intrauterine fetal death (IUFD) after 24 + 0 weeks of gestation. Perinatal death rate was calculated as the number of IUFDs and the number of intrapartum and neonatal deaths within the first week of postnatal life per total number of fetuses alive at 24 weeks. Neonatal death rate was defined as death within the first 28 completed days of delivery per total number of liveborn infants. Mode of delivery was considered as induced when labor-inducing medication was given, amniotomy was performed or a planned Cesarean delivery was carried out. Gestational age at delivery was based on the CRL of the largest fetus at the 11–14-week scan. Mean birth weight was calculated from the sum of the infants' individual weights. Intertwin weight disparity was calculated by dividing the difference in birth weight by the weight of the larger twin.

TTTS was defined as an unequal exchange of blood between twins, leading to at least Stage I in the Quintero staging method12.

Comparisons of categorical variables were made using the chi-square test or the Fisher's exact test if there were five or fewer expected observations in one or more cells. P-values for all hypotheses tested were two-sided, and statistical significance was set at P < 0.05. Continuous variables were compared by the Independent Samples t-test or the Wilcoxon rank sum test, as appropriate.

Statistical analysis was performed using SPSS, version 15.0 (SPSS, Chicago, IL, USA).

The study was approved by the Danish Data Protection Agency and by the National Board of Health.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The total study population comprised 2043 twin pregnancies. Five pregnancies (all DC) were excluded for loss to follow-up because of emigration before delivery and therefore outcome details could not be obtained. This left 2038 pregnancies, of which 1757 (86.2%) were DC twins and 281 (13.8%) were MC twins.

Table 1 shows the baseline characteristics for DC and MC twin pregnancies. The two groups differed significantly in mean maternal age and in mode of conception. Mothers of DC twins were older (32.0 vs. 31.0 years), and the pregnancies were more likely to be conceived by IVF or ICSI (363 (20.7%) vs. 17 (6.0%)).

Table 1. Baseline characteristics
CharacteristicDichorionic pregnanciesMonochorionic pregnanciesPTotal
  • Data are given as n (%) or as mean ± SD.

  • *

    Maternal age at time of nuchal translucency scan.

  • In-vitro fertilization and intracytoplasmic sperm injection.

  • Mean gestational age at time of nuchal translucency scan. BMI, body mass index.

n1757 (86.2)281 (13.8) 2038 (100)
Maternal age* (years)32.0 ± 4.331.0 ± 4.50.00131.9 ± 4.3
Prepregnancy BMI  0.23 
 < 18.5 kg/m255 (3.1)11 (3.9) 66 (3.2)
 18.5–24.9 kg/m21041 (59.2)176 (62.6) 1217 (59.7)
 25.0–29.9 kg/m2327 (18.6)45 (16.0) 372 (18.3)
 > 29.9 kg/m2188 (10.7)21 (7.5) 209 (10.3)
 Missing data146 (8.3)28 (10.0) 174 (8.5)
Smoking status  0.93 
 Non-smoking1471 (83.7)225 (80.1) 1696 (83.2)
 Smoking193 (11.0)29 (10.3) 222 (10.9)
 Missing data93 (5.3)27 (9.6) 120 (5.9)
Parity  0.19 
 Nulliparous982 (55.9)138 (49.1) 1120 (55.0)
 Parous723 (41.1)121 (43.1) 844 (41.4)
 Missing data52 (3.0)22 (7.8) 74 (3.6)
Mode of conception  < 0.001 
 Assisted reproductive technologies363 (20.7)17 (6.0) 380 (18.6)
 Spontaneous1394 (79.3)264 (94.0) 1658 (81.4)
Gestational age (weeks)12.8 ± 0.612.7 ± 0.60.0112.8 ± 0.6

Table 2 summarizes fetal and neonatal death according to chorionicity. Pregnancy termination was more likely to be performed in MC cases (n = 13, 4.6%) compared with DC (n = 6, 0.3%). TTTS was diagnosed in 10% of the MC cases (n = 28). Chromosomal anomalies and psychosocial indications accounted for the DC terminations, while 11 MC pregnancies were terminated because of TTTS (10 pregnancies between weeks 12 + 0 and 19 + 0, and one pregnancy at 23 + 6 weeks). Fourteen MC pregnancies were treated for TTTS by selective fetoscopic laser coagulation of anastomosing vessels or cord occlusion between weeks 17 + 0 and 24 + 5.

Table 2. Outcome of pregnancy according to chorionicity
VariableDichorionic pregnancies (n (%))Monochorionic pregnancies (n (%))PTotal (n (%))
  • *

    In one case, intrauterine fetal death (IUFD) after 24 weeks occurred in the remaining twin after selective feticide.

  • Number of IUFDs and/or early neonatal deaths within 7 days after delivery per total number of live fetuses at 24 weeks.

  • Infant deaths within first 28 days after delivery per total number of liveborn infants.

  • §

    Excluding infants which were delivered before 24 weeks.

Pregnancies1757 (86.2)281 (13.8) 2038 (100)
Pregnancy terminations6 (0.3)13 (4.6)< 0.00119 (0.9)
Pregnancies with selective feticide17 (1.0)8 (2.8)0.0225 (1.2)
Pregnancies with fetal loss < 24 + 0 weeks46 (2.6)23 (8.2)< 0.00169 (3.4)
 One fetus13 (0.7)9 (3.2)< 0.0122 (1.1)
 Both fetuses21 (1.2)8 (2.8)0.0529 (1.4)
 At least one fetus34 (1.9)17 (6.0)< 0.00151 (2.5)
Pregnancies with fetal loss ≥ 24 + 0 weeks    
 One fetus11 (0.6)4 (1.4)*0.1415 (0.7)
 Both fetuses1 (0.1)2 (0.7)0.053 (0.1)
 At least one fetus12 (0.7)6 (2.1)*0.0318 (0.9)
Perinatal death (fetuses/infants lost)47/3430 (1.4)16/503 (3.2)0.00663/3933
Neonatal death (infants lost)42/3417 (1.2)9/495 (1.8)0.2951/3912 (1.3)
Birth of at least one liveborn infant1729 (98.4)256 (91.1)< 0.0011985 (97.4)
Birth of two liveborn infants1688 (96.1)239 (85.1)< 0.0011927 (94.6)
At least one infant alive after 28 days1717 (97.7)253 (90.0)< 0.0011970 (96.7)
Both infants alive after 28 days1658 (94.4)233 (82.9)< 0.0011891 (92.8)
Both infants alive at 28 days if two live fetuses at 24 weeks§1657/1700 (97.5)231/243 (95.1)0.061888/1943 (97.2)

Seventeen (1.0%) DC pregnancies were reduced from two fetuses to one; most indications were chromosomal anomalies or severe malformations.

In MC pregnancies, the rate of spontaneous fetal loss in both second and third trimesters was more than threefold higher than the comparable rate in DC pregnancies (17 (6.0%) vs. 34 (1.9%) for at least one fetus in the second trimester and six (2.1%) vs. 12 (0.7%) in the third trimester). Among the 1779 pregnancies with two live fetuses at 32 weeks of gestation, IUFD of at least one fetus occurred in one (0.5%) of the MC pregnancies and in five (0.3%) of the DC pregnancies (P = 0.54). Perinatal death rate, calculated as number of fetuses, was 16 (3.2%) in MC cases and 47 (1.4%) in DC cases (P = 0.006). There was no difference in the rate of neonatal death between the two groups (42 (1.2%) for DC vs. nine (1.8%) for MC (P = 0.29).

A total of 1688 (96.1%) DC pregnancies and 239 (85.1%) MC pregnancies resulted in two liveborn infants (P < 0.001), and 1729 (98.4%) and 256 (91.1%) resulted in at least one liveborn infant, respectively (P < 0.001). Among pregnancies in which both fetuses were alive at 24 weeks of gestation, both infants were alive 28 days after delivery in 1657 (97.5%) of the DC cases and in 231 (95.1%) of the MC cases (P = 0.06). Table 3 provides the gestation and possible reason for fetal death after 24 weeks of gestation. For DC pregnancies, the main reason was unknown; fetuses were found dead at the routine scan. For MC pregnancies, the main reasons for fetal death were unknown or consequences of TTTS.

Table 3. Gestational age (GA) at, and possible reasons for, fetal death after 24 weeks' gestation
FindingDichorionic GAMonochorionic GA
  • *

    Found dead at routine scan.

  • Intrauterine fetal death of both twins. IUGR, intrauterine growth restriction; TTTS, twin-to-twin transfusion syndrome.

Unknown*24 + 524 + 2
 29 + 227 + 0
 32 + 235 + 4
 33 + 0 
 33 + 1 
 34 + 0 
IUGR26 + 3 
 28 + 3 
 29 + 0 
TTTS 25 + 0
  25 + 0
  31 + 5
Trisomy 13/1833 + 0 
 33 + 5 
Premature onset of labor24 + 1 

To assess whether mode of conception affected the spontaneous fetal loss rate, we performed analyses of fetal loss in twin pregnancies conceived by ART compared with spontaneously conceived twin pregnancies. We did not find any statistically significant differences in either DC or MC twins. The rate of miscarriage/IUFD in the second trimester was nine (2.5%) in DC pregnancies conceived by ART vs. 25 (1.8%) in DC pregnancies of spontaneous conception (P = 0.40), and two (11.8%) vs. 15 (5.7%) in MC pregnancies (P = 0.27).

Gestational age at birth, mode of delivery and birth weight were analyzed in the 1927 twin pregnancies with two liveborn infants (Table 4). There was no difference in mode of delivery, categorized as spontaneous vs. induced, between DC and MC twin pregnancies. Median gestational age at delivery was higher in DC twins compared with MC twins (260.5 days vs. 256.5 days, P < 0.001). While 998 (59.1%) of the DC twins were delivered at 37 weeks or later, this was the case for only 106 (44.4%) of the MC twins (P < 0.001). Two hundred and fifty (14.8%) DC twins and 49 (20.5%) MC twins were born before 34 weeks of gestation (P = 0.02).

Table 4. Gestational age at delivery, mode of delivery and birth weights for twin pregnancies with two liveborn infants
VariableDichorionic pregnanciesMonochorionic pregnanciesPTotal
  1. Data are given as n (%), mean ± SD or median.

Mode of delivery  0.06 
 Spontaneous709 (42.0)85 (35.6) 794 (41.2)
 Induced or elective979 (58.0)154 (64.4) 1133 (58.8)
Gestational age at birth (days) 
 Mean ± SD253.5 ± 18.6248.5 ± 20.9 252.9 ± 19.0
 Median260255< 0.001260
Gestational age at delivery  < 0.001 
 22 to 27 + 6 weeks33 (2.0)9 (3.8) 42 (2.2)
 28 to 31 + 6 weeks96 (5.7)16 (6.7) 112 (5.8)
 32 to 33 + 6 weeks121 (7.2)24 (10.0) 145 (7.5)
 34 to 36 + 6 weeks440 (26.1)84 (35.1) 524 (27.2)
 > 36 + 6 weeks998 (59.1)106 (44.4) 1104 (57.3)
Birth weight (g) 
 Mean ± SD2526.9 ± 551.92384.9 ± 579.6 2509.3 ± 557.2
 Median2620.02467.0< 0.0012603.8
Intertwin birth-weight disparity of ≥ 25%147 (8.7)18 (7.5)0.55165 (8.6)
Missing birth-weight values23 (1.4)4 (1.7) 27 (1.4)

Birth weights were available for 1900 (1665 DC and 235 MC) twin pairs. As expected according to the difference in gestational age at delivery, median birth weight was lower in MC twins compared with DC twins (2467 g vs. 2620 g, P < 0.001). There was no difference regarding intertwin birth weight disparity of more than 25% between DC and MC twins (147 (8.7%) vs. 18 (7.5%)).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

In this large national study of twin pregnancies, we found that the loss rate of at least one fetus in MC twin pregnancies was more than threefold that of DC twin pregnancies (8.2% vs. 2.6%), with most deaths occurring before 24 weeks of gestation. Despite the lower gestational age at birth and the higher rate of preterm birth in MC twins, neonatal mortality within the first 28 days following delivery was not correlated to chorionicity. Both infants were alive at 28 days after delivery in 94.4% of all DC cases and in 83.0% of all MC cases, but amongst pregnancies with two live fetuses at 24 weeks, the proportion with two live infants at 28 days after delivery was as high as 97.5% in DC twins and 95.1% in MC twins. The overall survival rate for MC twins was thus clearly lower than for DC twins, but leveled out after 24 weeks of gestation. Our rates of at least one fetal loss and neonatal death are comparable with the rates reported by others4, 18 for both DC and MC pregnancies. The population in the Sebire study18 was collected from one highly specialized center from 1992 to 1996. This study found a higher total fetal loss rate than ours for MC pregnancies (17.6% vs. 10.3%, P = 0.08), a difference that may be explained by increased surveillance and more aggressive use of invasive fetal therapy with time.

In a prospective study of solely MC pregnancies (n = 202), Lewi et al.19 reported that survival of two infants was 85.1%, which is in accordance with the survival we found for MC infants (83.0%). However, in our study, 4.6% of the MC pregnancies were terminated before 24 weeks, mainly because of severe TTTS, compared with only 0.5% in Lewi's study. This difference may be explained by the different recruitment methods in the two studies. Lewi et al. only included women followed at two highly specialized centers from 11–14 weeks' gestation, while in our study, cases were followed at 14 different centers. This could delay referral of cases to the national center performing laser therapy. In our study, the prospective risk of death after 32 weeks was similar in MC and DC pregnancies; there therefore does not seem to be any indication for elective delivery of MC twin pregnancies at 32 weeks.

To our knowledge this is the largest cohort of twin pregnancies with an early ultrasound chorionicity determination described to date. The enrolment of pregnancies was made possible through Denmark's unique use of the same ultrasound software program in all departments of obstetrics and gynecology and the reporting of all outcomes to the MBR. While other studies excluded pregnancies with complications such as congenital malformations, chromosomal defects and treatment for TTTS, we chose few exclusion criteria in this study, and thus expect very little selection bias. However, a selection exists, as our population solely consists of twin pregnancies with two live fetuses at 11–14 weeks' gestation. As 14 of the 21 Danish departments of obstetrics and gynecology contributed data to this study, this population is very likely to be representative for the general Danish twin population.

A limitation of our study is that it is based on register data. Inherent in register-based studies is the risk of incorrect data transfer through either human or computer error. To minimize this risk, one person manages all data entries into the MBR. In the present study all cases of fetal loss were also reviewed in patient files.

In our study, 13.8% of the twin pregnancies were MC, and among the ART pregnancies 4.5% were MC. Similar results were reported in a recent Danish-Swedish study4, while Sebire et al.18 reported 22% MC pregnancies in 1997. The difference may be explained by the increasing use of ART and thus the higher rate of DC placentation.

In three cases, chorionicity was determined as MC at 11–14 weeks' gestation, but later in gestation the twins were found to be of different genders—and thus DC. These pregnancies were included in the analysis as MC, based on the first assessment of chorionicity. This raises a question about the accuracy of chorionicity determination. Carroll et al.20 reported that basing dichorionicity determination on a combined use of the lambda sign and/or two separate placentae had a sensitivity of 97.4% and a specificity of 100%, while the T-sign had 100% sensitivity and 98.2% specificity in predicting monochorionicity. These are the methods currently used in Denmark, and our data are in accordance with these figures. Gender assessment should therefore be carried out for all twin pregnancies classified as MC to further minimize the risk of incorrect chorionicity determination.

Our findings provide important information for couples with twin pregnancies at the time of the 11–14-week scan. They can be counseled about the risk of fetal and neonatal loss, and it can be stressed that in 83% of MC pregnancies and 94% of DC pregnancies, both infants will be alive 28 days after delivery. Furthermore, if there are two live fetuses at 24 weeks there will be two live infants at 28 days in 97.5% and 95.1% of DC and MC pregnancies, respectively. The survival rate in MC twin pregnancies seems to be higher than previously reported18. This could be a result of the national first-trimester risk assessment program, better surveillance methods and interventions, as well as effective centralization of complicated cases and intensive prenatal and postnatal care. However, we have not investigated the long-term outcome, and knowing that twin pregnancy complications such as preterm birth and TTTS can cause neurodevelopmental impairment, there is a strong incentive to conduct follow-up studies on morbidity and development in later life.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Source of funding: Danish Agency for Science, Technology and Innovation.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES