Introduction
Pregnancy risks, clinical management and subsequent outcomes are very different for monochorionic and dichorionic twin pregnancies (and monochorionic, dichorionic and trichorionic triplet pregnancies). Currently, there appears to be considerable variation and uncertainty in the practice of assigning chorionicity for twin and triplet pregnancies, leading to the GDG prioritising this question for review. Diagnostic accuracy of various methods for determining chorionicity in twin and triplet pregnancies at different gestational ages was sought.
Description of included studies
Fourteen studies investigating diagnostic accuracy of the following characteristics (as determined by an ultrasound scan) for determining chorionicity were identified for inclusion:38–51
membrane thickness
number of membrane layers
number of placental sites and lambda/T-sign
composite measures based on the above characteristics and others (number of placental masses, number of gestational sacs, concordant/discordant fetal sex and number of fetal poles).
Only two studies included triplets, and one of these included only one triplet pregnancy, meaning that sensitivity, specificity, positive predictive values (PPVs) and negative predictive values (NPVs) and likelihood ratio statistics could not be calculated using the triplet data in the study.50
Six prospective cohort studies reported findings for using membrane thickness to determine chorionicity in twin pregnancies.38;39;42;45–47 Thresholds for determining monochorionicity ranged from 1.0 mm to 2.0 mm, and some studies reported results for different thresholds within the same publication. One study was conducted in the UK,39 one in Belgium45 and four in the USA.38;42;46;47
Four prospective cohort studies reported on using the number of placental masses and a lambda or T-sign for determining chorionicity in twin pregnancies.38;39;45;49 One study was conducted in the UK,39 one in Belgium,45 one in the USA38 and one in Canada.49
One prospective cohort study reported on using the number of membrane layers to determine chorionicity in twin pregnancies.48 This study was conducted in Canada.
One prospective cohort study conducted in the USA reported on using the number of placental sites to determine chorionicity in twin pregnancies.43.
Seven studies reported findings for a mixture of methods for determining chorionicity in twin and triplet pregnancies.39–41;44;49–51 Five studies were prospective cohort studies of twin pregnancies,39;41;49–51 one was a retrospective cohort study of twin pregnancies40 and one was a prospective cohort study of triplet pregnancies.44 Two studies were conducted in the UK,39;41 one in France,44 one in Canada49 and three in the USA.40;50;51
Published health economic evidence
No published health economic evidence was identified and this question was not prioritised for health economic analysis.
Evidence profiles
Evidence profiles for this question are presented in to .
GRADE summary of findings for scans performed at 11–14 weeks of gestation.
GRADE summary of findings for scans performed at more than 14 weeks of gestation.
GRADE summary of findings for scans performed before 11 weeks of gestation or over a wide range of gestational ages with no mean age reported.
presents data from scans performed at 11–14 weeks of gestation, which is when the first ultrasound scan is performed in general UK practice. presents data from scans performed after 14 weeks of gestation, which best represents the gestational age at which women would be scanned if they missed the scan at 11–14 weeks. presents data from scans performed before 11 weeks of gestation, and from studies that reported data for a wide range of gestational ages without reporting the mean gestational age at the time of the scan; these data are less applicable to UK practice.
Results for twin pregnancies are expressed in terms of detection of monochorionicity. For example, diagnostic accuracy values for the lambda sign are reported as absence of the sign (which suggests monochorionicity) rather than presence of the sign (which suggests dichorionicity).
Results for triplet pregnancies are expressed in terms of detection of a monochorionic or dichorionic triplet pregnancy, rather than a trichorionic pregnancy.
Evidence statement
Evidence was identified for a variety of methods used to determine chorionicity from ultrasound scans in twin and triplet pregnancies.
The sensitivity and specificity of the methods used to determine chorionicity from ultrasound scans is generally high. Over half of the reported methods achieved both a sensitivity and specificity over 90%.
At a mean or median gestational age of 11–14 weeks at the time of scan, diagnostic accuracy statistics were reported for membrane thickness (low and moderate quality evidence), the number of placental masses and lambda/T-sign (very low quality evidence), and two different composite methods (low quality evidence). The strongest likelihood ratios were reported for a composite method involving lambda/T-sign and number of placental masses with or without concordant/discordant fetal sex. The sensitivity for this test was also high.
For a mean or median gestational age of more than 14 weeks at the time of scan, results were reported for the use of membrane thickness (very low quality evidence), the number of placental sites (moderate quality evidence) and two different composite methods (very low and moderate quality evidence). Composite methods (number of placental masses and lambda/T-sign, and concordant/discordant fetal sex with or without membrane thickness) showed the strongest likelihood ratios. The highest sensitivity was reported when membrane thickness was included in the composite method.
Some studies reported findings for a gestational age of less than 11 weeks or over a wide range of gestational ages with no mean age reported. Results were reported for membrane thickness (very low to moderate quality evidence), number of membrane layers (moderate quality evidence), the number of placental masses and lambda/t-sign (low quality evidence), and composite methods (low to moderate quality evidence). The composite methods showed the strongest likelihood ratios and high sensitivity. These methods used membrane thickness and number of placental masses, with or without lambda/T-sign, number of gestational sacs and number of fetal poles.
The GDG is aware that the evidence presented may be biased due to analysis after the study concluded for patterns that were not specified before the study, particularly in studies that examined individual methods such as membrane thickness. In these studies, it is not clear how a clinician determining chorionicity on one measure alone (such as subjectively thin or thick membrane) would not be influenced by other aspects of the ultrasound scan (such as the number of gestational sacs).
Evidence to recommendations
Relative value placed on the outcomes considered
Sensitivity is the percentage of pregnancies found to be monochorionic at placental examination that were predicted to be monochorionic at scan (true positive). One hundred minus sensitivity (100 − sensitivity) is the percentage of pregnancies found to be monochorionic at placental examination that were predicted to be dichorionic at scan (false negative).
Specificity is the percentage of pregnancies found to be dichorionic at placental examination that were predicted to be dichorionic at scan (true negative). One hundred minus specificity (100 − specificity) is the percentage of pregnancies found to be dichorionic at placental examination that were predicted to be monochorionic at scan (false positive).
PPV is the percentage of pregnancies predicted to be monochorionic by the scan that were confirmed at placental examination to be monochorionic. One hundred minus PPV (100 − PPV) is the percentage of pregnancies predicted to be monochorionic by the scan result that were confirmed at placental examination to be dichorionic.
NPV is the percentage of pregnancies predicted to be dichorionic by the scan that were confirmed at placental examination to be dichorionic. One hundred minus NPV (100 − NPV) is the percentage of pregnancies predicted to be dichorionic by the scan that were confirmed at placental examination to be monochorionic.
The positive likelihood ratio (LR+) shows how much the odds of a pregnancy being monochorionic increase when a scan predicts monochorionicity. The negative likelihood ratio (LR−) shows how much the odds of a pregnancy being monochorionic decrease when a scan predicts dichorionicity.
The GDG prioritised likelihood ratios and sensitivity when considering the evidence for different methods of predicting chorionicity. They considered a sensitivity of less than 75% to be an imprecise test, and this is reflected in the GRADE profiles for this review question.
Trade-off between clinical benefits and harms
Determination of chorionicity is required to correctly stratify perinatal risk according to the type of twin or triplet pregnancy. Since pregnancy risks, clinical management and subsequent outcomes are very different for monochorionic and dichorionic twin pregnancies (and monochorionic, dichorionic and trichorionic triplet pregnancies), accurately determining chorionicity is very important.
Monochorionic twin pregnancies have a higher risk of developing complications, including feto-fetal transfusion syndrome (FFTS), fetal growth problems, structural abnormalities and overall perinatal loss compared with dichorionic twin pregnancies. The assessment of chorionicity is easier in the first trimester than in later pregnancy and so it is important to assess and document chorionicity clearly at this gestational age. There is benefit in identifying true positives as women with monochorionic pregnancies will require additional fetal surveillance. Women can make decisions fully informed of risks and appropriate management of monochorionicity can be implemented.
Identification of true negatives (women with dichorionic pregnancies) will result in a saving of time and money by avoiding unnecessary additional interventions. False positives will result in additional and unnecessary monitoring, anxiety and cost in women with dichorionic pregnancies.
False negatives have the least desirable outcome, as monochorionic pregnancies will be monitored less, increasing the likelihood of missing serious complications. Furthermore women with false negative test results will not be informed about these potential risks and the consequences.
The trade-off between clinical benefits and harms is unaffected by the choice of methods for determining chorionicity since any measurements would be taken during a single ultrasound scan appointment.
Trade-off between net health benefits and resource use
There is no cost difference between the methods themselves (except that composite methods might take more time for measurements to be conducted) as they can be done at the same ultrasound scan. A method that is more accurate will be more cost effective than less accurate methods if it means fewer women with dichorionic pregnancies receive unnecessary extra monitoring. The GDG emphasised that these scans will tie in to the existing NICE guidance for dating pregnancy and screening, and so the extra costs will be minimal.
Quality of evidence
The quality of evidence was summarised separately for scans done at different times.
For scans at 11–14 weeks:
membrane thickness: quality ranged from low to moderate and was mainly moderate
number of placental masses and lambda or T-sign: quality was very low
composite measures: quality was low.
For scans at more than 14 weeks:
membrane thickness: quality was very low
number of placental sites: quality was moderate
composite methods: quality was very low and moderate.
For scans at less than 11 weeks or at a wide range of gestational ages:
membrane thickness: quality was very low to moderate
number of membrane layers: quality was moderate
number of placental masses and lambda or T-sign: quality was low
composite measures: quality was moderate to low.
Other considerations
Only one study reported on diagnosing chorionicity in triplet pregnancies and this study evaluated only one method. The GDG assumed that the diagnostic accuracy of methods for determining chorionicity were similar for twin and triplet pregnancies. The GDG is aware that current practice for determining chorionicity involves a composite of methods and there are differences across England and Wales in timing of ultrasound scans. If a twin or triplet pregnancy is diagnosed before 11 weeks of gestation, determining chorionicity immediately using a composite of the number of placental masses, the presence of a lambda or T-sign and membrane thickness is as effective as waiting for the 11 weeks 0 days to 13 weeks 6 days scan. There is no evidence that the use of three-dimensional scans improves the accuracy of chorionicity determination. From a practical point of view it makes sense to perform estimation of gestational age, chorionicity and fetal trisomy screening at the same first-trimester ultrasound scan and the best interval for all three is 11 weeks 0 days to 13 weeks 6 days.
The GDG recognised the importance of assigning nomenclature to fetuses (for example upper and lower, or left and right) and documenting this clearly to ensure consistency throughout pregnancy.
The GDG also recognised the importance of training and support from senior colleagues to ensure that ultrasonographers can identify the presence of a lambda or T-sign accurately and confidently. In view of the potential consequences of failure to determine chorionicity at the time of diagnosis of the twin or triplet pregnancy (especially failure to identify monochorionic pregnancies correctly) the GDG’s recommendations include the possibility of seeking advice from a senior colleague or referral for specialist advice (from a healthcare professional who is competent in determining chorionicity by ultrasound scan).
The GDG’s discussions highlighted that many women with twin and triplet pregnancies are told that the risks associated with such pregnancies depend on zygosity whereas in fact the risks are dependent on chorionicity, and so the GDG identified this as a specific issue to be covered in training.
The GDG also recognised the importance of maternity networks (proposed in the NHS White Paper ‘Equity and excellence: liberating the NHS’**) in establishing appropriate care pathways for all twin and triplet pregnancies, regardless of chorionicity. Since maternity networks are not yet in place throughout England and Wales, the GDG has used the term ‘networks’ in its recommendations, in accordance with the Department of Health guidance.† The GDG considered that special consideration should be given to monochorionic monoamniotic pregnancies (see Chapter 9 for further details).