Objective
By convention, twins are labeled as twin A or twin B at the time of birth, depending on their birth order. However, antenatally twins are labeled as A or B according to their location relative to the birth canal. Antenatal labeling is usually established by prenatal ultrasound. Most often, fetuses known in utero as twin A are also first born and thus retain their label. However, this is not always the case, such that the twin A newborn was previously labeled twin B in utero and vice versa. This label switch at the time of birth results in confusion among caregivers and some distress to parents who often build a relationship with the babies based on their label. More importantly, the label switch of twins at birth can be a patient safety issue because it may lead to unnecessary medical interventions or medical errors (eg, wrong baby started on antibiotics for antenatal diagnosis of vesicoureteral reflux or wrong baby transferred to another hospital). Finally, the label switch of twins at birth can lead to substantial error in outcome research in linking antenatal intrapartum and neonatal follow-up data and affect results in studies in which the knowledge of antenatal characteristics and birth order is of essence. Ideally, the antenatal and the postnatal labeling of twins should not be conflicting and should have a standardized approach that minimizes the chances for confusion.
Study Design
Two prior studies have addressed the presumed rate of twin label switch at birth. In both studies, the aim was to determine the accuracy of antenatal ultrasound to predict the birth order. In the study by Chasen et al, fetuses were labeled in the second trimester at the time of genetic amniocentesis and were compared with the birth order at the time of delivery at the same institution. In that study, the rate of presumed twin label switch at birth was 10% and was independent of the mode of delivery. The most significant limitation of that study was the interval between ultrasound and delivery.
Study Design
Two prior studies have addressed the presumed rate of twin label switch at birth. In both studies, the aim was to determine the accuracy of antenatal ultrasound to predict the birth order. In the study by Chasen et al, fetuses were labeled in the second trimester at the time of genetic amniocentesis and were compared with the birth order at the time of delivery at the same institution. In that study, the rate of presumed twin label switch at birth was 10% and was independent of the mode of delivery. The most significant limitation of that study was the interval between ultrasound and delivery.
Results
D’Antonio et al reported an incidence of presumed twin label switch at birth of 37% in a retrospective study using a large regional database that included 9 hospitals over a 10 year period. Limitations of that study included the use of sonographic estimated fetal weight and sex discordance as a proxy for labeling. In addition, the study could not correct for the mode of delivery. In utero switching may also occur, with a twin originally designated as B subsequently becoming the presenting twin. In those cases, the fetus continues to be called twin B, even if it is more proximal to the birth canal. To address some of the limitations of the previous studies, we performed an institutional review board–approved retrospective review of prospectively collected data on complicated monochorionic twin pregnancies referred for possible fetal therapy, some of which also delivered at our institution (University of Miami Institutional Review Board protocol 20130553). A total of 367 patients were referred, of which 45 delivered at our institution. Fetuses were clearly labeled with ultrasound in the second trimester using the fetal position, intrauterine location, estimated fetal weight, amniotic fluid volume, and other individual sonographic characteristics. Patients were followed up closely and serially with weekly ultrasounds until the time of delivery. Fifteen cases were excluded because of monoamnionicity (n = 3) or for fetal demise of 1 or both twins (n = 12). The median interval between the last ultrasound and the time of delivery was 0 days, with a range of 0–5 days. Of the 30 patients eligible for analysis, 4 were label switched at birth, for a rate of 13.3%. The mean gestational age at delivery was no different between the label-switched and the non–label-switched groups (27.4 vs 30.9 weeks, P = .079). The median interval from the last ultrasound to delivery was also no different between the groups (0 days, P = .387, Mann-Whitney U test). Only 1 patient was delivered vaginally, and all mismatched twins were delivered by cesarean delivery. The reasons for the label switching at birth are shown in the Table .
