Introduction

Fetal growth restriction (FGR), defined by the American College of Obstetricians and Gynecologists (ACOG) as an estimated fetal weight (EFW) of <10th percentile of the EFW for the gestational age, complicates approximately 10% of pregnancies and is a leading cause of neonatal morbidity and mortality. Infants with birthweights below the 10th percentile are predisposed to multiple neonatal complications including hypoglycemia, hyperbilirubinemia, hypothermia, intraventricular hemorrhage, necrotizing enterocolitis, sepsis, respiratory distress syndrome, and death. ^, In addition to the increased risk for adverse perinatal outcomes, FGR can also impact long-term health outcomes, including an increased risk for the development of cardiovascular and endocrine diseases as well as neurologic impairment. ^, Given the increased risks associated with pathologic FGR, a correct prenatal diagnosis is essential for the antenatal management of these pregnancies, which includes increased antenatal surveillance and, in certain cases, iatrogenic preterm delivery.

Despite the prevalence of FGR, our ability to correctly identify those fetuses with pathologic FGR and, in turn, those most at risk for poor outcomes, can be affected by a multitude of factors. One of these factors is the variability in growth references used to detect fetal growth. Currently, twin growth is measured against population-based, singleton growth references. Compared with their singleton counterparts, twins are at a greater risk of being diagnosed with FGR mainly because twin growth trajectories deviate from those of singleton growth trajectories around 30 to 32 weeks gestational age. The percentage of twins with birthweights classified as <10th percentile ranges from 18% to 46% when singleton growth references are used, and decreases to 13% to 17% when twin-specific growth references are implemented. ^{, ,} It is unclear whether this differing growth trajectory in twins is a normal physiological adaptation to a multiple gestation or if it is caused by pathologic growth restriction that occurs if the intrauterine environment is incapable of sustaining the equivalent growth of twin fetuses compared with a singleton as a pregnancy progresses. ^, Various twin-specific growth references have been published, including the 2016 Eunice Kennedy Shriver National Institute of Child Health and Human Development’s (NICHD) Fetal Growth Studies—Twins, which was a prospective, longitudinal evaluation of fetal growth in dichorionic twin gestations. ^, These investigators and others have called for studies assessing the perinatal outcomes following use of these twin-specific fetal growth references instead of the singleton growth references that are typically used for the assessment of normal twin growth and which may overestimate the risk for morbidity and mortality. ^,

The objective of this study was to compare the NICHD’s twin-specific growth reference with the Hadlock singleton growth reference used at our institution. We compared the relative frequencies of FGR (EFW of <10%) identified by each reference and the adverse neonatal outcomes associated with this diagnosis in dichorionic twin gestations.

Materials and Methods

This was a retrospective cohort study of dichorionic, diamniotic twin pregnancies at ≥32 weeks’ gestation delivered at the Medical University of South Carolina (MUSC) between 2004 and 2019 with serial growth ultrasounds and neonatal outcomes data available for analysis. Cases of singleton pregnancies, monochorionic twin gestations, delivery at <32 weeks’ gestation, fetal demise, with known aneuploidy, known fetal anomaly, outbound delivery, or lack of a growth ultrasound at ≥32 weeks’ gestation were excluded. This study was approved by the MUSC Institutional Review Board (Pro00103996).

Eligible subjects were identified, and the maternal and neonatal variables were extracted from the MUSC Perinatal Information Network System (PINS) database. The PINS database is a validated research quality database operated collaboratively by the Departments of Obstetrics and Gynecology, Pediatrics, and Public Health Sciences at MUSC. All maternal and low-risk newborn inpatient medical records were abstracted by 3 specially trained data abstractors following delivery. All high-risk newborn records were abstracted by a dedicated nurse with 5 years of clinical experience in the neonatal intensive care unit (NICU) and 10 years of experience in abstraction of high-risk newborn medical records. The proportion of specific agreement reflects the correct identification of a condition when it exists; for the PINS database, this exceeds 97% with interrater reliability exceeding 98%.

Ultrasound data were obtained from our institutional ultrasound database software (Viewpoint, GE Healthcare, Milwaukee, WI). The EFW was determined using 4 elements of fetal biometry (biparietal diameter, head circumference, abdominal circumference, and femur length) and was calculated using the Hadlock formula. It should be noted that the Hadlock formula used to calculate EFW in our study was different from that used in the NICHD study, which did not include bronchopulmonary dysplasia (BPD) in its measurements. Approximately 100 twin pairs did not have chorionicity exported from Viewpoint to our database. In all of those cases, the chorionicity data were obtained by direct review of the patient’s ultrasound report in Viewpoint or electronic medical record in EPIC (Epic Systems Corporation, Verona, WI). If chorionicity was not documented, any cases with infants of 2 separate sexes were determined to be dichorionic gestations. In cases with 2 infants of the same sex, the placental pathology was reviewed to determine the chorionicity. All serial growth ultrasound examinations, which occurred every 4 weeks throughout gestation starting at around 20 weeks, were performed at 4 separate outpatient ambulatory sites and were interpreted by maternal-fetal medicine specialists. Our ultrasound department performs approximately 22,000 ultrasound examinations per year and is accredited in obstetrics and fetal echocardiography by the American Institute of Ultrasound in Medicine. All of our sonographers are registered diagnostic medical sonographers.

Based on their respective last fetal growth ultrasound before delivery, each individual fetus was assigned an EFW growth percentile based on the Hadlock singleton growth reference routinely used at our institution and the NICHD’s twin-specific growth reference for dichorionic twins. Then, each twin was classified into 1 of the following 3 groups: growth restricted (EFW of <10%) according to both the Hadlock and NICHD growth references, growth restricted according to the Hadlock growth reference only, or not growth restricted according to either growth reference. Adverse neonatal outcomes were compared between the fetuses in each of the 3 groups.

A matching process was used to ensure that the EFW percentiles assigned based on the final antenatal ultrasound examination corresponded to the correct neonatal outcomes obtained from PINS for each individual newborn in a twin pair. The records were first matched based on the maternal medical record number, name, date of birth, and pregnancy time point (predominately year) with duplicate records excluded. Once the twin sets were correctly identified in both data sets, the individual fetus match was determined based on the birthweight, with the larger birthweight infant being matched to the larger EFW measurement and the smaller birthweight to the smaller EFW measurement. For cases with identical birth weights recorded for both infants, the medical records were reviewed for quality control and revised accordingly. In the event of true identical birthweights, the infants’ head circumference followed by length measurements at gestational age were used to break the “larger” vs “smaller” twin ties. Similarly, for identical EFW measurements on the last ultrasound, the fetuses’ head circumferences followed by abdominal circumferences and then femur lengths were used to break ties.

The primary outcome was a composite of neonatal morbidities stratified as mild or severe. Mild neonatal morbidity included oxygen supplementation or a continuous need for positive airway pressure for <72 hours, hypoglycemia, hypocalcemia, hyperbilirubinemia, or intraventricular hemorrhage (IVH) grade I or II. Severe neonatal morbidity included necrotizing enterocolitis (NEC) grade 2A+, IVH grade III or IV, BPD, mechanical ventilation, or neonatal death. These neonatal outcomes were also analyzed individually.

Secondary outcomes included NICU admission, NICU length of stay (LOS) of >5 days, and a 5-minute Apgar score of <7. The maternal covariates included race, age, diabetes (gestational or pregestational vs none), mode of delivery (vaginal delivery vs cesarean delivery), and body mass index (kg/m ² ) at delivery (ordinal category of <25 kg/m ² , 25–29.9 kg/m ² , 30–34.9 kg/m ² , 35–39.9 kg/m ² , ≥40 kg/m ² ).

The maternal demographic and obstetrical characteristics were summarized descriptively. A bivariate and multivariable generalized linear mixed model (GLMM)-derived logistic regression was used to assess the adverse neonatal outcomes according to the FGR-defined groups. For each outcome, the lowest risk category (eg, no mild composite neonatal morbidity) was used as the referent category. The analyses were conducted in PROC GLIMMIX in SAS version 9.4 (SAS Institute Inc, Cary, NC) with the odds of neonatal morbidity estimated using a logistic link function to account for the paired data and to estimate subject-specific risks. A random-effects component was incorporated to account for the correlations within twin pairs and multiple maternal pregnancies. Multivariable analyses, with maternal covariate adjustments as described above, produced subject-specific adjusted odds ratio (aOR) and 95% confidence interval (CI) estimates that were used to assess the morbidity outcomes between FGR groups via planned, pair-wise comparisons. Because of the large number of tests conducted, the Holm-Bonferroni method was used to correct for family-wise error. Significance was defined by a 2-tail P value of <.05. We chose the maternal covariates carefully in an effort to prevent over adjustments by controlling for specific pregnancy complications (eg, preeclampsia, gestational age, etc.) that have a tendency to be on the causal pathway between FGR and adverse neonatal morbidity outcomes because these causal intermediates may have led to biased results.

Results

There were 1460 dichorionic, diamniotic twin infants (730 twin pairs) that met the study criteria. Of these, 8.1% (n=118) of the twins were deemed growth restricted by both the NICHD and Hadlock growth references. Another 8.8% (n=129) were identified as growth restricted by the Hadlock growth reference only. The remaining twin fetuses (83.1%; n=1213) were not considered growth restricted by either growth reference, because there were no fetuses identified as growth restricted by the NICHD’s reference only. These data are presented in Figure . The average time between the last ultrasound and delivery was 14±9.6 (median 13) days. The percentage difference (measured as ([weight _{larger twin} –weight _{smaller twin} /weight _{larger twin} ])×100) for birthweights was 12% (±10%) and 10% (±10%) for the EFW measurements, with birthweight vs EFW differences strongly positively correlated (Pearson rho=0.6; P <.0001). The FGR group assignment was concordant for 77.5% of the twin pairs with discordance noted in 22.5% (2.9% with 1 twin classified as FGR by both standards and 1 classified as FGR by the Hadlock reference only; 8.4% with 1 twin classified as FGR by both references and 1 classified as not FGR by either standard; and 11.2% with 1 twin classified as FGR by the Hadlock reference only and 1 classified as not FGR by either references). The demographic and obstetrical characteristics of the 730 maternal participants included in this cohort are presented in Table 1 . Our patients were cared for at an academic teaching center and were racially and ethnically diverse, relatively evenly divided between nulliparous and parous women, and were predominantly obese and privately insured ( Table 1 ).

Classification of 1460 infants by Hadlock and NICHD growth references

The solid and dotted lines indicate the cutoff for EFW <10th percentile for the Hadlock and NICHD growth references, respectively.

EFW , estimated fetal weight, NICHD , the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Shea et al. Twin-specific growth references for dichorionic diamniotic twins. Am J Obstet Gynecol 2021.

The primary outcomes of interest are summarized in Table 2 in addition to the bivariate and multivariable GLMM results. The pair-wise comparisons of the primary outcomes between the FGR groups are presented in Table 3 . Compared with twins with no FGR as determined by either reference, those with FGR according to both references were more likely to experience both mild (aOR, 2.38; CI, 1.38–4.13; P =.001) and severe (aOR, 2.82; CI, 1.16–6.88; P =.016) composite neonatal morbidity. Compared with twins with FGR according to the Hadlock reference only, those with FGR according to both references were more likely to experience mild (aOR, 2.03; CI, 1.00–4.14; P =.034), but not severe (aOR, 3.70; CI, 0.72–18.90; P =.11) composite neonatal morbidity. There were no significant differences in the proportions of mild or severe composite neonatal morbidity between the twins with FGR according to the Hadlock reference only and twins with no FGR according to either growth reference (aOR, 1.17; CI, 0.70–1.97; P =.462 and aOR, 0.76; CI, 0.17–3.37; P =.663, respectively). When assessed as individual outcomes, the twins with FGR according to both growth references were more likely to experience hypoglycemia compared with those with FGR according to the Hadlock reference only (aOR, 2.72; CI, 1.26–5.86; P =.004) and twins with no FGR according to either reference (aOR, 2.97; CI, 1.69–5.21; P <.001). In addition, twins with FGR according to both growth references were more likely to experience hyperbilirubinemia compared with those with FGR according to the Hadlock reference only (aOR, 3.44; CI, 1.10–10.77; P =.19) and twins with no FGR according to either reference (aOR, 3.06; CI, 1.47–6.38; P =.001). No significant differences were found for these individually-assessed outcomes between twins with FGR by the Hadlock reference only and twins with no FGR by either growth reference. No significant differences were found for IVH grade I or II, BPD, and mechanical ventilation between the 3 groups. Hypocalcemia, IVH grade III or IV, NEC, and neonatal death were unable to be assessed independently owing to lack of sample.

Table 2

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Long-term effectiveness and safety of open Burch colposuspension vs retropubic midurethral sling for stress urinary incontinence—results from a large comparative study
2. Services and payer mix of Black-serving hospitals and related severe maternal morbidity
3. Predicting cesarean delivery for failure to progress as an outcome of labor induction in term singleton pregnancy
4. Intracervical balloon catheter for labor induction after rupture of membranes: a systematic review and meta-analysis

Stay updated, free articles. Join our Telegram channel

Join