Publication Description
The use of birth weight standards to define small for gestational age (SGA) may fail to identify neonates affected by poor fetal growth since they include births associated with sub-optimal fetal growth. Our objective was to compare intrauterine versus birth weight-derived standards to define newborn SGA to predict neonatal morbidity and mortality.
This was a secondary analysis of a multi-center observational study of 118,422 births. Liveborn singleton, non-anomalous newborns born at 23-41 weeks were included. Those with missing gestational age estimation or without a 1
or 2
trimester ultrasound to confirm dating, birth weight, or neonatal outcome data were excluded. Birth weight percentile was computed using an intrauterine standard (Hadlock, Radiology 1991) and a birth weight-derived standard (Olsen, Pediatrics 2010). We compared the test characteristics of SGA (birth weight <10
percentile) by each standard to predict a composite neonatal morbidity and mortality outcome (death prior to discharge, NICU admission >48 hours, respiratory distress syndrome, sepsis, necrotizing enterocolitis, grade III or IV intraventricular hemorrhage, or seizures). Severe composite morbidity was analyzed as a secondary outcome and was defined as death, NICU admission >7 days, NEC, grade 3-4 IVH, or seizures. The areas under the curve (AUC) using receiver-operating characteristic methodology and proportions of the primary outcome by SGA status were compared by gestational age category at birth (<34, 34
-36
, ≥37 weeks).
Of 115,502 mother-newborn dyads in the parent study, 78,203 (67.7%) were included, with the majority of exclusions occurring because of missing or inadequate dating information, multiple gestations, or delivery outside the gestational age range. The primary composite outcome occurred in 9.5% (95% CI 9.3-9.7) and the severe composite outcome occurred in 5.3% (5.1-5.4). SGA was diagnosed by intrauterine and birth weight-derived standards in 14.8% and 7.4%, respectively (p<0.001). Neonates considered SGA only by the intrauterine standard experienced the primary outcome more than twice as often as those considered non-SGA by both standards (18.4% vs 7.9%, p<0.001). For prediction of the primary outcome, SGA by the intrauterine standard had higher sensitivity (29% vs 15%, p<0.001) but lower specificity (87% vs 93%, p<0.001) than by the birth weight standard. Both standards had weak performance overall, though the intrauterine standard had a higher AUC (0.58 vs 0.53, p<0.001). When sub-analyzed by gestational age at birth, the difference in AUCs was only present among preterm deliveries 34 to 36 competed weeks. Neither standard demonstrated any discrimination for morbidity prediction among term births (AUC = 0.50 for both). When prediction of severe morbidity was compared, the intrauterine still had better overall prediction than the birth weight standard (AUC 0.65 vs 0.57, p<0.001), though this also varied by gestational age at birth.
Among non-anomalous neonates, neither intrauterine nor birth weight-derived standards for SGA accurately predicted neonatal morbidity and mortality, with no discriminatory ability at term. SGA intrauterine standards performed better than birth weight standards.