Diagnostic accuracy of small-for-gestational-age status for infant mortality and school-age outcomes of live births <28 weeks' gestation: a cohort study

ObjectivesTo determine the diagnostic accuracy of small-for-gestational-age (SGA; <10th centile) status for infant mortality and adverse school-age outcomes in infants born extremely preterm (EP; <28 weeks' gestation). DesignGeographical cohort studies. SettingThe state of Victoria, Australia. PatientsFor mortality, live births 22-27 weeks' gestation from 2009 to 2017 offered active care after birth. For school-age outcomes, survivors to 8 years' corrected age born in 1991-1992, 1997 or 2005. ExposuresSGA <10th centile on four commonly used growth references: three derived from neonatal data (Fenton, UK-WHO and Intergrowth Newborn Size) and one from fetal data (Intergrowth Estimated Fetal Weight). Main outcome measures(a) Infant mortality; (b) major neurodevelopmental disability, and poor performance on tests of IQ, academic achievement, motor function, and executive function. ResultsInfant mortality data were available for 2040 infants, and neurodevelopmental data for 499 children. Diagnostic accuracy of SGA status was low overall and varied with the growth reference. Positive predictive values for infant mortality ranged from 18% to 21%, only marginally higher than its 18% prevalence. Compared with a prevalence of 17%, positive predictive values for major neurodevelopmental disability ranged from 30% to 38% for the neonatal growth references but was only 20% for Intergrowth Estimated Fetal Weight. SGA status was also associated with lower IQ, poor academic achievement and poor motor performance. ConclusionsAmong infants born EP, the diagnostic accuracy of SGA status was low for both infant mortality and adverse neurodevelopmental outcomes at school age, but importantly varied with the growth reference used to identify SGA status.