Ultrasound prediction of perinatal outcome: the unrecognised value of sibling data

Objective To identify high-risk fetuses at the first routinely performed ultrasound examination by making use of information from the mother's previous pregnancy. Design A population-based cohort study. Setting Norway, 1999–2009. Population All singleton first live births and their second-born siblings registered in the Medical Birth Registry of Norway (166 786 eligible sibling pairs). Methods Odds ratios were calculated by logistic regression. Main outcome measures Very small for gestational age (vSGA; birthweight ≤−1.96 standard deviations) and perinatal death (stillbirth at ≥22 weeks of gestation or death within 28 days of life). Results Small fetal size at ultrasound (i.e. a fetus smaller than expected by last menstrual period, LMP) is only weakly predictive of vSGA or perinatal death; however, if the firstborn sibling was vSGA at birth, ultrasound measures in the next pregnancy become strongly informative of risk. The smaller the fetal size on ultrasound, the higher its risk of vSGA (3–18%; Ptrend < 0.0001) and perinatal death (4–19 per thousand, Ptrend = 0.012). In contrast, if the first baby was not vSGA, small fetal size on ultrasound is uninformative. Conclusions When the firstborn baby is vSGA, discrepancies between fetal size on ultrasound and LMP become highly predictive of risk of vSGA and perinatal mortality in the second-born infant. The value of combining these routinely collected clinical data has not previously been recognised.