Neonatal Health and Long-Term Outcomes: Twin-Based Evidence From Linked Scandinavian Registers

This study examines the long-term consequences of neonatal health using three birth anthropometric measures — weight, length, and head circumference — and analyzes their differential associations with outcomes across the life course. Using linked Swedish and Norwegian population registers with follow-up into adulthood, we compare birth measures of singletons and twins and further estimate within-twin-pair fixed-effects models that control for shared genetic and family environmental factors. Our samples comprise up to approximately 679,000 singletons and 10,400 twins in Sweden, and 347,841 singletons and 6,762 twins in Norway, followed into adulthood. We find that different birth measures capture distinct aspects of fetal development with varying predictive validity across outcome domains. Birth length emerges as the strongest predictor of health outcomes (infant mortality, adult height), consistent with its role as a cumulative indicator of skeletal growth throughout gestation. Birth weight retains stronger associations with cognitive and educational outcomes (grade point average [GPA], years of schooling) in conditional models. Head circumference shows weak and inconsistent associations across specifications. Using within-pair head circumference differences to classify twins as symmetrically versus asymmetrically growth restricted, we find that symmetric intrauterine growth restriction is associated with substantially worse long-term outcomes than asymmetric restriction. In Norwegian twin pairs, growth restriction is associated with lower general cognitive ability (measured at military enlistment) relative to the unaffected co-twin; this within-pair deficit is more than twice as large for symmetric as for asymmetric restriction. We also document significant intergenerational transmission of birth anthropometrics, suggesting both environmental and genetic contributions. Finally, our register-based design with near-complete population coverage minimizes selection bias and attrition while providing statistical power for within-pair analyses across multiple outcomes. Exclusive reliance on birth weight may obscure important information about early-gestation developmental insults.