Weight changes and variability in the first week of life: associations with mortality in extremely preterm newborns

In extremely preterm newborns (ELGANs, < 28 weeks' gestational age [GA]), early weight trajectories reflect fluid balance, yet thresholds for early risk stratification are lacking. We aimed to evaluate whether percentage weight change (ΔW%) from birth weight (BW) and week-1 weight variability were independently associated with death by postnatal day 7 and to identify clinically relevant thresholds.

We performed a retrospective single-center secondary analysis of the Golden Week Program (University of Alabama at Birmingham NICU, 2014-2021). Newborns < 28 weeks' GA with BW ≥ 400 g and receiving active treatment were included; exclusion criteria were death ≤ 48 h, major congenital anomalies, or incomplete weight data. Multivariable logistic regression was performed, adjusted for GA, 5-min Apgar score, and BW z-score; optimal cutoffs were derived using the Youden Index.

Among 805 neonates, 55 (6.8%) died by day 7. Non-survivors had higher day-3 ΔW% than survivors (median - 5.3% vs. - 10.7%; p < 0.007). Day-3 ΔW% was independently associated with mortality (aOR 1.05; 95% CI 1.02-1.08), with an optimal cutoff of - 6.4% (sensitivity 57%, specificity 72%). Among variability metrics, total fluctuation (mean absolute day-to-day change in ΔW%) showed the best discrimination (aOR 1.45; 95% CI 1.29-1.65; AUC 0.84), with an optimal cutoff of 5.3% (sensitivity 76%, specificity 74%).

ELGANs who do not lose 5-7% of BW by postnatal day 3 or exhibit an average day-to-day weight fluctuation > 5% of BW are at increased risk of early mortality. These weight-derived metrics may support early risk stratification but require prospective validation.