Weaver Lucinda J, Gentle Samuel J, Nakhmani Arie, Rahman Fazlur, Ambalavanan Namasivayam, Shukla Vivek V, Stoops Christine, Askenazi David, Travers Colm P
Department of Pediatrics, University of Alabama at Birmingham, 1700 6Th Avenue South, WIC, Suite 9380, Birmingham, AL, 35233, USA.
Department of Electrical and Computer Engineering, University of Alabama at Birmingham, Birmingham, AL, USA.
Pediatr Nephrol. 2025 Sep 24. doi: 10.1007/s00467-025-06962-4.
Measures of early postnatal fluid balance may be associated with severe intraventricular hemorrhage (sIVH) and/or death in extremely preterm infants in the first postnatal week.
A single-center, retrospective cohort study including actively treated inborn infants weighing ≥ 400 g and 22-27 weeks' gestation from 2014-2021. Longitudinal mixed effect models compared daily fluid balance covariates including serum sodium, percent weight change, total fluid intake, urine output, and fluid balance (daily weight - birth weight /birth weight × 100) among infants with and without sIVH or death, during the first seven postnatal days. Multiple regression and machine learning models were developed to predict sIVH and/or death. Variables that were incorporated into the models included measures of fluid balance, gestational age, birth weight, antenatal corticosteroids, multiples, and sex.
We included 932 infants with mean ± SD gestational age of 25w2d ± 11d and birth weight of 746 ± 212 g of whom 195 (20.9%) had sIVH and/or death. Lower percentage weight change (p < 0.001), higher total fluid intake (p = 0.007), higher sodium (p = 0.007), and positive early fluid balance (p < 0.001) were associated with sIVH and/or death even after adjustment for baseline characteristics. The area under the receiver-operating curve (AUC) for regression models predicting sIVH and/or death incorporating baseline characteristics improved after adding fluid balance measures from 0.75 to 0.80, while the AUC for machine learning models improved from 0.72 to 0.84.
In extremely preterm infants, early fluid status measures were associated with risk of sIVH and/or death. The addition of fluid status measures improves the performance of models predicting sIVH and/or death.
出生后早期液体平衡的指标可能与极早产儿出生后第一周内的严重脑室内出血（sIVH）和/或死亡有关。
一项单中心回顾性队列研究，纳入了2014年至2021年期间出生体重≥400g、孕周为22 - 27周且接受积极治疗的活产婴儿。纵向混合效应模型比较了出生后前七天内发生和未发生sIVH或死亡的婴儿的每日液体平衡协变量，包括血清钠、体重变化百分比、总液体摄入量、尿量和液体平衡（每日体重 - 出生体重/出生体重×100）。开发了多元回归和机器学习模型来预测sIVH和/或死亡。纳入模型的变量包括液体平衡指标、孕周、出生体重、产前使用糖皮质激素、多胎情况和性别。
我们纳入了932例婴儿，其平均±标准差孕周为25周2天±11天，出生体重为746±212g，其中195例（20.9%）发生了sIVH和/或死亡。即使在调整基线特征后，较低的体重变化百分比（p<0.001）、较高的总液体摄入量（p = 0.007）、较高的钠水平（p = 0.007）和早期正液体平衡（p<0.001）仍与sIVH和/或死亡相关。在纳入基线特征的情况下，预测sIVH和/或死亡的回归模型的受试者工作特征曲线下面积（AUC）在加入液体平衡指标后从0.75提高到0.80，而机器学习模型的AUC从0.72提高到0.84。
在极早产儿中，早期液体状态指标与sIVH和/或死亡风险相关。加入液体状态指标可提高预测sIVH和/或死亡模型的性能。
