Bibi Shawana, Singh Rachana, Breeze Janis L, Nelson Jason, Kraft Walter K, Davis Jonathan M
Tufts Clinical and Translational Science Institute, Boston, MA, United States.
Cleveland Clinic Children's Hospital, Case Western Reserve University Lerner College of Medicine, Cleveland, OH, United States.
Front Pediatr. 2025 Aug 11;13:1527276. doi: 10.3389/fped.2025.1527276. eCollection 2025.
Development and validation of a clinical prediction model for receipt of pharmacotherapy for Neonatal Abstinence Syndrome (NAS).
Data from three cohorts included opioid exposed neonates ≥37 weeks gestation. Primary outcome was the receipt of pharmacotherapy utilizing a modified Finnegan Neonatal Abstinence Scoring System (FNASS). A stepwise multivariable logistic regression model was built and internally validated.
Of 698 infants included, 430 received pharmacotherapy. The final model included seven predictors of receipt of pharmacotherapy: gestational age, exposure to maternal breast milk, type of maternal opioid medication, and exposure to heroin, cocaine, benzodiazepines, and/or antipsychotic medications. The model had an AUROC of 0.68 (95% CI: 0.64-0.72; optimism corrected 0.65).
Our prediction model was parsimonious and identified seven predictors associated with the need for PT. Larger cohort studies are needed to more definitively establish risk of significant NAS requiring pharmacotherapy.
开发并验证一种用于预测新生儿戒断综合征（NAS）药物治疗接受情况的临床预测模型。
来自三个队列的数据纳入了孕周≥37周的阿片类药物暴露新生儿。主要结局是使用改良的芬尼根新生儿戒断评分系统（FNASS）进行药物治疗。构建了逐步多变量逻辑回归模型并进行内部验证。
在纳入的698例婴儿中，430例接受了药物治疗。最终模型纳入了七个药物治疗接受情况的预测因素：胎龄、母亲母乳接触情况、母亲阿片类药物类型以及海洛因、可卡因、苯二氮卓类药物和/或抗精神病药物的接触情况。该模型的曲线下面积（AUROC）为0.68（95%可信区间：0.64 - 0.72；校正乐观估计后为0.65）。
我们的预测模型简洁明了，识别出了七个与药物治疗需求相关的预测因素。需要开展更大规模的队列研究，以更明确地确定需要药物治疗的严重NAS的风险。
