欢迎访问《中国循证儿科杂志》官方网站，今天是
English Version
中国循证儿科杂志

中国循证儿科杂志 ›› 2024, Vol. 19 ›› Issue (6): 451-455.DOI: 10.3969/j.issn.1673-5501.2024.06.008

• 论著 • 上一篇    下一篇

NICU新生儿医源性皮肤损伤风险预测模型的构建与验证

吴宏丽1，周旋2，丁静1，李园芳1，刘琦2，张晓婷2，廖卫华1   

  1. 南方医科大学南方医院广州，510515；1新生儿科，2 儿科
  • 收稿日期:2023-12-07 修回日期:2024-06-12 出版日期:2024-12-25 发布日期:2024-12-25
  • 通讯作者: 廖卫华

Construction and validation of a risk prediction model for iatrogenic skin injury in NICU neonates

WU Hongli1, ZHOU Xuan2, DING Jing1, LI Yuanfang1, LIU Qi2, ZHANG Xiaoting2, LIAO Weihua1   

  1. Southern Medical University, Nanfang Hospital, Guangzhou 510515, China; 1 Department of Neonatology, 2 Department of Pediatrics
  • Received:2023-12-07 Revised:2024-06-12 Online:2024-12-25 Published:2024-12-25
  • Contact: LIAO Weihua

PDF

118

摘要： 背景新生儿重症监护室（NICU）中患儿容易发生医源性皮肤损伤，目前缺乏相关的风险预测模型。 目的探讨影响NICU新生儿医源性皮肤损伤的危险因素，以此构建风险列线图并验证。 设计病例对照研究。 方法纳入南方医科大学南方医院2020年1月至2023年12月入住NICU≥24 h、入院日龄≤28 d、在本院出生的新生儿；排除入NICU前已存在皮肤破损或皮肤感染，患先天性皮肤病或易导致皮肤损伤疾病（畸胎瘤），外科手术后，转院、非医嘱出院的新生儿。根据新生儿病历中记录的是否存在医源性皮肤损伤分为医源性皮肤损伤组和非损伤组。基于文献复习和2轮德尔菲专家咨询法最终确定纳入本文分析的影响因素。在病历系统中回顾性采集新生儿基本信息、临床信息及围生期孕母信息。运用单因素和多因素分析发生医源性皮肤损伤的影响因素，构建预测模型并行外部验证。 主要结局指标模型对NICU新生儿医源性皮肤损伤的预测性能。 结果631例进入本文分析，损伤组170例（26.9%），非损伤组461例。建模组442例，验证组189例。建模组和验证组一般情况比较，差异均无统计学意义。单因素及多因素分析显示，出生胎龄、出生体重、非纯母乳喂养、使用纸质医用黏胶布、医用黏胶使用不规范、去除医用黏胶清洁剂类型为75%乙醇溶液、使用无创机械通气、使用约束带、药物外渗均是影响NICU新生儿医源性皮肤损伤的独立影响因素。建模组和验证组ROC的AUC分别为0.929（95%CI：0.900~0.958）和0.941（95%CI：0.908~0.974）。建模组和验证组校准曲线H-L拟合优度检验显示，χ2分别为14.649和8.670，P分别为0.066和0.371。临床决策曲线结果显示该模型对高风险患儿进行及时干预可以使大部分患儿受益，具有较高的临床适用性。 结论NICU新生儿医源性皮肤损伤的发生率较高且受多种因素影响，构建的预测模型具有较高的临床适用性。

关键词: NICU, 医源性皮肤损伤, 危险因素, 列线图, 早期干预

Abstract: BackgroundNeonates admitted to the neonatal intensive care unit (NICU) are at high risk of iatrogenic skin injury. However, no established risk prediction models are currently available. ObjectiveTo identify risk factors associated with iatrogenic skin injury in NICU neonates, construct a nomogram-based risk prediction model, and validate its performance. DesignCase-control study. MethodsNeonates born at Nanfang Hospital, Southern Medical University, and admitted to the NICU between January 2020 and December 2023 for ≥24 hours within 28 days of birth were included. Exclusion criteria were pre-existing skin damage or infection before NICU admission, congenital skin diseases or conditions predisposing to skin injury (e.g., teratomas), post-surgical status, transfer to another hospital, or discharge against medical advice. Based on the presence or absence of iatrogenic skin injury recorded in medical records, neonates were categorized into the injury group and non-injury group. Risk factors were determined through literature review and two rounds of Delphi expert consultation. Data on neonatal demographics, clinical characteristics, and perinatal maternal factors were retrospectively collected from the medical record system. Univariate and multivariate analyses were conducted to identify independent risk factors, which were used to develop a predictive model and externally validate it. Main outcome measuresPredictive performance of the risk model for iatrogenic skin injury in NICU neonates. ResultsA total of 631 neonates were included in the analysis, with 170 cases (26.9%) in the injury group and 461 cases in the non-injury group. The modeling cohort included 442 neonates, while the validation cohort comprised 189 neonates. There was no significant difference in baseline characteristics between the modeling and validation groups. Univariate and multivariate analyses identified the following independent risk factors for iatrogenic skin injury: gestational age at birth, birth weight, non-exclusive breastfeeding, use of paper-based medical adhesive tape, improper use of medical adhesives, removal of adhesives with 75% ethanol solution, use of non-invasive mechanical ventilation, use of restraint bands, and drug extravasation. The area under the ROC curve (AUC) for the prediction model was 0.929 (95%CI: 0.900-0.958) in the modeling cohort and 0.941 (95%CI: 0.908-0.974) in the validation cohort. The Hosmer-Lemeshow goodness-of-fit test yielded χ2 values of 14.649 (P=0.066) and 8.670 (P=0.371) for the modeling and validation groups, respectively, indicating good calibration. The clinical decision curve analysis demonstrated that early intervention for high-risk neonates based on the model could benefit most patients, suggesting high clinical applicability. ConclusionThe incidence of iatrogenic skin injury in NICU neonates is relatively high and is influenced by multiple factors. The developed prediction model demonstrates high clinical applicability and may help guide early preventive interventions.

Key words: NICU, Iatrogenic skin injuries, Risk factor, Nomogram, Early Intervention