天津医药

天津医药 ›› 2025, Vol. 53 ›› Issue (10): 1037-1042.doi: 10.11958/20250625

• 临床研究 • 上一篇    下一篇

ICU患者早期肠内营养治疗发生误吸的列线图风险预测模型建立与验证

汪海霞(), 何飞△(), 朱丛梅, 王静   

  1. 合肥市第二人民医院重症医学科（邮编 230011）
  • 收稿日期:2025-02-20 修回日期:2025-07-21 出版日期:2025-10-15 发布日期:2025-10-12
  • 通讯作者: △E-mail：1977335804@qq.com
  • 作者简介:汪海霞（1981），女，主管护师，主要从事危重症患者营养管理方面研究。E-mail：17775357975@163.com
  • 基金资助:
    安徽省中医药传承创新科研项目(2024CCCX073)

Establishment and validation of a column chart risk prediction model for aspiration in early enteral nutrition therapy of ICU patients

WANG Haixia(), HE Fei△(), ZHU Congmei, WANG Jing   

  1. Intensive Care Unit, the Second People's Hospital of Hefei, Hefei 230011, China
  • Received:2025-02-20 Revised:2025-07-21 Published:2025-10-15 Online:2025-10-12
  • Contact: △E-mail：1977335804@qq.com

RichHTML

3

PDF (PC)

12

摘要：

目的 探讨重症监护病房（ICU）患者早期肠内营养支持治疗时发生误吸的风险因素，建立相应的列线图风险预测模型并进行验证。方法 收集2022年6月—2024年5月在我院ICU实施早期肠内营养支持的348例患者资料，根据患者是否发生误吸分为误吸组（74例）与非误吸组（274例）。收集患者年龄、性别、体质量指数（BMI）、糖尿病史、气管插管或机械通气状态、入ICU后24 h内血浆白蛋白（ALB）水平、疾病类型（重症肺炎/脑卒中/感染性休克）、意识水平[格拉斯哥昏迷量表（GCS）评分]、急性生理与慢性健康状况（APACHEⅡ）评分、鼻饲管插入深度、输注量、营养风险（NRS2002评分≥3分为高风险）及营养方式（经鼻胃管/鼻肠管）等指标。运用Logistic回归分析误吸的危险因素，利用R软件构建列线图风险预测模型。选择2024年6月—2025年1月在ICU接受早期肠内营养支持的72例患者进行该模型的外部验证。结果 Logistic回归分析筛选出年龄（OR=2.701，95%CI：1.633~4.467）、APACHEⅡ评分（OR=2.125，95%CI：1.133~3.987）、意识水平（OR=2.826，95%CI：1.617~4.940）、鼻饲管插入深度（OR=1.101，95%CI：1.006~1.136）以及营养风险（OR=8.996，95%CI：5.017~16.132）为误吸的独立影响因素（均P<0.05）。基于多因素Logistic回归分析，本研究构建了以上述5项因素为指标的列线图预测模型。该模型通过累加各指标的分值，直观转换为患者发生误吸的风险概率，并在内部验证（AUC=0.860，校准曲线斜率=0.930）及外部验证（AUC=0.831）中均表现出良好的预测性能。此外，决策曲线分析显示，该模型在不同风险阈值下均具有显著的临床净获益，进一步支持其在实际应用中的有效性。结论 本研究构建的列线图模型在区分度和准确度方面表现较好，可为临床针对ICU患者早期肠内营养治疗时发生误吸的风险进行个体化预测提供参考依据。

关键词: 肠道营养, 误吸, 列线图, 重症监护病房, 预测模型

Abstract:

Objective To investigate the risk factors of aspiration during early enteral nutrition (EEN) support treatment in patients in intensive care unit (ICU) and establish and validate the corresponding nomogram risk prediction model. Methods A total of 348 ICU patients who received EEN between June 2022 and May 2024 were enrolled and divided into the aspiration group (n=74) and the non-aspiration group (n=274) based on the occurrence of aspiration. Clinical data were collected included age, sex, body mass index (BMI), history of diabetes, endotracheal intubation/mechanical ventilation status, plasma albumin (ALB) levels within 24 h after admission to ICU, disease type (severe pneumonia/stroke/septic shock), consciousness level (Glasgow Coma Scale, GCS), APACHE Ⅱscore, nasogastric tube insertion depth, infusion volume, nutritional risk (NRS2002 score ≥3 indicating high risk), and nutrition mode (nasogastric/nasointestinal tube). Logistic regression was used to identify risk factors of aspiration, and a nomogram prediction model was constructed using R software. External validation was performed on 72 EEN-treated ICU patients admitted between June 2024 and January 2025. Results Logistic regression identified age (OR=2.701, 95% CI: 1.633-4.467), APACHE Ⅱ score (OR=2.125, 95%CI: 1.133-3.987), consciousness level (OR=2.826, 95%CI: 1.617-4.940), nasogastric tube insertion depth (OR=1.101, 95%CI: 1.006-1.136) and nutritional risk (OR=8.996, 95%CI: 5.017-16.132) were independent risk factors for aspiration (all P<0.05). A nomogram incorporating these factors was developed, converting cumulative scores into individualized aspiration risk probabilities. The model demonstrated strong predictive performance in internal validation (AUC=0.860, calibration curve slope=0.930) and external validation (AUC=0.831). Decision curve analysis (DCA) confirmed significant clinical net benefits across risk thresholds, supporting its practical utility. Conclusion The nomogram model exhibits good discrimination and accuracy, providing a valuable tool for individualized aspiration risk assessment in ICU patients receiving EEN.

Key words: enteral nutrition, respiratory, aspiration, nomogram, intensive care unit, predictive model

中图分类号: