ICU经口气管插管患者 3个月非计划再住院风险模型的构建 [中文引用][英文引用]

目的 引入灰度共生矩阵，构建重症监护（intensive care unit，ICU）经口气管插管患者 3 个月非计划再住院的风险模型，为减少非计划再住院提供依据。方法 采用便利抽样法，选取 2023 年 11 月— 2024 年 5 月河南省郑州市某三级甲等医院呼吸重症监护病房 260 例经口气管插管患者作为研究对象。根据 3 个月内是否非计划再住院将其分为再住院组和未再住院组。收集患者的一般临床资料、实验室指标和拔管后第 1 天、第 7 天股直肌超声指标（灰度共生矩阵），采用多因素 Logistic回归分析确定经口气管插管患者 3 个月非计划再住院的危险因素，采用 Logistic 回归和 Framingham 危险评分函数构建风险模型。采用受试者操作特征曲线（receiver operating characteristic curve，ROC）下面积（area under ROC curve，AUC）和 HosmerLemeshow 评价模型的准确性。结果 最终纳入 224 例。经口气管插管 ICU 患者 3 个月非计划再住院的发生率为 35.71%（80/224）。年龄≥ 60 岁、营养风险筛查≥ 3 分、休克指数≥ 1.0、机械通气时长≥ 251h、拔管后第 7 天股直肌横截面积≤ 1.41cm2、拔管后第 7 天角二阶距≤ 0.71 和拔管后第 7 天股直肌占股四头股的比率变化率为 0 是经口气管插管患者 3 个月再住院的危险因素（均 P<0.05）。ROC 曲线分析显示，气管插管患者 3 个月非计划再住院风险评分模型的 AUC 为 0.791（95%CI：0.707~0.875，P<0.001），敏感度为 75.02%，特异度为 67.33%。Hosmer-Lemeshow 结果显示 χ 2=2.581（P=0.630）。最佳临界值为 3 分。结论该风险评分模型具有良好的预测效能，可为临床医护人员评估经口气管插管患者的 3 个月非计划再住院的风险提供参考。

Objective To identify predictive indicators for unscheduled rehospitalisation within 3 months in patients of oral endotracheal intubation（OEI）with ultrasound gray level co-occurrence matrix（GLCM）and develop a corresponding risk assessment system for the purpose to reduce the event of unscheduled rehospitalisation. Methods A total of 260 OEI patients who underwent extubation in a Tier-IIIA hospital between October 2023 and May 2024 were enrolled by convenience sampling. Patients were divided into a rehospitalisation group and a non-rehospitalisation group according to the event of unscheduled rehospitalisation within 3 months after discharge. Demographic data and laboratory test report, ultrasound morphological indicators and GLCM of rectus femoris muscle were collected on day-1 and day-7 after extubation. Multivariate logistic regression analysis and Framingham risk function were used to identify independent risk factors for unscheduled rehospitalisation within 3 months. A risk assessment system for unscheduled rehospitalisation within 3 months was subsequently developed. Predictive accuracy were evaluated using receiver operating characteristic （ROC）curve and area under the curve（AUC）, and Hosmer-Lemeshow test. Results Toally 224 patients were included. The incidence of unplanned rehospitalization within 3 months in patients with oral tracheal intubation was 35.71%（80/224）. The independent risk factors for unscheduled rehospitalisation within 3 months in OEI patients were identified as age ≥ 60, nutrition risk screening2002 ≥ 3, shock index≥1.0, duration of mechanical ventilation≥251 hours, rectus femoris cross-sectional area≤1.41cm2 , angular second moment≤0.71, the proportion change rate of ratio of rectus femoris on quadriceps femoeis for 0 on day-7 after extubation. The risk assessment system exhibited an AUC of 0.791（95%CI：0.707~0.875, P<0.001）, with a sensitivity of 75.02% and a specificity of 67.33%. The HosmerLemeshow value was 2.581（P=0.630）, and the optimal cut-off value was determined at 3. Conclusion The developed risk assessment system demonstrates a satisfactory predictive performance. It provides a valuable reference for clinical assessment of the patients who had unscheduled rehospitalisation within 3 months in OEI patients.