| 冯轶,李璇,闵汇刚.缺血性脑卒中病人气管插管临床预测模型的建立与验证[J].安徽医药,2025,29(4):710-716. |
| 缺血性脑卒中病人气管插管临床预测模型的建立与验证 |
| Establishment and validation of a clinical predictive model for tracheal intubation in patients with ischemic stroke |
| |
| DOI：10.3969/j.issn.1009-6469.2025.04.015 |
| 中文关键词: 卒中 脑梗死 列线图 风险因素 气管插管 预测模型 |
| 英文关键词: Stroke Cerebral infarction Nomogram chart Risk factors Tracheal intubation Prediction model |
| 基金项目:湖北省卫生健康委员会联合基金项目( WJ2019H167) |
|
| 摘要点击次数: 236 |
| 全文下载次数: 152 |
| 中文摘要: |
| 目的分析缺血性脑卒中病人需要气管插管干预的风险因素，并构建和验证列线图模型。方法回顾性分析 2001—2019年重症监护医学信息数据库( MIMIC-Ⅲ/Ⅳ)中 3 420例缺血性脑卒中病人的临床资料。 MIMIC-Ⅲ数据库中 1 264例病人，其中 70%的病人作为训练组，剩余 30%的病人和 MIMIC-Ⅳ数据库 2 156例病人分别作为内部和外部验证组。应用多因素 logistic回归模型构建气管插管风险因素列线图预测模型。使用受试者操作特征曲线(ROC曲线)及曲线下面积(AUC)，评估列线图的预测性能。采用 1 000次重复抽样的 bootstrap方法绘制校准曲线，反映实际概率与预测概率的一致性。借助决策曲线分析(DCA)评估该模型的临床实用性。结果训练组与验证组病人血管升压素使用率( 3.38%比 6.65%，P=0.009)，糖尿病患病率(31.98%比 23.94%，P=0.004)差异有统计学意义；尿素氮［( 8.12±5.90)mmol/L比( 8.50±6.51)mmol/L，P=0.314］、Elixhauser合并症指数( 6.40±7.10比 6.54±6.89，P=0.739)和年龄［训练组:(68.55±16.09)年比( 69.73±15.82)年， P=0.234］差异无统计学意义。多因素二元 logistic分析结果显示:年龄( OR=1.020，P=0.020)、尿素氮( OR=1.020，P=0.010)和 Elixhauser合并症指数( OR=1.150，P<0.001)是缺血性脑卒中病人气管插管的独立影响因素。 ROC曲线显示训练组 AUC=0.79、内部验证组 AUC=0.78和外部验证组 AUC=0.82，表明该列线图具有良好的预测能力。校准曲线和 DCA曲线在训练组、内部验证组和外部验证组中都具有准确性及临床获益。结论年龄、尿素氮和 Elixhauser合并症指数是缺血性脑卒中病人需要气管插管的风险因素，基于多种因素构建的列线图模型对缺血性脑卒中病人是否需要气管插管具有良好的预测价值。 |
| 英文摘要: |
| Objective To analyze the risk factors for endotracheal intubation intervention in patients with ischemic stroke and to con.struct and validate a nomogram model.Methods A retrospective analysis was conducted on the clinical data of 3 420 patients withischemic stroke from the Medical Information Mart for Intensive Care Ⅲ/Ⅳ (MIMIC-Ⅲ/Ⅳ) database, covering the period from 2001 to 2019. Among the 1 264 patients in the MIMIC-Ⅲ database, 70% were chosen as the training group, while the remaining 30% and 2, 156 patients from the MIMIC-Ⅳ database served as the internal and external validation groups. A multivariable logistic regression mod.el was used to construct a nomogram prediction model for the risk factors of endotracheal intubation. Receiver operating characteristic(ROC) curve and the area under the curve (AUC) were used to assess the predictive performance of the nomogram. Calibration curveswere drawn using a bootstrap method with 1,000 times of resampling to reflect the consistency between the actual probability and pre.dicted probability. Decision curve analysis (DCA) was used to assess the clinical utility of the model.Results The training group and validation group had statistically significant differences in the use of vasopressors (3.38% vs. 6.65%, P=0.009) and the prevalence of di. abetes (31.98% vs. 23.94%, P=0.004). No significant differences were found in urea nitrogen levels [(8.12±5.90) mmol/L vs. (8.50± 6.51) mmol/L, P=0.314], Elixhauser Comorbidity Index (6.40±7.10 vs. 6.54±6.89, P=0.739), and age [(68.55±16.09) years vs. (69.73± 15.82) years, P=0.234]. Multivariable binary logistic analysis results showed that age (OR=1.020, P=0.020), urea nitrogen (OR=1.020, P=0.010), and the Elixhauser Comorbidity Index (OR=1.150, P<0.001) were independent risk factors for endotracheal intubation in pa.tients with ischemic stroke. The ROC curve showed AUC values of 0.79 for the training group, 0.78 for the internal validation group,and 0.82 for the external validation group, indicating good predictive ability of the nomogram. The calibration and DCA curves demon.strated accuracy and clinical benefit across all groups. Conclusions Age, urea nitrogen, and the Elixhauser Comorbidity Index are risk factors for endotracheal intubation in patients with ischemic stroke. The nomogram model, based on multiple factors, provides agood predictive value for the need for endotracheal intubation in these patients. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
