当前位置: X-MOL 学术Anesth. Analg. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A Simple Risk Scoring System for Predicting the Occurrence of Aspiration Pneumonia After Gastric Endoscopic Submucosal Dissection
Anesthesia & Analgesia ( IF 5.7 ) Pub Date : 2021-10-21 , DOI: 10.1213/ane.0000000000005779
Kyemyung Park , Na Young Kim , Ki Jun Kim , Chaerim Oh , Dongwoo Chae , So Yeon Kim

ent gastric ESD for gastric neoplasm under anesthesiologist-directed sedation. Candidate risk factors were screened and assessed for significance using a least absolute shrinkage and selection operator (LASSO)-based method. Top significant factors were incorporated into a multivariable logistic regression model, whose prediction performance was compared with those of other machine learning models. The final risk scoring system was created based on the estimated odds ratios of the logistic regression model. RESULTS: The incidence of aspiration pneumonia was 1.5%. The logistic regression model showed comparable performance to the best predictive model, extreme gradient boost (area under receiver operating characteristic curve [AUROC], 0.731 vs 0.740). The estimated odds ratios were subsequently used for the development of the clinical scoring system. The final scoring system exhibited an AUROC of 0.730 in the test dataset with risk factors: age (≥70 years, 4 points), male sex (8 points), body mass index (≥27 kg/m2, 4 points), procedure time (≥80 minutes, 5 points), lesion in the lower third of the stomach (5 points), tumor size (≥10 mm, 3 points), recovery time (≥35 minutes, 4 points), and desaturation during ESD (9 points). For patients with total scores ranging between 0 and 33 points, aspiration pneumonia probabilities spanned between 0.1% and 17.9%. External validation using an additional cohort of 827 patients yielded AUROCs of 0.698 for the logistic regression model and 0.680 for the scoring system. CONCLUSIONS: Our simple risk scoring system has 8 predictors incorporating patient-, procedure-, and sedation-related factors. This system may help clinicians to stratify patients at risk of aspiration pneumonia after ESD....

中文翻译:

一种简单的风险评分系统,用于预测胃内窥镜黏膜下剥离术后吸入性肺炎的发生

在麻醉师指导的镇静下对胃肿瘤进行胃 ESD。使用基于最小绝对收缩和选择算子 (LASSO) 的方法筛选和评估候选风险因素的显着性。将最重要的因素纳入多变量逻辑回归模型,将其预测性能与其他机器学习模型的预测性能进行比较。最终的风险评分系统是基于逻辑回归模型的估计优势比创建的。结果:吸入性肺炎的发生率为1.5%。逻辑回归模型显示出与最佳预测模型相当的性能,极端梯度提升(接受者操作特征曲线下的面积 [AUROC],0.731 对 0.740)。估计的优势比随后用于开发临床评分系统。最终评分系统在测试数据集中显示 AUROC 为 0.730,风险因素包括:年龄(≥70 岁,4 分)、男性(8 分)、体重指数(≥27 kg/m2,4 分)、手术时间(≥80 分钟,5 分)、胃下三分之一处的病变(5 分)、肿瘤大小(≥10 毫米,3 分)、恢复时间(≥35 分钟,4 分)和 ESD 期间的去饱和(9分)。对于总分在 0 到 33 分之间的患者,吸入性肺炎的概率在 0.1% 到 17.9% 之间。使用额外的 827 名患者队列进行外部验证,逻辑回归模型的 AUROC 为 0.698,评分系统的 AUROC 为 0.680。结论:我们的简单风险评分系统有 8 个预测因子,包括患者、手术和镇静相关因素。
更新日期:2021-12-20
down
wechat
bug