胃肠道肿瘤患者术后低蛋白血症的相关危险因素分析和预测模型建立

Analysis of related risk factors and establishment of prediction model for postoperative hypoalbuminemia in patients with gastrointestinal tumors

目的:旨在筛选胃肠道肿瘤术后低蛋白血症的围术期相关危险因素,并建立相应预测模型。方法:回顾性分析2022年12月至2023年6月在我院行胃肠道肿瘤根治术的244例患者临床资料,运用Logistic回归分析对术后低蛋白血症的相关因素进行单因素和多因素分析,筛选出独立危险因素;并建立Nomogram预测模型,评估其预测价值。结果:研究结果显示,共计125例患者术后出现低蛋白血症,占比54.8%。多因素Logistic回归分析显示:术前白蛋白水平(OR 0.70,P<0.01)、BMI(OR 0.79,P<0.01),术中激素(OR 0.92,P<0.01)、白蛋白使用量(OR 0.72,P<0.01),术毕血糖(OR 1.54,P<0.01)和复合硬膜外麻醉(OR 0.38,P<0.05)为术后低蛋白血症发生的独立危险因素。预测模型ROC曲线下面积为0.848(95%CI 0.800~0.896),预测价值较好。结论:术前白蛋白水平、BMI,术中激素、术中白蛋白使用量,术毕血糖和复合硬膜外麻醉是术后低蛋白血症发生的独立危险因素,基于上述变量的Nomogram列线图预测模型具有较高的预测效能,可为临床决策提供有效参考。

Objective:To identify the perioperative risk factors for hypoalbuminemia after gastrointestinal tumor surgery and to establish a predictive model.Methods:A retrospective analysis was conducted on the clinical data of 244 patients who underwent radical resection of gastrointestinal tumors at our hospital between December 2022 and June 2023.Logistic regression analysis was used to conduct univariate and multivariate analysis to isolate independent risk factors for postoperative hypoalbuminemia.A Nomogram prediction model was established to evaluate its prognostic efficacy.Results:Postoperative hypoalbuminemia was observed in 125 patients(54.8%).Multivariate Logistic regression analysis identified several independent risk factors:Preoperative albumin level(OR 0.70,P<0.01),BMI(OR 0.79,P<0.01),intraoperative hormone(OR 0.92,P<0.01),albumin dosage(OR 0.72,P<0.01),blood glucose at the end of operation(OR 1.54,P<0.01)and combined epidural anesthesia(OR 0.38,P<0.05).The predictive model's area under the ROC curve was 0.848(95%CI 0.800~0.896),indicating a highpredictive value.Conclusion:The preoperative albumin level,BMI,intraoperative hormone and albumin use,blood glucose at the end of surgery and combined epidural anesthesia are significant independent risk factors of postoperative hypoalbuminemia.The Nomogram prediction model based on the above variables has high prediction efficiency and can serve as an effective tool for clinical decision-making.