Abstract:

Objective To develop and validate a risk prediction model for non-cuffed catheter（NCC） dysfunction in hemodialysis patients，aiming to provide a reference for early clinical identification and warning. Methods A prospective study design was adopted. A total of 569 patients with indwelling NCC from the hemodialysis center of a tertiary hospital in Nanchang between December 1，2023 to May 20，2024，were included as a modeling cohort. An additional 172 patients from the hemodialysis center of a tertiary hospital in Changsha，enrolled between May 30 to October 20，2024，formed a validation cohort. Data were collected on general patient characteristics，dialysis information，catheterization details，and clinical parameters. The risk prediction model was constructed using a combination of variables identified through univariate analysis，Lasso regression，logistic regression，and the Boruta algorithm. Model performance was evaluated accordingly. Results The incidence of NCC dysfunction in hemodialysis patients was 44.94%. A total of 5 common predictors were identified by both algorithms，including age，ultrafiltration volume，catheter insertion site，catheter indwelling time，and C-reactive protein. The area under the receiver operating characteristic curve（AUC） was 0.720 for internal validation and 0.766 for external validation. The Brier scores for curve calibration were 0.213 and 0.203，respectively. The decision curve analysis showed clinical benefit within risk threshold ranges of 22%~82% and 22%~96%，respectively. Conclusion The risk prediction model developed in this study demonstrates good predictive performance and can serve as a screening and assessment tool for identifying the risk of NCC dysfunction in hemodialysis patients.

Key words: Hemodialysis, Non-Cuffed Catheter, Dysfunction, Risk Prediction Model, Nursing Care

摘要：

目的 构建并验证血液透析患者无涤纶套透析导管（non-cuffed catheter，NCC）功能不良风险预测模型，为临床早期识别和预警提供参考。 方法 采用前瞻性研究设计，便利选取2023年12月1日—2024年5月20日南昌市某三级甲等医院血液净化中心569例留置NCC患者为建模组，选取2024年5月30日—10月20日长沙市某三级甲等医院血液净化中心172例留置NCC患者为验证组。收集患者的一般资料、透析资料、置管资料和临床资料，基于单因素分析、Lasso回归和Logistic回归分析的常规算法及Boruta算法的共同变量构建风险预测模型并评价模型性能。 结果 血液透析患者NCC功能不良发生率为44.94%，2种算法共同变量为年龄、超滤量、置管位置、留置天数、C反应蛋白。该模型在内部与外部验证中的受试者操作特征曲线下面积分别为0.720和0.766；校准曲线布里尔分数分别为0.213和0.203；临床决策曲线的风险阈值分别为22%~82%、22%~96%。 结论 该研究构建的风险预测模型具有良好的预测性能，可作为血液透析患者NCC导管功能不良风险筛查评估工具。

关键词: 血液透析, 无涤纶套透析导管, 功能不良, 风险预测模型, 护理