首页|期刊导航|汕头大学医学院学报|患者自控静脉镇痛术后恶心呕吐风险预测模型的构建与验证

患者自控静脉镇痛术后恶心呕吐风险预测模型的构建与验证OA

Construction and validation of a risk prediction model for postoperative nausea and vomiting after patient-controlled intravenous analgesia

中文摘要英文摘要

目的:探讨患者自控静脉镇痛(PCIA)术后恶心呕吐(PONV)的独立影响因素,构建并验证风险预测模型,为临床个体化预防决策提供量化工具.方法:回顾性收集2023年5月至2025年12月于乐山市中医医院接受全身麻醉或椎管内麻醉并术后使用PCIA的1 533例患者资料.采用Boruta算法、LASSO回归及文献证据综合筛选变量,纳入多因素logistic回归构建预测模型.采用受试者操作特征曲线下面积(AUC)、校准曲线及决策曲线分析评估模型区分度、校准度及临床实用性,采用Bootstrap法(1 000次)进行内部验证.结果:115例(7.5%)患者发生PONV.多因素logistic回归显示,女性(OR=2.647,95%CI:1.592~4.561)、头颈及会阴部手术(OR=2.802,95%CI:1.330~5.613)、较高的动态疼痛评分(OR=2.018,95%CI:1.255~3.232)、术中使用苯二氮䓬类药物(OR=1.929,95%CI:1.246~3.022)是PONV的独立危险因素;较高的血小板水平(OR=0.996,95%CI:0.993~0.999)、较大的PCIA泵注总量(OR=0.988,95%CI:0.982~0.993)、术中使用依托咪酯(OR=0.600,95%CI:0.381~0.926)是独立保护因素(P均<0.05).模型训练集AUC为0.724(95%CI:0.675~0.773),Bootstrap校正后AUC为0.722(95%CI:0.698~0.746);校准曲线显示模型拟合良好(Hosmer-Lemeshow P=0.373);决策曲线分析显示在0~0.56阈值范围内模型具有临床净获益.根据最佳截断值(0.081),可将患者分为高危组(预测风险≥8.1%)和低危组(<8.1%),高危组PONV发生率(14.0%)是低危组(3.8%)的3.7倍.结论:本研究构建的PCIA患者PONV风险预测模型整合了患者特征、手术因素及PCIA操作参数,具有良好的区分度和校准度,可为临床个体化预防决策提供量化工具.

Objective:To identify the independent influencing factors for postoperative nausea and vomiting(PONV)in patients receiving patient-controlled intravenous analgesia(PCIA)and to develop and validate a risk prediction model,providing a quantitative tool for individualized clinical preventive decisions.Methods:Data of 1 533 patients who underwent general anesthesia or neuraxial anesthesia and received postoperative PCIA at Leshan Hospital of Traditional Chinese Medicine from May 2023 to December 2025 were retrospectively collected.Variables were screened using the Boruta algorithm,LASSO regression,and literature evidence,and then incorporated into multivariate logistic regression to build the prediction model.Model discrimination,calibration,and clinical utility were evaluated using the area under the receiver operating characteristic curve(AUC),calibration curve,and decision curve analysis.Internal validation was performed using the Bootstrap method(1 000 resamples).Results:PONV occurred in 115 patients(7.5%).Multivariate logistic regression showed that female sex(OR=2.647,95%CI:1.592-4.561),head,neck and perineal surgery(OR=2.802,95%CI:1.330-5.613),higher dynamic pain score(OR=2.018,95%CI:1.255-3.232),and intraoperative use of benzodiazepines(OR=1.929,95%CI:1.246-3.022)were independent risk factors for PONV;while higher platelet count(OR=0.996,95%CI:0.993-0.999),larger total PCIA infusion volume(OR=0.988,95%CI:0.982-0.993),and intraoperative use of etomidate(OR=0.600,95%CI:0.381-0.926)were independent protective factors(all P<0.05).The model achieved an AUC of 0.724(95%CI:0.675-0.773)in the training set,and a Bootstrap-corrected AUC of 0.722(95%CI:0.698-0.746).The calibration curve indicated good fit(Hosmer-Lemeshow P=0.373),and decision curve analysis showed clinical net benefit within the threshold probability range of 0-0.56.According to the optimal cutoff value of 0.081(Youden index),patients were classified into high-risk group(predicted risk≥8.1%)and low-risk group(<8.1%).The actual PONV incidence was 14.0%in the high-risk group and 3.8%in the low-risk group,with a 3.7-fold higher risk in the high-risk group.Conclusion:The PONV risk prediction model for PCIA patients constructed in this study integrates patient characteristics,surgical factors,and PCIA operation parameters.It exhibits satisfactory discrimination and calibration,and can serve as a quantitative tool for individualized clinical preventive decisions.

钟佳芯;刘景天;郭貔

汕头大学医学院预防医学系,广东 汕头 515041||乐山市中医医院公共卫生科,四川 乐山 614000汕头大学医学院预防医学系,广东 汕头 515041汕头大学医学院预防医学系,广东 汕头 515041

医药卫生

术后恶心呕吐患者自控静脉镇痛危险因素预测模型

postoperative nausea and vomitingpatient-controlled intravenous analgesiarisk factorsprediction model

《汕头大学医学院学报》 2026 (3)

139-144,6

10.13401/j.cnki.jsumc.2026.03.003

评论