碘对比剂在肿瘤专科医院CT增强检查中急性不良反应预测模型的构建与评估

Development and evaluation of a predictive model for acute adverse reactions to iodinated contrast media in enhanced CT examination

目的探讨碘对比剂相关的急性不良反应(acute adverse reaction,AARs)的风险因素,构建碘对比剂AARs列线图预测模型,并分析其预测效能,为检查前发现碘对比剂AARs高危人群并尽早干预提供参考依据。方法回顾性收集2021年1月—12月在本院行CT增强检查患者的数据,筛选出符合纳入标准的不良反应组242例,利用SPSS随机匹配未发生不良反应的242例,采用Logistic回归分析碘对比剂AARs的独立危险因素,通过R软件建立列线图预测模型,并利用受试者工作特征曲线(Receiver operating characteristic,ROC)、Hosmer-Lemeshow拟合优度检验进行评估。结果碘对比剂总的AARs的发生率为0.205%。多因素Logistic回归分析结果显示,过敏史、碘对比剂种类、进食距检查时间为碘对比剂AARs的独立危险因素(均P<0.05);碘对比剂AARs列线图预测模型预测碘对比剂AARs风险的ROC曲线下面积为0.869(95%CI:0.837~0.905),特异性为0.810,敏感度为0.777;列线图校准曲线斜率接近1,Hosmer-Lemeshow拟合优度检验χ^(2)=2.468,P=0.963。结论基于在肿瘤专科医院CT增强检查中影响碘对比剂AARs的3个因素构建的列线图预测模型具有良好的区分度和准确度,可为检查前干预提供科学的参考依据。

Objective To explore the risk factors of acute adverse reaction(AAR)associated with the administration of iodinated contrast media by developing a nomogram prediction model and further evaluating the performance of the model so as to provide a reference for earlier screening for high-risk population of AARs associated with iodinated contrast media before a CT examination and an early intervention against an AAR.Methods Data of the patients who received enhanced CT examinations in our hospital between January 2021 and December 2021 were retrospectively collected.A total of 242 patients who met the inclusion and exclusion criteria were included in the AAR group,and another 242 patients without AAR were randomly selected in the control group.The nomogram model was developed based on the independent risk factors of the AAR by Logistic regression.The receiver operating characteristic(ROC)curve and Hosmer-Lemeshow goodness-of-fit test were used to evaluate the performance of the model.Results The incidence of AARs was found at 0.205%.Three variates being found significantly associated with AARs were history of allergy,type of iodine contrast medium and time from last eating to a CT scan.This predictive model built on the basis of above mentioned three variates showed good calibration and discriminatory abilities,with an area under the curve(AUC)of 0.869(95%CI:0.837~0.905),where sensitivity was 0.777,sepcificity was 0.810,Nomogram calibration slope reach 1,Hosmer-Lemeshow godness of fit χ^(2)=2.468,P=0.963 Conclusions The nomogram prediction model based on the three factors affecting AAR that associated with iodinated contrast media in enhanced CT examination in a cancer hospital has good discrimination and accuracy.It provides scientific basis and reference for pre-examination intervention.