联合预测因子对利拉鲁肽致胃肠道不良反应风险的预测价值

Predictive value of combined predictors for the risk of gastrointestinal adverse reaction induced by liraglutide

目的探讨利拉鲁肽致胃肠道反应(GIAR)的危险因素并建立预测模型,同时验证该模型的准确率。方法回顾性分析2021年7月—2022年9月第九○九医院(厦门大学附属东南医院)使用利拉鲁肽治疗的149例2型糖尿病住院患者的信息,按照4∶1的比例随机分为研究集119例和验证集30例。将119例研究集患者的临床指标进行统计分析,筛选出利拉鲁肽致GIAR的独立危险因素,对Logistic模型方程进行转换建立联合预测因子,采用ROC曲线确定联合预测因子的曲线下面积(AUC)及最佳临界值。结果119例患者中,共有28例患者(23.53%)发生了GIAR,多因素Logistic回归分析表明晕车史、联合α-葡萄糖苷酶抑制剂(AGI)和性别是利拉鲁肽致GIAR的独立危险因素。对Logistic模型方程转换得出联合预测因子(Y联合预测因子)的计算公式,Y联合预测因子=X晕车史+1.165X联合AGI+3.285X性别,联合预测因子的AUC最大为0.741,约登指数(Youden)最大时(0.376)对应的最佳临界值为3.785。利用30例验证集患者的临床指标验证该模型预测的准确率为76.67%。结论晕车史、联合AGI和性别是利拉鲁肽致GIAR的独立危险因素,联合上述3项临床指标构建联合预测因子对利拉鲁肽致GIAR风险具有良好的预测价值。

Objective To explore the risk factors of liraglutide-induced gastrointestinal adverse reaction(GIAR),and establish a prediction model and verify the accuracy of it.Methods The information of 149 inpatients with type 2 diabetes who had been treated with liraglutide in the 909th Hospital/Dongnan Hospital of Xiamen University from July 2021 to September 2022 was retrospectively analyzed.In a ratio of 4:1,those patients were randomly divided into study group(119 cases)and verification group(30 cases).The clinical indexes of 119 patients in the study group were statistically analyzed,and the independent risk factors of GIAR induced by liraglutide were detected.Logistic model equation was transformed to establish combined predictors,and receiver operating characteristic(ROC)curve was used to determine the area under the curve(AUC)and optimal critical value of the combined predictors.Results Among 119 patients,28(23.53%)had GIAR.Multivariate logistic regression analysis showed that history of carsickness,combinedα-glucosidase inhibitor(AGI)and sex were independent risk factors of GIAR induced by liraglutide.Logistic model equation was transformed into a calculation formula of the combined predictors(Ycombined predictors).Ycombinedpredictors=Xhistoryofcarsickness+1.165XcombinedAGI+3.285Xsex.The maximum AUC of the combined predictors was 0.741.The optimal critical value was 3.785 when the Youden index was the highest(0.376).The accuracy of the model was 76.67%with the clinical indexes of 30 patients in the verification group.Conclusion History of carsickness,combined AGI and sex were independent risk factors of GIAR induced by liraglutide.Combining the above three clinical indexes to establish the combined predictors had great predictive value for the risk of GIAR induced by liraglutide.