120例利奈唑胺致血小板减少的危险因素分析及风险预测模型构建

Analysis of Risk Factors and Construction of Risk Prediction Model in 120 Cases of Thrombopenia Induced by Linezolid

目的:采用Logistic模型和受试者工作特征曲线(ROC),探讨接受利奈唑胺治疗的患者发生血小板减少的危险因素,并预测该不良反应发生风险。方法:回顾性收集并分析2017年6月至2022年5月昆明医科大学第一附属医院符合标准的接受利奈唑胺治疗的120例患者的临床资料。使用多因素Logistic回归法分析利奈唑胺致血小板减少的危险因素,建立Logistic模型,联合ROC曲线预测血小板减少发生风险。结果:120例患者使用利奈唑胺治疗期间,29例(占24.17%)发生血小板减少。多因素Logistic回归分析结果表明,血小板减少的独立危险因素有利奈唑胺谷浓度、基础血小板计数、白蛋白水平、肌酐清除率和用药时间,利用独立危险因素建立Logistic回归方程,方程变换后得到联合预测因子计算公式,联合预测因子=0.172×利奈唑胺谷浓度-0.019×基础血小板计数-0.184×白蛋白水平-0.023×肌酐清除率+0.222×用药时间,联合预测因子ROC曲线下面积(0.928,95%CI=0.972~0.984,P<0.001)优于其他指标,具有一定预测价值,Youden指数最大时(0.766)的切点为ROC曲线上的最佳界值(-2.63)。结论:利奈唑胺谷浓度、基础血小板计数、白蛋白水平、肌酐清除率和用药时间是血小板减少的独立危险因素。临床使用利奈唑胺时,可用独立危险因素计算联合预测因子,预测血小板减少的发生风险,便于及时调整治疗方案。

OBJECTIVE:To probe into the risk factors of thrombocytopenia in patients receiving linezolid treatment by Logistic model and receiver operator characteristics(ROC)curve,and to predict the risk of adverse drug reactions.METHODS:Clinical data of 120 patients who met the criteria and received linezolid treatment in the First Affiliated Hospital of Kunming Medical University from Jun.2017 to May 2022 were retrospectively collected and analyzed.Multivariate Logistic regression method was used to analyze the risk factors for thrombocytopenia induced by linezolid.A Logistic model was established and combined with ROC curves to predict the incidence of thrombocytopenia.RESULTS:Thrombocytopenia occurred in 29 cases(24.17%)of 120 patients during treatment with linezolid.Multivariate Logistic regression analysis showed that valley concentration,basal platelets,albumin,creatinine clearance and medication duration were independent risk factors for thrombocytopenia.A Logistic regression equation was established by using independent risk factors,and a joint prediction factor calculation formula was obtained after transformation.Joint predictor=0.172×valley concentration-0.019×basic platelet-0.184×albumin-0.023×creatinine clearance rate+0.222×medication duration.The area under ROC curve of the joint predictor(0.928,95%CI=0.972-0.984,P<0.001)was better than other indicators,and had certain predictive value.The cut-off point at the maximum of Youden index(0.766)was the optimal cut-off value on ROC curve(-2.63).CONCLUSIONS:Valley concentration,basal platelet,albumin,creatinine clearance,and medication duration were independent risk factors for thrombocytopenia.During clinical medication,the above factors could be incorporated into the calculation formula of joint predictive factors to predict the risk of thrombocytopenia,so as to timely adjust the medication regimen.