中文摘要：

目的：应用Logistic模型和接受者工作特征曲线（ROC）探讨接受利奈唑胺治疗的危重感染患者发生相关血小板减少症的危险因素，并对血小板减少症发生的风险进行预测。方法：收集利奈唑胺治疗的危重感染患者资料，分析接受利奈唑胺治疗的患者发生血小板减少症的危险因素，建立Logistic模型，绘制风险因素的ROC曲线并寻找患者发生血小板减少症的最佳界值，以预测血小板减少症发生的危险性。结果：151名患者（男性107名，女性44名）纳入研究，43名患者（28．5％）发生血小板减少症。多因素Logistic模型分析显示，年龄（≥65岁）（OR=3．52；95％CI，1．46～8．49；P=0．005）、体质量（OR=0．92；95％CI，0．88～0．96；P〈0．0001）、基础血小板值（PLT〈200×10^9／L）（OR=7．21；95％CI，2．61～19．91；P〈0.0001）等是发生血小板减少症的独立危险因素。根据Logistic模型分析的结果，分别用年龄、体质量、基础血小板值及联合预测因子构建ROC曲线，联合预测因子的ROC曲线下面积（0．847；95％CI，0．781～0．912；P〈0．0001）大于其他3个单一指标，风险预测价值较优，Y0uden指数最大（O．576）时的切点为ROC曲线上的最佳界值（466．89），对应切点的敏感度为79．1％，特异度为78．5％。结论：与利奈唑胺致血小板减少症相关的独立危险因素包括年龄（≥65岁）、体质量、基础血小板值。在临床实际工作中可将患者年龄、体质量及基础血小板值代入联合预测因子计算公式（Y联合=X_PLT-2．636＊X年龄＋7．091＊X体质量），计算联合预测因子，预测患者可能发生血小板减少症的风险，以便调整给药方案。

英文摘要：

OBJECTIVE To study the risk factors, and to predict the risks of linezolid-associated thrombocytopenia by logistic model and receiver operating characteristic curves in critically ill patients. METHODS Data of critically ill patients who were treated with linezolid were extracted. The risk factors of linezolid-associated thrombocytopenia were analyzed, logistic model was set up,and receiver operating characteristic（ROC）curve was drawn in order to find the best cut-off value to predict the risks of thrombocytopenia. RESULTS 151 patients（107 male and 44 female）were included in the study,and thrombocytopenia occurred in 43 patients（43/151,28.5%）. Based on a multivariate logistic regression analysis,only age（≥65 years） （OR = 3.52;95% CI, 1.46 - 8. 49;P = 0. 005） ,weight（OR = 0. 92; 95% CI,0. 88 - 0. 96; P〈0.000 1 ）, baseline platelet count（〈200 x 109/L） （OR = 7. 21 ;95%CI, 2. 61 - 19. 91 ;P〈0. 000 1 ）, were significant independent risks for linezolid-associated thrombocytopenia. The ROC area of the joint predictor, which was larger than the other three single indicators,was the more superior predictive value. When the Youden index was the largest（0. 576）, the best cut-off value was 466. 89;corresponding to the sensitivity 79. 1% and the specificity 78. 5 %. CONCLUSION The risk factors of linezolid-associated thrombocytopenia included age （≥65 years）, body weight, and baseline platelet count. In clinical practice, we could calculate the joint predictor by the age, weight, and baseline platelet value of the patients. Using calculation formula（Yjoint = X_PLT- 2. 636 ＊ Xage ＋ 7. 091 ＊ Xweight ）, we can predict the risk of the patients who may occur linezolid-associated thrombocytopenia, in order to adjust the dosing regimen, and enhancing patient medication monitoring.