头孢哌酮/舒巴坦致血小板减少列线图预测模型的建立与验证  

Establishment and validation of nomogram prediction model of cefoperazone/sulbactam-induced thrombocytopenia

作　　者：白荷荷[1] 彭莉蓉[1] 王园姬 聂晓静[1] 王金萍[1] 马莉[1] 王冠 BAI Hehe;PENG Lirong;WANG Yuanji;NIE Xiaojing;WANG Jinping;MA Li;WANG Guan(Dept.of Pharmacy,Xi’an Central Hospital,Xi’an 710003,China;Dept.of Pharmacy,Xi’an First Hospital,Xi’an 710002,China)

机构地区：[1]西安市中心医院药剂科,西安710003 [2]西安市第一医院药剂科,西安710002

出　　处：《中国药房》2024年第8期980-985,共6页China Pharmacy

基　　金：陕西省自然科学基础研究计划项目(No.2020JQ-933);陕西省中医药管理局秦创原中药创新研发项目(No.2022-QCYZH-023);西安市科技计划项目(No.22YXYJ0015);西安市卫生健康委员会科研项目(No.2023ms01)。

摘　　要：目的探讨头孢哌酮/舒巴坦致成人住院患者血小板减少的预测因子,建立列线图预测模型并进行验证。方法回顾性收集西安市中心医院2021年6月30日至2023年6月30日使用头孢哌酮/舒巴坦治疗的成人住院患者资料,按7∶3随机分为训练集和内部验证集。采用单因素/多因素Logistic回归分析筛选头孢哌酮/舒巴坦致血小板减少的独立预测因子,通过R4.0.3软件“RMS”包绘制列线图;采用受试者工作特征曲线及C-index值评估模型的预测效能;采用Hosmer-Lemeshow拟合优度检验评价模型的校正度。以相同标准,收集西安市第一医院同期使用头孢哌酮/舒巴坦治疗的成人住院患者的临床资料,对列线图预测模型进行外部验证。结果共纳入西安市中心医院患者1045例,其中头孢哌酮/舒巴坦致血小板减少患者67例,发生率为6.41%。排除假阳性患者后,最终纳入患者473例,其中训练集331例、内部验证集142例。多因素Logistic回归分析结果显示,患者年龄[OR=1.043,95%CI(1.017,1.070)]、估算的肾小球滤过率(eGFR)[OR=0.988,95%CI(0.977,0.998)]、基线血小板[OR=0.989,95%CI(0.982,0.996)]、营养风险[OR=3.863,95%CI(1.884,7.921)]和累计限定日剂量数(DDDs)[OR=1.082,95%CI(1.020,1.147)]是头孢哌酮/舒巴坦致血小板减少的独立预测因子(P<0.05)。训练集和内部验证集的C-index值分别为0.824[95%CI(0.759,0.890)]和0.828[95%CI(0.749,0.933)],Hosmer-Lemeshow检验的χ^(2)值分别为0.441、1.804(P值分别为0.802、0.406)。外部验证集中,C-index值为0.808[95%CI(0.672,0.945)],Hosmer-Lemeshow检验的χ^(2)值为0.899(P值为0.638)。结论患者年龄、基线血小板、eGFR、营养风险和累计DDDs是头孢哌酮/舒巴坦致血小板减少的独立预测因子;所建列线图预测模型具有良好的预测效能和外推性,有助于临床快速、准确地识别头孢哌酮/舒巴坦致血小板减少的潜在风险。OBJECTIVE To explore the predictive factors of cefoperazone/sulbactam-induced thrombocytopenia in adult inpatients,and to establish and validate the nomogram prediction model.METHODS Data of adult inpatients treated with cefoperazone/sulbactam in Xi’an Central Hospital from Jun.30th,2021 to Jun.30th,2023 were retrospectively collected.The training set and internal validation set were randomly constructed in a 7∶3 ratio.Singler factor and multifactor Logistic regression analysis were used to screen the independent predictors of cefoperazone/sulbactam-induced thrombocytopenia.The nomogram was drawn by using“RMS”of R 4.0.3 software,and the predictive performance of the model was evaluated by the receiver operating characteristic curve and C-index curve.Hosmer-Lemeshow goodness-of-fit test was used to evaluate the calibration degree of the model.Using the same standard,the clinical data of hospitalized patients receiving cefoperazone/sulbactam in Xi’an First Hospital in the same period were collected for external validation of the nomogram prediction model.RESULTS A total of 1045 patients in Xi’an Central Hospital were included in this study,among which 67 patients suffered from cefoperazone/sulbactam-induced thrombocytopenia,with an incidence of 6.41%.After the false positive patients were excluded,473 patients were included finally,including 331 in the training set and 142 in the internal validation set.Multifactor Logistic regression analysis showed that age[OR=1.043,95%CI(1.017,1.070)],estimated glomerular filtration rate(eGFR)[OR=0.988,95%CI(0.977,0.998)],baseline platelet(PLT)[OR=0.989,95%CI(0.982,0.996)],nutritional risk[OR=3.863,95%CI(1.884,7.921)]and cumulative defined daily doses(DDDs)[OR=1.082,95%CI(1.020,1.147)]were independent predictors for cefoperazone/sulbactam-induced thrombocytopenia(P<0.05).The C-index values of the training set and the internal validation set were 0.824[95%CI(0.759,0.890)]and 0.828[95%CI(0.749,0.933)],respectively.The results of the Hosmer-Lemeshow test showed thatχ^(

关 键 词：头孢哌酮/舒巴坦 血小板减少 药物不良反应 列线图预测模型 预测因子 

分 类 号：R978.1[医药卫生—药品] R969[医药卫生—药学]