血液病患者血小板输注无效的预测模型构建
Prediction model of platelet transfusion refractoriness in patients with hematological disorders
岳姝含 1黄秀兰 1曾彦 1雷乔 1何梦珍 1卢黎琦 1游时松 1张静薇1
作者信息
- 1. 成都市第二人民医院 输血科,四川 成都 610051
- 折叠
摘要
目的 探讨血液病患者血小板输注无效(platelet transfusion refractoriness,PTR)发生的危险因素,构建预测模型并对模型效能进行验证.方法 回顾性纳入 2021 年 12 月—2022 年 12 月成都市第二人民医院接受血小板输注治疗的血液病患者,对血小板输注效果进行判断,通过单因素及多因素Logistic回归对其进行风险因子筛选,并构建血小板输注无效预测模型,采用受试者操作特征(receiver operating characteristic curve,ROC)曲线、校正曲线和决策曲线分析(decision curve,DCA)分别评估模型的区分度、校准度和临床价值.结果 共计纳入血液病患者 334名,男性168 名,女性176 名,PTR发生率40.4%.单因素及多因素Logistic回归分析显示血小板输注量、红细胞输注量和中性粒细胞比值是PTR发生的危险因子(P<0.05).基于这些危险因子建立血液病患者PTR预测模型.模型曲线下面积为 0.837 7(95%CI:0.723～0.772),灵敏度为 58.52%,特异度为 89.95%;校正曲线显示S∶P为0.964,最大绝对差值Emax为 0.032,平均绝对差值Eavg为 0.009;DCA分析显示风险阈值范围在 0.2～0.9 之间时,该模型具有临床应用价值.结论 基于血小板输注量、红细胞输注量和中性粒细胞比值构建的PTR预测模型可为血液病患者血小板有效输注提供依据.
Abstract
Objective To explore the risk factors for platelet transfusion refractoriness(PTR)in patients with hemato-logical disorders,construct a prediction model and validate the model efficacy.Methods Patients with hematological disor-ders who received platelet transfusion therapy in the Chengdu Second People's Hospital from December 2021 to December 2022 were retrospectively included to judge the effectiveness of platelet transfusion and screened for risk factors by univariate and multivariate logistic regression.A prediction model for PTR was constructed using receiver operating characteristic(ROC)curve,calibration curve and decision curve(DCA)to assess the differentiation,calibration and clinical value of the model,respectively.Results A total of 334 hematological patients were included,including 168 males and 176 females,with a PTR incidence of 40.4%.Univariate and multivariate logistic regression analysis showed that platelet transfusion vol-ume,erythrocyte transfusion volume,and neutrophil ratio were risk factors for PTR(P<0.05).A prediction model for PTR in hematological patients was established based on these risk factors.The area under the model's curve was 0.8377(95%CI:0.723-0.772),the sensitivity was 58.52%,and the specificity was 89.95%.The calibration curve showed that the S∶P was 0.964,the maximum absolute difference Emax was 0.032,and the average absolute difference Eavg was 0.009.The DCA a-nalysis showed that the model had clinical application value when the risk threshold ranged from 0.2 to 0.9.Conclusion The PTR prediction model based on platelet transfusion volume,erythrocyte transfusion volume and neutrophil ratio can pro-vide a basis for effective platelet transfusion in hematological patients.
关键词
血小板输注不良/危险因素/预测模型Key words
platelet transfusion refractoriness(PTR)/risk factors/prediction model引用本文复制引用
基金项目
四川省自然科学基金面上项目(23NSFSC1745)
出版年
2024
国家科技期刊平台
NSTL
万方数据