单采血小板输注疗效影响因素分析及预测模型构建

Analysis of the influencing factors of apheresis platelet transfusion efficacy and construction of prediction model

目的:探讨临床进行单采血小板输注的效果及其影响因素,构建血小板输注疗效预测模型。方法:回顾性调查分析2016年1月至2021年8月输注血小板的病人治疗效果情况,根据输注是否有效进行分组,对血型、原发疾病、血小板输注次数、血小板计数及脾肿大情况等进行单因素分析,对血小板输注效果的独立影响因素采用logistic回归分析,对拟合疗效预测模型运用ROC进行评估。结果:200例病人共输注单采血小板次数为384例次,总输注有效率为77.80%。血小板输注疗效与病人的年龄、性别、原发疾病方面无相关性(P>0.05),logistics回归分析显示造成血小板输注无效的非免疫性危险因素有血型B(参照A型)、脾肿大、输注前血小板计数低、输注次数>3次(P<0.05~P<0.01);拟合的疗效预测模型AUC为0.828,95%CI:0.781~0.875;采用外部数据对新建立模型进行验证,AUC=0.782,95%CI:0.616~0.948。结论:临床上引起血小板输注无效的非免疫性危险因素较多;拟合的疗效预测模型具有较好的预测能力和效能,可指导临床科学合理使用血小板,提高血小板输注有效率及临床疗效。

Objective:To investigate the effects of apheresis platelet transfusion(PLT) in clinic and its influencing factors,and construct a prediction model of platelet transfusion efficacy.Methods:The treatment effects of platelet transfusion on patients from January 2016 to August 2021 was retrospectively investigated and analyzed.The patients were grouped according to whether the infusion was effective or not.The blood type,primary disease,platelet transfusion times,platelet count and splenomegaly were analyzed using the univariate analysis method.The logistic regression analysis was used to analyze the independent influencing factors of platelet transfusion efficacy,and the ROC was used to evaluate the fitted therapeutic effect prediction model.Results:The total number of apheresis platelet infusion was 384 times in 200 patients,and the total effective rate of which was 77.80%.There were not correlation between platelet transfusion efficacy and patients′ age,gender and primary disease(P>0.05).The results of logistic regression analysis showed that the non-immune risk factors of platelet transfusion refractoriness included the blood group B(referring to type A),splenomegaly,low platelet count before transfusion and infusion times >3 times(P<0.05 to P<0.01).The AUC of the fitted efficacy prediction model was 0.828,95%CI:0.781-0.875.The external data were used to verify the new model,the AUC=0.782,95%CI:0.616-0.948.Conclusions:Clinically,there are many non-immune risk factors that cause platelet transfusion refractoriness.The fitted therapeutic effect prediction model has good predictive ability and efficacy,which can guide the rational use of platelets in clinic,and improve the effective rate of platelet transfusion and clinical efficacy.