Analysis of factors related to platelet transfusion effect and analysis of improvement strategies for building prediction models
Objective:To explore the influencing factors of platelet transfusion effect,construct the relevant prediction model and propose the improvement strategy.Methods:The clinical data of 211 patients receiving platelet transfusion from January 2021 to July 2022 were retrospectively analyzed and divided into two groups according to the infusion effect:effective group(CCI≥4.5,n=125 patients)and ineffective group(CCI<4.5,n=86 patients).Clinical data of the two groups were collected,including gender,age,height,weight,platelet type,fever,bleeding,infection,number of transfusions,platelet count before initial transfusion,hemoglobin value before initial transfusion,and total number of input platelets.Univariate and multivariate Logistic regression were used to analyze the factors affecting the effectiveness of platelet transfusion of different types,and the infusion prediction model was constructed.Receiver operating characteristic curve(ROC)analysis model prediction efficiency.Corresponding improvement strategies are proposed according to relevant factors.Results:In the clinical data,the main platelet transfusion type was irradiated monopexic platelet,accounting for 72.50%(153/211),and the number of blood transfusions"<3"was more than"≥3",and some patients with fever,bleeding,infection and other symptoms.The results of single factor analysis showed that there were significant differences in infection,splenomegaly and infusion times between the two groups(P<0.05).The results of Logistics regression analysis showed that infection,splenomegaly and the number of blood transfusion were independent factors affecting the platelet transfusion effect(P<0.05).The combined prediction model Y=1.746* infection+1.253* splenic enlargement+1.554*number of blood transfusions-2.536 was obtained.The ROC curve was inserted into the model and it was found that the area under the prediction curve of the model was 0.811,the standard error was 0.031,and the 95%confidence interval was 0.751～0.871.The specificity,sensitivity and cut-off were 78.4%,73.3%and 0.440,respectively.Conclusion:Infection,splenomegaly and the number of blood transfusions are independent factors affecting the effect of platelet transfusion,and the prediction model built based on the above factors has certain reference value.For patients with platelet transfusion therapy,it is necessary to strengthen the prevention and control of infection during treatment,carry out appropriate management of splenomegaly,and wash red blood cells to reduce transfusion reaction and the production of isoantibodies.
万方数据
维普
