住院患者输血相关循环超负荷的危险因素及预测模型的构建与验证

Risk Factors and Predictive Models of Transfusion-associated Circulatory Overload in Hospitalized Patients

目的分析住院患者输血相关循环超负荷(TACO)的危险因素,构建预测TACO的模型,并验证其预测能力。方法选择2020年6月—2023年12月在本院住院期间进行输血的547例患者,依据有否出现TACO分成TACO组和非TACO组。回顾性收集17项资料进行Logistic回归分析,确定住院输血患者TACO发生的危险因素。依据预测指标构建预测TACO的模型,并利用ROC曲线、校准曲线及决策曲线对模型的预测效能(区分度、一致性及临床获益情况)做出评价。结果547例患者TACO发生率为8.04%(44/547)。TACO组、非TACO组年龄、合并心功能不全、输血原因、输血量及液体正平衡量等5项资料比较有统计学差异(P<0.05)。年龄≥65岁(OR=2.480)、合并心功能不全(OR=2.716)、血容量不足(OR=3.564)、输血量≥800 mL(OR=3.371)、液体正平衡量(OR=2.806)对输血患者TACO发生有显著影响。内部验证显示,ROC曲线下面积(AUC)为0.826(95%CI:0.748~0.905),校准曲线拟合较佳,拟合优度HL检验χ^(2)=7.737,P=0.356,阈值概率处于3%~78%时模型净获益率较高。结论根据年龄、合并心功能不全、血容量不足、输血量、液体正平衡量构建的模型可对输血患者TACO的发生予以有效预测。

Objective To analyze the risk factors of transfusion-associated circulatory overload(TACO)in hospitalized patients,construct a predictive model for TACO,and verify its predictive ability.Methods 547 hospitalization patients who underwent blood transfusion in our hospital from June 2020 to December 2023 were selected and divided into TACO group and non-TACO group based on the presence or absence of TACO.Seventeen items of data were retrospectively collected for logistic regression analysis to determine the risk factors for TACO occurrence.The model for predicting TACO was constructed based on predictive indicators,and the predictive efficacy(discrimination,consistency and clinical benefit)of the model was evaluate using ROC curves,calibration curves and decision curves.Results The incidence of TACO in 547 patients was 8.04%(44/547).There were statistical differences between the TACO group and the non-TACO group in terms of age,concomitant heart failure,reasons for blood transfusion,transfusion volume,and fluid balance(P<0.05).Age≥65 years old(OR=2.480),concomitant heart failure(OR=2.716),hypovolemia(OR=3.564),transfusion volume≥800 mL(OR=3.371),and fluid balance(OR=2.806)had a significant effect on the occurrence of TACO in transfusion patients.Internal validation showed that the area under the ROC curve(AUC)was 0.826(95%CI:0.748~0.905),the calibration curve fit was better,the goodness of fit HL test showedχ^(2)=7.737,P=0.356,and the net benefit rate of the model was higher when the threshold probability was 3%~78%.Conclusion A model constructed based on age,concomitant heart failure,hypovolemia,transfusion volume,and fluid balance can effectively predict the occurrence of TACO in blood transfusion patients.