关键词: 输血相关循环超负荷, 危险因素, 预测模型, 内部验证

Abstract: 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 showed χ²=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.

Key words: Transfusion-associated circulatory overload, Risk factors, Prediction model, Internal verification