Abstract: Objective To explore the effect of cardio-pulmonary and renal function on erythrocyte transfusion threshold in patients with chronic anemia, and provide reference basis for clinical pre-transfusion evaluation. Methods The basic information of 979 patients with chronic anemia undergoing blood transfusion in our hospital from 2020 to 2021 was collected, including the patient's sex, age, left ventricular ejection fraction, partial pressure of carbon dioxide, glomerular filtration rate estimation, hemoglobin concentration before blood transfusion and other indicators. Multiple linear regression analysis and propensity score matching were used to study the risk factors affecting the red blood cell infusion threshold. Results The multiple linear regression analysis of 979 patients with blood transfusion showed that the independent risk factors affecting the threshold of red blood cell transfusion were age and partial pressure of carbon dioxide, and the B values were -0.181 and 0.296, respectively, P<0.01. The R² of the regression model is 0.072, the VIF value is less than 2, and the DW value is 0.533. The independent variables of the model are good, there is no collinearity problem, and the residual conforms to the normal distribution. After score matching of 290 cases of renal failure group, 268 cases were matched successfully. The mean hemoglobin concentration before blood transfusion in the matched renal failure group and the normal renal function group were 57.74 and 68.88, respectively, and the F value was 24.90, P<0.01. Conclusion This study suggests that age, lung function and renal function can affect the red blood cell infusion threshold of patients, and the increase of age and renal failure can reduce the red blood cell infusion threshold; The increase of partial pressure of carbon dioxide will increase the threshold of red blood cell infusion; There was no statistical difference in the effect of left ventricular ejection fraction on red blood cell infusion threshold.

Key words: Erythrocyte infusion threshold, Linear regression, Tendency score matching