ObjectiveTo explore risk factors of blood transfusion during liver transplantation and construct its prediction model. MethodsThe patients underwent liver transplantation who met the inclusion and exclusion criteria of this study from March 2020 to December 2020 in the Beijing Youan Hospital of Capital Medical University were retrospectively collected. The univariate and logistic multivariate analysis were used to evaluate the risk factors of blood transfusion during liver transplantation and construct the prediction model for intraoperative blood transfusion. ResultsA total of 151 eligible liver transplantation patients were collected in this study, including 51 non-transfusion patients and 100 transfusion patients. The univariate analysis results showed that the differences of primary diagnosis, preoperative hemoglobin (Hb), platelet count, prothrombin time, international normalized ratio, Child-Turcotte-Pugh score, and end-stage liver disease (MELD) score were statistically different between them (P<0.05). The above variables selected by the univariate analysis were selected by stepwise method, then the preoperative Hb and MELD score were selected into the multivariate logistic regression analysis, the results showed that the preoperative Hb≤113 g/L and MELD score ＞14 increased the risk of blood transfusion during liver transplantation [Hb: OR=6.652, 95%CI (2.282, 19.392), P<0.001; MELD score: OR=16.037, 95%CI (6.336, 40.592), P<0.001]. The logistic regression model predicted the area under receiver operating characteristic curve was 0.873 [95%CI (0.808, 0.919), P<0.001], the sensitivity and specificity were 91.0% and 67.5%, respectively, Youden index was 0.674, the accuracy was 86.1%. ConclusionsResults of this study suggest that preoperative Hb ≤113 g/L and MELD score＞14 increase risk of blood transfusion during liver transplantation. Logistic regression model constructed according to preoperative Hb and MELD score has a better sensitivity and specificity of intraoperative blood transfusion.