Poor prediction of blood transfusion requirements in adult liver transplantations from preoperative variables

Study Objective: To assess the ability of preoperative information to predict intraoperative blood transfusion requirements in adult orthotopic liver transplantation Design: Retrospective review. Setting: Liver transplantation program in a referral center Patients: 583 sequential adult patients undergoing orthotopic liver transplantation Measurements: Preoperative variables with a previously demonstrated relationship to intraoperative transfusion were identified from the literature. These variables were then collected retrospectively from 583 consecutive liver transplantations. Relationships between these and intraoperative blood transfusion requirements were examined by both univariate analyses and multiple linear regression analysis. Results: Univariate analysis revealed significant associations between blood transfused and the following preoperative variables: age, gender, diagnosis, presence of grade 3 or 4 encephalopathy, pseudocholinesterase, creatinine, bilirubin, mean pulmonary artery pressure, activated partial thromboplastin time, and platelet count. Multiple linear regression analysis with correction for diagnosis identified age, creatinine, bilirubin, and pseudocholinesterase as independent predictors; for the final model r(2) = 0.22. Conclusion: Preoperative variables are poor predictors of intraoperative transfusion requirements even when significant associations exist, identifying a small proportion of the variability observed. A predictive approach based on this method would be too inaccurate to be of clinical use. The majority of the variability in transfusion requirements during liver transplantation most likely results from intraoperative and donor organ factors. (C) 2000 by Elsevier Science Inc.