RESUMO
BACKGROUND: Perioperative red blood cell transfusions are required in up to 23 per cent of patients undergoing hepatectomy. Previous research has developed three transfusion risk scores to assess risk of perioperative red blood cell transfusion. Here, the performance of these transfusion risk scores was evaluated in a multicentre cohort of patients who underwent hepatectomy and compared with that of a simplified transfusion risk score. METHODS: A database of patients undergoing hepatectomy at four specialized centres between 2008 and 2012 was developed. External validity was assessed by discrimination and calibration. Discrimination was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC). Calibration was evaluated by the degree of agreement between predicted and actual red blood cell transfusion probabilities. A simplified transfusion risk score using variables common to the three models was created, and discrimination and calibration were evaluated. RESULTS: There were 1287 patients included in this study, with 341 (26·5 per cent) receiving a red blood cell transfusion. Discriminative ability was similar between the three transfusion risk scores, with AUCs of 0·66-0·68 and good calibration. A new three-point risk score was developed based on factors present in all models: haemoglobin 12·5 g/dl or less, primary liver malignancy and major resection (at least 4 segments). Discriminative ability and calibration of the three-point model were similar to those of the three existing models, with an AUC of 0·66. CONCLUSION: The three-point transfusion risk score simplifies assessment of perioperative transfusion risk in hepatectomy without sacrificing predictive ability.