Assessment of the external validity of a predictive score for blood transfusion in liver surgery

Perioperative bleeding is a predictor of morbidity following liver resection. The transfusion-related score (TRS), which is derived from five variables (cirrhosis, preoperative haemoglobin level, tumour size, vena cava exposure and associated extraliver surgical procedure), has been proposed to predict the likelihood of transfusion in liver resection.

The purpose of this observational study was to evaluate the external validity of the TRS.

In a retrospective, monocentre, observational cohort study of patients undergoing elective liver resection surgery, data for transfused and non-transfused patients were compared by univariate analysis. The TRS was calculated for each patient. The frequency of transfusion was calculated for each score level. The accuracy of the TRS was evaluated using the area under the receiver operator characteristic curve (AUC).

A total of 205 patients submitted to liver resection were included. Of these, 48 (23.4%) patients received a blood transfusion. There was no significant difference between transfused and non-transfused patients in age, American Society of Anesthesiologists (ASA) score or cirrhosis. The AUC for the TRS was 0.68 (95% confidence interval 0.59-0.77). Among TRS items, only vena cava exposure and associated surgical procedures were significantly associated with risk for transfusion.

In the present population, the TRS appeared to serve as a weak predictor of perioperative transfusion. This study confirms that the external validity of the transfusion predictive score should be subject to further investigation before it can be implemented in clinical use.