Agreement of surgical blood loss estimation methods

Abstract

Background: Surgical blood loss is usually estimated by different formulae in studies of strategies aimed at reducing perioperative bleeding. This study assessed and compared the agreement of the main blood loss estimation formulae using a direct measurement of blood loss as the reference method.

Study design and methods: Eighty consecutive patients undergoing urologic laparoscopic surgery were studied. Only optimal conditions for the direct measurement of surgical blood loss were considered. Surgical blood loss was estimated by six formulae at four different postoperative time points. The agreement of the formulae was evaluated by the Concordance correlation coefficient (CCC) and Bland-Altman analyses. An analysis of the agreement's variability regarding different magnitudes of blood loss was also performed.

Results: Directly measured blood loss ranged from 200 to 2200 mL. The formulae studied showed poor agreement with the direct measurement of blood loss; 95% limits of agreement widely exceeded the criterion of ±560 mL. Significant biases were found, which for most of the formulae led to an overestimation of blood loss. For all formulae, agreement remained constant regardless of the amount of blood loss, with limits between -40 and +120% approximately. Among the formulae, the best agreement was achieved by López-Picado's formula at 48 hours (CCC: 0.577), with a bias of +283 mL and 95% limits of agreement between -477 and +1043 mL.

Conclusion: Formulae currently used to estimate surgical blood loss differ substantially from direct measurements; therefore, they may not be reliable methods of blood loss quantification in the surgical setting.