Determination of Perioperative Blood Loss: Accuracy or Approximation?

Various different interventions can be used to reduce surgical blood loss; however, there is no "gold standard" for accurately measuring the volume of perioperative blood loss, and this makes it difficult to assess the efficacy of these interventions.

We used data from a previous multicenter double-blind randomized clinical trial in patients undergoing total hip arthroplasty in which we compared 2 regimens for administering tranexamic acid versus placebo. We assessed direct measures (external blood loss) and indirect estimates (using the formulas of Bourke, Gross, Mercuriali, and Camarasa and a new formula we have developed) using analysis of variance to compare estimated volumes of blood loss among the study groups. In addition, intraclass correlation coefficients (ICCs) and Bland-Altman diagrams were used to compare the estimated volumes of blood loss obtained with each formula.

The mean estimated external blood loss was 909 ± 324 mL, and the mean estimates of blood loss calculated using the formulas of Gross, Bourke and Smith, and Camarasa were 1308 ± 555, 1091 ± 454, and 1641 ± 945 mL, respectively, whereas we obtained a value of 1511 ± 919 mL with the new formula at day 2. In all cases, the results favored the use of tranexamic acid (P < .0001). Comparing results of the new and other formulas, we found moderate-to-low agreement (in terms of ICCs) except for that of Camarasa (ICC: 0.992). The limits of agreement with the new formula ranged from -378 to 93 in the case of the comparison with Camarasa's formula and from -2226 to 959 for external blood loss, the difference depending on the magnitude of the estimate to a large extent.

Formulas that take into account both anthropometric and laboratory parameters are useful for evaluating the efficacy of interventions aiming to decrease blood loss but do not ensure that the values obtained are sufficiently accurate for absolute measuring.