Comparison of quantitative and calculated postpartum blood loss after vaginal delivery

Because there is no consensus on the method of assessing postpartum blood loss, the comparability and relevance of the postpartum hemorrhage-related literature are questionable. Quantitative blood loss assessment using a volumetric technique with a graduated collector bag has been proposed to overcome limitations of intervention-based outcomes but remains partly subjective and potentially biased by amniotic fluid or missed out-of-bag losses. Calculated blood loss based on laboratory parameters has been studied and used as an objective method expected to reflect total blood loss. However, few studies have compared quantitative with calculated blood loss.

This study aimed to compare the distribution of postpartum blood loss after vaginal delivery assessed by 2 methods-quantitative and calculated blood loss-and the incidence of abnormal blood loss with each method.

Data were obtained from the merged database of 3 multicenter, randomized controlled trials, all testing different interventions to prevent postpartum blood loss in individuals with a singleton live fetus at ≥35 weeks of gestation, born vaginally. All 3 trials measured blood loss volume by using a graduated collector bag. Hematocrit was measured in the eighth or ninth month of gestation and on day 2 postpartum. The 2 primary outcomes were: quantitative blood loss, defined by the total volume of blood loss measured in a graduated collector bag, and calculated blood loss, mathematically defined from the peripartum hematocrit change (estimated blood volume × [(antepartum hematocrit-postpartum hematocrit)/antepartum hematocrit], where estimated blood volume [mL]=booking weight [kg] × 85). We modeled the association between positive quantitative blood loss and positive calculated blood loss with polynomial regression and calculated the Spearman correlation coefficient.

Among the 8341 individuals included in this analysis, the median quantitative blood loss (100 mL; interquartile range, 50-275) was significantly lower than the median calculated blood loss (260 mL; interquartile range, 0-630) (P<.05). The incidence of abnormal blood loss was lower with quantitative blood loss than calculated blood loss for all 3 thresholds: for ≥500 mL, it was 9.6% (799/8341) and 32.3% (2691/8341), respectively; for ≥1000 mL, 2.1% (176/8341) and 11.5% (959/8341); and for ≥2000 mL, 0.1% (10/8341) and 1.4% (117/8341) (P<.05). Quantitative blood loss and calculated blood loss were significantly but moderately correlated (Spearman coefficient=0.44; P<.05). The association between them was not linear, and their difference tended to increase with blood loss. Negative calculated blood loss values occurred in 23% (1958/8341) of individuals; among them, >99% (1939/1958) had quantitative blood loss ≤500 mL.

Quantitative and calculated blood loss were significantly but moderately correlated after vaginal delivery. However, clinicians should be aware that quantitative blood loss is lower than calculated blood loss, with a difference that tended to rise as blood loss increased.