Hemorrhagic shock in intensive care: Variation in blood product administration during resuscitation

Hemorrhagic shock requires timely resuscitation to maintain organ perfusion and hemostasis until source control is achieved. Most resuscitation strategies have been studied outside the intensive care unit (ICU), and optimal approaches in critically ill patients remain unclear. With this study, we aim to characterize variation in resuscitation practices for hemorrhagic shock with (near-)massive transfusion across public ICU datasets.

Red blood cell (RBC) transfusions were extracted from four public ICU datasets (AmsterdamUMCdb, HiRID, MIMIC-IV, and SICdb). Transfusions were stratified by 24-h intensity: near-massive (6-10 RBC units) and massive (≥10 units). Clinical parameters from 48 h before and after transfusion were analyzed. Continuous variables were compared using Kruskal-Wallis tests, with Dunn's post hoc tests and Bonferroni correction.

In near-massive transfusion, median nadir hemoglobin differed across datasets: AmsterdamUMCdb 6.6 g/dL (interquartile range [IQR], 5.6-7.7), HiRID 6.8 (6.1-7.3), MIMIC-IV 7.2 (6.2-8.0), and SICdb 6.2 (5.5-6.7) (p < 0.001). In massive transfusion, nadir hemoglobin was 6.4 (5.6-7.1), 6.9 (6.3-7.3), 7.1 (6.0-8.1), and 5.3 (4.3-6.3), respectively (p < 0.001). Post-resuscitation hemoglobin also differed: 9.5 (8.7-10.3), 9.1 (8.4-9.6), 10.4 (9.4-11.6), and 7.7 (6.9-8.5) (p < 0.001). Crystalloid use varied widely (3350-12,865 mL, p < 0.001), as did the mean arterial pressure at transfusion (60-76 mmHg, p < 0.001).

Resuscitation practices vary substantially across ICU datasets, including differences in hemoglobin levels, fluid administration, and hemodynamic status at transfusion. These findings highlight the heterogeneity of real-world resuscitation strategies in hemorrhagic shock.