Effect of Transfusion on Mortality and Other Adverse Events Among Critically Ill Septic Patients: An Observational Study Using a Marginal Structural Cox Model*

Objectives: RBC transfusion is often required in patients with sepsis. However, adverse events have been associated with RBC transfusion, raising safety concerns. A randomized controlled trial validated the 7g/dL threshold, but previously transfused patients were excluded. Cohort studies led to conflicting results and did not handle time-dependent covariates and history of treatment. Additional data are thus warranted to guide patient's management. Design: To estimate the effect of one or more RBC within 1 day on three major outcomes (mortality, ICU-acquired infections, and severe hypoxemia) at day 30, we used marginal structural models. A trajectory modeling, based on hematocrit evolution pattern, allowed identification of subgroups. Secondary analyses were performed into each of them. Setting: A prospective French multicenter database. Patients: Patients with sepsis at admission. Patients with hemorrhagic shock at admission were excluded. Interventions: None. Measurements and Main Results: Overall, in our cohort of 6,016 patients, RBC transfusion was not associated with death (hazard ratio, 1.07; 95% CI, 0.88-1.30; p = 0.52). However, RBC transfusion was associated with increased occurrence of ICU-acquired infections (hazard ratio, 2.77; 95% CI, 2.33-3.28; p < 0.01) and of severe hypoxemia (hazard ratio, 1.29; 95% CI, 1.14-1.47; p < 0.01). A protective effect from death by the transfusion was found in the subgroup with the lowest hematocrit level (26 [interquartile range, 24-28]) (hazard ratio, 0.72; 95% CI, 0.55-0.95; p = 0.02). Conclusions: RBC transfusion did not affect overall mortality in critically ill patients with sepsis. Increased occurrence rate of ICU-acquired infection and severe hypoxemia are expected outcomes from RBC transfusion that need to be weighted with its benefits in selected patients.