PURPOSE: Anemia correlates with increased ICU mortality
but the use of erythropoietin (EPO) as a treatment remains debated. We sought to assess EPO use in ICU severe anemia. METHODS: A retrospective single-center study was conducted in four adult ICUs. Inclusion criteria were ICU stay ≥10 days (to limit immortality bias) and RBC transfusion within the first 48 h (an indication of severe anemia likely to justify EPO). EPO contraindication was an exclusion criterion. Univariate tests were followed by a multivariable analysis. RESULTS: Over a 28-month period, 190 patients (69 with EPO) were included. EPO subgroups displayed had a higher prevalence of hemorrhagic shock and surgical ICU admissions. EPO administration was significantly associated with a lower requirement for late RBC transfusions in trauma and non-trauma subgroups, with odds ratios [95 % confidence interval] of 0.29 [0.10-0.85] and 0.03 [0.004-0.18], respectively. In the EPO subgroup, the median hemoglobin level rose by 1.2 g/dL. Cox model showed a significant association with mortality at day 28 and 365. CONCLUSION: Our study supports the hypothesis whereby EPO administration in severely anemic ICU patients reduces late transfusion needs, with a potentially higher survival rate. Systematic EPO use post-RBC transfusion in ICU patients warrants further investigation.