Should We Continue Using Erythropoietin in the ICU?

Abstract & Commentary

By Andrew M. Luks, MD, Pulmonary and Critical Care Medicine, University of Washington, Seattle, is Associate Editor for Critical Care Alert.

Dr. Luks reports no financial relationship to this field of study.

Synopsis: This randomized, double-blind, placebo-controlled trial demonstrates that administration of erythropoietin once a week for three weeks does not reduce the incidence of red blood cell transfusion in a mixed population of critically ill patients but is associated with an increased incidence of thrombotic events and a possible decrease in mortality in trauma patients.

Source: Corwin HL, et al. N Engl J Med. 2007;357(10):965-976.

Previous trials have demonstrated that administration of recombinant erythropoietin to critically ill patients decreases the need for red blood cell transfusions and leads to higher hemoglobin values. Building on these trial results, Corwin and colleagues conducted a prospective, multi-center, randomized, double-blind trial to determine whether administration of a reduced dose of the erythropoietin was safe and effective in a mixed group of critically ill patients. They recruited medical, surgical and trauma patients who were in the ICU for at least 48 hours and had hemoglobin levels less than 12 g/dL. Patients were excluded if they were expected to leave the ICU within a subsequent 48-hour period or had either active thrombotic disease such as acute coronary ischemia or pulmonary embolism, or a history of the same.

Between 48 and 96 hours after admission, patients were randomized to receive 40,000 units of subcutaneous erythropoietin or placebo once a week for three weeks. The medication was not given if the patient's hemoglobin was above 12 mg/dL at the intended administration time. The primary end-point was the percentage of patients receiving red blood cell transfusion between days 1 and 29 of the study. Secondary end-points included the number of units transfused, mortality at days 29 and 140 and the change in hemoglobin values from baseline to day 29. The need for transfusion was at the discretion of each patient's treating physician but the investigators provided guidelines recommending that physicians target a hemoglobin value between 7 and 9 mg/dL and not transfuse patients with values above the upper limit of that range.

There were 1460 patients enrolled in the study — 733 in the erythropoietin group and 727 in the placebo group. The groups were well matched except for the fact that the trauma patients were significantly younger than either the medical or surgical ICU patients. In the erythropoietin group, 28% of patients received only one dose of the study medication, 32% of patients received two doses and 40% completed the entire three-week course.

There were no significant differences in the percentage of patients receiving transfusion (46% erythropoietin vs 48.3% placebo) or the number of units transfused in each group (4.5 + 4.6 vs 4.3 + 4.8 units). At day 29, the change in hemoglobin concentration was greater in the erythropoietin group than the placebo group (1.6 + 2.0 g/dL vs 1.2 + 1.8 g/dL) but by day 42, there were no significant differences in absolute hemoglobin concentrations. Mortality was lower among the patients treated with erythropoietin alfa (8.5% vs 11.4% at 29 days and 14.2% vs 16.8% at 140 days) with the largest differences in mortality seen in the trauma patients. There was an increased incidence of thrombotic events in the erythropoietin group (16.5% vs 11.5%, hazard ratio 1.4) with post-hoc analysis demonstrating that thrombosis largely occurred in those patients not on heparin prophylaxis.


The decision to use any treatment should always be driven by a consideration of the risk-to-benefit ratio. In light of the study by Corwin and colleagues, it does not appear that erythropoietin fares well in this analysis. Regarding the benefits, there is insufficient evidence of clinical utility; the study described above showed no significant differences in the use of red blood cell transfusions and no clinically meaningful changes in hemoglobin concentrations.

Although earlier trials by this group did show a benefit in this regard, it should be remembered that in one trial1 erythropoietin decreased the need for transfusion by a very modest one unit over the course of admission, a difference of questionable clinical benefit. Supporters of erythropoietin use might point to the mortality benefit demonstrated in the trial described above, but the observed difference was largely restricted to the trauma patients and further work is necessary to confirm what can best be viewed as a preliminary result, particularly in light of a recent meta-analysis which showed no mortality benefit to erythropoietin administration.2

There is also reason for concern on the risk side of the equation. Earlier trials have shown evidence of increased thrombotic complications with higher doses of erythropoietin and the study by Corwin and colleagues now shows that even with administration of lower doses, there is still an increased risk of thrombotic events, particularly in those patients not on heparin prophylaxis.

When viewed in conjunction with the questionable clinical benefit, the increased risk of thrombotic events should prompt reconsideration of erythropoietin use in the care of critically ill patients. Future work may yet demonstrate further evidence of a mortality benefit as well as the reasons for such an effect, but until this work is complete, routine use of this medication is not warranted.


  1. Corwin HL, et al. EPO Critical Care Trials Group. Efficacy of recombinant human erythropoietin in critically ill patients: a randomized controlled trial. JAMA. 2002;288:2827-2835.
  2. Zarychanski R, et al. Erythropoietin-receptor agonists in critically ill patients: a meta-analysis of randomized controlled trials. CMAJ. 2007; [Epub ahead of print].