Neuroanesthesia and Intensive Care

Canadian Journal of Anesthesia

, Volume 51, Issue 6, pp 621-622

First online:

Best evidence in critical care medicine: The use of recombinant human erythropoietin to reduce red cell transfusions in critically ill patients

  • Osama AlmuslimAffiliated with
  • , David LeasaAffiliated with

Structured abstract

Clinical issue

Allogeneic red blood cell (RBC) transfusions are common in critical care practice. Can recombinant human erythropoietin (rHuEPO) administration reduce the need of RBC transfusions and hence the associated risks?

Methodology (internal validity)

Prospective randomized multicentre clinical trial.

Computer generated randomization stratified by site.

Six hundred and fifty patients were randomized to rHuEPO and 652 patients to placebo.

Allocation concealment was by randomization at the data coordinator centre.

Double-blind placebo controlled trial by using similar syringes.

Cointerventions were similar in both groups.

Demographic and baseline characteristics were similar. Outcome assessor awareness of group allocation was not clear.

About 3% of patients were lost to follow-up.

All patients were included in the final analysis using the intention-to-treat principle.

Intervention and outcomes (results)

Weekly sc injection of rHuEPO (40,000 units) or placebo was started on intensive care unit (ICU) day three for a total of three or four doses.

All outcomes were assessed at the end of the 28th day from the first dose.

Primary efficacy outcomes

Patients receiving rHuEPO were less likely to undergo allogeneic RBC transfusions (60.4% placebovs 50.5% rHuEPO;P < 0.001); number needed to treat was 10 and the OR 0.67 (95% confidence interval [CI], 0.54–0.83).

Secondary efficacy outcomes

  1. 1)

    There was a 19% reduction in the total units of RBCs transfused in the rHuEPO group (1,963 units for placebo vs 1,590 units for rHuEPO).

  2. 2)

    The increase in hemoglobin from baseline to study end was greater in the rHuEPO group (mean [SD], 1.32 [2] g·dL−1 vs 0.94 [1.9] g·dL−1;P < 0.001).

  3. 3)

    There was a reduction in RBC units transfused per day alive in the rHuEPO group (ratio of transfusion rates, 0.81; 95% CI, 0.79–0.83;P = 0.04).

  4. 4)

    Mortality was not different between groups (14% in rHuEPOvs 15% in placebo;P = .61).

  5. 5)

    Adverse clinical events were not significantly different.


Applicability (external validity)

Inclusion criteria: hematocrit less than 38% in adult patients on their third day of ICU admission if they were expected to have at least four days ICU stay.

Exclusion criteria: renal failure on dialysis, uncontrolled hypertension, seizures, acute burns, acute coronary syndrome and acute gastrointestinal bleed. The included patients represented about one third of most ICU patients.

The study could not detect clinically important outcomes (mortality and morbidity) likely because of inadequate power. Lack of any obvious adverse events could be due to short follow-up.

In spite of the high dollar cost of rHuEPO, consideration of other factors like blood availability, transfusion complications and long ICU stay costs should be kept in mind.


The administration of 40,000 units of rHuEPO per week in long stay critically ill patients reduces allogeneic RBC transfusion and increases hemoglobin concentration. However, no effects on mortality or morbidities could be found in this study.