A cost-effectiveness analysis of erthropoietin in ICU patients

Introduction

Recombinant human erythropoietin (rhEPO) has been shown to decrease the packed red blood cell (PRBC) transfusion requirement by about 20% among ICU patients admitted for longer than 48 hours. Despite this, its use has not been widely instituted, in part due to its cost.

Methods

A decision analytic Markov process was designed to model anemia management strategies for patients admitted to the ICU for > 48 hours. The model compared two strategies: standard care versus standard care plus administration of rhEPO, 40,000 U weekly. Probabilities, costs and outcomes were derived from longitudinal studies of ICU patients, current Canadian medical costs and a randomized trial of rhEPO versus placebo for ICU patients. The model measured ICU costs arising from administration of PRBC and rhEPO, long-term costs of infection and effectiveness in reducing infection. The base-case data included: rhEPO cost $439/40,000 U, PRBC cost $616/U, median ICU length of stay 4 days, overall probability of PRBC transfusion 0.6 per patient per admission and mean number PRBCs transfused 3 U per patient. Long-term costs of infection (hepatitis C or HIV) were estimated at $200,000 per infection. All costs are expressed in US dollars.

Results

The mean cost per patient using the rhEPO strategy was $1030 vs $1323 for no rhEPO. Fewer units of PRBCs were administered in the rhEPO arm, resulting in a lower infection rate. rhEPO use was therefore both less expensive and more effective (i.e. it was the dominant strategy). One-way sensitivity analyses revealed that administration of rhEPO remained the preferred strategy provided that the PRBC cost was greater than $186/U or that the rhEPO cost was less than $1455/40,000 U. These results can be generalized: provided that the cost of rhEPO < 2.4 × cost 1 U PRBC, the rhEPO strategy will be preferable. Further, the advantage of treatment with rhEPO was maintained down to an expected transfusion rate of 1.1 U PRBC per patient per admission. Decreasing the probability of infection or increasing the long-term cost of infection did not significantly affect the model.

Conclusion

The management of chronic anemia among ICU patients admitted > 48 hours using rhEPO appears to be cost-effective, provided that the rhEPO cost is less than 2.4 times the cost of 1 U PRBC. This effect is retained at low expected transfusion rates and with changes in the cost or risk of infection arising from blood administration.