BACKGROUND: Epoetin alfa and darbepoetin alfa are erythropoiesis-stimulating agents (ESAs) indicated for the treatment of anemia in chronic renal failure, including patients on dialysis and patients not on dialysis. Clinical experience demonstrates that the dose conversion ratio (DCR) between epoetin alfa and darbepoetin alfa is nonproportional across the dosing spectrum. However, previous calculations of the dose relationship between epoetin alfa and darbepoetin alfa, described in previous work as the "dose ratio" (DR), (a) used cross-sectional designs (i.e., compared mean doses for patient groups using each ESA) and were therefore vulnerable to confounding or (b) did not adjust for the nonproportional dose relationship. DRs reported in the literature range from 217:1 to 287:1 epoetin alfa (Units [U]):darbepoetin alfa (micrograms [micrograms]). Payers may need a single DCR that accounts for the nonproportional dose relationship to evaluate the economic implications of converting a nondialyzed patient population with chronic kidney disease (CKD) from epoetin alfa to darbepoetin alfa. OBJECTIVE: To estimate a single mean maintenance DCR between epoetin alfa and darbepoetin alfa in subjects with CKD not receiving dialysis, using methods that take into account the nonproportional dose relationship between the 2 ESAs. METHODS: This was a post-hoc analysis of a subset of patients enrolled in an unpublished, open-label, single arm phase 3 clinical trial (ClinTrial. gov identifier NCT00093977) that was completed in 2006. Although the clinical trial enrolled both dialyzed and nondialyzed patients, the present study used a patient subset comprising nondialyzed patients with CKD previously receiving weekly or every-other-week (Q2W) epoetin alfa who were switched to Q2W darbepoetin alfa to maintain hemoglobin (Hb) levels between 11.0 and 13.0 grams per deciliter. A population mean DCR was estimated using 2 methods: (a) a regression-based method in which the log-transformed (natural logarithm) mean weekly darbepoetin alfa dose over the evaluation period of the study (weeks 25 to 33) was regressed on the log-transformed (natural logarithm) weekly epoetin alfa dose over the 2-week screening period; and (b) a mean ratio method in which the DCR was calculated for each individual patient and then averaged for the study population to give a population-level DCR. Sensitivity analyses estimated the DCR in various subgroups. RESULTS: Of 1,127 patients enrolled in clinical trial NCT00093977, 567 patients on dialysis were excluded. Of the remaining 560 patients, 104 received weekly or Q2W epoetin alfa, were switched to Q2W darbepoetin alfa, received at least 1 non-zero dose of darbepoetin alfa during the evaluation period, and were included in the DCR calculation for the present study. Analysis of the log-log plot for the regression-based method indicated 2 or more possible regression lines with separate slopes. However, based on our a priori analysis plan to estimate a single DCR for the patient sample, the estimated sample mean maintenance DCR in the regression analysis was 330.6 U epoetin alfa to 1 micrograms darbepoetin alfa. In the mean ratio analysis, the DCR was 375.6 U:1 micrograms. Sensitivity analyses in which DCRs were calculated for different subgroups with different baseline differences identified a variable DCR range of 302-380 U:1 micrograms. CONCLUSIONS: The methodology used in estimating the DCR accounts for the nonproportional dose relationship between epoetin alfa and darbepoetin alfa and may represent an advance over the methods used in previous research. The mean maintenance DCR between the 2 ESAs exceeds a threshold of 300 U:1 micrograms, which is greater than previously reported DRs. This methodology provides payers the means to compare ESA doses in CKD patients not receiving dialysis.