Response to: Coyne DW et al. (2008) Comment on “Does ferric gluconate lower epoetin requirements in hemodialysis patients with high ferritin levels?” Nat Clin Pract Nephrol [doi:10.1038/ncpneph0965]

Original article: Agarwal R et al. (2008) Does ferric gluconate lower epoetin requirements in hemodialysis patients with high ferritin levels? Nat Clin Pract Nephrol 4: 418–419

Coyne et al. defend the use of the statistical methods used in their paper reporting the results from the Dialysis Patients' Response to IV Iron with Elevated Ferritin (DRIVE) II study. They clarify that they used analysis of covariance (ANCOVA)—not an unpaired t-test—to analyze the epoetin doses, which is encouraging. They state that the primary interest of DRIVE II was the epoetin dose at 12 weeks and not the trajectory of change. Why this should be so is unclear; to me, the journey is just as important, if not more so, than the destination.

As discussed elsewhere, analysis of variance (ANOVA) is the traditional model; it is easy to understand and simple to interpret, but it may not be valid in complex real-world situations.1 In the context of taking many measurements from the same patient, problems with the assumption of sphericity, unit of analysis, lack of consideration for different types of change, and missing data are often encountered. Multilevel modeling, a newer and more sophisticated method for the analysis of change, overcomes these limitations and provides a better framework for understanding the true nature of change without making assumptions about missing data. Mixed models should replace the last observation carried forward method as the default method in the analysis of longitudinal data.2

I agree that epoetin responsiveness is associated with death and that iron can improve epoetin responsiveness. However, whether improving epoetin responsiveness will result in a reduced number of deaths is not known. Coyne et al. propose, and I agree, that longer randomized controlled trials assessing the safety of intravenous iron in relationship to the benefits of sparing epoetin and raising hemoglobin are now urgently needed.3