Trial Design and Oversight
We conducted this prospective, randomized, open-label, blinded end-point,14 controlled trial at 50 sites in the United Kingdom. The trial protocol15 (available with the full text of this article at NEJM.org) was approved by the relevant health authorities and institutional review boards, and all the patients provided written informed consent. An independent data and safety monitoring committee performed regular safety surveillance. Data were entered into an electronic case-report form by the investigators (see the Supplementary Appendix, available at NEJM.org) and were analyzed at the Robertson Centre for Biostatistics, University of Glasgow, in the United Kingdom.
This was an academic investigator–led trial. The trial was funded by Kidney Research UK, which was supported by an unrestricted grant from Vifor Fresenius Medical Care Renal Pharma (which also provided iron sucrose for the trial, free of charge). Vifor Fresenius Medical Care Renal Pharma had no input into the trial design or the data collection or analysis. However, the company was kept abreast of the progress of the trial by regular study reports and newsletters. No confidentiality agreements regarding the data were in place.
The initial draft of the manuscript was written by the first author and revised by all the authors. Medical writing assistance was provided by a professional medical writer, funded by Kidney Research UK (supported by Vifor Fresenius Medical Care Renal Pharma). The authors had access to the final trial results and take responsibility for the accuracy and completeness of the data, for the fidelity of the trial to the protocol, and for the decision to submit the manuscript for publication.
Adults with end-stage kidney disease in whom maintenance hemodialysis had been initiated no more than 12 months before the initial screening visit, who had a ferritin concentration of less than 400 μg per liter and a transferrin saturation of less than 30%, and who were receiving an erythropoiesis-stimulating agent were eligible to participate. Any iron therapy that had been prescribed previously was discontinued at the screening visit. The full eligibility criteria are provided in the protocol.
Randomization, Treatment, and Follow-up
Using a Web-based randomization system, we randomly assigned participants, in a 1:1 ratio, to receive a regimen of high-dose intravenous iron administered proactively or a regimen of low-dose intravenous iron administered reactively; patients were then evaluated monthly. Randomization was stratified according to vascular access (dialysis catheter vs. arteriovenous fistula or graft), diagnosis of diabetes (yes vs. no), and duration of hemodialysis treatment (<5 months vs. ≥5 months).
The ferritin concentration and transferrin saturation were measured monthly (usually during the first week of the month), and these values determined the monthly dose of iron sucrose to be administered intravenously during the subsequent week of hemodialysis (usually the second week of the month). In the high-dose group, 400 mg of iron sucrose per month, to be administered intravenously, was prescribed to the patients, with safety cutoff limits (ferritin concentration of 700 μg per liter or a transferrin saturation of 40%) above which further intravenous iron administration was withheld pending repeat testing 1 month later. Patients in the low-dose group received a monthly dose of 0 mg to 400 mg of iron sucrose as required to maintain a minimum target ferritin concentration of 200 μg per liter and a transferrin saturation of 20%, in line with accepted clinical guidelines (for details of the iron-dosing regimen, see the Supplementary Appendix). Iron therapy was temporarily withheld if the trial team identified an active infection that was deemed by the investigator to be sufficient to contraindicate the use of intravenous iron. Therapy was restarted when it was judged by the investigator to be safe to do so.
Clinicians selected the dose of erythropoiesis-stimulating agent that would be sufficient to maintain a hemoglobin level of 10 to 12 g per deciliter.16 Apart from the dose of erythropoiesis-stimulating agent, the trial teams treated patients according to standard practice.
Trial End Points
The primary end point was the composite of nonfatal myocardial infarction, nonfatal stroke, hospitalization for heart failure, or death from any cause, assessed in a time-to-first-event analysis; definitions of the end-point events are provided in the Supplementary Appendix. The first secondary end point consisted of the components of the primary end point, including first and repeat events, which were analyzed as recurrent events. Other secondary efficacy end points included death from any cause; the composite of fatal or nonfatal myocardial infarction, fatal or nonfatal stroke, or hospitalization for heart failure; and each of the three subcomponents of that end point, all assessed in time-to-first-event analyses. An independent committee whose members were unaware of the trial-group assignments adjudicated these events according to prespecified criteria. Additional secondary efficacy end points included the dose of erythropoiesis-stimulating agent, the incidence of blood transfusion, and two quality-of-life measures (the European Quality of Life–5 Dimensions [EQ-5D] questionnaire and the Kidney Disease Quality of Life instrument).
Safety end points included vascular access thrombosis, hospitalization for any cause, and hospitalization for infection, each assessed in a time-to-first-event analysis, and the rate of episodes of infection. Laboratory tests, including the hemoglobin level, serum ferritin concentration, and transferrin saturation, were assessed monthly. Data on serious adverse events were collected prospectively, and events were coded with the use of the Medical Dictionary for Regulatory Activities (MedDRA), version 15.1. Data on nonserious adverse events, other than infection and vascular access thrombosis, were not collected.
In the initial sample-size calculations, we assumed a 3-year event rate of 40% in the low-dose group and a 10% loss to follow-up (including loss to follow-up due to kidney transplantation). Thus, we estimated that a sample of 2080 patients who had 631 primary end-point events would provide the trial with 80% power to assess the noninferiority of high-dose iron to low-dose iron, with a noninferiority limit for the hazard ratio of 1.25.
Summary statistics are provided as numbers and percentages, as mean values with standard deviations, and as median values with interquartile ranges (25th to 75th percentile). Treatment effects were estimated as the effect in the high-dose group as compared with (or minus) the effect in the low-dose group, with adjustment for the stratification variables at randomization. The primary end point was analyzed first in terms of noninferiority in the intention-to-treat population, which included all the patients who had undergone randomization validly, with a supporting analysis in a per-protocol population that excluded patients with a major protocol violation. Analyses were censored at the date of kidney transplantation, withdrawal of consent, loss to follow-up, or transfer to home or peritoneal dialysis, whichever came first. Noninferiority was also assessed in a sensitivity analysis that included only patients who were currently receiving treatment, with data censored after patients discontinued the trial drug. The time-to-first-event analyses were conducted with the use of cause-specific Cox proportional-hazards models, including the stratification variables and the treatment variable. The noninferiority analysis tested the null hypothesis that the hazard ratio for the treatment effect was at least 1.25 against the alternative that the hazard ratio was less than 1.25, with a required one-sided significance level of 0.025. If noninferiority was established, a two-sided superiority test (Wald statistic) was carried out with no penalty regarding the P value.
The incidence of death from any cause and a composite of myocardial infarction, stroke, or hospitalization for heart failure as recurrent events was analyzed with the use of the proportional-means model of Lin et al.17 and described in the form of mean frequency functions (method of Ghosh and Lin).18 Other statistical methods and details regarding statistical assumptions are described in the Supplementary Appendix. The results for the secondary end points are reported as point estimates and 95% confidence intervals with no adjustment for multiple comparisons, so the confidence intervals should not be used to infer definitive treatment effects. The cumulative doses of intravenous iron were compared between the treatment groups with the use of Wilcoxon rank-sum tests. The statistical analysis plan is available with the protocol at NEJM.org.