Relative risk of death in UK haemodialysis patients in relation to achieved haemoglobin from 1999 to 2005: an observational study using UK Renal Registry data incorporating 30,040 patient-years of follow-up

Abstract

Much controversy has been generated in recent times over the optimal target haemoglobin in chronic kidney disease patients receiving erythropoiesis-stimulating agent therapy. This has arisen from the paradoxical conclusions obtained from large retrospective or epidemiological studies versus interventional randomized controlled trials. Data from haemodialysis patients in the UK Renal Registry from 1999 to 2005 were analysed year by year for the relative risk of death at different haemoglobin concentrations, compared with a reference of 10-11 g/dl. The population size varied from 2291 in 1999 to 8209 in 2005. The data were analysed by chi-square tests, and a multivariate analysis was performed. Across the years 1999 to 2005, there was a consistent relationship between the haemoglobin achieved and the risk of death (P < 0.0001). In 2005, the relative risk (RR) of death = 1.32 for a haemoglobin (Hb) of 9-10 g/dl; RR = 0.44 for Hb > 13 g/dl. The relationship between Hb and the RR of death is nevertheless remarkably consistent across the 7 years of study, with an S-shaped correlation (polynomial) between an Hb range of <9 g/dl and an Hb range of >13 g/dl (P < 0.0001). Multivariate analysis also showed age, time on dialysis and diabetes to be strongly predictive of death across all 7 years analysed (P < 0.0001 in all cases). There is a significant relationship between achieved haemoglobin and mortality across the 7 years analysed, with no increase in risk seen with higher Hb levels.