Mortality in incident haemodialysis patients: time-dependent haemoglobin levels and erythropoiesis-stimulating agent dose are independent predictive factors in the ANSWER study

Background. Although the association between low haemoglobin levels and mortality is well established in haemodialysis patients, data are conflicting regarding levels >12 g/dl. In addition, divergent results have been reported on the relation between erythropoiesis-stimulating agents (ESAs) and mortality. Methods. This was a multicentre, observational, prospective, 24-month study, which recruited Spanish incident haemodialysis patients (N = 2310). Univariate and multi-variate time-dependent Cox regression models examined the longitudinal association of mortality with haemoglobin and ESA dose; adjustment was made for iron deficiency and other confounders. Results. After adjusting for age, functional status, body mass index, albumin levels, catheter as vascular access, previous history of cardiovascular disease, neoplasia, and ESA dose, mortality decreased with increasing haemoglobin. Adjusted hazard ratios relative to the reference category (11-12 g/dl) and 95% confidence intervals were: 1.36 (1.01-1.86) for <= 10 g/dl, 1.03 (0.75-1.42) for 10-11 g/dl, 0.93 (0.68-1.26) for 12-13 g/dl and 0.69 (0.49-0.97) for >13 g/dl. Independent of haemoglobin, patients on sustained ESA doses of 1-4000 IU/week and 8001-16 000 IU/week had better survival than non-treated (reference) patients, with adjusted hazard ratios of 0.61 (0.41-0.90) and 0.68 (0.49-0.94), respectively. No significant difference was found for doses of 4001-8000 IU/week or >16 000 IU/week, adjusted hazard ratios of 0.87 (0.63-1.20) and 0.89 (0.63-1.28), respectively. Conclusions. Higher haemoglobin levels are associated with lower mortality in Spanish incident haemodialysis patients, regardless of ESA dose, comorbidity, vascular access or malnutrition. No increase in mortality occurs for high ESA doses, independent of haemoglobin levels.