Developing a classification system for haemoglobin management in patients with end-stage renal disease on haemodialysis: a secondary data analysis

BACKGROUND Ongoing discussion on anaemia management and target haemoglobin (Hb) levels in patients on haemodialysis with erythropoietin treatment require a systematic approach in evaluating current practice. Aim of the present study was to develop a new classification system to easily monitor Hb trajectories and categorise patients on haemodialysis. METHODS Routine data from five dialysis centres in the USA collected between 2010 and 2016. Data were anonymised and only those from patients with fortnightly Hb values were included in the analysis. Entries on blood parameters and medication were standardised to achieve overall comparability. Data from each patient was grouped in periods of 120 days. Hb values above or below the target level of 10-12 g/dL were counted for each period. Periods were then assigned to Hb-classes according to the number of Hb values out of range per period: Hb-class I with 0-2, Hb-class II for 3-5 and Hb-class III for ≥6 values out of range. RESULTS Records from 3349 patients with fortnightly Hb values, information on haemodialysis data, laboratory parameters correlated to red blood cells and data on medication with erythropoiesis-stimulating agents (ESAs) were available. Patients were 64.4±15.9 years old; 55.0% were men. Statistical analysis revealed significant differences between Hb-classes in all of the examined parameters, except erythrocytes mean corpuscular volume and C reactive protein above the threshold, with more critical values in higher Hb-classes. The usage of ESAs showed a mean difference between Hb-class III and Hb-class I of 6.4 units/day and kilogram body weight in a 120-day period. CONCLUSION Our classification system allows an easily achievable overview of the patients' responsiveness and performance of Hb management. Integrated into a disease management programme or continuous quality improvement, the classification delivers an instant appraisal without complex statistical or mathematical processing.