Hemoglobin and Clinical Outcomes in Hemodialysis: An Analysis of US Medicare Data From 2018 to 2020

Rationale & Objective: Anemia management in patients treated with maintenance dialysis remains a challenge. We sought to update information in this area by evaluating the association between hemoglobin and various outcome and utilization measures using data-rich Medicare sources. Study Design: Observational cohort study using data from the Consolidated Renal Operations in a Web-enabled Network and Medicare claims. Setting & Participants: We studied 371,250 prevalent patients treated with hemodialysis, covering 3,326,072 patient-months in 2019. Exposure: Monthly patient hemoglobin concentrations. Outcomes: We examined several outcomes, including mortality, all-cause hospitalization, cause -specific hospitalization, and emergency department utilization in the month following the exposure measurement. Analytical Approach: For each monthly observa-tion period, we calculated unadjusted and adjusted (for demographics and comorbid condition) hazard ratios using Cox regression. Results: The hemoglobin concentration was <10.5 g/dL for 40% of observations. We found an inverse association between mortality and hemoglobin measured over a range from <9 g/dL (HR, 2.53; 95% CI, 2.45-2.61; P < 0.0001, refer-ence = 10.5-11 g/dL) to 11-11.5 g/dL (HR, 0.92; 95% CI, 0.89-0.9 6; P < 0.0001). Mortality risk started to increase at hemoglobin levels >11.5 g/dL. All-cause hospitalization, cause -specific hospitalization (including cardiovascular, infection, and several subcategories including coronavirus disease 2019 hospitalization), and emergency department utilization were inversely associated with hemoglobin concentration, with risk reduction stabilizing at hemoglobin levels of approximately 11.5-12 g/dL and higher. Limitations: As with prior observational studies, the observed associations are not necessarily causal. Conclusions: In a large US hemodialysis population, there were better clinical outcomes at higher hemo-globin concentrations over short exposure and follow-up periods, consistent with other observational studies that generally used longer exposure and follow-up times. Mortality risk increased at hemoglobin concentrations >11.5 g/dL, consistent with findings from erythropoiesis-stimulating agent clinical trials. The apparently beneficial short-term effects associated with higher hemoglobin concentrations suggest that hemoglobin measurements capture unmeasured elements of patient risk.