Dialysate potassium and risk of death in chronic hemodialysis patients

• Published in: Journal of nephrology Vol. 23; no. 1; p. 33
• Main Authors: Al-Ghamdi, Ghassan, Hemmelgarn, Brenda, Klarenbach, Scott, Manns, Braden, Wiebe, Natasha, Tonelli, Marcello
• Format: Journal Article
• Language: English
• Published: Italy 01.01.2010
• Subjects: Aged; Cohort Studies; Dialysis Solutions - adverse effects; Dialysis Solutions - therapeutic use; Female; Humans; Inflammation - blood; Inflammation - epidemiology; Kidney Failure, Chronic - mortality; Kidney Failure, Chronic - therapy; Male; Malnutrition - blood; Malnutrition - epidemiology; Middle Aged; Potassium - adverse effects; Potassium - blood; Potassium - therapeutic use; Prospective Studies; Registries; Renal Dialysis - methods; Retrospective Studies; Risk Factors
• ISSN: 1121-8428

Few data guide the prescription of dialysate potassium (dK) in hemodialysis, which is usually prescribed empirically on the basis of predialysis serum potassium levels. This was a retrospective cohort study of prospectively collected data. We studied all patients initiating chronic hemodialysis in the Northern Alberta Renal Program (NARP) between January 2001 and December 2006. Data on demographic, clinical and treatment characteristics as well as the dates of death or transplant were extracted from the NARP database. We aimed to examine the relation between dialysate potassium level and all-cause death. During the study, 515/1,267 of patients (41%) died. The frequency of dK of 0 or 1 mEq/L, 2, 3 and 4 mEq/L was 6%, 40%, 51% and 3%, respectively. In our base model, which considered dK as a categorical exposure, the hazard ratios associated with 0 or 1 mEq/L, 2, 3 and 4 mEq/L were 1.13 (95% confidence interval [95% CI], 0.78-1.63), 1 (referent), 1.29 (95% CI, 1.07-1.56) and 1.74 (95% CI, 1.09-2.77), respectively. When markers of inflammation or malnutrition were adjusted for separately, the association between dK and mortality was attenuated but remained significant. After simultaneous adjustment for markers of inflammation and malnutrition, the risk of death associated with the higher dK categories was attenuated, and the overall trend was eliminated. Analyses using dK as a time-varying covariate found similar results. Although unadjusted and partially adjusted models suggested a graded association between higher dK and the risk of all-cause death, this association was apparently due to confounding by factors suggesting malnutrition and inflammation. The relative paucity of data on the association between dK and clinical outcomes despite the biological importance of potassium suggest that further studies are needed.