Cognitive Impairment, Frailty, and Adverse Outcomes Among Prevalent Hemodialysis Recipients: Results From a Large Prospective Cohort Study in the United Kingdom

RATIONALE & OBJECTIVE: Frailty and cognitive impairment are common in hemodialysis recipients and have been associated with high mortality. There is considerable heterogeneity in frailty reporting, with little comparison between commonly used frailty tools and little exploration of the interplay between cognition and frailty. The aims were to explore the relationship between frailty scores and cognition and their associations with hospitalization and mortality. STUDY DESIGN: Prospective cohort study SETTING & POPULATION: Prevalent hemodialysis recipients linked to national datasets for hospitalization and mortality. PREDICTORS: Montreal Cognitive Assessment (MoCA), Frailty Phenotype, Frailty Index (FI), Edmonton Frailty Scale, and Clinical Frailty Scale (CFS) were performed at baseline. Cognitive impairment was defined as MoCA scores of <26, or <21 in dexterity impairment, <18 in visual impairment. OUTCOMES: Mortality, hospitalization. ANALYTICAL APPROACH: Cox proportional hazards model for mortality, censored for end of follow-up. Negative binomial regression for admission rates, censored for death/end of follow-up. RESULTS: In total, 448 participants were recruited with valid MoCAs and followed up for a median of 685 days. There were 103 (23%) deaths and 1,120 admissions of at least one night. Cognitive impairment was identified in 346 (77.2%) participants. Increasing frailty by all definitions was associated with poorer cognition. Cognition was not associated with mortality (HR, 0.99; 95% CI, 0.95-1.03; P = 0.41) or hospitalization (IRR, 1.01; 95% CI, 0.99-1.04; P = 0.39) on multivariable analyses. There were interactions between MoCA scores and increasing frailty by FI (P = 0.002) and Clinical Frailty Scale (P = 0.005); admissions were highest when both MoCA and frailty scores were high, and when both scores were low. LIMITATIONS: As frailty is a dynamic state, a single cross-sectional assessment may not accurately reflect its year-to-year variability. In addition, these findings are in maintenance dialysis and may not be transferable to incident hemodialysis. There were small variations in application of frailty tool criteria from other studies, which may have influenced the results. CONCLUSIONS: Cognitive impairment is highly prevalent in this hemodialysis cohort. The interaction between cognition and frailty on rates of admission suggests the MoCA offers value in identifying higher risk hemodialysis populations with both high and low degrees of frailty.