Development and validation of a predictive model for cognitive impairment in patients undergoing maintenance hemodialysis: A retrospective, cross-sectional study

Cognitive impairment is a common and clinically relevant complication among patients undergoing maintenance hemodialysis. We developed and internally validated a simple risk score based on routinely available clinical and laboratory parameters to identify prevalent cognitive impairment in this population. In a retrospective, cross-sectional study of 260 stable adult patients from 2 dialysis centers, cognitive impairment was defined by a Montreal Cognitive Assessment score < 26. Multivariable logistic regression identified 6 independent predictors: older age, lower educational attainment, history of cerebrovascular disease, higher serum uric acid, lower serum calcium, and lower serum sodium. The model showed strong discrimination (area under the receiver operating characteristic curve 0.86; 95% confidence interval 0.81-0.90) and acceptable calibration (Hosmer-Lemeshow P = .48). We derived a simplified 0-20 point score that stratified risk from 8.5% in the low-risk group to 89.1% in the high-risk group (P for trend < .001). Higher scores were concurrently associated with greater interdialytic weight gain, poorer phosphorus control, increased falls, and longer hospital stays (all P < .001), indicating broader vulnerability beyond cognition alone. This pragmatic 6-item score may help clinicians flag patients who warrant targeted cognitive monitoring and multidisciplinary support in busy dialysis settings. External validation across diverse cohorts is needed before clinical implementation.