A prescription-based model for predicting post-hemodialysis urea and electrolytes: a real-world validation study in acute hemodialysis

Abstract

Background: Urgent-start hemodialysis often involves severe biochemical abnormalities but is associated with a high risk of complications. Currently, no tool has been validated to predict post-hemodialysis urea and electrolytes in acute settings from prescription data alone. Methods: In this ambispective, two-phase study, we developed and validated a model to predict post-hemodialysis blood urea nitrogen and electrolytes using only prescription parameters in patients with urgent hemodialysis indications and those at risk for dialysis disequilibrium syndrome, severe hyponatremia, or hyperkalemia. The model was validated through statistical assessment of correlation and agreement. The model was integrated into a free, user-friendly web application (Adequator app®). Results: The development cohort included 303 treatments in 42 chronic hemodialysis patients. Dialyzer clearance, which was calculated via formal urea kinetic modeling, demonstrated excellent correlation and agreement with the proposed model (r = 0.99; 95% CI: 0.99 to 0.99) and a bias of 0.23 ± 1.2 ml/min (95% CI: −2.17 to 2.63). The validation cohort included 44 urgent hemodialysis sessions in 30 patients with chronic kidney disease or acute kidney injury and severe electrolyte abnormalities or at risk of dialysis disequilibrium syndrome. The predicted post-hemodialysis, blood urea nitrogen, sodium, and potassium levels strongly correlated with the measured values: r = 0.96 (95% CI: 0.93 to 0.98), r = 0.96 (95% CI: 0.94 to 0.98), and r = 0.84 (95% CI: 0.73 to 0.91), respectively. Agreement was also high, with biases of 0.4 ± 9.4 mg/dL (95% CI: −19.0 to 18.1) for blood urea nitrogen, −0.6 ± 1.6 mEq/L (95% CI: −3.8 to 2.5) for sodium, and 0.17 ± 0.4 mEq/L (95% CI: 0.6 to 0.9) for potassium. Among patients at risk of dialysis disequilibrium syndrome, 92% achieved the target blood urea nitrogen reduction (< 40%) without severe neurological events. The sodium correction remained below 6 mEq/L, and potassium normalized in all the cases. Conclusions: A prescription-based model can reliably predict post-Hemodialysis urea and electrolyte values in urgent dialysis settings. The Adequator App HD-Predictor (https://adequatorapp.com/hd-predictor) offers a non-invasive, accessible tool to guide individualized prescriptions, enhancing safety and precision in high-risk clinical scenarios.