Phosphate removal model: An observational study of low‐flux dialyzers in conventional hemodialysis therapy

Abstract Precise assessing phosphate removal by hemodialysis ( HD ) is important to improve phosphate control in patients on maintenance HD . We reported a simple noninvasive model to estimate phosphate removal within a 4‐hour HD . One hundred sixty‐five patients who underwent HD 4 hours per session using low‐flux dialyzers made of polysulfone (1.2 m 2 ) or triacetate (1.3 m 2 ) were enrolled. Blood flows varied from 180 to 300 m L /min. Effluent dialysate samples were collected during the 4‐hour HD treatment to measure the total phosphate removal. Predialysis levels of serum phosphate, potassium, hematocrit, intact parathyroid hormone, total carbon dioxide ( TCO 2 ), alkaline phosphatase, clinical and dialysis characteristics were obtained. One hundred thirty‐five observations were randomly selected for model building and the remaining 30 for model validation. Total amount of phosphate removal within the 4‐hour HD was mostly 15–30 mmol. A primary model (model 1) predicting total phosphate removal was Tpo 4 = 79.6 × C 45 (mmol/ L ) − 0.023 × age (years) + 0.065 × weight (kg) − 0.12 × TCO 2 (mmol/ L ) + 0.05 × clearance (m L /min) − 3.44, where C 45 was phosphate concentration in spent dialysate measured at the 45 minute of HD and clearance was phosphate clearance of dialyzer in vitro conditions offered by manufacturer's data sheet. Since the parameter TCO 2 needed serum sample for measurement, we further derived a noninvasive model (model 2): Tpo 4 = 80.3 × C 45 − 0.024 × age + 0.07 × weight + 0.06 × clearance − 8.14. Coefficient of determination, root mean square error, and residual plots showed the appropriateness of two models. Model validation further suggested good and similar predictive ability of them. This study derived a noninvasive model to predict phosphate removal. It applies to patients treated by 4‐hour HD under similar conditions.