Clinical evaluation of a model for prediction of end-dialysis systemic ionized calcium concentration in citrate hemodialysis.

Background/aim: Citrate anticoagulation in hemodialysis (HD) is increasingly drawing attention in the nephrology community. One of the major deterrents to a more widespread use are the monitoring requirements for fear of systemic calcium derangements. Means of accurately predicting systemic ionized calcium (iCa) may help to overcome this challenge. We have previously presented a mathematical model of regional citrate anticoagulation (RCA) to address this need. Here, we present a refined model and show results in an independent validation cohort of maintenance HD patients on Citrasate®, a calcium- and citrate-containing dialysate.

Methods: A hybrid RCA model was developed, comprising the previously published 'native' RCA model and a statistical correction based on levels of alkaline phosphatase as a marker of bone turnover. The model was validated in 120 patients on Citrasate, a dialysate containing 0.8 mmol/l citrate and 1.125 mmol/l calcium. Systemic iCa was measured at the beginning and end of one HD treatment in each subject. Serum iCa predictions were compared between our previously published model and the new hybrid model.

Results: On average, the hybrid model predicted end-HD systemic iCa with an error (predicted - measured) of 0.028 mmol/l, compared to -0.051 mmol/l with the previously published model. There were only 4 subjects out of the 120 analyzed in whom the prediction error was <-0.1 mmol/l, and only 6 in whom the error was >+0.1 mmol/l (max: +0.13 mmol/l).

Conclusion: This study demonstrates that the novel hybrid model is an improvement over the previously published model and that it is capable of predicting end-dialysis systemic iCa levels with improved accuracy and precision even in a citrate dialysis setting which was much different from the original derivation cohort.