Determining the adequacy of sodium balance in hemodialysis using a kinetic model

The importance of sodium balance avoiding intradialytic cardiovascular instability and interdialytic hypertension and pulmonary edema is well known. An early analytical single-pool kinetic model created to evaluate sodium balance in hemodialysis, using flame photometry to determine plasma and dialysate sodium concentrations, has been shown to have a level of imprecision of +/- 2.8 mEq/l in predicting end-dialysis sodium plasma water concentrations (NaPWt). The ionometric determination of sodium concentrations seems to be more accurate and refers to the activity of the sodium capable of crossing dialysis membranes. On the basis of the theoretical premises of the model mentioned above, we developed a computerized single-pool kinetic model which makes it possible to calculate the ionized dialysate sodium activity (NaDI) required to reach a pre-established target of end-dialysis blood sodium activity (NaBI). Thirty-seven non-diabetic and anuric patients undergoing regular thrice-weekly hemodialysis were given their usual dialysis treatment, with NaDI at the usual value for each patient (range 137-147 mEq/l) and kept constant throughout dialysis. At the beginning and end of the session, NaDI and NaBI were measured in quadruplicate by means of a Nova-1 device (Direct Potentiometry, Pabisch Instruments). The validity of this kinetic model was tested by considering the difference between predicted and observed (P-O) NaBI at the end of dialysis [(t)]. P-O NaBI(t) was -0.37 +/- 0.42 mEq/l, which was statistically different from 0 (p < 0.001). When P-O NaBI(t) was plotted against ONaBI(t), it was more negative at the higher values of ONaBI(t). P-O intradialytic sodium removal (Nag) was -12.5 +/- 17.8 mEq/session, which was also statistically different from 0 (p < 0.001). The imprecision of this kinetic model was less than 0.84 mEq/l, as estimated by doubling the SD of P-O NaBI(t) (0.42 = 0.84 mEq/l). Although the reasons for its inaccuracy especially at higher ONaBI(t) values remain to be clarified, these data are the expression of a satisfactory clinical model.