Calculations of mKt/V and eKt/V provide complementary measurements of dialysis dose from the perspectives of the dialyzer and patient, respectively. The optimal method for calculating these indices remains unclear.OBJECTIVE
To identify the most accurate methods for calculating mKt/V and eKt/V.DESIGN AND INTERVENTION
Stable end-stage renal disease patients who were undergoing hemodialysis three times weekly were chosen at random for this cross-sectional study. Blood urea nitrogen (BUN) levels were measured before the midweek dialysis session (BUN1), 1-2 min afterwards (BUN2) and 30 min afterwards (BUN3). Dialysate flow rate was set at 500 ml/min. The cohort was divided into anuric patients whose BUN2 was measured 1 min or 2 min postdialysis (groups 1 and 2, respectively) and oliguric patients whose BUN2 was measured 2 min postdialysis (group 3). Reference values for mKt/V and eKt/V calculated by use of two-pool and single-pool urea kinetic modeling (UKM), respectively, were compared with estimates obtained using 16 blood-based single-pool methods (3 iterative methods-nspUKM, nspUKM2m and spUKM-and 13 noniterative methods). Details of some of the methods used are provided as Supplementary Information online.OUTCOME MEASURES
Accuracy of the methods was expressed as limits of agreement. Sensitivity of the methods to the effects of residual diuresis, time of postdialysis blood sampling and between-group variation was expressed as an 'intergroup sensitivity parameter' (SM).RESULTS
In total, samples from 98 patients were analyzed. The reference value of eKt/V was less than that of mKt/V in all cases, and the magnitude of this difference increased as dialysis dose increased. The best method for calculating mKt/V was nspUKM, as indicated by the narrow limits of agreement (−0.08 ± 0.58%) and the low sensitivity of the limits of agreement to variation between groups (SM −0.58 and −0.12 for group 2 vs group 1; 1.55 and 0.30 for group 3 vs group 1). When methods requiring a BUN3 measurement were excluded, nspUKM2m had the best accuracy (limits of agreement −0.03 ± 1.44%), low sensitivity (SM 1.21 and 0.11 for group 2; 0.75 and 0.08 for group 3), and its SD showed a weaker association with mKt/V than the SD of spUKM. Among the noniterative formulae, the Prado 1 equation (which requires BUN3) and the Prado 2 equation (which does not require BUN3) were the most accurate predictors of mKt/V (limits of agreement −1.65 ± 1.26% and −1.93 ± 2.09%, respectively). Their SDs showed no significant correlation with mKt/V and all SM values for these methods were <1. Of the formulae used to estimate eKt/V (all noniterative), the Daugirdas 1 equation had the greatest accuracy among those requiring BUN3 (limits of agreement −2.42 ± 1.05%; all SM values <1). Of the methods not requiring BUN3, the Daugirdas 2 equation most closely approximated the reference eKt/V (limits of agreement 1.74 ± 0.79%), with SM values of −1.34 and −0.17 for group 2 and −1.26 and −0.06 for group 3. The commonly used Lowrie formula underestimated both indices by around 16%, but with low SD.CONCLUSION
To calculate mKt/V, either the Prado 1 or the Prado 2 equation should be used depending on whether or not BUN3 is known; similarly, eKt/V should be calculated using the Daugirdas 1 or Daugirdas 2 equations.