The post: pre dialysis plasma urea nitrogen ratio to estimate K·t/V and NPCR: validation

Abstract

The ability of empirical formulae to predict K· t/V based on the ratio (R) of the postdialysis (C_t) to predialysis (C₀) plasma urea nitrogen levels was tested. In 256 patients, 336 three-point modeling sessions were performed. The K· t/V and NPCR were derived by in-terative solution of equations for V and G according to the variable volume single-pool model. The modeled K· t/V values were compared to K· t/V predicted from the formula: K· t/V= - In (R - 0.008·t - UF/W), where R is the C_t/C₀ ratio, t the session length (h), UF the ultrafiltrate volume (liters) and W the postdialysis weight (kg). Further, the ratio-derived K· t/V was used in conjuction with the Gotch nomogram for 3/week dialysis to estimate NPCR (NPCR-R/N); the latter value was compared to the NPCR from standard 3-point modeling (NPCR-MOD).

The two K· t/V values were quite similar, although statistically separable: modeled K· t/V, 0.97±0.22 (SD), ratio-derived K·t/V 0.96±0.23, p<0.001. The mean percent error was -0.73%±2.5 (range - 11 to +4.8), and the correlation coefficient was 0.994, slope 1.01, int - 0.016. Modifications of the prediction formula which incorporated UF/V instead of UF/W, and/or which weighted the UF/V term according to the expected K· t/V, produced only slight improvement in accuracy. The two values of NPCR were less similar, but still highly correlated: NPCR-MOD, 1.04± 0.26 g/kg/day NPCR-R/N, 0.94± 0.23, p< 0.001; mean percent error, -8.2± 16; r=0.78, slope = 0.68, int = 0.23.

The results suggest that, in a routine 3/week dialysis population, the K· t/V can be predicted with great accuracy from R, t, UF and W. Using the midweek C₀ value, the NPCR can be derived as well. (Int J Artif Organs 1989; 12:420-7)