Modeling urea kinetics with two vs. three BUN measurements. A critical comparison.

Hemodialysis urea modeling usually requires three BUN measurements, C1, C2, and C3 (3-BUN), to determine urea generation rate (G) and volume of urea distribution (V) from dVC/dt = f(C, V, G). We eliminated the requirement for C3 by modifying the algorithm to generate a periodic solution to the equations describing C1 and C2 (2-BUN). During 156 consecutive dialyses in 37 patients there were no differences in V or G by the two methods: V = 38.3 +/- 1.3 L, G = 6.35 +/- 0.22 mg/min (2-BUN); V = 37.9 +/- 1.3 L, G = 6.08 +/- 0.23 mg/min (3-BUN). Theoretic analysis showed that transient departure from a constant protein intake produces greater errors in average G determined by the 3-BUN model than by the 2-BUN model. We compared predicted predialysis BUN values to measured predialysis BUN (mg/dl) at two days, one week and two months following the modeled dialysis. The 2-BUN method accurately predicted C3 in 156 studies: r = 0.90, mean error 6.8 +/- 6.0% (SD). Prediction of predialysis BUN at one week was less accurate for both methods: mean error 10.4 +/- 7.1% (2-BUN), 16.0 +/- 10.7% (3-BUN). At 2 months, the correlation between measured and calculated predialysis BUN fell to 0.59 (2-BUN) and 0.65 (3-BUN). Deviations from predicted BUN increased with time regardless of the model; no clear advantage of either method was seen at longer time intervals. Computer CPU time required to calculate V and G remained under 1 sec for the 2-BUN method.(ABSTRACT TRUNCATED AT 250 WORDS)