Lack of precision of indirect estimates of body water affects urea kinetic analysis in chronic peritoneal dialysis.

To test the precision of estimates of body water and urea clearance in peritoneal dialysis (PD), we compared, in 925 PD patients who underwent formal urea kinetics studies, estimates of V and Kt/V urea obtained by the use of the Watson, Hume, and Sahlgrenska anthropometric formulas and two novel formulas, one (Vcreat) computed using fat-free mass (FFM) estimated from creatinine kinetics as 0.73 x FFMcreat, and the other (VBMI) calculated as 0.73 x FFM(BMI) where FFM(BMI) was obtained by the Gallagher formula, which estimates body composition as a function of body mass index (BMI). Comparisons by twos were performed using the paired t-test and the Wilcoxon sign rank test with the Bonferroni correction for multiple (n=10) comparisons. The results for V (liters) were Watson, 36.7 +/- 7.1; Hume, 37.3 +/- 7.3; Sahlgrenska, 36.8 +/- 7.6; Vcreat, 32.2 +/- 9.8; and VBMP 37.2 +/- 7.8. With the exception of V(BMI) and V(Hume) which did not differ, all other values differed (p < 0.001) from one another regardless of whether a parametric or nonparametric comparison was performed. The results for weekly total Kt/V urea were Watson, 2.05 +/- 0.57; Hume, 2.03 +/- 0.57; Sahlgrenska, 2.06 +/- 0.59; from Vcreat 2.42 +/- 0.71; and from V(BMP) 2.03 +/- 0.58. All of those values differed from one another (p < 0.001) by both methods of comparison. Using cut-off values (1.50, 1.75, and 2.00) as indices of adequate total weekly Kt/V urea, the discrepancies between any two estimates by the five studied formulas varied in the range 1.1% - 34.2%. Despite numerically close mean values, estimates of V based on various anthropometric formulas differ substantially and cause substantial discrepancies in the classification of Kt/V urea as inadequate or adequate. This lack of precision, added to the known lack of accuracy of the estimates, confounds the interpretation of the clinical relevance of urea kinetic estimates in PD.