Measurement of total body water and urea kinetic modelling in peritoneal dialysis

Studies of the effect of Kt/V (urea) on prediction of outcome in patients on peritoneal dialysis have shown conflicting results. We performed this study to examine the effects of the measurement of V by varying techniques on the calculation of Kt/V, using body water estimated by deuterium oxide dilution (D2O dilution) as the criterion method for estimation of V. Studies were performed in 20 peritoneal dialysis patients. Kt was calculated from 24-hour dialysate and urine collections and V estimated by D2O dilution, Watson formulae, 58% of body weight, bioelectrical impedance (BIA) and 73% of fat-free mass estimated by DEXA. V was also measured in 35 healthy controls. Hydration, expressed as body water by D2O dilution as a percentage of fat-free mass estimated by DEXA did not differ between peritoneal dialysis patients 71.0 (4.9)% and a healthy control group 71.1 (5.0)%. Mean Kt/V using D2O dilution was 2.14 (0.36). The other techniques resulted in a significantly lower Kt/V; Watson equations 2.01 (0.35), p < 0.005, BIA 1.93 (0.31), p < 0.0001, DEXA 2.06 (0.28), p < 0.05, 58% body weight 1.83 (0.38), p < 0.0001. Limits of agreement of Kt/V by the simpler techniques compared with D2O dilution [mean difference of (other techniques -D2O dilution) as % of mean values +/- 95% limits of agreement] were Watson equation -5.9 +/- 15.3%, BIA -10.1 +/- 15.5%, DEXA -3.4 +/- 13.5% and 58% body weight -9.9 +/- 23.5%. Differences in Kt/V from estimates using D2O dilution were significantly negatively correlated with body fat for 58% body weight (r = -0.80, p < 0.0001) and the Watson formulae (r = -0.49, p < 0.05) but not for BIA or DEXA. We conclude that clinically significant variation in Kt/V may occur due to the estimation of V and may account for the uncertainty of the value of Kt/V as a predictor of outcome in peritoneal dialysis patients. Estimating V by BIA and DEXA did not have any benefit over the Watson formulae in terms of agreement with D2O dilution, though did avoid systematic errors related to body fat. Estimation of V as a fixed proportion of body weight is clearly inferior to the other techniques.