Normalization of clearances in peritoneal dialysis using a formula for body water derived from an end-stage renal disease population

Objective:To compare body water (V) estimates from the Chertow formula (V-C ), which was derived in an endstage renal disease population, to V estimate s from the Watson formulas (V-W) in continuous ambulatory peritoneal dialys is (CAPD) patients. To identify CAPD patients in whom V-C is preferred to V -W for clearance studies. Design: Retrospective analysis of clearance studies. Setting: Dialysis units of four academic medical centers. Participants: 302 subjects on CAPD. Intervention: 613 clearance studies by standard methods. Main Outcome Measures: Comparisons between V-C and V-W, and between urea cl earance normalized by V-C [(Kt/V-C)(ur)] and V-W [(Kt/V-W)(ur)]. Results: V-C exceeded V-W by 3.5 +/- 1.6 L (p < 0.001), or 9.6% on average. This degree of overestimation of V-W is in the range of body water estimat es found in CAPD subjects with severe volume overload (> 5% of body weight) in previous studies. Total (Kt/V-W)(ur) exceeded total (Kt/V-C)(ur) by 8.6 %. By linear regression, V-C = -0.589 + (1.112 x V-W), r = 0.983. V-W excee ded V-C in only 12 studies. Young age, short height, low body weight, and l ow prevalence of diabetes characterized the studies with V-W > V-C. Total ( Kt/V-W)(ur) was adequate (greater than or equal to 2.0 weekly) in 276 studi es. Among these, 74 studies had inadequate total (Kt/V-C)(ur) (<2.0 weekly) . By logistic regression, the predictors of inadequate (Kt/V-C)(ur), when ( Kt/V-W)(ur) was adequate, included the presence of diabetes, great height, and long duration of CAPD. Conclusions: V-C provides estimates of body water exceeding those provided by V-W in a great majority of CAPD patients. Consequently, approximately 25 % of the clearance studies that are adequate when V-W is used as the normal izing parameter may be inadequate when V-C is used. V-C may provide a more appropriate estimate of body water than V-W in CAPD patients with volume ov erload.