Estimating Urea Clearance in Patients on Continuous Ambulatory Peritoneal Dialysis: A Multivariate Analysis

Abstract

The purpose of this study was to determine if Kt/V urea in continuous ambulatory peritoneal dialysis (CAPD) could be estimated by a multivariate model based upon simple clinical observations. The study included 439 clearance studies in 301 CAPD patients followed in 8 dialysis centers. Weekly urea clearance, 24 h urine volume and 24 h drain volume were normalized to body water by the formulae of Watson (Kt/V, UV/V and DV/V respectively). Adequate dialysis was defined as Kt/V > or = 2.0 weekly. Subjects at 2 units were used to derive the models, while others were used for model validation. Stepwise multiple linear regression was performed on the derivation set (DS) to identify the clinical variables that correlated with Kt/V. The model was then used to estimate Kt/V for the validation set (VS). In the DS, 110 clearance studies were performed in subjects with residual renal function. Multiple linear regression showed that weekly Kt/V was defined by the expression: Kt/V=1.48 + 24.1 (UV/V) + 2.92(DV/V) - 0.049 (serum creatinine) (r=0.750, p<0.001). In 204 VS studies, the correlation between estimated and measured Kt/V was 0.633. There were marked differences in the proportion of adequately dialyzed patients when Kt/V estimated from the formula shown was <2.0, between 2.0 and 2.3, and >2.3 weekly (7.9%, 54.7% and 79.7%, respectively; p<O.001). In the 33 studies done in DS anuric patients, regression analysis showed the following: Kt/V=0.46 + 2.59 (DV/V) + O.009(age) (r=0.562; p=0.003). In 92 VS studies in anuric subjects, there was strong correlation between estimated and measured Kt/V (r=0.740). Again, there were marked differences in the frequency of adequate dialysis in anuric patients with estimated Kt/V <2.0, between 2.0 and 2.3, and >2.3 weekly (8.1%, 68.8%, and 100%, respectively; p<0.001). The risk of low Kt/V can be estimated by multivariate linear models requiring only simple clinical measurements.

Prediction of Single‐Pool Kt/V Based on Clinical and Hemodialysis Variables Using Multilinear Regression, Tree‐Based Modeling, and Artificial Neural Networks

“…However, when the same variables were evaluated in a multivariate analysis, some were found to be good predictors of spKt/V. Similarly, in peritoneal dialysis patients, simple clinical observations were shown to predict Kt/V, when a multivariate linear regression technique was employed (14). A formula was derived that employed urine volume, body water, 24 h drain volume, age, and serum creatinine to predict weekly Kt/V.…”

The Prescription of Peritoneal Dialysis

“…In one study (6), none of the size indicators tested (weight, height, BSA, V ) had any effect on D / P Urea . In another study (24), multivariate analysis identified Dv/ V and daily urine volume as the only predictors of total Kt / V Urea above 1.70 weekly in CAPD patients.…”