Prediction model of renal recovery by recursive partitioning analysis.

<p>Prediction model developed by recursive partitioning analysis to estimate the risk class of renal recovery based on two split variables: mode of extracorporeal therapy (HCO-HD <i>vs</i>. conv. HD) and serum uric acid values (<10.4 mg/dl <i>vs</i>. ≥10.4 mg/dl) before therapy initiation. With the application of HCO therapy, the rate of renal recovery doubled from 29% to 64% of patients. Further assessment of uric acid values predicts the probability of renal recovery more precisely, resulting in a medium-risk class (40.0% renal recovery) or low-risk class (71.9% renal recovery).</p>