Influence of range of renal function and liver disease on predictability of creatinine clearance

Several formulas for predicting creatinine clearance (Ccr) are used for adjusting drug dosages but limited data are available on their accuracy in patients with significant renal impairment or concurrent disease. We measured 144 Ccr in 103 patients and compared results using four predictive methods. Of nine common diseases in these patients, liver disease was associated with a large (p less than 0.02) prediction error (overprediction). After data from eight patients with liver disease were removed, there was good overall correlation between predicted and measured Ccr (r2 = 0.91 for each method) but only two of the methods (I and IV) were consistently accurate in all ranges of renal function. Methods for predicting Ccr should not be used in patients with liver disease.