Superiority of disease-specific over conventional formula in predicting creatinine clearance from serum creatinine in patients with liver cirrhosis.

Abstract

To determine if liver cirrhosis may influence the accuracy of formulas estimating creatinine clearance (Clcr) from serum creatinine, we compared the measured and estimated Clcr in 95 male (group A) and 47 female cirrhotic patients (group B). The Clcr values of group A and B patients were estimated with the equations obtained from 93 male (group C) and 86 female patients (group D) who were free of liver disease: the mean estimation errors (+/- SD) for the group C and D patients with the equations obtained from their own population data were 5 +/- 24% and 4 +/- 22%, respectively. However, these equations significantly (p less than 0.01) overestimated the Clcr of cirrhotic patients: the mean estimation errors for the group A and B patients were 35 +/- 43% and 14 +/- 27%, respectively. In contrast, the mean estimation errors for the group A and B patients using cirrhotic patient-specific equations obtained from their own population data were 5 +/- 33% and 3 +/- 25%, respectively. The accuracy of these disease-specific formulas was also confirmed in a prospective manner in 43 male (group E) and 21 female cirrhotic patients (group F): the mean estimation errors for the group E and F patients were 2 +/- 32% and -7 +/- 29%, respectively. We conclude that the Clcr values of cirrhotic patients, particularly male patients, may not be accurately estimated with equations of formulas deriving from liver disease-free patients, but should be estimated using their own population data.