We agree with Dr. Matz that our equation is more complicated than the Cockcroft-Gault equation. We had suggested that clinical laboratories collect information on age, sex, and race and perform the computations and report predicted GFR in addition to serum creatinine, albumin, and blood urea nitrogen levels. Alternatively, the prediction equation can be programmed into hand-held calculators or computers, which are now widely used by physicians.

We thank Dr. Goldwasser for mentioning studies that document the higher creatinine excretion rate in black persons than in white persons. Analyses of factors associated with changes in creatinine excretion during the course of chronic renal disease using the Modification of Diet in Renal Disease (MDRD) Study database are planned.

Dr. Matz points out that high-dose cimetidine can be used to inhibit tubular secretion of creatinine, thereby increasing serum creatinine levels and reducing the overestimation of GFR by creatinine clearance estimated using the Cockcroft-Gault equation. We did not use this method of estimating GFR in the MDRD Study, so we cannot compare the accuracy of this method to the accuracy of the method of estimating GFR using the MDRD Study prediction equation. However, in the original report of van Acker and colleagues, creatinine excretion was not fully inhibited by large doses of cimetidine. Nor does the proposed method take into account racial differences in creatinine excretion.

Drs. Ando and Hasegawa discuss newer, enzymatic assays for creatinine that report lower normal values in serum because they do not cross-react with noncreatinine chromogens, as does the kinetic alkaline picrate assay. Consequently, laboratories using these enzymatic methods would report systematically higher values for predicted GFR when using the MDRD Study equation. Similarly, they would also report higher values for creatinine clearance when using the Cockcroft-Gault equation, whether or not cimetidine was used to inhibit creatinine secretion. In principle, this overestimation could be partly reduced by multiplying the value for serum creatinine by a factor that reflects the ratio of the concentration of creatinine chromogens to total chromogens (creatinine chromogens plus noncreatinine chromogens), which is approximately 0.9 at normal levels of GFR. However, this approach could also introduce errors because this ratio varies among patients and decreases as GFR declines. The most accurate approach would be to measure creatinine (and albumin) in a study sample by using a variety of laboratory methods and to derive prediction equations with alternative coefficients for each laboratory method.