Can creatinine clearance in children clear the indexing GFR with BSA from the charge of errors?

Can creatinine clearance in children clear the indexing GFR with BSA from the charge of errors? We read with interest the article of Delanaye and colleagues stating that indexing glomerular filtration rate (GFR) with body surface area (BSA) required mathematical prerequisites to be satisfied, even though many studies reported they were not [1]. Although not the gold standard to measure GFR, creat-inine clearance (Ccr) is one of the methods most frequently indexed by BSA in hospital. Despite the many reports referred to by the authors, we wished to make certain if the prerequisites were not satisfied because indexing was our routine practice. Ccr was measured with serum and 24-h urine in 74 children between 2 and 12 years old. The values of Ccr before (Ccr-b) and after (Ccr-a) indexing with BSA were plotted respectively with BSA, and the correlation coefficient and regression equation were calculated in each plot. Creatinine was measured by Jaffe method, and BSA was calculated by Mosteller method [2]. Contrary to our worries, the correlation coefficient between BSA and Ccr-b was high enough to satisfy the authors' first prerequisite. It was also quite different from the value in article [3] as referred (Figure 1A). It seemed as if the low correlation coefficient of the study was drawn from many patients, with chronic kidney disease included among healthy subjects. The y-intercept between BSA and Ccr-b was very close to 0, one of the prerequisite values the authors presented. The slope of regression curve was far from 1, the authors' other insisting value, but it is hard for the slope to reach this without the same magnitude of two variables, i.e. BSA and Ccr-b. We thought this was an authors' linguistic flaw. Finally, the relationship between BSA and Ccr-a disappeared almost totally, and the authors' last prerequisite was fulfilled (Figure 1B). Turner and Reilly (article [4] referred to by the authors) already accepted that GFR indexing with BSA was good enough, albeit incomplete. They proved that all the assumptions about slopes and y-intercepts of the regression curves between BSA and GFR before and after indexing were satisfied statistically. But they argued the regression model of their own was better than indexing with BSA because their slope was completely 0 [4]. We were afraid that the above completeness might have resulted from applying the regression model to the same subjects who participated in establishing the model. In conclusion, it was …