Skip to main content
editorial
. 2016 Aug 3;11(9):1518–1521. doi: 10.2215/CJN.07530716

Figure 1.

Figure 1.

Differences between eGFR and measured GFR (mGFR) are only unbiased across levels of eGFR on the logarithmic scale. Here, we derive eGFR to estimate mGFR in 1093 subjects from an earlier study (10) using a least squares regression model (ln mGFR regressed on ln serum creatinine, ln serum creatinine squared, age, and sex). We then assess whether bias changes as GFR changes in the same dataset with a linear fit. (A) eGFR−mGFR across levels of mGFR. (B) eGFR−mGFR across levels of (eGFR+mGFR)/2. (C) eGFR−mGFR across levels of eGFR. (D) ln eGFR−ln mGFR across levels of ln eGFR.