eGFR: Remember, 'e' Stands for 'Estimated'

Jeffrey S. Berns, MD


January 29, 2013

This feature requires the newest version of Flash. You can download it here.

Hello. This is Jeffrey Berns, from the Perelman School of Medicine at the University of Pennsylvania in Philadelphia, and Editor-in-Chief of Medscape Nephrology.

I saw a patient in the office last week who was quite distressed because he was told by his physician that he now had stage 3 CKD (chronic kidney disease). His creatinine had been slightly abnormal for a number of years, and it had not changed at all during those years. The only thing that really had changed over the past several years was that he was now several years older than he was before, and because of his age, his MDRD (Modification of Diet in Renal Disease formula)-calculated estimated GFR (glomerular filtration rate) now placed him in the category of stage 3 CKD. He was quite upset, very worried about how many stages there were and when he was going to progress through them and need dialysis or a transplant, or just die.

I spent a fair amount of time talking with him about the implications of CKD, what stage 3 CKD meant and the different parts of stage 3, and what the absence of proteinuria in his case meant. I think he left the office feeling quite relieved, as was his wife, about where he was in terms of his kidney function.

One of the points of our discussion -- and this is an issue I talk quite a lot about with my fellows and others -- is that the calculated eGFR with the MDRD equation, or the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation, is an estimate. We need to keep that in mind. A point I make frequently, and have discussed here on Medscape, is the confidence interval of any calculated eGFR number reported by a lab. The following Figure is from an article published in the ClinicalJournal of the American Society of Nephrology a year or so ago that compared MDRD and eGFR along the x-axis, with inulin-measured GFR along the y-axis.[1] Take a look.

Figure. Comparison of MDRD-calculated eGFR vs inulin-calculated eGFR shows potential for wide variations in both.
Modified with permission from: Botev R, et al. Clin J Am Soc Nephrol. 2009;4:899-906.

I have placed 2 red arrows along where the eGFR is: 45 mL/min per 1.73 m2. As you can see, the inulin clearance ranges quite widely at that eGFR estimate, anywhere from as high as 90 mL/min per 1.73 m2 to around 20 or 25 mL/min per 1.73 m2.

In this patient, who was entirely asymptomatic and felt well, and who had a normal hemoglobin and normal metabolic parameters, was his eGFR really 60 or 59 or 58? Was it 100? Or was it something lower than that? When you see patients who have eGFRs of 30 or 45, is their true GFR twice that or half that? You cannot be certain because it could be either one.

My personal opinion is that labs should not be reporting a single number for calculated eGFR by MDRD or CKD-EPI equations but should probably report the 95% confidence interval, or perhaps the 25th to 75th percentile interquartile range. We need more emphasis on the imprecision of this number in an individual patient so that practitioners can consider that [imprecision] when they think about whether the patient needs to be referred for transplant or prepared for dialysis, or when counseling patients about the implications of their CKD "number."

This figure was from an article in the ClinicalJournal of the American Society of Nephrology[1] describing the CKD-EPI equation, which reached very similar conclusions, if you analyze the confidence interval of any calculated serum creatinine concentration.

In my opinion, we should stop reporting single eGFR numbers and report some range around them to better appreciate the value of this calculation and what it means clinically. Thanks for listening. This is Jeffrey Berns, Editor-in-Chief of Medscape Nephrology.