Thanks for comment. The eGFR equation yields a number with considerable variability. Many don't appreciate that and follow it with the idea that it is precise and accurate, that it is a rate not a concentration, and that the decimal point matters. None of these aspects are correct. In an individual, the serum creatinine is usually the best way to follow the serial evolution of the renal function. Cystatin C does not help much.
As the estimate "eGFR" derives from the plasma (or serum) creatinine, another issue to consider is where creatinine comes from: creatine phosphate, key source of energy for muscles.
When muscles or atrophied (wasting diseases, advanced age) or inactive (quadriplegia), less creatine phosphate is utilized, so less creatinine is generated. Thus a low plasma creatinine in someone with little muscle or inactive muscle can greatly OVERESTIMATE true GFR. I learned that from a quadriplegic man who remained unconscious long after sedative drugs ought to have been excreted - but whose eGFR was about 20ml/min/1.73m2.
I've always been puzzled why it is so hard for doctors/nurses/patients to collect 24 hour urine specimens to measure true creatinine clearance. When I've done it myself a few times, the measured clearance exceeded my eGFR. If one misses the bucket or forgets, one can start over with an empty container (no need to be sterile) and sample the plasma creatinine when one turns in the 24 hour urine.
Another issue in healthier people is that vigorous exercise should raise plasma creatinine, so if one wants to score a lower estimate (eGFR), one should rest after exercise and dilute the plasma with plenty of water.
Perhaps one of the commenters explained the real reason for the popularity of eGFR: money. Setting entirely arbitrary limits for a marker that varies so much allows people to be categorized as Stage 1 CKD or similar ... even though their kidneys are perfectly adequate for their eventual needs to filter ...
Good points to make. However, it's important to emphasize that the serum creatinine has different meanings for different people depending on various individual physiologic characteristics (race being one of the least important). It's not a bad idea to have a calculation that takes that into account, or at least teach us how to do it for individual patients.
Once established for an individual patient who is stable however, wouldn't eGFR be a somewhat more precise way to follow renal function over time and so more useful than just the serum creatinine.
Yes, I agree that eGFR is not really different from the s creatinine. "eGFR" has the patina of units that imply a rate (ml/min) whereas the simple s creatinine has the units of a concentration (mg/dl). And rates would seem more useful than mere concentrations.
But eGFR is not a rate, it is a number derived from the s creatinine.
As clinicians attempt to bridge the communication gap with patients, labs could help by listing the value with a confidence interval or at least a +/- 30%. Like, when my wife says that I didn't clean the dishes well, I point out that 90% of the dishes were quite clean and yes, a chunk of oatmeal may have been left on one plate and one of the plates still smells like egg. This makes her assessment incorrect. Though I've never managed to convince her of this. I hope I have better luck with my online friends, here.
I see this as a semantic issue. eGFR is the same thing as Creatinine. It is simply a formula that tries to normalize creatinine for “the individual”. They could just change the name from eGFR to “normalized creatinine”.
Thx for your comment. At least we (all of us) should remember that the "e" stands for estimate. Note the statistics, and the P30 metric. We are accepting an estimate that is as much as 30% inaccurate. It's an estimate.
I think the EGF should be completely abolished. I was the practicing the nephrologist for 40 years and saw the implementation of this to mine for chronic renal insufficiency.,which increased the payments from Medicare with a classification of renal disease according to severity. Also causes terrible anxiety amongst patients and physicians who do not understand. The start of the avarice of data mininb.
Yes, I agree with your statement. To which I add that a recent meat meal will temporarily elevate one's serum creatinine level. Regarding cystatin C, it could be a better marker of renal function but might be affected by inflammatory states and also states of high cortisol.
My main point is that the "e" in eGFR stands for estimate.
Thanks for this, Dr. Cohen. My understanding is that eGFR is also typically calculated from creatine levels which vary according to muscle mass. If you have a high muscle mass for your BSA creatine estimated eGFR will be underestimated. Vice versa, low muscle mass for BSA and eGFR overestimated.
Supplementing with creatine also effects the eGFR and kidney function better estimated by Cystatin-C eGFR for these two circumstances
And the use of the 1.73m^2 to index the eGFR to body surface area creates a further drop in the reported eGFR. Last week, a patient was worried about his eGFR of 58 ml/min. De-indexing to his actual BSA yielded an eGFR of 75 ml/min.
Removing the race coefficient (because “race is a social construct”) is a denial of physiologic differences, and may have unintended consequences. Same with M/F “normals” for Hgb/Hct.
That’s an interesting comment. So how do we determine what “race” someone is? And how do we quantify that? Is it based on the content of melanin in the skin, and that quantification is somehow associated with distinct physiologic differences? I’m sincerely curious on how the use of “race” can be applied practically in the clinical setting, because I have been unable to find a reliable method.
In practice, race is determined the same way that it is generally done across American society - by self-identification. If someone says they are black, then they are black. If they say they're white, then they're white. People check a box on a form. And, more to the point, that was the way it was done in the MDRD and CKD-EPI studies. So if one wanted to be pedantic, one could say "among those who self-identified as African American, their measured GFR was higher given the same age, sex, and creatinine as those who self-identified as non-African American."
Long known that those of black African descent have higher levels of creatinine in their blood, originally attributed to higher muscle mass, but that has not borne out as the reason. Using the wrong “normal values” may help, because they might have their tendency to develop CKD picked up earlier, but may hurt in that many drugs are not available for use if eGFR is below a certain level. And soon, doctors will forget how to interpret the tests. Changing physiologic parameters based on social constructs is the opposite of science.
Can you point me toward the studies about higher creatinine levels? I think many of us are skeptical, because it has been shown that race-based GFR estimates were used to delay access to life-saving treatment to African Americans, and most “facts” about race-based practices had no actual scientific evidence. But that doesn’t mean none of it is true. But also, it’s hard to believe that people from such a huge geographic area would have such similar genetics that would then differ so widely from groups not so far removed from them.
The links Fred sent are good. In the MDRD and CKD-EPI equations, there were studies done of thousands of people where various demographic and laboratory data were compared with measured GFR. Then the researchers attempted to create an equation to estimate GFR from the data. They found that a reasonable estimate could be done with age, sex, race, and creatinine. These were not pre-specified, and other lab values were considered. Race was only included because it was a particularly strong predictor of GFR, not because the researchers thought a priori that it was important.
From the paper cited below regarding the CKD-EPI:
"The coefficient of 1.16 reflects that mGFR was 16% higher in blacks than nonblacks with similar age, sex, and creatinine in the dataset used to develop the equation. The CKD-EPI development dataset included 2601 blacks (31.5% of the total population), of whom the largest proportion was from the AASK, in which race was self-reported and all study participants were residents of the United States"
Theoretically, we're all of African descent (although I don't know if that ancestor was originally "black"?). I know--splitting hairs.
I don't want to strike a political tone here, since I have nothing good to say about either major political party. But, I'd ask: Is Obama black? Because he's 50% black? But he's not white? He is 50% white. Point is, there's a whole spectrum from totally black to totally white. Same with other mixtures. Where's the demarcation line? And why?
But let's take an example to see what this means with the eGFR in practice. Imagine a 55-year-old woman with a creatinine of 1.7.
The original 2009 CKD-EPI equation would estimate her GFR as 39 ml/min/1.73m2 if black, and 33 ml/min/1.73m2 if white.
The new 2021 CKD-EPI equation, where race was removed, would estimate the GFR as 35 ml/min/1.73m2 for all women.
Multiple points here:
1. No matter the situation, the stage of CKD (this is whole other fake construct) would be the same (CKD stage 3b).
2. It would be reasonable for a black woman to think that her actual GFR is higher than 35 and for a white woman to think her actual GFR is lower than 35.
3. As the author of the piece points out, it's also all weirdly splitting hairs because all of these are not great estimates at best. We don't really know whether the true GFR is 33 or 35 or 39, and it does not usually make much clinical difference within that range.
Well, due to historical reasons related to race-based slavery, American society, in contrast with much of Latin America has traditionally followed a "one drop" rule, where those with mixed ancestry are more often classified as black. In general, Americans who identify as white are very high in European ancestry whereas those who identify as black has a much wider range of European vs African ancestry.
Women are being harmed by being assigned the same “normal” levels of hemoglobin as males, because early lung disease and chronic hypoxia are being missed. Males and females are physiologically different.
Thanks for comment. The eGFR equation yields a number with considerable variability. Many don't appreciate that and follow it with the idea that it is precise and accurate, that it is a rate not a concentration, and that the decimal point matters. None of these aspects are correct. In an individual, the serum creatinine is usually the best way to follow the serial evolution of the renal function. Cystatin C does not help much.
As the estimate "eGFR" derives from the plasma (or serum) creatinine, another issue to consider is where creatinine comes from: creatine phosphate, key source of energy for muscles.
When muscles or atrophied (wasting diseases, advanced age) or inactive (quadriplegia), less creatine phosphate is utilized, so less creatinine is generated. Thus a low plasma creatinine in someone with little muscle or inactive muscle can greatly OVERESTIMATE true GFR. I learned that from a quadriplegic man who remained unconscious long after sedative drugs ought to have been excreted - but whose eGFR was about 20ml/min/1.73m2.
I've always been puzzled why it is so hard for doctors/nurses/patients to collect 24 hour urine specimens to measure true creatinine clearance. When I've done it myself a few times, the measured clearance exceeded my eGFR. If one misses the bucket or forgets, one can start over with an empty container (no need to be sterile) and sample the plasma creatinine when one turns in the 24 hour urine.
Another issue in healthier people is that vigorous exercise should raise plasma creatinine, so if one wants to score a lower estimate (eGFR), one should rest after exercise and dilute the plasma with plenty of water.
Perhaps one of the commenters explained the real reason for the popularity of eGFR: money. Setting entirely arbitrary limits for a marker that varies so much allows people to be categorized as Stage 1 CKD or similar ... even though their kidneys are perfectly adequate for their eventual needs to filter ...
Tom Perry MD, FRCPC
Vancouver
Good points to make. However, it's important to emphasize that the serum creatinine has different meanings for different people depending on various individual physiologic characteristics (race being one of the least important). It's not a bad idea to have a calculation that takes that into account, or at least teach us how to do it for individual patients.
Once established for an individual patient who is stable however, wouldn't eGFR be a somewhat more precise way to follow renal function over time and so more useful than just the serum creatinine.
What about the Cystatin C?
Thx for your comment. I agree.
Life was simpler and less stressful before eGFR was added to routine labs.
That is all.
(and I have no sense that population health is better since).
THANK YOU for this essay!!
Yes, I agree that eGFR is not really different from the s creatinine. "eGFR" has the patina of units that imply a rate (ml/min) whereas the simple s creatinine has the units of a concentration (mg/dl). And rates would seem more useful than mere concentrations.
But eGFR is not a rate, it is a number derived from the s creatinine.
As clinicians attempt to bridge the communication gap with patients, labs could help by listing the value with a confidence interval or at least a +/- 30%. Like, when my wife says that I didn't clean the dishes well, I point out that 90% of the dishes were quite clean and yes, a chunk of oatmeal may have been left on one plate and one of the plates still smells like egg. This makes her assessment incorrect. Though I've never managed to convince her of this. I hope I have better luck with my online friends, here.
I see this as a semantic issue. eGFR is the same thing as Creatinine. It is simply a formula that tries to normalize creatinine for “the individual”. They could just change the name from eGFR to “normalized creatinine”.
Thx for your comment. At least we (all of us) should remember that the "e" stands for estimate. Note the statistics, and the P30 metric. We are accepting an estimate that is as much as 30% inaccurate. It's an estimate.
I think the EGF should be completely abolished. I was the practicing the nephrologist for 40 years and saw the implementation of this to mine for chronic renal insufficiency.,which increased the payments from Medicare with a classification of renal disease according to severity. Also causes terrible anxiety amongst patients and physicians who do not understand. The start of the avarice of data mininb.
Yes, I agree with your statement. To which I add that a recent meat meal will temporarily elevate one's serum creatinine level. Regarding cystatin C, it could be a better marker of renal function but might be affected by inflammatory states and also states of high cortisol.
My main point is that the "e" in eGFR stands for estimate.
Thanks for this, Dr. Cohen. My understanding is that eGFR is also typically calculated from creatine levels which vary according to muscle mass. If you have a high muscle mass for your BSA creatine estimated eGFR will be underestimated. Vice versa, low muscle mass for BSA and eGFR overestimated.
Supplementing with creatine also effects the eGFR and kidney function better estimated by Cystatin-C eGFR for these two circumstances
Do you agree?
I have found most patients and even some/many clinicians also forget eGFR is measured in ml/min, and not as a %.
And the use of the 1.73m^2 to index the eGFR to body surface area creates a further drop in the reported eGFR. Last week, a patient was worried about his eGFR of 58 ml/min. De-indexing to his actual BSA yielded an eGFR of 75 ml/min.
I agree. But I wanted to be brief and cover the essential point that "e" stands for estimate.
Removing the race coefficient (because “race is a social construct”) is a denial of physiologic differences, and may have unintended consequences. Same with M/F “normals” for Hgb/Hct.
That’s an interesting comment. So how do we determine what “race” someone is? And how do we quantify that? Is it based on the content of melanin in the skin, and that quantification is somehow associated with distinct physiologic differences? I’m sincerely curious on how the use of “race” can be applied practically in the clinical setting, because I have been unable to find a reliable method.
In practice, race is determined the same way that it is generally done across American society - by self-identification. If someone says they are black, then they are black. If they say they're white, then they're white. People check a box on a form. And, more to the point, that was the way it was done in the MDRD and CKD-EPI studies. So if one wanted to be pedantic, one could say "among those who self-identified as African American, their measured GFR was higher given the same age, sex, and creatinine as those who self-identified as non-African American."
Long known that those of black African descent have higher levels of creatinine in their blood, originally attributed to higher muscle mass, but that has not borne out as the reason. Using the wrong “normal values” may help, because they might have their tendency to develop CKD picked up earlier, but may hurt in that many drugs are not available for use if eGFR is below a certain level. And soon, doctors will forget how to interpret the tests. Changing physiologic parameters based on social constructs is the opposite of science.
Can you point me toward the studies about higher creatinine levels? I think many of us are skeptical, because it has been shown that race-based GFR estimates were used to delay access to life-saving treatment to African Americans, and most “facts” about race-based practices had no actual scientific evidence. But that doesn’t mean none of it is true. But also, it’s hard to believe that people from such a huge geographic area would have such similar genetics that would then differ so widely from groups not so far removed from them.
The links Fred sent are good. In the MDRD and CKD-EPI equations, there were studies done of thousands of people where various demographic and laboratory data were compared with measured GFR. Then the researchers attempted to create an equation to estimate GFR from the data. They found that a reasonable estimate could be done with age, sex, race, and creatinine. These were not pre-specified, and other lab values were considered. Race was only included because it was a particularly strong predictor of GFR, not because the researchers thought a priori that it was important.
From the paper cited below regarding the CKD-EPI:
"The coefficient of 1.16 reflects that mGFR was 16% higher in blacks than nonblacks with similar age, sex, and creatinine in the dataset used to develop the equation. The CKD-EPI development dataset included 2601 blacks (31.5% of the total population), of whom the largest proportion was from the AASK, in which race was self-reported and all study participants were residents of the United States"
Perhaps I have not been sufficiently skeptical. Too busy to research at the moment, but FYI:
https://pubmed.ncbi.nlm.nih.gov/32393465/
https://pubmed.ncbi.nlm.nih.gov/34554658/
Theoretically, we're all of African descent (although I don't know if that ancestor was originally "black"?). I know--splitting hairs.
I don't want to strike a political tone here, since I have nothing good to say about either major political party. But, I'd ask: Is Obama black? Because he's 50% black? But he's not white? He is 50% white. Point is, there's a whole spectrum from totally black to totally white. Same with other mixtures. Where's the demarcation line? And why?
But let's take an example to see what this means with the eGFR in practice. Imagine a 55-year-old woman with a creatinine of 1.7.
The original 2009 CKD-EPI equation would estimate her GFR as 39 ml/min/1.73m2 if black, and 33 ml/min/1.73m2 if white.
The new 2021 CKD-EPI equation, where race was removed, would estimate the GFR as 35 ml/min/1.73m2 for all women.
Multiple points here:
1. No matter the situation, the stage of CKD (this is whole other fake construct) would be the same (CKD stage 3b).
2. It would be reasonable for a black woman to think that her actual GFR is higher than 35 and for a white woman to think her actual GFR is lower than 35.
3. As the author of the piece points out, it's also all weirdly splitting hairs because all of these are not great estimates at best. We don't really know whether the true GFR is 33 or 35 or 39, and it does not usually make much clinical difference within that range.
Well, due to historical reasons related to race-based slavery, American society, in contrast with much of Latin America has traditionally followed a "one drop" rule, where those with mixed ancestry are more often classified as black. In general, Americans who identify as white are very high in European ancestry whereas those who identify as black has a much wider range of European vs African ancestry.
The vast number of patients know their race, and if it’s not clear, or mixed, that needs to be a topic of discussion during test interpretation.
Women are being harmed by being assigned the same “normal” levels of hemoglobin as males, because early lung disease and chronic hypoxia are being missed. Males and females are physiologically different.