The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
Race and Kidney Transplants - A Brief Rejoinder to Dr. Stanley Goldfarb
My brief rejoinder to his response to my earlier post on this subject.
In my last post, I put up a reply by Dr. Stanley Goldfarb to a post I wrote on "Race, 'Wokeness,' and Kidney Transplant Shortages," which was in part a critique of Dr. Stanley Goldfarb's article on the same subject, published by the City Journal. In this post, I offer a brief rejoinder to Dr. Goldfarb.
In his response, Dr. Goldfarb write that he does not support "a race-based formula for determining kidney function," which in turn would affect prioritization for kidney transplants. Rather, he says he merely rejects claims that the race-based formula recently replaced by the Organ Procurement and Transplantation Network (OPTN) and the United Network for Organ Sharing (UNOS), was racist, because it disadvantaged African-Americans. He also object to the retroactive application of the new, race-neutral formula to patients already on the transplant waitlist.
I appreciate Dr. Goldfarb's clarification of his views. However, it seems to me that, in his original article, he did in fact object to the displacement of the old race-conscious formula. Indeed, he described it (along with another policy change) as "perhaps the most dangerous victory for wokeness in health care to date." Even in his reply, he continues to support the use of the race-based system for patients already on the waitlist. At the very least, his position strikes me as far from an unequivocal rejection of racial discrimination.
As noted in my previous post, the use of racial criteria here can perhaps be defended on the ground that this is one of the very rare cases where there are genuinely significant physical differences between racial and ethnic groups that affect their chances of a successful transplant. But, as also noted there, such an argument at least requires strong evidence that there is no other comparably accurate way to assess patients. In addition, as discussed in my earlier piece, racial discrimination is particularly indefensible in a situation where we can easily eliminate kidney shortages for patients of all races, simply by legalizing organ markets.
While Dr. Goldfarb and I may continue to differ on the issue of racial discrimination, there are some points of agreement between us, as well, also covered in my previous piece. For example, we agree it would be a mistake to reduce the priority given to kidney donors in the waitlist, in order to promote racial "equity."
Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
There is a problem with giving priority to kidney donors on the waitlist.
I assume by this that you mean that a person who signed a donor card or gotten that red dot you put on your drivers license as a consent to donate gets priority, not just that small number of people who have performed an actual living kidney donation and come to need a kidney in return.
If there is currently a big priority boost for people who have attached a donor declaration to their state ID record or some such, then anyone who contracts a condition that may decades hence force a kidney transplantation [think diabetes] should mark themself as a donor immediately. Diabetics are likely not going to have their organs donated, so it doesn't mean anything, but it gets you priority under this regime.
-dk
I'm pretty sure that being an actual live kidney donor is the criteria for getting a boost on the waitlist.
This would presumably be a small number of people as only a very small percentage of the population have donated a kidney while alive.
My support for the preference is based on my assumption that it applied only to those who had actually donated a kidney, not to those merely signing the card. That would be subject to gamesmanship to jump the list and is a very minor contribution to society that doesn't justify prioritizing it over other virtuous acts and it's a dangerous place to start deciding who is a more deserving person.
Give a kidney, get a kidney. Just say you'll give one, get in line.
A kidney donor only has one left. In exchange for this risk, they go to the front of the line.
You are incorrect (and the prior article said so explicitly. Priority for kidney donors is only to those who in fact have give a kidney already, are down to just one and that one is now failing. There is no bump in priority for merely carrying an organ donor card.
Thank you, Ilya!
This: “At the very least, his position strikes me as far from an unequivocal rejection of racial discrimination.”
And this: “But, as also noted there, such an argument at least requires strong evidence that there is no other comparably accurate way to assess patients.”
But so many here push back on that latter idea. As if there’s nothing especially dangerous or immoral about racial discrimination in the United States of America.
And it should be repeated: Self-identified race frankly is not a valid scientific method of categorizing people for purposes of calculating eGFR. At best, there are correlations to other relevant factors. It's immoral not to find and measure those factors as failing to do so penalizes individuals (or privileges individuals) based on their race rather than anything that actual has medical salience.
The 2009 CKD-EPI equation did not use self-identified race as a factor. If you think it did, you should read your sources more critically. https://pubmed.ncbi.nlm.nih.gov/19414839/ is the abstract of the original publication of that equation. https://pubmed.ncbi.nlm.nih.gov/17332152/ is another method used to estimate GFR; it is typically abbreviated MDRD.
The problem with "it's immoral not to find and measure those factors" is that it is quite expensive to GFR directly. eGFR is part of a normal annual physical exam, so estimates like MDRD or CKD-EPI are typically used as an initial screening test. There are more accurate estimates that use cystatin C as an additional variable, and those are recommended when more precision is needed, but not all labs report cystatin C. Other tests -- particularly for blood and urine albumin levels -- can augment eGFR as a screening method.
If you don't think race has actual medical salience, you should probably survey the medical literature more. It is an extremely common proxy variable. It is not perfect, but it is better than waiting for someone to discover and validate a globally better estimation method or an enormously cheaper measurement method. As an aside, the 2021 CKD-EPI formula still uses sex as a proxy variable.
Actual practice is not even consistent at a more basic level, though. For example, the National Kidney Foundation says eGFR >= 90 is normal, and 60 <= eGFR 60", and say eGFR < 60 is chronic kidney disease. (The latter, at least the ones I see, do not say how eGFR of exactly 60 is interpreted.)
Michael P,
You have no idea what you are talking about. The formula uses self-identified race (from your link “…x 1.212 (if black)…”). Here’s an medical article that says it clearly:
“The 2009 Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation is the most used equation to estimate glomerular filtration rate (GFR), with race being a factor thereof, increasing GFR by 16% in self-identified Black persons compared with non-Black persons.” (https://academic.oup.com/ckj/article/16/2/322/6779793)
The links you provide only confirm the formula uses race. How do you think they determine the race of the patient? Genetic testing? Nope. Extensive family history and the one drop rule? Nope. It’s self-identification, as my link confirms.
Of course it is hard to find a good measure for kidney function. That doesn’t justify racial discrimination. Particularly where, as it inevitably is, the race of an individual is either their self-identification or arbitrary assignment via phenotypical features like skin color.
“If you don’t think race has actual medical salience, you should probably survey the medical literature more.”
You understand that it is almost never used in the medical literature anymore, right? This is because it is an incredibly crude and unscientific method. It can be used for crude screening measures, though that is being abandoned in many cases because, again, it discriminates against individuals based on their membership in a cultural (not genetic or other scientifically valid) group. Basing placement on the transplant list on self-identified race is immoral. Which is what has been done for 20 years. Which is what Stanley supports.
If self-identified race is used, why the ranting about clinicians attributing race based on phenotypes?
Please try to be consistent.
So you admit you're wrong?
And, yes, although the formula is based on self-identified race and typically applied using self-identified race, sometimes clinicians applying the formula check the box based on their own subjective criteria rather than asking the patient. It's not that complicated. No inconsistency.
I do not admit I am weong. I believe that when someone shows they can write "self-identified race" or the like, they show they can mean something different when they write "race" alone.
Why do you say the formula is based on self-identified race? It says "if black". If someone chooses to apply that based solely on self-identification, that is on them -- it is not a flaw in the formula.
And at least now you admit that you are wrong, and that clinicians can override the self-identification.
"Why do you say the formula is based on self-identified race?"
Because that's what literally every study says it's based on.
Because that's how it is applied (except, as I have consistently noted, when physicians make their own guess which is no better than self-identification).
Because what method do you think they use other than self-identification (or equally arbitrary categorization by physician based on phenotypic features like skin color)?
You're just wrong.
Professor Somin, I don't think you fairly characterized Dr. Goldfarb's position when you wrote, Even in his reply [Dr. Goldfarb], he continues to support the use of the race-based system for patients already on the waitlist. At the very least, his position strikes me as far from an unequivocal rejection of racial discrimination. I really have a problem with how that was put, you can infer multiple meanings from your characterization.
This is what Dr. Goldfarb actually wrote (the part I believe you were referring to).
The previous formulas that required a separate calculation for African Americans have been labeled as racist. That is simply and demonstrably untrue. They were verified in multiple clinical studies with hundreds of patients.
This looks like a straightforward factual statement to me.
Do better.
re: organ markets. Yes. Where is that 'My body, my choice crowd'? 🙂
It has been zero days since Somin nuked an entire comment thread to memory-hole leftist hot takes and people agreeing with Dr. Goldfarb. How libertarian!
I retract the above. I did not notice this was a different post on the exact same topic.
https://www.nejm.org/doi/full/10.1056/NEJMoa2103753 says that using race is more accurate than ignoring race:
Using generic ancestry as a different proxy did maybe-worse (higher average error, but not statistically significant) than plain race. Using cystatin C without race had the best average error of those four methods, but statistically overlapped the race-without-cystatin-C formula.
Point for Dr. Goldfarb, I think.
A technical point, but the way things are going, it's probably best to go with the blood tests if they're equivalent, because you're not going to get a lot of nutcases "identifying as" having a particular blood test result. Whereas if the race hucksters get people thinking that there's racial discrimination going on, you might get a significant number of people deliberately misstating their race.
As long as the consequences fall on the nutjob, why care. If they're not AA and ID as black have them pay for the needless testing or if they are black and ID as not they risk their own health. Reality being others will pay for both cases you're right about the hucksters and the need for a cheap identifier.
"The estimation of GFR with the use of cystatin C generated similar results while eliminating the negative consequences of the current race-based approaches."
So there is a method that is as accurate as a method which uses self-identified race as a factor, and you think that's a point for Dr. Goldfarb?
Seriously?
Two methods: one involves explicit racial discrimination, the other does not. They are equally accurate, so let's lean into racial discrimination? Wtf?
(And now you're citing studies which confirm what you deny elsewhere, that the old formula uses self-identified race for the plus factor.)
The 2021 recommended formula doesn’t use cystatin C. Yes, that’s a point for Goldfarb.
The study does subgroup analysis based on self-identified race. That doesn't mean eGFR uses the same variable.
"The study does subgroup analysis based on self-identified race. That doesn’t mean eGFR uses the same variable."
Just admit you're wrong. The old formulas use self-identified race.
Keep clinging on to that one weird misreading -- it's literally the only point you have been able to bitterly cling to.
Lol, I see you no longer dispute that the "race" factor is based on self-identified race.
And funny you claim it's the only point I've been clinging. I've made several and quoted Ilya's statements that made those same points. I then added a bit about the unscientific basis for the categorizations. You then latched onto "self-identified race" and disputing that was used as the hill you wanted to die on.
You're dead. The old formulas' race components use self-identified race. End of story. You're wrong.
Piling a spectacularly stupid misreading on top of your previous misreading doesn't make you look any less stupid.
You admitted that clinicians can and would override self-identified race for the purposes of eGFR calculation. End of story: it doesn't use self-identified race, you were wrong, and you refuse to admit the conclusion even though you admitted the facts.
You admitted that clinicians can and would override self-identified race for the purposes of eGFR calculation. End of story: it doesn’t use self-identified race, you were wrong, and you refuse to admit the conclusion even though you admitted the facts.
You're blindingly stupid.
I didn't "admit" any such thing.
I pointed out that typically, physicians ask patients their race to determine whether the race factor gets added, i.e., it's self-identification / self-reported race that is used. Sometimes clinicians don't ask, but make their own subjective categorization based on phenotypic features like skin color which is no better.
What didn't happen is that study subjects were assigned to Black or non-Black category based on an objective, rigorous definition of "race". What doesn't happen is that the race-based factor is implemented by using an objective, rigorous definition of "race" such that the person getting the plus factor is actually in some relevant category (genetically, culturally, something) with the study participants.
But you don't care about the principle, the fact that the development and implementation of the formulas were based on self-reported race (or physician-assigned race based on each physician's idiosyncratic definition of that term), and that such unscientific categorizations are necessarily overinclusive and under inclusive of whatever underlying factor for which the formulas' developers thought race was a proxy. You don't care that this necessarily results in the discrimination against individuals based on their self-reported race (which is cultural not biological) or their skin color.
You're desperate to be right about something. So you've moved the goalpost from the formula doesn't use self-identified race, but that it typically, but doesn't exclusively, use self-identified race.
What a great point, Michael! Completely irrelevant to the substance of my criticisms, a fact I was the first to point out (you just consistently denied the formulas used self-identified race, which you seem now to admit they do), but other than that, great point.
https://pubmed.ncbi.nlm.nih.gov/36544336/ - the new equation is worse for Koreans.
https://www.tuftsmedicalcenter.org/Research%20Clinical%20Trials/Institutes%20Centers%20Labs/Chronic%20Kidney%20Disease%20Epidemiology%20Collaboration/GFR%20Estimating%20equations/Creatinine%20Based%20Equations - everyone recognizes the biases and reduced accuracy of the 2021 recommendations.
The 2021 formula was chosen because of wokeness and "equity", not for accuracy.
Now you’re on to Koreans and the 2021 formula. You keep trying to start a new debate on a new topic. Let’s review.
My claim: The old formula unfairly disadvantages Black patients and Goldfarb is curiously unconcerned with discrimination against Black patients.
Fact: The 2009 CKD-EPI (with race-based adjustment) overestimates GFR in Black patients (which necessarily disadvantages Black patients in treatment options and on the transplant list). You cited one study, among many, that show that.
Point me. (Since you decided we were doing points.)
Claim: The MDRD and 2009 CKD-EPI (with race-based adjustment) use sloppy categorization of people by race typically using self-identification (but of course, as I have noted many times, also sometimes the equally sloppy eye-balling of the physician) and this is a bad way to do medicine, particularly given 300 years of sloppy and malicious categorization of people by race in the United States.
Michael P’s counterpoint: “The 2009 CKD-EPI equation did not use self-identified race as a factor.”
Fact: The 2009 CKD-EPI equation did and does use self-identified race as a factor.
Michael P’s new counterpoint: Sometimes it also uses a physician’s subjective, non-rigorous categorization which itself is often based on the patient’s self-reported race.
Fact: I already said that and it changes nothing about the argument that that is a sloppy, unscientific way to do medicine that has a real (and in this case realized) danger of unfairly disadvantaging people who self-report as or are perceived as Black.
Point me.
Michael P’s updated new counterpoint: I want to instead argue about how accurate the 2021 CKD-EPI is.
Fact: I don’t.
I want someone to develop and for medical institution to use the best possible formula which doesn’t arbitrarily and unfairly disadvantage Black patients. The MDRD and 2009 CKD-EPI (with race-adjustment) do disadvantage Black patients, so they don’t qualify. I’m entirely open to the idea that we can do better than the 2021 CKD-EPI too. In fact, I’m sure of it. And from what I’ve seen especially with something as important to individual patients as placement on the transplant list, that probably involves a formula using serum cystatin C.
(See, e.g., https://www.nejm.org/doi/full/10.1056/NEJMoa2203769)
https://pubmed.ncbi.nlm.nih.gov/35915546/ says that the new CKD-EPI formula is worse (less accurate) than the old one among French patients. Unless someone wants to argue that Frenchness is medically significant on its own, that seems like another point for Goldfarb.
https://snucm.elsevierpure.com/en/publications/impact-of-chronic-kidney-disease-epidemiology-collaboration-ckd-e says that the new formula has a higher chance of putting Asian patients in a lower-disease category too, but (paraphrasing) that’s okay because they probably don’t really need kidney care.
"another point for Goldfarb"
You just randomly assign him points. You are aware that: "nearly all European countries do not use the race coefficient for the 2009 CKD-EPI equation." It's not at all clear from the article you cite whether they used the race coefficient or had any self-identified Black patients in their 489 study subjects. Thus, this article, at best, supports not using the race coefficient as, presumably, they followed the European practice of not using the race coefficient and this study suggest that this was more accurate than the new formula. (It also showed the EKFC formula which does not use race was as good as the 2009 CKD-EPI.)
So, European countries reject the race coefficient (the study you cite doesn't involve any study of that), Goldfarb is arguing to continue using the race coefficient, but it's a point for Goldfarb?
Tell another joke, because you're hilarious.
First you complain that they don't have Black patients in their study, and then you think it's telling that they don't apply a correction for Black patients?
You might want to look at the biases described at https://www.nejm.org/doi/full/10.1056/NEJMe2114918 -- the 2009 formula was essentially unbiased for non-Black patients, and supposedly overestimated GFR for Black patients. The 2021 formula has a similar magnitude of bias for Black patients, just in the opposite direction -- and adds a bias that overestimated GFR for non-Black patients. That's excused only because the biases average out for the "right" racial mix of patients.
“First you complain that they don’t have Black patients in their study, and then you think it’s telling that they don’t apply a correction for Black patients?”
First, you’re really dumb. I didn’t complain about the European study not having Black patients. I pointed out that it wasn’t clear whether they did or not (so it’s of questionable relevance to what we’re discussing) and it’s also probable they didn’t apply the race-based plus factor (which would make it completely irrelevant to our discussion).
I had no complaint, just observations that eviscerate the study’s usefulness for your purposes which is in support of the race-based plus factor.
There doesn’t need to be a second, because it’s pretty clear you aren’t equipped for this discussion. But it’s worth pointing out, you accept all the findings at your link except the one that “supposedly overestimated GFR for Black patients”. All the others are fact, but a finding proving what you’ve disputed elsewhere that the old formula is biased against Black patients, well, that’s “supposedly.”
Please, keep citing articles that contradict points you’ve been making in this thread. Specifically, you’ve claimed the old formula was validated as accurate, but here you are citing an editorial that cites studies showing that the old formula is literally biased against Black patients by overestimating their GFR (which disadvantages them on the transplant list).
You’re setting the bar for stupidity.
If a study doesn't include Black people, it's irrelevant to the point at hand? You could at least try to not live down to the "Only Black Lives Matter" stereotype. I cited that study because it shows how Europeans recognize that the 2021 CKD-EPI equation is terribly biased. See also https://www.medscape.com/viewarticle/987711 .
But you would rather police others' language than admit that you have been defending a racially biased estimator that -- as Dr Goldfarb observed -- was selected specifically to produce racially biased results. And that is the real problem. You harp about "not settled" studies that suggest bias in the 2009 CKD-EPI equation, and studiously ignore that the 2021 equation has worse biases.
You claimed it was immoral to not identify and use better estimation methods. By your own standard, the 2021 CKD-EPI is immoral -- and less moral than the 2009 CKD-EPI, which used an additional factor that improves its accuracy. EKFC is more moral by your standard, but not qualified for use in Black patients. EKFC was developed based on an entirely non-Black patient population; you harped on how it was better because it did not consider race, but you did not know that it could not.
No, you keep harping on about how you are sure the 2009 CKD-EPI formula is based on self-identified race, even though you acknowledge that doctors can ultimately determine the patient's race for that equation.
“If a study doesn’t include Black people, it’s irrelevant to the point at hand?”
Are you special? Because when we’re discussing whether the 2009 CKD-EPI formula is biased against Black patients by overstating their kidney function, a study that doesn’t have anything to say about accuracy vis-a-vis Black patients is as irrelevant as you can get.
“You harp about “not settled” studies that suggest bias in the 2009 CKD-EPI equation”
Yes, because that’s the point Goldfarb misrepresented and that I argued and you chose to engage. I never claimed anything about the 2021 CKD-EPI (other than, as Goldfarb would have it, it “isn’t more accurate” which is also to say it isn’t more inaccurate or he would say that and you helpfully cited studies making that point).
You’ve just continually tried to change the subject, which effort has gotten more intense since your own goal of citing a study that made my point.
EKFC is more moral by your standard, but not qualified for use in Black patients.
Another attempt to change the subject? I never claimed it was or wasn’t. I don’t think one parenthetical aside pointing out that the EKFC is as accurate as the 2009 CKD-EPI is “harping” on anything.
you keep harping on about how you are sure the 2009 CKD-EPI formula is based on self-identified race, even though you acknowledge that doctors can ultimately determine the patient’s race for that equation.
It is. And I have frequently, though not always, pointed out the obvious that sometimes doctor’s check the box rather than asking the patient. What point do you think you’re making? That every time I say self-identified I don’t also write a paragraph about how the physician’s might also subjectively assign a race which is not more scientific than self-identification?
Yes, the MDRD and the 2009 CKD-EPI use self-identified race or the equally arbitrary and unscientific categorization by physicians using varied, subjective criteria but presumably usually either patient responses (self-identification) or observation of phenotypic features like skin color.
I’m not sure how that helps you. But, yes, I didn’t write that paragraph the first time I wrote “self-identification” or each time thereafter.
Now, how does this affect the point that the race factor included in the MDRD and 2009 CKD-EPI are based on subjective assignment to unscientific racial categories rather than some rigorously defined genetic or dietary or skeletal muscle mass or other factor for which “race” is purportedly a proxy?
You may only care about one single question. The rest of us care about things like whether the new equation is less accurate than the old one (which it is, as Goldfarb pointed out) and whether we have better estimators (only with additional tests).
The rest of us care about things like whether the new equation is less accurate than the old one (which it is, as Goldfarb pointed out)
No. As you pointed out, the new equation is not less accurate for Black patients. As you also pointed out, the old equation is demonstrably unfair to Black patients. But neither you nor Goldfarb care about that, apparently.
It's not that I care about only one question. It's that you and Goldfarb don't care about that question.
The new formula has just as much error for Black people as the old one, in its own validation data — just in the opposite direction. The new formula has more error for everyone else.
(I will also note that, in its own development data set, the new equation was significantly biased about Black patients, whereas the 2009 equation was unbiased on that data. That implies some population or systemic difference[s] between the development and validation data, but it's hard to say how much of each.)
But we know you don’t care about anyone else.
“But we know you don’t care about anyone else.”
You can keep saying that, but you’re the one that has shown not an iota of concern that, as your own study showed, the old formula is unfair to Black patients. For you, that doesn’t matter. From which I can only surmise that, for you, Black lives don’t matter.
But Black lives do matter, your white grievance whining be damned.
https://pubmed.ncbi.nlm.nih.gov/23901055/ - the 2009 formula was validated in heart failure patients, as well as broader populations
https://www.ahajournals.org/doi/full/10.1161/CIRCHEARTFAILURE.112.968545 - the 2009 formula was validated extensively
And, of course, the 2009 equation is still held as the gold standard by the KDIGO CKD Guideline.
Your first study says nothing about accuracy of the 2009 formula vis a vis Black patients which is what we (or at least I) am debating. Otherwise, the study merely says the 2009 formula is better than the 1999 MDRD.
Completely irrelevant.
Your second study also says nothing about accuracy of the 2009 formula vis a vis Black patients. It does say it’s better than the MDRD in “for younger people, white individuals, and women”. Not Black people though. And that’s the debate we’re having.
Completely irrelevant.
The 2012 KDIGO CKD Guideline (which is the Guideline they have on their website) still uses the 2009 formula?
Completely irrelevant.
Yes, we know you think only Black Lives Matter. No need to repeat yourself.
The whole discussion began regarding whether the old formula was unfair to Black patients. You helped me show that it is unfair to Black patients. But you don’t care. You'd rather use it anyway. Ergo, apparently Black lives don’t matter to you.
In many cases race and sex are irrelevant and should be treated as such but sometimes it is relevant and should be included; a wise man tries to discern the difference while a fool blithely ignores it. Here Ilya once again reveals himself as a fool.