Race-Based Diagnosis Part 3 PDF

8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 ADVERTISEMENT PERSPECTIV...

8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 ADVERTISEMENT PERSPECTIVE I N T E N T I O N TO T R E AT  Race-Based Diagnosis, Part 3 — ITT Episode 35 Published July 31, 2024 N Engl J Med 2024;391: e9 DOI: 10.1056/NEJMp2407611 VOL. 391 NO. 5  AUDIO INTERVIEW Race-Based Diagnosis, Part 3 — ITT Episode 35 25m 17s DOWNLOAD As race-based diagnostic tools, such as pulse oximeters that function poorly on darker skin, continue to lead to inequitable care, a growing movement is working to weed them out of U.S. health care. Listen to this episode. “Intention to Treat” is a biweekly podcast offering a behind-the-scenes look at the complicated, perplexing, and fascinating issues facing medicine today. Subscribe to the “Intention to Treat” podcast: Apple Podcasts Podbean Spotify YouTube Music Rachel Gotbaum: Welcome to “Intention to Treat” from the New England Journal of Medicine. I’m Rachel Gotbaum. Today we continue our conversation about race-based algorithms and race-based diagnoses, how they became accepted as science in American medicine, and the challenge to change them. Joel Bervell: My name is Joel Bervell. I’m a fourth-year medical student at Washington State University. Myself and one other student were the first two Black students at my medical school. During my first year of medical school, I was thinking a lot about what it meant to be a Black medical student. When I would see patients that looked like me, I realized that I’d never seen them represented in my medical courses. And then Covid hit. And at the time I came across an article about this device called a pulse oximeter, and it was from the New England Journal of Medicine and it was called “Racial Bias in Pulse Oximetry [Measurement].” This study showed that Black patients were three times as likely to have inaccurate, overestimated oxygen saturation levels when compared to other patients. And I was shocked because I’d just recently finished my pulmonology unit, had never heard anything about this, even though we know that pulse oximeters are literally used every day in hospitals to help triage, to help understand who should get care, who shouldn’t get care, to help diagnose and understand the severity of an infection like Covid. My first thought was, “How is this not being taught? This seems like something that is so important, that’s integral, that every single medical professional should be taught because it is literally life or death.” And at the time, the FDA was actually telling people, “Get this pulse oximeter and use it at home so you can understand what your own oxygen saturation levels are.” And there was no warning on there that they may not work equally on darker skin tones. Studies later came out that showed that Black patients were more likely to be turned away from hospitals because their oxygen looked more normal even when it wasn’t because of these pulse oximeters. This research has been out there for decades. https://www.nejm.org/doi/full/10.1056/NEJMp2407611 1/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 You can go back to the original article that talked about pulse oximeters, and there’s a line in there that says, “Skin pigmentation may impact the results of this device.” So there’s been research out there, but we’ve ignored it for a long time. So for me, I was shocked that I’d never heard about this, and I kind of did what any millennial/Gen Z, I call myself a zillennial, would do, and I took to TikTok and I made a 30-second video about it, and I told people that they should just watch out, and especially providers to understand how it may not be an end all, be all when it comes to diagnosing patients. Twenty-four hours passed, and that video got over half a million views. And there was doctors, and nurses, and PAs saying, “I’ve used this device every single day, and I never knew this disparity existed.” And then there was patients saying, “I wonder if this is why my dad, my mom, my sister, my brother, my son was sent home and then died at home when they should have actually been admitted to the hospital.” Olaseni Bello: My name is Olaseni Bello, and I live in San Diego, California. I’m 43 years old. I was getting my MBA at UC Berkeley Haas, and I was in the process of building a health tech startup doing quite a bit of research and came across a social media post talking about the pulse oximeter. And at the time I found myself with the UK variant. This was January 2021. And I ordered a pulse oximeter because I wanted to be best armed and know that if I’m going to go to the hospital, let me really need to go to the hospital. It came from Amazon, you can clip it on your finger, and then it does the blinking-red-light thing and gives a digital number. There was nothing about, “Hey, by the way, if you’re a person of color, you should understand that this may not accurately read your oxygen levels.” No, there was nothing like that. And yes, I know I’ve watched a video that tells me it’s not the most accurate read, but I’m not in a position to make a judgment call on whether this thing will work on myself or not. This is what is being used widely by the medical profession. So when I order mine, it doesn’t mean that I fully trust it, but it’s the standard. It’s all that’s available. I was taking my temperature, trying to just rest. About day 5, day 6, I called a friend, and I said, “I need to go to the hospital.” I didn’t think I would make it another night, and my breathing was very labored. As you can imagine, the hospital is just overwhelmed. There were tents. There were tents outside the hospital with patients. And what they’re trying to do is make a quick decision on how ill you are. And the doctor comes out and I’m trying my best, I’m actually having a hard time speaking. And of course, the next thing, the situation is, you measure my oxygen levels. Puts a device on me, he looked at it, and he said, “You’re going to be fine.” He’s like, “You’re not that sick.” I was like, “That actually doesn’t measure my oxygen well. That actually is probably not an accurate reading.” And I think he was kind of surprised that I said that. And he was like, “No, this device is reading well, there’s nothing wrong with this device.” I was like, “No, because of my skin color.” That’s the point I was trying to make. And he was adamant that, yeah, I was going to go home. And I was scared because I could see that I was losing that argument. I called my uncle, who’s a doctor, and he’s talking to this doctor, and the doctor took a second, he paused, and he’s like, “Okay, we’ll get you in.” I was outside close to 8 hours in the tent outside because that’s just where the intake was happening. And I’m just doing my best to hang on. And I get moved into the actual brick-and-mortar physical room, and then I was moved shortly after that to the ICU. Now, if I moved to the ICU, what would’ve happened if I had gone back home based off the pulse oximeter or based off of that initial assessment? And because of how labored my breathing was, they wanted to intubate me. They also gave me a laptop, an iPad to speak to my family, which is not per se a good sign if you’re really reading the tea leaves here. So I’m looking at this as in, “If they’re giving you an iPad, then there’s really uncertainty about how you’re going to turn out.” Rachel Gotbaum: Olaseni Bello spent 8 days in the hospital. He says one reason he survived: that TikTok video. Olaseni Bello: I can say that the video of the pulse oximeter helped save my life. There’s no question about that. This video armed me with information that helped me be an advocate and push back and helped get me admitted. Absolutely. And I wrote a personal note to the individual who created the video. Joel Bervell: They sent me a direct message and said, “Thanks to your video, I’m here today.” Rachel Gotbaum: Soon after he posted on TikTok about the pulse oximeter, Joel Bervell was invited to meet with the FDA to talk about his video. A few months later, the agency issued an advisory about the device’s reduced accuracy in people with darker skin. But so far, the FDA has not required manufacturers to warn patients or clinicians about their product’s serious limitation. We are joined again by two doctors who have been pioneering change of these race-based practices and race-based algorithms. Dr. Michelle Morse is chief medical officer at the New York City Department of Health and Mental Hygiene, and Dr. Darshali Vyas. She’s a pulmonary and critical care fellow at Massachusetts General Hospital. So Dr. Morse, these race-based algorithms seem to be embedded in so many aspects of medical care. Do we understand how that happened and how they became a tool in clinical practice? https://www.nejm.org/doi/full/10.1056/NEJMp2407611 2/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 Michelle Morse: I do think that medicine in the United States has had a very complex history when it comes to the ways it either upholds White supremacy or counteracts it. That is evidenced by the eugenics movement, by different ways in which White male physicians themselves were a part of a society that considered Black people to be inferior socially and biologically. And that history is not a history we like to talk about. I think in medicine, we like to consider it to be far behind us, outdated. And in fact, when you look at these algorithms and the ways in which they use race, they connect directly back to that history. And what’s clear is that those historical explanations still find their way into the current use of clinical algorithms. Rachel Gotbaum: So Dr. Vyas, your initial research was about how the VBAC calculator gave Black women who had had a previous C-section a higher risk score for a vaginal birth. How is that an example of how racist beliefs are adopted into medical care? Darshali Vyas: The VBAC tool I think really reflects how these ideas of essential difference between races get ingrained into modern-day clinical tools. For example, in obstetrics, there is a really troubling history of the ways in which ideas of inadequacy in laboring and in delivering get reified and copied forward. So even now in obstetrics, there’s the Caldwell and Moloy’s classification of pelvic structures, which is still in the most recent version of Williams Obstetrics textbook. And this classification reduced a large breadth of anatomic variation down into four subtypes of pelvic shape, which were infused with racialized ideas of adequacy and normalcy around pelvic shape. Rachel Gotbaum: So tell us what you mean by that, how these pelvic shapes we still teach today were based on perceived historical racial differences. Darshali Vyas: So for example, they called the gynecoid pelvis, defined as mostly found in a White woman, described as ideally suited for childbirth. And then by contrast, they called the anthropoid pelvis, which was noted to be narrower, more common in non-White woman, and less suited for childbirth. And the anthropoid pelvis was described as degraded or animalized that was seen in the lower races. And this sort of parallel that was drawn between anthropoid and animal reflected intentions to cast Black women as anatomically deficient for this act of giving birth. And there’s a very similar history for Latinx women as well, dating back into the early 1900s. Mexican women, for example, who were in the Indigenous parts of Mexico, received extremely high rates of cesarean section to resolve their so-called faulty anatomy, despite having lots of complications and deaths from these procedures. And so we argue that the VBAC tool, for example, by systematically subtracting from Black and Hispanic women’s likelihood of successful vaginal birth, echoes this troubling history of them as inadequate laborers. I’m not saying that providers are sitting in the office actively thinking about the race and pelvic subtype of a woman delivering, but I do argue that these notions become infused into the modern-day clinical tools that we use, and they really carry forward this idea of different coefficients for different bodies. Rachel Gotbaum: So Dr. Vyas, you had pushback when you presented your findings on your VBAC research. You didn’t get them changed right away. Talk to us about the pushback and why there was pushback. Darshali Vyas: I think part of the pushback is that when something has been done a certain way for a long time, I think it is hard to question. It’s almost like if you pull at a thread, the whole thing unravels. And so the more times that you ask why, and you’re pushing back on why these correction factors ended up there to begin with, I think it really challenges not just that individual tool, but it really challenges the whole way that we conceive race within medicine. So I think a lot of the initial challenges around reconsidering the use of race in these tools was based on just fundamentally different understandings of how we should use race and how we can understand the ways that Black and Brown patients in this country have different outcomes in their health using race and predictive modeling. But I think in general, people worried that if we took race out of some of these calculators or predictive tools, then we may be subjecting Black and Brown patients to worse outcomes. So for example, if we have the empiric data to show that Black and Brown women have more challenges in delivering vaginally or have worse maternal health outcomes, if we take race out of it, are we subjecting Black and Brown patients to a higher risk? And when they do all of the data and then say, “Here are the patients who had a successful vaginal birth,” what characteristics did they have that made them more likely to have a successful vaginal birth? And more times than not, race is going to show up as having a signal in that data analysis because we all know, unfortunately, there are still persistent racial inequities in this country. Rachel Gotbaum: So how do you think about these inequities and how predictive tools could actually help eliminate them? Darshali Vyas: I think it’s really interesting that when they did that analysis, for example, for VBAC, not only is there a signal that shows up for race, there’s also a signal that shows up for all kinds of other factors. So for example, for VBAC, the researchers found that marital status and insurance type also correlated with a successful vaginal birth. But those variables were never built into the tool, and it doesn’t seem like researchers ever considered building it into the tool. https://www.nejm.org/doi/full/10.1056/NEJMp2407611 3/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 And to me, that really highlights the reasons that challenging race in this tool shifts our whole conception of race. Because researchers would never build in marital status or insurance into a predictive tool because there’s a general recognition that those signals are social and political capital. If we really believed that race was also a social and political variable, as we’ve learned in my medical school class, then we would’ve done the same exact thing for the race variable as we did for marital status and insurance. We didn’t, and I think that highlights the fascination with thinking about race as a biological, physiological concept, and that’s what we were challenging. And I think that’s why it caused so much pushback. Rachel Gotbaum: Dr. Morse, I’d like to talk about your work and how you’re trying to get this changed in New York City hospitals. Michelle Morse: I think there is tremendous work to do in this space, and part of the reason that we developed the Coalition to End Racism in Clinical Algorithms, also known as CERCA, in New York City is because what we also know is on average, it takes somewhere between 10 and 14 years for cutting-edge research to become mainstream practice. And our feeling was, if we have these concerns about the impacts of using race variables in clinical algorithms in ways that conflate race with biology, if we have these concerns and we know that it may be delaying care or harming care for people of color, then we need to do something about it urgently. We can’t wait to make these changes across all of the health systems, clinics, hospitals, and health care delivery settings across our country. And part of also what the challenge is is that our health care system and public health systems are very siloed. The electronic medical records are not unified. The clinical training and practices are not centralized. And pathology labs that really are responsible for the reporting of laboratory results across health systems, clinics, and hospitals, how do we get them involved? They’re all very decentralized just because of the way that our health care system is constructed. And so that’s another reason that we launched the coalition in the fall of 2021 because we also said, “Well, each health system doesn’t have to figure this out on their own.” And there were nine hospitals and health systems across New York City, including the six largest health systems, that said, “Yes, we see this problem. We are going to come to the table and agree to these three main goals.” And the first was to reexamine the use of race in three priority clinical algorithms: kidney function, lung function, and vaginal birth after cesarean section. Rachel Gotbaum: And what happened? Michelle Morse: So the members of the coalition agreed to end the use of race in those clinical algorithms, adopt a race- neutral version of those algorithms. The second was that they agreed to measure the impact on racial health inequities of changing those algorithms. And then the third was to develop patient-engagement plans, to make sure that patients whose care was impacted by the use of these algorithms or even delayed or harmed by the use of these algorithms actually were aware of it and were prioritized for getting the care that they should have gotten or that had been delayed because of the use of the algorithm. And that’s what we see as the role of the New York City Public Health Department. In this we’re the convener, right? We are agnostic to hospital or health system. Our best role in many ways is to bring them together and to say, “This is a really meaningful health-equity intervention to make these changes,” and then make sure that patients whose care is impacted are both aware and at the table to make sure that these changes continue. Darshali Vyas: One of the things I wanted to add, we talked a little bit about how the VBAC calculator officially changed in May of 2021, and that was, I think, the first example of a tool systematically removing the race adjustment in response to an equity concern, which is amazing. But a lot of these tools are very different. So whereas the VBAC calculator is a website that you access online, and once that coding has changed, the whole calculator has changed, most of the tools that use race in this way are not just online calculators that can be instantly changed. And so things like the kidney- function measurement that uses race, a lot of those are lab-based or pulmonary function tests. And so even though the professional societies release updated guidelines or recommendations that say we should stop using race, there is a real disconnect between the professional societies changing their guidelines and things actually changing on the ground at the hospital level. So just because PFTs are now changed and we should technically all be using a race-free version of pulmonary function tests, that doesn’t mean that the pulmonary function test labs at a hospital in New York or in Massachusetts necessarily are using the new and updated version. And so that’s where I think groups like CERCA are really the model of how you bridge that divide. Rachel Gotbaum: So despite the efforts of doctors like yourselves, it still seems that implementation of race-neutral algorithms is haphazard. Dr. Morse, what needs to happen here? Michelle Morse: I think health care organizations really need clear guidance to their clinicians about how to talk to patients about this. And we need to be able to explain the historic use of race in these algorithms as well as the changes that are being made in a way that patients know that health care institutions are trying to do better and actually want to deliver more equitable care. Another thing that I think is imperative is that payers, whether that is Medicare, Medicaid, or managed care organizations, we need to see changes in behavior across health systems. https://www.nejm.org/doi/full/10.1056/NEJMp2407611 4/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 And because the lever of payment is so powerful in driving behavior of clinicians and health care systems, that work also needs to happen. And perhaps even penalties for organizations that are still using outdated forms of the clinical algorithms, because this actually has material impact on the lives of patients of color. And then the third thing is that health care organizations are still struggling to just collect self-identified race/ethnicity data. We can’t assess racial equity in health care if we don’t have reliable race/ethnicity data. And that’s something that I think is very important for us to be clear about in this conversation. We are not saying that we should stop collecting race/ethnicity data. In fact, we’re really saying the opposite. We need to know what the impact of our policies and clinical decisions are on racial health outcomes. Darshali Vyas: I wanted to underscore we should absolutely continue to study racial disparities to understand how race affects our patients’ health. I think where we exert caution is using race in predictive tools when race-based medicine could actually be functioning to keep non-White patients from accessing resources compared to White patients. So rather than building race into tools like the VBAC calculator, using our knowledge about race-based inequities to guide interventions that are specific in targeting the actual barriers to accessing care, whether it’s transportation or vouchers to get to doula appointments or to have a midwife or actually doing the more granular data, digging one step deeper to understand not just that racial inequities exist in our outcomes, but why they exist, and then using that more granular data to design effective interventions that actually target the reasons why. Michelle Morse: There are a number of ways that we can actually use variables like race/ethnicity to make these changes. At Brigham and Women’s Hospital, we found that systematically, Black and Latinx patients, with all other things being equal, were much less likely to go to the cardiology specialty service when they came into the hospital with heart failure. So we used the electronic medical record and a clinical algorithm to try to fix that trend. So that best-practice alert is used in the electronic medical record to say, “Hey, reminder, this patient with heart failure is from a group that historically has been excluded from the cardiology service. You should really consider where you send that patient because of that historical exclusion.” Those are the kinds of ways that we can both consider race/ethnicity, we would call that race consciousness, consider how systems of racism have impacted care, and then we can counteract it with specific interventions that are meant to reverse a historical pattern of exclusion of Black, Latinx, or other communities of color. Rachel Gotbaum: Thank you both so very much. Michelle Morse: Thank you. Darshali Vyas: Thank you. Thank you for doing this. Rachel Gotbaum: That’s Dr. Michelle Morse, chief medical officer at the New York City Department of Health and Mental Hygiene, and also Dr. Darshali Vyas. She’s a pulmonary and critical care fellow at Massachusetts General Hospital. We had help from our managing editor, Debra Malina. Our engineer is Adam Straus. For more on the impact of race algorithms, go to our Race and Medicine page at NEJM.org. Next time, for patients who can no longer communicate because of disease and injury, new technology that can read brain signals and translate them into spoken language. Speaker 6: This is a step in demonstrating just how quickly we might be able to restore fluent and easy communication for somebody with ALS. Rachel Gotbaum: That’s next time on “Intention to Treat” from the New England Journal of Medicine. I’m Rachel Gotbaum. NOTES All interviewees report having no potential financial conflicts of interest. ADVERTISEMENT https://www.nejm.org/doi/full/10.1056/NEJMp2407611 5/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 PHYSICIAN JOBS AUGUST 3, 2024 Internal Medicine Cortland, New York Internal Medicine - $100k Starting Bonus! - Cortland, NY Hematology / Oncology Kentucky Bowling Green, Kentucky - Hematology/Oncology Psychiatry East Providence, Rhode Island Adult Psychiatrist Internal Medicine Vermont Internal Medicine Physician Opening in Central Vermont - Signing Bonus Geriatrics Seattle, Washington State Geriatric Medicine Physician Cardiology Fort Myers, Florida Heart Failure Specialist (Physician) MORE PERSPECTIVE PERSPECTIVE AUG 03, 2024 Preparing for the Next Pandemic — Expanding and Coordinating Global Regulatory Capacity S. Halabi and G.L. O’Hara PERSPECTIVE AUG 03, 2024 Collective Trauma and Commemoration — A Moment of Silence, Please G.J. Makari and R.A. Friedman PERSPECTIVE AUG 03, 2024 Family History: Unknown A.R. Abruzzo https://www.nejm.org/doi/full/10.1056/NEJMp2407611 6/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 YOU MAY ALSO LIKE EDITORIAL | JUL 31, 2024 From Evidence to Policy — Finding Authoritative Sources of Information on Health Eric J. Rubin and others CLINICAL DECISIONS | MAY 16, 2024 To Preround or Not to Preround Abigail Schubach and others PERSPECTIVE | MAY 9, 2024 All’s Fair in Love and Work Tessa Adžemović PERSPECTIVE | JUL 25, 2024 Should I Laugh at That? Coping in the Setting of Serious Illness Danielle Chammas MORE FROM VOL. 391 NO. 5 ORIGINAL ARTICLE AUG 01, 2024 CME Belantamab Mafodotin, Bortezomib, and Dexamethasone for Multiple Myeloma V. Hungria and Others ORIGINAL ARTICLE AUG 01, 2024 Autoantibodies Targeting Nephrin in Podocytopathies F.E. Hengel and Others ORIGINAL ARTICLE AUG 01, 2024 Belantamab Mafodotin, Pomalidomide, and Dexamethasone in Multiple Myeloma M.A. Dimopoulos and Others ARTICLE CATEGORIES RESOURCES Research Authors & Reviewers Reviews Submit a Manuscript Clinical Cases Subscribers Perspective Institutional Administrators Commentary Media Other Advertisers Browse all Articles Agents Current Issue Permissions https://www.nejm.org/doi/full/10.1056/NEJMp2407611 7/8 8/3/24, 12:59 PM Race-Based Diagnosis, Part 3: ITT Episode 35: New England Journal of Medicine: Vol 391, No 5 Issue Index Reprints NEJM CareerCenter ABOUT US SUBSCRIPTIONS About NEJM Subscribe NEJM Group Renew Products & Services Activate Subscription Editors & Publishers Create Account Advertising Policies Manage Account Contact Us Pay Bill Accessibility Institution Subscriptions FAQs Special Content Help Site Feedback STAY CONNECTED FOLLOW US Email Alerts Facebook Create Account X (formerly Twitter) Apps Instagram NEJM CareerCenter Youtube Podcasts LinkedIn RSS Feed Remote Access JOURNALS The New England Journal of Medicine NEJM Catalyst Innovations in Care Delivery NEJM Evidence NEJM AI Copyright © 2024 Massachusetts Medical Society. All rights reserved. Electronic ISSN 1533-4406. Print ISSN 0028-4793. The content of this site is intended for health care professionals. Copyright Terms Privacy Policy Cookie Preferences https://www.nejm.org/doi/full/10.1056/NEJMp2407611 8/8