Racial and ethnic disparities: What can we learn from RWE?

Biographies

Mike

Kathleen Maignan, AGPCNP-BC, MSN, OCN

Senior Clinical Director

Research Clinicians
Linkedin

Kathleen Maignan, AGPCNP-BC, MSN, OCN joined Flatiron in February 2017, bringing an extensive background centered in hematology-oncology nursing and clinical research. As a Senior Clinical Director on the Research Clinicians team, Kathleen uses her expertise to help answer important research questions using the data available from our Flatiron network. Kathleen also continues to practice as a Nurse Practitioner focused on Bone Marrow and Stem Cell transplantation.

Read more

Mike

Lura Long

Senior Manager

Partner Solutions
Linkedin Twitter

Lura Long leads the Partner Solutions team at Flatiron Health, managing efforts to equip data scientists with information and tools to effectively use Flatiron real-world data. In prior roles at Flatiron, Lura contributed to functionality within OncoEMR ® to increase patient access to patient assistance programs and managed early efforts to identify patients for clinical trials using electronic health record information.

Lura came to Flatiron by way of the Michael J. Fox Foundation where she managed digital recruitment and website development for patient reported outcomes studies, trial matching and patient engagement programs. She’s driven to close inequities in healthcare outcomes via software and evidence based research.

Read more

Mike

Kelly Magee, FNP-BC, MS

Senior Clinical Director

Research Clinicians
Linkedin

Kelly Magee, FNP-BC, MS is a nurse practitioner who serves as Senior Clinical Director on the Research Clinicians team at Flatiron Health. In this role, Kelly focuses on the development of novel approaches for using electronic health record data to understand the experiences of people with cancer.

Prior to joining Flatiron Health, Kelly was Nurse Practitioner Clinical Program Director at Memorial Sloan Kettering Cancer Center, leading nurse practitioners across the institution and maintaining a clinical practice caring for critically ill medical and surgical inpatients in the Advanced Care Unit. Prior to that she worked as a Staff Nurse at Maimonides Medical Center in Brooklyn, NY.

Read more

Mike

Alex Gorstan

Director

Research Marketing
Twitter

Alex Gorstan is a marketing director at Flatiron Health and the host of this episode of The Real World, In Practice podcast. He partners with experts across the organization to drive the measured and meaningful application of real-world evidence in oncology. Before joining Flatiron, Alex worked in Product Management and Software Engineering at companies including Opower, Johns Hopkins Medical Institute and CareFirst Bluecross Blueshield

Special thanks to Tanya Elshahawi, Lesley Plotkin, Dan Reyes and Chad Michael Snavely for their work on this episode.

Click here to listen to the previous episode, "How COVID-19 is changing cancer care, health policy and research—for good"

Read more

In this episode:

  • 07:51

    Are our data fit-for-purpose to answer questions about racial disparities?

  • 12:34

    Have we been naive to think that the standard “Table 1” of patient characteristics accurately characterizes our study patient population?

  • 15:55

    Can we apply real-world data to evaluate the impact of policy interventions?

  • 22:57

    And finally, if race is a driver of patient outcomes, how do we fix it?

Transcript

00:02 Alex

Alex Gorstan This is The Real World, In Practice, a podcast by Flatiron Health. I'm Alex Gorstan. If I asked you, what's the best thing that we can do to improve outcomes for people with cancer, what would you say? Invest more in novel therapeutics? Broaden trial inclusion criteria? Standardize broad based genomic sequencing? What if we also thought about resolving the disparities that we see for people of color or different ethnic backgrounds? What would that impact be?

00:38 Kathleen

Kathleen Maignan My name is Kathleen Maignan. I'm a senior clinical director at Flatiron working on the research side of the organization.

00:45 Alex

Alex Gorstan Kathleen is part of a team at Flatiron working to answer questions about race and ethnicity, whether a person's ethnic background or racial identity affects the care they receive and what we can do about it.

01:02 Kathleen

Kathleen Maignan I'm a nurse practitioner by training. I still see patients at Weill Cornell. Before I worked in any sort of clinical capacity, I actually had my first experience in oncology at a tech startup when I worked at Medidata back when it was a startup about 15 years ago. And around that time is when I really started to get interested in clinical research and what data can tell.

I sometimes find it astounding that people aren't interested in disparities. If you say you're interested in any sort of cancer research, our day to day is focused on outcomes, how do patients do? How can we get them to live longer, live better, live well?

And we know that there's a huge part of our population that doesn't do as well as others, and that's a huge thing. How can you get equity across the board? And when you know for sure that you have patients that don't perform as well, it's important to understand why, what is it about them? What is it about us that isn't meeting their needs?

02:16 Alex

Alex Gorstan For Kathleen, it was her lived experience with these issues that led her to seek answers.

02:21 Kathleen

Kathleen Maignan I had a family member, very close family member who was like a grandfather to me who passed away from multiple myeloma. And I noticed during the trajectory of his care, he wasn't always offered the best care available. He was an immigrant from Haiti who spoke perfect English. I mean, he worked for the LIRR for 30 years, but it was often dismissed because he was older. He was a black man, he was an immigrant.

It wasn't until they would open up his insurance profile and say, "Oh wow, this guy has really great insurance," that they would pay him much attention. But other than that, there was no effort to really make sure that he understood his disease, understood his prognosis and what he could do.

03:10 Alex

Alex Gorstan The assumptions and the expectations that others had for him led him to assume the worst and he began to prepare his family for his death.

03:18 Kathleen

Kathleen Maignan I had gotten a new camera and he comes down the stairs one day, we're eating dinner and he's like, "Okay, I got two new suits, let me know which one I should wear to my funeral." So he was planning his outfit and wanted pictures to make sure that we had really good pictures for him to put up at the funeral home. But that should never have been the conversation, it should have been like, "There's treatment for this and we can actually take care of you."

03:47 Alex

Alex Gorstan When he brought his charts back for Kathleen to look at, she's saw the same thing the doctors did, but she arrived at a very different prognosis. She knew that with the care she had seen given to others that he could live far longer with this. So she took him on as her own patient. And under her care, he lived five years with this diagnosis. Ultimately he passed away, but it ignited her desire to make things right for more people like him.

04:14 Kathleen

Kathleen Maignan When it comes down to my myeloma, African-Americans are overrepresented. They represent at least 20% of patients with myeloma. But then when you look at upfront trials, prospective trials, they represent sometimes 1% to 4% of the trial population. So how can I say for sure to a family member that I feel that this is a safe drug when I don't even know how this drug performs on patients that look like myself or any of my family members.

04:52 Alex

Alex Gorstan So many people in the industry recognize the limitations of clinical trial data. The question now is whether or not a database of real-world data, data gathered from routine care outside of clinical trials, can actually address these limitations.

05:06 Kathleen

Kathleen Maignan We actually have an ability to look at outcomes for all of our patients that are represented in our database. We have about 20% of our patients that are African-American. It's a disservice if we're not actually asking these questions because not enough people are. Real-world data is great in that, yes, it can help us fill gaps when we're looking at drugs in a post-marketing setting, drugs that have already been approved, we can see how patients perform on those drugs once they get access to those drugs.

05:39 Alex

Alex Gorstan But does real-world data solve the problem of under-representation of African-Americans in clinical trials?

05:45 Kathleen

Kathleen Maignan Absolutely not. We still need to know how patients perform in a prospective trial.

05:56 Alex

Alex Gorstan Prospective clinical trials allow us to account for multiple factors that can otherwise complicate analysis, like ensuring consistent timing of assessments or drug administrations. And randomized prospective studies have the power to erase the effects of unmeasured confounder. So they play an essential role. Yet when Kathleen was faced with the reality of clinical trials, she sought answers from real-world data and her work published in 2018 at the American Society of Hematology revealed a troubling insight.

06:26 Kathleen

Kathleen Maignan The abstract that I wrote looked at use of a novel therapy called Daratumumab in patients comparing African-American patients to white patients. And what I found is that even in the post-approval setting, uptake for African-American patients was much slower than white patients. So what does that really mean?

Newer therapies are not getting to African-American patients as quickly as it would white patients. And we looked at all the different factors that could affect this. We looked at payer information, we looked at where in the country these patients lived, we looked at comorbidities. And for this particular drug, co-morbidities isn't a huge issue, it's great for patients with all kinds of organ dysfunction, but the only difference was race.

And so as much as I want to know how African-American patients perform on Daratumumab, if they're not getting it as often, how can we really answer that question? And so, yes, real-world data can help, but it's not the only thing, we still need to figure out why people are not enrolled on trials and we need to figure out why they're not getting newer therapies when they become available.

07:51 Alex

Alex Gorstan Doing good research means using data that's fit to answer the particular question you're studying. Working with real-world data, Kathleen and her team have come against a problem, how do you categorize someone's identity?

08:04 Kathleen

Kathleen Maignan We have our five set categories, but we also live in America, people are not just one race.

08:11 Alex

Alex Gorstan If you're multi-racial, if you fit in a race category that doesn't exactly check off a box, that can throw things off. So to understand more, we wanted to speak with someone who knew the data inside and out.

08:23 Lura

Lura Long My name is Lura Long and I'm focused every day on making it easier and better to work with Flatiron data.

08:29 Alex

Alex Gorstan Lura is the daughter of a public defender and singer songwriter activist and she's another member of the racial and ethnic disparities team. She grew up hearing the stories of people who had very different experiences than she was afforded, of people being treated unfairly.

And as she led her career towards the intersection of healthcare and data, she finds herself now at get another intersection asking, how can the stories she heard in her youth be retold and validated with data? And again, this starts with getting the data right.

09:00 Lura

Lura Long Let's focus on one data point. Let's say the data point we're focusing on is a patient reporting that they are black American. So a patient would report that often in the context of an intake form when they first go to a doctor's office, for example. So they go to their first appointment, they're handed a paper, they fill out a lot of information about themselves. And one of those points is, what do you identify as racially and as your ethnicity? So a patient may report I'm black, I'm not Hispanic in that form. That form then will get often transcribed into a software. Often that is the EHR software, it could also be a billing software such as a practice management system.

09:41 Lura

Lura Long And from there, you have hundreds and hundreds and hundreds and hundreds of identities that patients have reported as their race, for example, and then for ethnicity typically you only see two, so you see Hispanic or non-Hispanic or you have a nul available for the patient. So from there, Flatiron's job as part of the processing approach is to take all of the hundreds of identities and map them to common terms. And you want to do that with terms that are relied upon in industry.

After the data is reported by a patient and is recorded by the physician in the software and we arrive at the list of hundreds of racial identities that then need to be mapped, at this stage, of course we do de-identify the records of those individual patients so you're no longer able to recognize who the individual in any identifiable way before moving forward with processing the data for use in a research context.

10:43 Lura

Lura Long And there's actually a really interesting tension there that I think about a lot, which is that there are so many nuances to an individual's identity, and there are so many different identities within each of these kind of higher level categories. So for example, Asian, I'm sure there's literature out there that confirm that the lived experience of people within those different identities that might fall under Asian is likely not the same, but for the purposes of research, both to have de-identification standards that are top priority for us and also to ensure that you have a large enough cohort to be statistically meaningful, we have to combine these terms into these top level categories, which means you do lose some of that really interesting granularity by the time the data gets to a researcher

11:33 Lura

Lura Long So I was doing a lot more direct action work when I was younger and I really became frustrated because I was so angry by the stories I was hearing and I knew just fundamentally how true the difference in experience is for different populations and yet I didn't feel like I was seeing that action in politics or in society to address those things and I saw data work as a way to potentially drive more impact. Our working group is focused on identifying all the ways that Flatiron could have an increased impact on the field of race and ethnicity disparities research.

So I'm currently thinking a lot about, what do we know about our race and ethnicity information? What do we know researchers want to understand about race and ethnicity? And is there more we can do to increase the granularity of the data that we're able to provide and provide more transparency about where the data comes from so that it's more actionable for researchers?

12:34 Alex

Alex Gorstan If you're at all familiar with peer-reviewed research, you've probably seen it a hundred times. It's Table One, a fairly mundane attempt at summarizing the people included in the research study. Despite its routine role in research papers, it often brings to the surface a failure to address something that's really quite deep and culturally important, the fact that much research struggles to include patients that look like the rest of us and that when we try to characterize variations in that group, we don't paint a picture that's as vibrant as it is in reality.

13:04 Lura

Lura Long I find it frustrating that there's a lot of data upstream about so many different variations that likely do impact differences in experience that we're not able to deliver in a research context and I think that speaks to the importance of scale. So as you get a larger and larger, data set, and as we're seeing potentially being able to revisit now, as we have, so many more records than we did when we initially built our first datasets, you'd think you'd have enough patients with a certain background to be able to, in a very confident de-identified fashion, deliver that really detailed identity through to research so it can be acted upon is important.

13:48 Alex

Alex Gorstan And it's not just these issues that make Lura's job more difficult.

13:51 Lura

Lura Long There's so many aspects of racism. So there's power systems set up or not set up to treat people fairly and equally. So questions such as, are patients of different racial identities receiving the newest treatment close to when that newest treatment was released? You can observe that, is it enough that you observe that there's a difference and that you're able to control for other factors that could be driving that difference and you're able to isolate that the difference is tied to racial identity, is that enough to say we should do something about this?

I would say, yes, it is demonstrably true that people have different experiences. There is data to back that up. There's a difference between providing descriptive information that tells you something versus information that also can feel really actionable. When you seek to understand something about a patient's lived experience or a populations experience, try to answer a question that people can act on.

14:48 Alex

Alex Gorstan And the fact is, people are taking actions. In 2019, some research authored by members of this team got some pretty big exposure as big as it gets in the oncology world.

15:04 Kelly

Kelly Magee Yeah, it was really exciting.

15:06 Alex

Alex Gorstan This is Kelly.

15:10 Kelly

Kelly Magee Hi, my name is Kelly Magee, I'm a nurse practitioner by background.

15:13 Alex

Alex Gorstan Kelly was part of the study team who coauthored research looking at the impact of the Affordable Care Act on racial disparities for cancer patients. It was accepted for presentation on the main stage ASCO in 2019.

15:25 Kelly

Kelly Magee It is obvious when you're in healthcare the disparities faced on all different types of levels by patients, everything that dictates people's health status, access to nutrition, access to opportunities, to exercise, smoking rates, access just to healthcare, and then ability even to manage issues that are facing them once they've been diagnosed. We looked at the impact of the Affordable Care Act on racial disparities and time to treatment for cancer patients.

It was really kind of a, I don't know, landmark moment for me to see that Flatiron Health data and real-world data like ours could be used to evaluate the impact of policy interventions.

16:23 Alex

Alex Gorstan Flatiron focuses a lot on drug development and other therapies so it was uncharted territory for what real-world data had to say about policy effectiveness.

16:32 Kelly

Kelly Magee Yeah so we found that states that did expand Medicaid eliminated the disparities between black and white patients in terms of time from diagnosis to time to treatment start. So states where they expanded access to care were able to reduce and eliminate the disparities as compared to states that did not do that.

16:59 Alex

Alex Gorstan That's huge, yeah.

17:00 Kelly

Kelly Magee Yeah, huge. That was our big headline finding.

17:03 Alex

Alex Gorstan I should say that again, that was huge. In states where these policies weren't in place, on the average, treatment started sooner for white patients than it did for black patients and policy makers with the stroke of a pen were able to erase that treatment disparity for black people with cancer.

17:22 Kelly

Kelly Magee Of course, we also learned a lot along the way of just how to do these studies and even just saying that you may have noticed that we had to limit to white versus black patients. So one consideration in all of this is even in large data sets like ours, you can end up with small subsets of patients of specific races and ethnicities that are too small to power understanding those types of differences effectively.

That gets into some of the work we've been focusing on this year in terms of maximizing capture of patients; race and ethnicity data, and how we handle that from point of care to our research data sets so that we can support ideally some of those comparisons by having larger representation of all different populations.

18:14 Kathleen

Kathleen Maignan So yes, we're making progress. This research has been able to show measurable results way beyond just pointing a finger at the problem, results that other policymakers can look to as they work to improve healthcare for everyone. So it's encouraging and we're making progress on the long road.

18:33 Lura

Lura Long One of the areas that's really still challenging and the field still hasn't addressed in a unified framework is how to actually assess whether the data you're using is appropriate for your population of interest. And as a result, researchers are hearing a lot of feedback in the form of responses to their publications from journals, responses to their funding applications in the academic research space and even in the regulatory context, hearing questions about how researchers have gone about answering this question of is the cohorts that they're using representative of their population of interest and then therefore are their findings generalizable to that population.

Part of our focus area in 2020 is to work closely with external and internal researchers to better equip partners to walk through that decision-making process. I think it will make a big difference for people to feel that they at least have a starting point and there's a common language for talking about these things

19:38 Kelly

Kelly Magee Publishing on racial disparities and just continuing to show it just kind of doesn't feel like enough to me and I'm sure that's true for a lot of people. It's at this point very well known and well described. So taking it to the next level, which to me can manifest in a few different ways and was one reason that the Affordable Care Act project was so important was that it actually showed the impact of an intervention versus just describing a situation.

20:12 Kelly

Kelly Magee There was a recent Health Affairs blog about the responsibility of researchers and publications around when we do publish racial disparities data, not avoiding the topic of systemic racism in describing our findings. So a lot of papers you read, the tradition unfortunately has really just been to sort of comment on their societal factors or kind of vague references to unmeasured confounders that might be influencing why we have found a racial disparity in a given project, and we really have a responsibility to start naming what's going on and using the words racism, and systemic racism in those publications.

21:08 Kelly

Kelly Magee I think the big question of 2020, right, is, is this lasting change? What will stick around? And I think for me, the parts that have been reassuring are some of the professional organizations that have started to come out more vocally, particularly, there was a column in the New England Journal of Medicine that spoke very explicitly about this and it's kind of unheard of for professional societies in the past to have taken such clear stances.

21:43 Kathleen

Kathleen Maignan Representation matters, period.

So if you think about lasting impact, lasting impact comes from people in positions of power actually using their influence and their power to do more and do better. Unless we have, we the collective we, have representation where it matters, these changes aren't going to last a long time. I mean, I think you need people who have a vested interest in serving the underserved in order to keep this going.

And so I would love to see people actually getting paid to do equity research, which is not far-fetched when you work at a research company, I did not get paid to write the ASH abstract, neither did Kelly get paid to do ASCO plenary, that came from passion, it came from interest, it came from just working after hours on things that motivated them. Imagine who and how many people you can motivate by actually making this part of their job. And we actually have the power to do that. And so that's exciting.

22:57 Kathleen

Kathleen Maignan Race is the category nobody really wants to accept as a primary driver in patient outcomes unfortunately, just due to the fact that it's a very hard problem to solve, right? It's easy to say, well, if you give a patient this drug or you treat them in this particular way that we can influence an outcome, but when race is the driver and an outcome, how do you fix that? How do you fix the way they are perceived? How do you fix the way providers treat patients that don't look like them? Those are big questions.

23:35 Kathleen

Kathleen Maignan And it's also hard sometimes except the fact that you may be a driver in some of this, you as a provider, that's a hard prospect to face when you know that your intent really is to do well by patients, but you might be the source of your patient's not getting the treatment that they deserve. Yeah, that's really hard because there is no easy fix. But again, until you acknowledge that the problem exists and part of that is actually doing this research, I don't think you can fix it. So gathering data, while it doesn't sound very sexy and it's not the fastest route, it is absolutely essential. It's where you start.

24:32 Alex

Alex Gorstan To learn more about what real-world data is teaching us about the experiences of every cancer patient visit rwe.flatiron.com. Thanks for listening.