This month’s podcast focuses on structural racism.
Welcome to the Healthy Intersections podcast for November, 2023. This month’s podcast focuses on structural racism in the United States. Joining us to talk about the new Structural Racism Effect Index (SREI) is Dr. Zach Dyer, lead author on the analysis. Check out the dashboard at SREIndex.com and the journal article in Health Affairs.
Audio here (also available on your favorite podcast streaming service).
Transcript below!
Transcript – Healthy Intersections, Nov 2023
Lisa Lines (LL):
Hello, and welcome to Healthy Intersections for November 2023. I’m Lisa Lines and the host of this podcast. I’m also a senior health services researcher at RTI International, an independent nonprofit research institute. And I’m also one of the editors of The Medical Care Blog, and that’s where we’re hosted. So, if you’re listening to this on Spotify or Apple Podcasts or anywhere that you get your fine podcasts, I want to encourage you to go to our blog to see the video that comes along with this audio. We’re going to be looking at some interactive map displays today. So, I hope that you all will go to our website, themedicalcareblog.com to see the video. And subscribe, so that you’ll hear when we come out with a new episode.
And with that, I’d like to welcome our guest on the podcast this month, Zach Dyer. Welcome, Zach. Thank you so much for joining us today.Zach Dyer (ZD):
Thank you for having me. I’m excited to be here.LL:
Well, I’m so excited to talk to you. For the audience, just so you know, we are both people who did our PhDs with Arlene Ash, who is one of the leading lights in health services research and applied statistics and mathematics and risk adjustment and absolutely one of my heroes. One of the best experiences I could have possibly asked for to have gone through my PhD with her. And Zach, you also worked with Arlene for your PhD at UMass Chan Medical School. Tell us tell us how you ended up there, and how did you get to that program and end up working with Arlene?ZD:
I grew up in Worcester, grew up going to UMass Memorial as my hospital where my, you know, pediatrician was. And I always knew I wanted to go to medical school. But after undergrad was really, or during undergrad, I got interested in public health and decided I wanted to dig into that a little more before I applied to med school. I did my MPH in Boston, which brought me back.
I was commuting from Worcester, so I was back in Worcester, did not expect to be, you know, I thought it was just a brief time I’d be back while I was doing that degree. And then as part of that I ended up doing a practicum with the Worcester Division of Public Health, which turned into a job, which turned into me being there for about 5 years.
And as I was applying to medical school, I knew a lot of the researchers at UMass in the Clinical and Population Health Research program, which is the PhD program we went through. And yeah, I think there was a lot of hemming and hawing for a little bit. But I ended up deciding to stay. And I’m very glad that I did. I get to be connected to all the work that I was doing before and also do new things and learn new things.
And I should say I’m an MD/PhD student. I finished my PhD a little more than a year ago, so just finishing up the MD now. As I was starting my PhD work, I’d done a lot of community health work at the time in Worcester and felt like I knew that area of public health population health really well. And I said, you know, I want to get into bigger policy questions and ended up working with the MassHealth ACO evaluation team, and Arlene is part of that. And that’s the road my research was heading on for a while and still applying the social determinants/health equity lens that I’ve been working in.Then the pandemic hit and I shifted my focus a little bit. I think we’re having a lot of conversations about changes to “the system”, all systems. And I said, you know, I really want to pursue this for my dissertation research. And Arlene was very supportive and it lined up with the work that she was doing. And that’s how I ended up working with Arlene. And I’m so glad that I did. I’ve learned so much from her and continue to learn so much from her. And I think that the work we’ve done together is, it was great and something I’m very proud of.
LL:
As you should be! It is great work. It is fantastic work. And so not only were you doing your PhD during the pandemic, you were also working at the Department of Public Health there in Worcester, right.ZD:
It was an awkward time for me because, and I guess maybe one of the reasons why I think I shifted my focus a little bit was that, you know, I was in medical school up until fall. I was doing the medical school portion of the MD/PhD up until fall of 2019, then started the PhD portion. I was in my first year when the pandemic hit, hadn’t really been engaged in Worcester Public Health for a little while.
And I suddenly found myself in this position where I wasn’t part of the medical system that I had been in very recently to be able to, you know, help out with what was happening there.
And I wasn’t directly in the public health system, and I felt very– I’m a doer — So I felt very paralyzed in that, I wasn’t able to help out during the pandemic. And I ended up finding other things.But you know, I think it was that discomfort that led to me shifting my focus and thinking a lot more about equity and antiracism. And then after I finished my PhD, I did go back to the Division of Public Health for a year. I took a leave of absence from school, went back for a year, and helped them transition out of the pandemic into our new normal.
LL:
Quite a journey. And I’m sure that your work is really profoundly influenced both by, what we’ve all been through, but also the experience of just, you know, seeing it right up front in front of you. You know, I’m also a doer, and I think the impetus to do something is a really, really beneficial and powerful place to come from. And you know, in my case, growing up poor and growing up in really desperate circumstances informed my wider feeling about, you know, inequity and poverty and the influences on us, as we’re trying to be healthy but are blocked by so many things in the places where we live. There can be great assets and then a lot of not-so-great things.So, I wonder, as we start to transition and talk about your work some more, can you give us an idea about how you came up with this piece of work that we’re going to be focusing on today? I would like you to introduce it.
ZD:
Sure. Arlene and her research team had developed the Neighborhood Stress Score (NSS) and it’s something that MassHealth uses in their Risk Adjustment payment models, and it’s a pretty simple measure of neighborhood stress. It uses 7 variables from Census. And we are comparing it to the Area Deprivation Index which is widely used. It’s available across the country. The NSS is just available in Massachusetts.
And we’re noticing that there are really big differences in different areas. And one was in Boston. I’m familiar with Worcester, so I can look through Worcester neighborhoods and say, this makes sense, but this doesn’t make sense. I had also studied in Boston, knew some of the neighborhoods, especially around Boston Medical Center, saying this, what I’m seeing isn’t making sense for these two different scores in either direction, what is behind this? And we started to dig into it.
And at the same time, you know, we were having a lot of conversations about racism and antiracism and how we often let the root causes of the social determinants of health go unspoken. So many different things happening at the same time we’re looking at these measures.
You know, it’s during the pandemic, we’re having a lot of conversations about equity and racism and trying to really, you know, commit ourselves to looking at things differently. And one of the thoughts that I had during that time was, you know, if we approached these measures from a place of recognizing that racism has shaped so much of the distribution of these social determinants.
I bet we’ll do a better job of, you know, tracking health outcomes or you know, well-being and prosperity that that if we can, you know, start from that place and move forward from there and use that as our tool or our framework to pull in different information, different data about neighborhoods, then we’ll have a better sense of what’s going on.
That’s when we started talking about the tool that we now use, that was part of my dissertation research and is now published and available, which is the Structural Racism Effect Index or SREI. Either way it’s a mouthful but… it started in 2020 and it’s great in 2023 to see it out in the world and being used.LL:
I have that same feeling. We’ve been on a very parallel journey in fact, as you may know, and Arlene and I have been in conversation along the way. I think she said that the idea to compare to life expectancy was shared between our projects. We’ve been doing similar work in terms of collecting as much data as possible and creating an index across the whole US. So, let’s go ahead and have a look at the map that you’ve released. I mean I’m so impressed by your publication in Health Affairs. We’ll put a link to that in the description for the podcast and hopefully folks will also be able to go to see the SREIndex.com.ZD:
I’ll go ahead and pull it up here. This is the, the homepage for the Structural Racism Effect Index. It’s at SREindex.com and it’s a very simple website at this time. There’s a map and there’s the data to download and some extra information about methods. There’s a link to the Health Affairs article here.
I think if anyone wants to understand how we built this, that’s the best source of information. But I’ll go ahead and pull up the map here.Something I didn’t talk about is that, one of the conversations that we were having thinking about COVID is how important neighborhood is and how the resources that are available in your neighborhood are so important to your overall health and prosperity and really shapes a lot of health outcomes. We’re seeing that very clearly in COVID where two towns or two neighborhoods right next to each other are having vastly different outcomes. And you know, the thing I always say is that it’s to see a infectious disease, a novel viral illness, follow the exact same pattern as chronic disease prevalence tells you something that that should raise an alarm. You know, there’s something else going on here. We believe that neighborhood is really the unit where a lot of the different policies, you know, racist policies and decisions about how resources are distributed end up. Neighborhood is so important to that conversation.
I’m going to pull up Worcester, of course. And so we the map here, I’ll say for those of you who can’t see it, the lighter areas—I’ll zoom out—the lighter areas on the map are where there are the greatest effects of structural racism.So we see a lot in the Southeast United States, but really everywhere in the country you’ll find different pockets. And the darker purple areas are where there are the least effects. We think of it as resource rich. Those areas that are more deeply colored correspond to the resources that are available.
If we zoom in on Worcester here, you can see there’s a pretty wide variation in the different scores, although of course, there’s a wider variation other places in the country. But if we click here on maybe, oh yeah, on Union Hill, this neighborhood here, it’s going to pull up information about, it’s going to give you the overall score percentile. And then there are nine different domains that are a part of the Structural Racism Effect Index.
For those of you who are listening and can’t see it on the screen, it’s built environment, criminal justice, education, employment, housing, income and poverty, social cohesion, transportation, and wealth.
We got those domains from looking at the structural racism literature and asking how are other people measuring this really difficult-to-measure concept. What sources of data are people looking at? What are the different areas? Trying to make as an exhaustive list as possible and seeing what buckets they fall into. And these nine areas are what we were able to find data to back it up with.And there are a lot of, you know, indexes, indices out there that that do similar things. But some of the most novel domains that we have are criminal justice, for sure, and you know, maybe social cohesion. But our approach to other domains is also a little different.
On the website, when you click, you’ll get the information on the scores in those nine domains, the scores, the score for the SREI, the national percentile where it ranks, and it also gives you life expectancy and then the ethnic racial composition of that area on the website as well. It’s available for almost every census tract in the country—I think 99% of the population is covered by an area that’s scored. And all of the data is available to download from our website.
LL:
Fantastic. And it’s coming from publicly available data. There’s no proprietary anything in here. And what you’ve done, your methods are pretty straightforward. You’ve oriented everything in the same direction. So that higher means worse. Is that correct?ZD:
Yeah, that’s correct.LL:
And so, the higher the SREI and that ranked percentile score, I assume that’s what that is, the higher the disadvantage.
ZD:
We say the higher the effects of structural racism or the more compounded the effects of structural racism.LL:
Got it. I’m curious now, to look at yours and ours, side by side. I imagine we would probably see very similar things, because we do have very similar domains and similar measures, but your methods are quite different. So tell us about your methods again, I thought they were pretty straightforward, but so for a lay audience, how did you put this together?ZD:
We first identified those different domains by looking at the literature and seeing what other people were measuring when they thought about structural racism. We came up with those 9 domains and then we said, OK, what’s all of the publicly available data that we can find that fit into one of these domains? We started with, I think the most commonly used source of information for these type of indices, which is the American Community Survey from the Census Bureau. And there are a lot of reasons to like that. There’s a lot of information in there and information is available at the census tract level and that’s the proxy for neighborhood that we use.19:04
I don’t know if you use census tract or block group in yours, I don’t remember.19:08
But those are usually the two choices that people go with for sure.LL:
We use tract. I mean the big thing is they don’t change except every 10 years.ZD:
Yeah, they don’t change. The American Community Survey gives us their estimates. For the most part they’re good estimates, but they’re estimates. So the smaller geography you get the broader or the less sure those estimates are. So we want to get as narrow and granular as possible while still feeling like the data was pretty, pretty accurate and pretty stable. So we looked at the American Community Survey, put all of those different variables into the buckets of our nine domains.Then we looked at other Federal data sets, and we found a lot of good information in there from measures of segregation that the federal government’s using to retail job availability or foreclosure risk is something that Housing and Urban Development, the federal government, is measuring and reporting. Then the last place we went was to see, you know, in these nine domains, are there other publicly available sources of data that people are reporting on in the literature. And it didn’t yield too many, but it yielded, I think some really important ones.
One is eviction data, which is included in the housing domain. And the other, criminal justice data, comes from the federal government, but some of it comes via another organization. So we ended with I think 42. Some of them had, you know, enough data issues that we just couldn’t end up using them. One of them was that I really wanted to because, you know, it’s an area that I think is pretty central to when you’re talking about structural racism – that was voting data and there just wasn’t a good way to include it. You know, we started with census tracts. You won’t find anything by the census tract. And then there are precincts. We tried to go at the precinct level and couldn’t find it there. One state or a collection of states might have good data. The most consistent source we saw, it was at the county level, and we really tried not to use county level data. I think we did end up using it for some of the criminal justice data but for the most part tried to avoid county level data. But it’s one that I still, you know, every now and then I still just I’ll spend the night, you know, trying to see if I can, if I can work with the data enough to get it workable to be able to put into it. But I haven’t so far.
LL:
So I wonder if you can talk a little bit about your definition of people of color for Your Structural Racism Effect Index.ZD:
It’s a good question and one that we talked about internally a lot as especially as we were as my dissertation was getting prepared and then when we were going to publish in Health Affairs. So one thing I should say is that for the most part, the SREI is race-neutral in that there is no explicit measure of demographics in it. And that’s different than some other indices like the SVI, which has, you know, percent Black and percent Hispanic as reported by the census and, you know, maybe percent people with a disability. We intentionally did not want to include that data in our index.
The only place where there is some information about the racial composition or ethnoracial composition of a neighborhood is when you look at segregation. So there is a composite segregation measure that looks at three different measures of segregation. One is comparing White and Black populations. One is comparing White and Latine populations. And one is a more general type of segregation measure. It just looks at how different ethnoracial composition is between areas.So that one variable is the one place where there is some information. But if you just looked at that number, you wouldn’t really be able to tell exactly what’s happening in a neighborhood.
In the validation of the paper, we did define people of color. We wanted to make sure that our index correlated in some way to the number or proportion of an area that was people of color. Because it’s measuring structural racism. We’d expect that to be highly correlated to the ethnoracial composition of a neighborhood, but not always.
And to do that we defined it, and in the paper it gets very specific about what categories, using the census data. But it is anyone who identifies as Black alone or with another some other race. Anyone who identifies as Native American, American Indian, or indigenous alone or at some other race and anyone who identifies as Hispanic or Latine as the ethnicity in the census. And there’s a lot of conversation about that and I’ll, I, you know, I’ll let you ask the question about it.I’ll start by just saying when we think about the long history of structural racism, I think there are a lot of different types of, you know, structural racism doesn’t look the same for all populations and for everyone and it affects everyone in different ways, including, you know, White Americans.
24:53
We wanted to start from two, I’ll say starting places of racist policies and that being slavery and American colonialism, imperialism and how those policies perpetuated in different ways throughout history.25:10
And that’s where we started.25:11
But I will let you ask a very important question.LL:
So I live in Long Beach, CA. We have Little Cambodia here and, you know, here in in California during World War 2, you know, the Japanese were evicted from their homes and businesses and interned [in camps]. So, you know, there are people here who have suffered from the racism inherent in, you know, wars. We’ve invaded a lot of Asian countries and bombed them and there was certainly a lot of racism involved in that. So in my mind, I do think that, for inclusivity and for fairness, it would be better to consider people of color as a more inclusive definition, including people of Asian descent.
Because, you know, I mean that’s it’s a very broad brush. And I think even people that we might consider “high status” or “White-equivalent” or whatever the nonsense is about that, there are still effects of structural racism inherent in our history, in the Chinese Exclusion Act, in the racist determination that Indian people or Sikhs were not White, you know, so many different things in our history that I would be remiss if I did not say that. I do think it is an omission in the paper. And I am curious about you know, how you ended there. I mean what was the deciding factor? Was it just because of, you know, the 1619 project?ZD:
Yeah, I it had a lot to do with it. And I’m glad you asked the question because it lets me talk about something I didn’t get to talk about in the paper a lot and talked a little bit more in my dissertation which is like the categorization of ethnoracial groups is, I think, at the heart of this question and this problem of who we are including and who are we not including.
I’ll also say, personally the term “people of color” I think is not always useful, and if I could in every place in the… actually I don’t know in the Health Affairs paper… I think we only talk about people of color once or twice. It’s definitely in one of the graphics. I prefer to be much more explicit and say Black, Latino, and indigenous populations.The ethnoracial categorization that we use — it doesn’t come from the census. I’m blanking on where it actually comes from, the different part of the federal government. But that’s the basis that everyone uses. When we think about race, it’s not great. It’s not very, reflective of how ethnoracial groups or racialization actually happens in the country. And there’s even I think this is one of the lines that sadly got cut out of the paper as you’re cutting words.
28:29
But the Census Bureau itself released a white paper over 10 years ago.LL:
I never use that term [white paper] anymore. We published a blog post about it and we’re like, no, we shouldn’t be using that term anymore.ZD:
It was maybe 5 to 10 years ago. It said, you know, we’re not doing this right, and here’s a recommendation to be better at it. One was to stop thinking of Hispanic/Latine as an ethnicity separate from race, because it didn’t match people’s identity. That’s not really how people are identifying and ended up that they were, a lot of the people who are checking “other” was because of that distinction, and we weren’t properly catching that.Another was that Middle Eastern, Northern African and that’s an area that you I think appropriately point out that excluding Asian from the category “people of color” doesn’t make sense. And I also think that Middle Eastern/North African, excluding them from “people of color” also doesn’t make sense. And it’s hard to know exactly what box in the census is being checked. And I think in that paper they actually go into that a lot of it is White, some of it is other, some of it’s Black. It’s hard to know.
I think the term Asian is so broad and includes such a huge portion of the world who have vastly different experiences just like White, just like Black, just like Latine, that it’s really hard.I wish we had more granular data or that could be more true to the history of how especially if we’re looking at what is happening in the US and of what the history is of people, how they came to the US and what that story was. And I think the broad brush that we do paint is OK, for the most part. We know what the history of the majority of Black Americans is and what that history has been, and I’ll say there’s some notable exceptions.
And what I pulled up on the map, Worcester, is one of them. There are two distinct Black populations in Worcester. One is recent African immigrants and the other are descendants of American slavery. Those are very different histories. They might experience similar interpersonal racism on a day-to-day basis but have had very different of generational histories and lives. And I think when we—I mean, I’m guilty of it too, in this paper, saying these are the categories we’re using—but when we categorize people so easily, I think we miss a lot of that difference and can obscure some of the injustice that continues to happen. Latine is another, there’s actually a lot of different histories there, that that I think we see that have similar threads. It overlaps a lot with American Indian, indigenous history, that is part of those original sins that that I was talking about, of slavery and American colonialism/imperialism.
All of that to say is, that’s how we landed. And I think anything that we do with our current data will be imperfect. And that is, you know, we landed on an imperfect answer to a question that is, yeah, it was difficult to meet. We talked about it a lot. But I’m glad you asked the question.
LL:
Well, I’m glad we had the limited discussion that we have time for. Unfortunately, we don’t have unlimited time. But I wanted to say that, you know, I mean in terms of race and ethnicity, you know, one of the things that I think is really, really tricky nowadays, anyone can take a DNA test and know exactly what their DNA says that their ancestry is. There are a lot of surprises that have come out for people. A lot of people who were told that they had Native American ancestry, for example, you know, doesn’t come out. A lot of people who weren’t told, it does come out.I mean, if you really look at the DNA, we’re 99.9% the same anyway. This race thing, it’s a, it’s a social construct and it’s a way that our country has used to police behavior and enforce caste distinctions and do all kinds of awful things in the name of this idea. I’m with a lot of other people in that, when it comes to the healthcare system, we have so far to go in dealing with the effects of structural racism and seeing things with a wider lens and really understanding the causes of the causes. Being Black, being African American, that is not a risk factor. That’s not the risk factor. The risk factor is the racism. And I’m glad that you all have really focused on structural racism for the SREI. It’s really great work. When you think about like all of the things that go into what we’re talking about with this, with this issue of structural racism, I think a lot of people really feel American is in a sense a multiethnic identity. The growth of the multiracial category on the census I think is another recognition that it’s not Black and White, it’s not even Black, White, and Brown. It’s so much more complex and nuanced. So I’m glad we had this chance to talk about that.
And I think one maybe other thing that we could chat about really in the last couple of minutes of our conversation is just this idea of people imputing race and ethnicity. I know you work with a lot of data and a lot of times I work with data where you know, survey data or other data where we have linkages between different kinds of data sets.
And a lot of times there’s missing race and ethnicity. In the Massachusetts All Payer Claims Database work that we did, 60% of people were missing race and ethnicity. Now it’s I think down to 50%. So in that situation, a lot of people will do things like create an algorithm that will guess at, you know, probabilities that this person is one thing or another which, we’ve had the conversation, Arlene and I, and we’ve had this conversation on the blog and in the podcast. But I want to just give you an opportunity to talk about that as well because I bet you’ve thought about it.
ZD:
I have. And it I think links a lot to the conversation that we were just having, and I am going to bring it back to that. One of the biggest questions I have when that happens is what are you using that data for? What is the purpose of having that data about that person? A lot of times we’ll include race and ethnicity in a model or in the way we’re in research. I think we do not properly discuss “race as a risk factor”.It’s people’s experience of racism. That is really the risk factor that we’re talking about.
And access to resources that people have. We’re using race as a proxy for that. It is inherently problematic, but I think if you ask the question of, what is this information for? What am I am I hoping to achieve with it?There are possibly other solutions out there. One solution might be to look at the neighborhood someone lives in. Say not like, you know, what is the ethnoracial composition of this neighborhood, but what are the resources that are available in this neighborhood? Use a tool like the SREI to say,
there’s a high concentration of the effects of structural racism in this neighborhood. This is a risk factor. The experience of structural racism is a risk factor for anyone who lives in this neighborhood.That I think is the power of a tool like this, that you don’t have to confuse those two concepts. You can include both in a model. When we know where someone lives. This is the experience of living in their neighborhood. Also for whatever reason, you have to conceptually figure out why you’re including someone’s ethnoracial group in that model and be and I think be able to justify it. I think we always just throw it in and it’s fine. But you really, I think, need to conceptually justify why you’re including that. You can use both. I think, if there’s something, if there’s a tool available to tell you about the experiences of racism in terms of resources, then that should be used. You shouldn’t just have to rely on imperfect data about a racial group and then make it up when you when you don’t have that available to you.
LL:
Yeah, it has very big equity implications to make it up or to guess, and guessing wrong definitely has implications. We don’t want to, you know, aim for health equity and end up with more inequity. And that’s exactly what that approach can actually be to the unintended consequence in my view. And I think the research, the evidence base would actually back that up.
ZD:
And I think it perpetuates an idea that ethnoracial group is inherently linked with these other things that we’re looking at, whether it’s the outcomes that are used for imputing or whatever else people are using that it and doesn’t look the same everywhere.LL:
On that note, thank you again for taking the time out of your very busy schedule. Doctor-Doctor Dyer.ZD:
I’m happy to chat and be able to make this connection between our parallel work that’s happening though.LL:
Absolutely. All right. Thanks for listening, and viewing, and see you next month.
Lisa M. Lines