EPITalk: Behind the Paper

Racial, Rural, & Occupational Inequities during the COVID-19 Pandemic

Annals of Epidemiology

Send us a text

PhD candidate Hannah Zadeh and Dr. Martha Carvour tackle the methodological complexities of modeling racial, rural, and occupational disparities during the first wave of the COVID-19 pandemic. Their paper, “Epidemiological approaches to multivariable models of health inequity: A study of race, rurality, and occupation during the COVID-19 pandemic” can be found in the June 2024 (Vol. 94) issue of Annals of Epidemiology. 

Read the full article here:
https://www.sciencedirect.com/science/article/pii/S1047279724000589?via=ihub

Episode Credits:

  • Executive Producer: Sabrina Debas
  • Technical Producer: Paula Burrows
  • Annals of Epidemiology is published by Elsevier.



Patrick Sullivan:

Hello, you're listening to EPITalk: Behind the Paper, a monthly podcast from the Annals of Epidemiology. I'm Patrick Sullivan, E ditor-in-Chief of the journal, and in this series we take you behind the scenes of some of the latest epidemiologic research featured in our journal. Today we're talking with Ms. Zadeh and Dr. Carver about their article " Epidemiological Approaches to Multivariable Models of Health Inequity a Study of Race, rurality and Occupation During the COVID-19 Pandemic. You can find the full article online in the June 2024 issue of the journal at wwwanalystofepidemiologyorg. Hannah Zada is a PhD candidate at the University of Iowa Department of Sociology and Criminology and a current NSF GRFP fellow. Their research centers on the use of race and statistics in medicine and they're currently working on a dissertation project about the history of risk prediction in medicine over the 20th century. Welcome, H annah and A Dr. Martha Carver is an assistant professor of internal medicine, infectious diseases and epidemiology at the at the University of Iowa, carver C College of Medicine and College of Public Health. Her research combines epidemiological and community-engaged research methods to promote health among people with diabetes and improve the outcomes of diabetes-related infections, including COVID-19, with a focus on designing and implementing equitable systems in healthcare and public health. Reading your bios, you're exactly the two people I want to be talking to today, because I love where your professional focus is and I'm so glad you could join us today.

Martha Carvour:

Thank you so. much. It's great to be here. Yeah, happy to be here.

Patrick Sullivan:

Great. Thanks. So we're going to talk about this article that you submitted to Annals of Epidemiology t hat really captures up a lot of very current things like, obviously, the COVID-19 pandemic, but also you're really working on approaches to multivariable modeling of health inequity and just taking the complexity of how these inequities occur and really representing it through modeling methods. So I'm excited to talk to you about this. Just starting out with what the purpose of the study was, what research question you were aiming to answer?

Martha Carvour:

Sure, yeah. Thank you so much, D r. Sullivan, for the opportunity to be here today. We're really excited to talk more about this paper. So I can talk sort of briefly about the impetus behind the study and that was really what we were seeing initially in that first year, especially in the first six months of the pandemic, in a public health context and in a clinical context. Clearly we were seeing health inequities. In our own state, in Iowa, we were seeing inequities particularly in rural areas and rural counties, and many of those inequities were particularly affecting people in frontline occupations meat packing, meat processing. There were highly publicized reports of this across lots of states and lots of regions, but many rural areas and rural counties have substantial industrial reliance on some of these frontline occupations and so we were seeing that this was one of the factors contributing to these inequities. We also saw, in rural contexts as well, the persistence of otherwise well-documented inequities by race and ethnicity. So really we set out to understand the intersections of these different factors in health inequity and COVID and understand more about what happens with racial and occupational inequities in rural contexts like here in Iowa and in other largely rural states.

Patrick Sullivan:

So you really described then sort of these multiple layers of exposures and epi terms and inequities, and I wonder if you could tell us about what some of the methodologic hurdles were when you're really trying to understand some of the structural characteristics that may be associated with disparities. And you talk a little bit about this in the introduction, but what were some of those methodologic hurdles?

Hannah Zadeh:

Thanks. Yeah. So in our introduction we kind of lay out three of the common methodological problems that we sometimes see in these kinds of analyses. And the first one a really big one and one that I'm really interested in coming from, like the sociology of race, is the way that race is modeled in a lot of these statistical analyses. So you know, even though in the past several years at least, it's coming to be more solidified in the mainstream that the truth that race is not a biological variable, it's not a biological characteristic, it's a social, structural level system of racism, we still see, despite that understanding, we still see methodologically sort of a dissonance between that idea and how race is actually modeled. We still see race modeled as an individual level variable. So we were really interested in how do we operationalize racism really in these models? And one way that we were interested in doing that is by using county level data to focus on where racism is occurring, via the proportion, racial proportion in different counties. So that was kind of a first thing is modeling structural racism through these statistical analyses.

Hannah Zadeh:

A second thing was thinking about how this period immediately after the pandemic began in the United States, period immediately after the pandemic began in the United States.

Hannah Zadeh:

It's kind of a unique period for thinking about what kinds of resources counties have in terms of health infrastructure, which is everything from counties' ability to distribute PPE to be able to deliver remote education to students, everything like that.

Hannah Zadeh:

And so we were really interested in sort of this first initial period, this first six months, as a way of sort of looking at this as sort of a test of like what kinds of resources counties have. Because, you know, our kind of thinking is from the literature that counties with higher proportions of non-white people are maybe going to have fewer county resources because of how we know that systemic racism works in the United States. And then a third thing, lastly, is just that we have seen that there is a general paucity of statistical epidemiological analyses that are looking at the intersection between racism and workplace-related processes that affect health in rural areas. I think there's still sort of this idea that rural places are only populated by white people and that, like, these racial disparities aren't and these racial processes aren't happening in rural areas, and we know that that's not true. And then I think there hasn't been, as I think, as much attention as we would like to sort of the health-related processes happening due to processes happening in the workplace.

Patrick Sullivan:

Yeah, I mean thank you for that. You really have sort of put some meat on the bones around the statement in your introduction, which is that race variables aren't representing biological mechanism in this case. Maybe in some cases there may be some very specific cases they do, like sickle cell or some other biological mechanism, but here race variables are really representing structural racism, you say, in the form of inequitable resource distribution or access. So you've really sort of talked a little bit about you know how that plays out. And then I think the kinds of variables that you brought into the analysis animate that idea of inequitable resource distribution when you get into the analysis. So I think it's just so important, especially when dealing with race and ethnicity as social determinants of health, to ask that next layer down of questions and really operationalize it in ways that are changeable. You know that could be addressed and maybe we'll get into some of those outcomes. So in this research then talk a little bit about your study design and sort of your unit of analysis here, because you relied on some population data sources. So what was the design and what was the unit of analysis?

Martha Carvour:

Sure, yeah, so this study really emphasized the county level. So we measured exposures and outcomes on the county level. And we did that for two reasons. One is we have more county level data available to look at social and structural reserves, as I think Hannah outlined really nicely. Part of what we were thinking is, every county going into the pandemic, that early initial shock period had some set of social and structural reserves and our hypothesis is that the outcomes, or the disparities in COVID-related outcomes, could depend in some way on those social and structural reserves. So we were really focused on the county level for that reason. The other reason is also connected with what Hannah mentioned, which is we wanted to get away from this idea that everything we were looking at was exclusively an individual level outcome. So what is it about these structural factors? Thinking in public health terms, what can we do to make a big impact right, to have an impact on a county level or population level, by identifying those group level or county level types of characteristics? What are those social and structural reserves that might be amenable to intervention?

Martha Carvour:

So we did a county level analysis.

Martha Carvour:

We had focused entirely on rural counties in the US and we matched counties that were identified as case counties, which were counties with disparities, so at least one standard deviation above the mean for COVID incidents or for COVID mortality.

Martha Carvour:

We did models of each and then control counties and those controls were both within the same state as the cases we recognized. Of course there's a lot of state to state variation and different types of policies and social and structural reserves, so we had controlled for that and we also matched on the rural-urban continuum code defined by the USDA to try to have direct comparisons of the social and structural variables that we were looking at. Then we used a form of hierarchical logistic regression modeling. And this is, I think, one of the aspects of the study that I, as an epidemiologist, felt like I learned the most from which was really splitting this deliberately into two phases or two sections. One phase the first phase was focused on who was disparately affected, so what are the demographic characteristics of counties that experienced disparities? And then, as the second phase, what are the social and structural factors that seem to be connected to those disparities, to remind us that those are the modifiable factors in a potential public health intervention.

Patrick Sullivan:

Great. So, just building on that, I'm looking at your table four, which are the final adjusted models, and I'm not sure if that's where you would go next. There are some really for me unexpected associations with both incidence and mortality. So what were some of the main findings or key takeaways from these analyses that you're describing?

Hannah Zadeh:

I can talk a little bit about that. For COVID-19 incidents. We found that hospital closures, hospital size and industry reliance on mining or manufacturing, as well as baseline metric related to smoking, all correlated with county disparities. We also found evidence of interactions between racial disparities and industry reliance disparities and industry reliance. And so industry reliance is one of the variables that we were, I think, really interested to see what we would find in this analysis and just as a little bit of background on that variable. So this is from the USDA's Economic Research Service and it's basically measuring counties' economic dependence on certain industries. So more precisely it's if a county takes over 23% of its earnings or if over 60% of county residents are employed in those industries, and so we were interested in if a county is really sort of economically dependent on a certain industry, like what kinds of COVID-19 outcomes might that be associated with? So we did find for counties in which mining and manufacturing, respectively, are predominant. We found that was significant for incidents. Like I was saying, we found evidence of interactions between racial disparities and industry alliance. More specifically, we found that in counties with higher proportions of Hispanic residents that had an economic reliance on manufacturing, there tended to be higher racial disparities in COVID-19 incidents. Also, for incidents, we found that disparities persisted in counties with higher proportions of Black or Hispanic residents. So that was incidents. And then for COVID-19 mortality, we found that a couple of variables made it into the model. So the proportion of residents in the county classified as unemployed or disabled was one, and then the proportion classified as quote unquote obese was one. The proportion reliant on public transportation was one, and a variable that was an index of poverty segregation, all correlated with non-metropolitan county disparities, all correlated with non-metropolitan county disparities. And you know, maybe we can talk about this later. But, as we sort of said in the beginning, like we were using a lot of these variables that are commonly modeled as individual level variables and trying to sort of contextualize them more and think of them as signs of county level resources, right. So I think one thing that we see a lot in the literature is a focus on these sort of quote unquote lifestyle associated variables and sort of a push to sort of shift the locus of blame onto individuals. And we're sort of interested in taking these kinds of data because that is the kind of data that is publicly available and recontextualizing it as a sign of county level resources. So that's just a bit of background on why we included some of these variables in the initial pool, and that was some of the findings. Dr. Carvour, I don't know if you have anything to add.

Martha Carvour:

No, I think that's a great summary, H hannah, thank you, and I would really reinforce that piece about what we can understand about these variables on a county level or sort of a group level.

Martha Carvour:

HOne of the things we talked about as an example is the public transit, which was something we didn't really expect to see sort of inversely associated, because it makes perfect sense just in terms of transmission models and the way that we were reading about other risk factors for exposure to COVID or to SARS-CoV-2, being in environments where you might have more exposure to other individuals, including certain transportation modes, where that might be more likely.

Martha Carvour:

It has the opposite effect, right, it's more likely to drive at least incidents, and we saw this was inversely correlated with mortality. And the way we're kind of putting that together is thinking that this could be a sign of infrastructure, transportation infrastructure to get access to healthcare, for example, from rural areas where there are often geographic barriers to care, and so that's, I think, a different issue in terms of what this variable represents on a county level versus what the exposure might represent for an individual, and that's an example, too, that has borne out in community-based discussions, you know, with community partners. We hear this time and again about the issue with, very specifically with, transportation, which I think was helpful to kind of add to the context of what we were seeing with these publicly available data.

Patrick Sullivan:

Or you know, higher access to public transit may also be confounded with other sort of positive actions in communities.

Patrick Sullivan:

And I think one of these kinds of analyses I'm a fan of, and so they let you get at some sort of big issues within communities, but also have the disadvantage that you can't really link the people, the individuals who are using, for example, public transportation systems. So there may be some need to supplement with some other kinds of studies. But yeah, I was struck by I encourage everybody to look at table four, particularly in the mortality column, like some of the, maybe because of small cell sizes, but there are some impressive adjusted odds ratios and maybe some findings that might not be what you might've predicted. So an interesting table. So can you talk just in general about some of the strengths and limitations of the approach? We've touched on these a little bit in terms of the sort of ecological level approach, but what do you see as the strengths and any of the limitations that you would want to highlight?

Hannah Zadeh:

I can get us started. So I think one of the strengths of this study and something that was, I think, born out of a lot of conversations that we had with co-authors was, like I was talking about in the beginning, thinking about how to operationalize race in epidemiological models. Right, because, you know, we definitely do want to be continuing to track the way that racism is affecting public health, but we want to be doing that in a way that is really, you know, wedding a structural theory of race to methodology. You know, I think I really always like to go back to Ruth Wilson Gilmore's definition of structural racism, right, because I feel like structural racism is something that is increasingly like it's kind of become like a buzzword. That is increasingly like it's kind of become like a buzzword, and Ruth Wilson Gilmore defines it as the state sanctioned or quote the state sanction and or extra legal production and exploitation of group differentiated vulnerability to premature death. And I just feel like that definition always sort of grounds me or grounds a discussion, like when we're thinking about what in our methodology we're trying to represent, right, like it's very serious. And in some of our conversations we talked about how, you know, we were frustrated with some of the literature where it doesn't seem like sometimes folks are applying the same level of like methodological rigor to really representing racism in these models. So that was something that I think is a strength of the study is we were trying to sort of explore approaches to doing that. You know here, what we were doing was, you know, using the data we have to at the county level, model a proportion of residents of a given racial identity and sort of use that as a way to sort of follow where these resource distribution mechanisms of racism are happening. So I think that's one strength, and we had a couple.

Hannah Zadeh:

There's always limitations to any study.

Hannah Zadeh:

I think that one thing that was really sort of a constant thing that we were trying to navigate is the availability of data.

Hannah Zadeh:

So, like we were talking about a little bit before, a lot of the publicly available data that is available is this sort of data that is often thought about as individual level data like smoking, like BMI, even income, and so that was something that we were aware of.

Hannah Zadeh:

I know that at one point we were really interested in looking at how local union density would be shaping COVID-19 transmission and mortality, especially when we know that the workplace and that workers' power within their workplace is something that really impacted their ability to access PPE and necessary protections, and that data we just weren't able to find. So it would be very interested if others are able to get that data and to do those kinds of analyses. And then I think a last limitation is that, especially in rural areas, the county level is like pretty big and there can be a lot that's happening in a given county. A lot of like dynamics city to city or town to town, that like are just sort of being like all mixed up and smoothed over when you're looking at the county level. So I think that, like future analyses that are getting down to a more localized level would be really helpful to have.

Patrick Sullivan:

Yeah, I mean I think the idea that sometimes ecological analysis at a broader level can be hypothesis generating. But then you know you could think about smaller administrative areas if data is available. The data are available, or you know, sometimes I think like the next step could be as, on such a different methodologic path, to say, like key informant interviews with around you know some of the manufacturing or I don't know the public transportation, so you can just see how this is like. This rich hypothesis generation and the ecological approach is great for that, and then it sort of answers several important questions and then raises a bunch more. You know that you might need to explore with different methods, so great. So in light of that, what do you think the implications are for this idea of advancing health equity? You sort of talked some about inequities, but what should be done with these data to advance conversations about improving health equity through policy or through some kind of public health practice?

Martha Carvour:

Yeah, I think there are a couple of things that I know Hannah and I have talked about, and we've also had a lot of, I think, really helpful and rich discussions with co-authors, who I also want to acknowledge for a huge contribution to the way that we thought about each of these variables and kind of put these results together. I think there are two sort of takeaways for us, and one of those is methodological and one of those is sort of this practical public health piece in rural settings. On the methodological side, I think using the phased analysis that we did here allowed us to interrogate some of our own assumptions about what race variables represented. I think we're used to in epidemiological models, we're very used to sort of adjusting for certain socially constructed demographic variables, and in this case we tested some of the assumptions about what does race mean? What is it actually measuring here? Where does it fit into a directed acyclic graph? Are our assumptions? Do they appear to be valid, based on what we're seeing with the analysis and the order in which you put variables into the model? And we found that, as expected, there's a lot of variability in what the race constructs measure, especially on the county level like this, and so really thinking carefully about questioning those assumptions and thinking carefully about what are the mechanisms, socially and structurally, that we're trying to measure, because that's going to speak to how we can actually improve outcomes, reduce inequities, really thinking critically about more mechanistic pieces.

Martha Carvour:

The practical public health piece is that we really want to work with community partners, work with other collaborators, really work in community settings to learn more about what the social and structural factors are that are important, how to think about intervening on these modifiable factors to improve outcomes.

Martha Carvour:

We've been working in this area, motivated in part by the research here that Hannah has led and that we've been doing as a group to ask those questions at the community level right and conduct qualitative, quantitative research, interventional research, and start to answer some of those more specific questions.

Martha Carvour:

A standout piece of that is what's still happening with occupational disparities after that initial wave where so many frontline workers were affected.

Martha Carvour:

We don't talk nearly as much about frontline workers or hear as much about frontline workers today as we did several years back when this data was collected or when the study was sort of devised. And yet we know that there are long-term impacts on many of the frontline workers in those communities many persistent inequities in access to care for things like long COVID or just other health conditions. So this is something that I think has motivated us to see frontline workers as a part of the public health workforce in rural communities, the people who are packing the food, delivering food this is part of what sustains life and health in a public health emergency and so ensuring that we have equitable access to resources. Much like I had, I was quite privileged to have access as a frontline healthcare worker to different types of occupational protections, and I still have those occupational protections to this day as a healthcare worker, and those may not exist for others who were also on the frontlines.

Patrick Sullivan:

Yeah, I think this idea of extending frontline workers to say that that does include some very highly trained, you know, colleagues, but also there's as critical capacity for transportation, for food, for other things, and I think some of the discussion about frontline workers, you know, did focus more in medical settings.

Patrick Sullivan:

So I really appreciate you reframing that. So we're going to transition now to a piece of the podcast that to me is always at least interesting, which is called Behind the Paper, Dr. Carvour always someone who remains grateful for the mentorship that I got throughout my career and now see my role as a mentor as the most important thing that I do for public health. But I want to talk a little bit just about, even in the way I think you've prepared for this we're on Zoom, I'll just tell Carvour listeners we're on Zoom, so that body language about who's going to answer isn't quite so evident. But even down to the level I think of preparing for this, it's clear that you've been thoughtful about how you're going to take these roles. So I really just want to talk a little bit about the interaction and your work together on this and how you sort of saw those roles and how you work together obviously collaborated in this, so either one of you can start. But what was that working relationship like and how did you handle the roles? Sure, yeah, I can start. We were just talking about beforehand that we sort of started. . We Carvour met each other in March 2020. And I was saying that my Zoom meeting with Dr Carver D was. Carvour first Zoom meeting I'd . ever had in my life. And look at us now.

Hannah Zadeh:

So, yeah, I H was in the University of Iowa what they call a pre-MSTP medical scientist training program as an undergrad and it was supposed to be for summer 2020, but it got shut down because of the pandemic. But Dr Carver graciously agreed to still work with me on a project anyway, and we had a lot of conversations early on about racial health disparities and health equity. And it was cool because I think we sort of got to have some conversations sort of in real time as we were observing the tragedies all across the US, all across the world, that were happening as a result of the COVID-19 pandemic, and we were also observing, I think, some of the things that were sort of being overlooked in the media and also, I think, in the scientific conversations around the pandemic. And I feel very lucky to have been able to have those conversations with Dr Carver, as well as all of our co-authors throughout this project. And yeah, I'll stop there, dr Carver, if you have anything to add.

Martha Carvour:

Yeah, thank you, hannah. I mean I think I would echo a lot of what you've just said. I think this is a project that has gone through its own phases, as the pandemic has, and really recognizing the significance and the importance of the types of questions and in many ways, even though it was a long project, as any research project can be, it's also an urgent topic, right. And so thinking about how we can still be balancing that with the practical public health work and the practical community engagement along the way.

Martha Carvour:

Hannah has really approached this as a leader, intellectually, collaboratively, has engaged a lot of co-authors from different disciplines to bring their perspectives, co-authors from different disciplines to bring their perspectives.

Martha Carvour:

I think that's really important, especially for scholars sort of in training and in development. I think we're all sort of in training . always Carvour and I advice- that's the good thing about the field, but I think, really having that chance to lead. and to ask those tough questions, to find answers to those tough questions and to work through a process, even when there are challenges and in Hannah's case, over several years and other professional milestones, and so it's wonderful to sort of see it come to this point, to see that we continue to have discussions about not only this project but about the profession, about the direction of the profession, and I think those are, for me, those are some of the most important and gratifying parts of doing this work. I think the research, of course, is crucial. The ability to have an impact is crucial, but one of the ways to make the biggest impact is through mentorship of other really, really skilled scholars like Hannah.

Patrick Sullivan:

Thank you and I'm going to ask you one more question and then Hannah, I'm going to come back to you which is Dr Carver, what advice so you describe how this really productive, professional, scientific relationship developed and how impactful that's been? What advice would you have for early career researchers who come in wanting this kind of experience? How do they go about, you know, opening up those relationships or putting themselves in a position to be in this kind of mentoring, collaborative relationship? Yeah, I think that's such an important question. I think I know from talking with so many students, especially who have been in school or in different training programs over the last several years, this has also been a really tough several years to be a student and be thinking about what the future looks like and be thinking about how to get into a program or how to get research experience. So, first and foremost, you're not alone and don't be patient with yourself. Be patient with the process and seek out people who want you to learn, want you to succeed, who you can, you feel comfortable with and feel like you can advocate for yourself with. I think one of the key skills to learn in a mentoring relationship that I learned also as a student is how to communicate about what I need or what I'm, what I'm, what's not clear to me. So it's just such a great opportunity to do that in a place that you should have support and and I think that's really really important.

Martha Carvour:

Thanks. I Hannah the other piece of advice I would really put out there is there's no substitute for critical thinking. There are a lot of great methods, there are a lot of great resources, but you really do have to think critically about the data that's put in front of you, the questions that you're asking, and I think, if you have in mind a career trajectory or a research topic that's of interest to you, there may also be a different way that you're approaching that or a different way that you're thinking about that, and I think that it can actually be something that the profession needs right now. We need people who think differently and critically, and so don't give up on that piece. Definitely it takes hard work and dedication and great mentors, but don't give up on thinking critically for yourself. Thanks.

Patrick Sullivan:

And Hedda, I'm going to give you the last word, which is what advice would you have for maybe more seasoned or later career colleagues in terms of mentoring and working with students or younger researchers? Yeah, I think like I've. been really, Carvour really lucky in the mentors that I've had throughout my career so far and it's something that I think sort of took for granted until like the past couple of years. .

Hannah Zadeh:

But I mean, like Dr Carver was saying, students, especially in the past couple of years, have been dealing with a lot, and I think like, especially if we want to, like you know, make research something that is accessible to everyone including, you know, people who are coming from low-income backgrounds, you know, who don't have like family connections to research, or even to higher education, people who are trying to care for dependents while they're pursuing a career in research, we really need mentors who are like willing to be flexible with students and patient with students, even coming down to like nitty gritty of like scheduling stuff and students having to miss meetings, stuff like that. I think like that kind of like patience and flexibility is really important to like do the work of making this profession accessible and open to all kinds of students. And again, I just have been so lucky with Dr Carver and other mentors that I have to really see that in practice.

Patrick Sullivan:

And you'll have a chance to pay it . forward Carvour So keep your notes. So, do you have any last thoughts that you'd like to share with our listeners?

Hannah Zadeh:

I Carvour? one thing that you know, Dr Carver, and I talk about a lot still to this day is the pandemic is not over. We've talked in this conversation about how the implications of long COVID and you know there are still thousands of deaths every week in the United States, and so like, even though this project was just focusing on that initial period. Yeah, my final thought is that the pandemic is still ongoing and more of this critical thinking about health equity needs to be done.

Patrick Sullivan:

Dr Carver.

Martha Carvour:

No, I think that's great. Thank you very much and thanks Hannah. M. Z.. T P t o j o t A C i o E. F a t o t p o t r t a f i t e a m f t j y c v u www. annalsofepidemiology. o

Patrick Sullivan:

All right, that brings us to the end of this episode. Thank you again, ms Sada, and Dr Carver, for joining us today. It was so nice to have you on the podcast, thank you, thank you so much. Today was so nice to have you on the podcast. Thank you, thank you so much. I'm your host, patrick Sullivan. Thanks for tuning in to this episode and see you next time on EpiTalk, brought to you by Annals of Epidemiology, us online at wwwannalsofepidemiologyorg.

People on this episode