Subscribe
Share
Share
Embed
Health Affairs Editor-in-Chief Alan Weil interviews Meera Kotagal from Cincinnati Children's Hospital discussing her and colleagues' paper in the December 2022 issue examining 15 different area-based socioeconomic deprivation indices.
Each week, Health Affairs' Rob Lott brings you in-depth conversations with leading researchers and influencers shaping the big ideas in health policy and the health care industry.
A Health Podyssey goes beyond the pages of the health policy journal Health Affairs to tell stories behind the research and share policy implications. Learn how academics and economists frame their research questions and journey to the intersection of health, health care, and policy. Health policy nerds rejoice! This podcast is for you.
00;00;00;01 - 00;00;28;28
Alan Weil
Hello and welcome to “A Health Podyssey”. I'm your host, Alan Weil. When people conduct research on the effects of a policy or programmatic intervention, part of what they do is attempt to control for factors other than the intervention that could affect the outcome. In health care with growing understanding of the relationship between social factors and health, one relevant factor is the characteristics of the community in which people live.
00;00;29;08 - 00;00;57;09
Alan Weil
For certain interventions, you might expect someone living in a resource-poor neighborhood to be affected differently by the intervention than an identical person living in a higher resource neighborhood. But it turns out there's no one way to define a neighborhood or to describe a neighborhood. And so as a result, various scholars have created indices of what's called “socioeconomic deprivation” to capture a number of dimensions and collapse them into a single metric
00;00;57;09 - 00;01;29;10
Alan Weil
more useful for analysis. The nature and uses of deprivation indices is the topic of today's episode of A Health Policy. I'm here with Meera Kotagal, Director of Trauma Services and Director of the Pediatric Surgical Equity Fellowship at Cincinnati Children's Hospital. Dr. Kotagal and colleagues published a paper in the December 2022 issue of Health Affairs, examining 15 different area-based socioeconomic deprivation indices.
00;01;29;26 - 00;01;47;26
Alan Weil
They found important differences across indices and raised some questions about the variables included and excluded given the intended uses of these indices. We'll discuss these findings and the broader topic of community characteristics in today's episode. Dr. Kotagal, welcome to the program.
00;01;48;23 - 00;01;50;12
Meera Kotagal
Thank you so much, and thank you for having me.
00;01;50;28 - 00;02;18;23
Alan Weil
So this is a really complicated topic, but I also think it's a really interesting one, and so I’m hoping in our conversation we can sort of get past the complexity and get into why this is so important. Let's just start with some of the basics beyond what I was able to cover in the introduction, what actually is an area-based socioeconomic deprivation index, help us out with area, and index, and maybe even socioeconomic and deprivation if you have the chance.
00;02;19;05 - 00;02;42;17
Meera Kotagal
Yeah, absolutely. I think it's really an important topic that's been raised increasingly in the last 5 to 10 years as people try to understand the ways in which neighborhood context influences health. And I think there's two big components to area-based and socioeconomic deprivation index. The first is around what does area mean and at what level are you really measuring this sort of context of where people live?
00;02;42;25 - 00;03;00;27
Meera Kotagal
And there are lots of different types of areas you can measure it at a national level. You can measure at a state level. In these indices, for the most part, that is about understanding, is it measured at the census tract level or at the zip code level? And we can talk more about why that matters and why that's important when you understand how you're using the tool.
00;03;01;17 - 00;03;30;29
Meera Kotagal
Socioeconomic deprivation indices really are a way to measure what the context is in which people live. And that includes understanding their neighborhood poverty, education levels, all the things that we think may be social determinants of health and how they might be relevant to impacting both health outcomes and the success of health interventions. And so indices are a composition of multiple measures that we try to put together into one single tool and use that to understand socioeconomic deprivation at a neighborhood level.
00;03;31;19 - 00;03;52;04
Alan Weil
So we're trying to capture a very complex concept in a oversimplified way. And I think it's important for the listeners to understand why would you want to collapse? Why not just keep it complex and get education separate from poverty, separate from whether there are grocery stores? Why would you want to have a single index?
00;03;52;14 - 00;04;15;28
Meera Kotagal
So I think there's a lot of reasons to do that. There are two main ones that I think are particularly important. One is that it allows people to use one tool in their analysis. So one data point when they're creating their analyses in order to understand that broad, complex context. And that helps with collinearity. That helps assessing sort of the complex dimensions and simplifies the analysis a little bit, which I think is helpful.
00;04;16;09 - 00;04;38;19
Meera Kotagal
The other is that in in many ways, these indices are more comprehensive than any one single measure. So if I go to your neighborhood and I look at the median income, and I look at that in isolation compared to thinking about income, education, level, vacant housing, primary language, you know, all of those pieces give me a much richer understanding of what that context looks like.
00;04;38;29 - 00;04;55;09
Meera Kotagal
And as a result, they tend to correlate better with health outcomes. So a complex socioeconomic deprivation index is a better predictor and a better, you know, sort of association with health outcomes that any one single measure. So that's the reason to do that. But it's important to understand what those indices contain.
00;04;55;27 - 00;05;09;08
Alan Weil
Right. And so it's a really useful analytical tool. At the outset, in this paper, you went out and said, what indices are people using? What did you find? How many? Where do they come from?
00;05;09;25 - 00;05;29;06
Meera Kotagal
Yeah. So we you know, we started this project really because we wanted to better understand we were trying to decide why did we pick this index over that index when we were doing our own studies. And as we looked through the literature to better understand it, we felt like there was a big gap in people's ability to understand there's many indices and why would you pick one or another?
00;05;29;15 - 00;05;51;20
Meera Kotagal
So in our initial search, we actually did a systematic scoping review and we found over 3000 articles, which we then narrowed down to a little over a thousand abstracts once we excluded duplicates and things like that. And then that got us down to a total of 44 potential indices. We then reviewed each of those indices to ensure that they actually assessed area deprivation, that they weren't too geographically broad.
00;05;51;20 - 00;06;08;18
Meera Kotagal
We weren't focused at the state level or the national level and that they weren't only limited to one particular state. So looking just at Massachusetts or just at a certain county. And that led us to a total of 15. So we this review really covers those 15 indices and understanding those as well as possible.
00;06;09;08 - 00;06;28;12
Alan Weil
And so just to be clear, what happens is some researcher somewhere is trying to answer some question and they say, gee, I think in order to answer this, I need to understand the characteristics of my community or the communities in which that I'm studying. And they go out and look and say, who's come up with tools to help me capture that?
00;06;28;12 - 00;06;40;21
Alan Weil
And they either use one off the shelf that they've seen published or they say, none of these meet my needs, I'm going to make my own. And that's sort of how we get 15 of these, is that that how indices get formed?
00;06;41;02 - 00;07;01;00
Meera Kotagal
Yeah, I think so. And you know, we really limited this to indices that were used in the literature more than once. So we weren't getting just the I decided today to use this index, but at the same time, there were a lot of these were developed for different purposes. And so as you go through and learn about each index, you realize that it has different components intentionally.
00;07;01;10 - 00;07;25;13
Meera Kotagal
So some of them included, like the Social Vulnerability Index, really was focused on trying to understand how vulnerable communities were to disaster. And it recognized that communities that have a higher proportion of minority individuals are more vulnerable to disaster for a host of structural racism and racism-related reasons, as well as other things above just their socioeconomic deprivation.
00;07;25;21 - 00;07;36;27
Meera Kotagal
And so for that reason, SVI includes race, but understanding the origins of those indices is sort of important in understanding how to use them and what assumptions we make when we use them.
00;07;37;13 - 00;07;57;23
Alan Weil
Right. So this is really the topic I hope we'll focus on for the rest of our conversation, which is they vary, not surprisingly, they vary on a number of dimensions. And part of what your paper does is help us think about why would you choose one over another, which is basically, as you say, what you what motivated the project in the first place.
00;07;58;01 - 00;08;26;17
Alan Weil
So let's just start with some of the dimensions along which they vary. And at the outset you noted area, whether as you said, as zip code or census tract, and also you just introduced the question of whether or not race would be included. Just give me a sense, if you could, of both geographic areas and how those are defined and some of the key dimensions of what's included and excluded.
00;08;26;17 - 00;08;29;16
Alan Weil
And then I'm sure we're going to dig deeper into both of those topics.
00;08;30;05 - 00;08;47;11
Meera Kotagal
Yeah, I think it's those are really important questions. And let's talk about the latter one first, which is what are the components that go into indices? And I think there are really a couple of dimensions at which you need to understand that. The first is what are the variables that go into that? What elements did somebody choose to include when they were developing their indices?
00;08;47;11 - 00;09;17;25
Meera Kotagal
And that can include, you know, measures of income or poverty, measures of education and employment, household structure, race and ethnicity, transportation, language barriers, or primary language. And some most measures include some measure of income, poverty, education and employment, but some also include other dimensions. So understanding what are those variables that go into the index? The second really important dimension to understand is how are those variables selected and combined, and how does that come together into forming an index?
00;09;17;25 - 00;09;41;27
Meera Kotagal
And some are done...those are selected empirically or based on theoretical considerations. The SVI is a great example of that. Some where they really used expert consensus to identify what factors should go into the SVI. Some are developed with statistical methods that sort of look at, okay, if I'm taking a certain sample and I want to assess how much of the variation in that sample is due to which factors, and those that that are the result of more variation
00;09;41;27 - 00;10;11;28
Meera Kotagal
I'm going to include-- they're going to be weighted to have more of an impact on the index. So some have factor loading as a strategy to determining what goes into those indices. And then the last is what's the output measure where you get something on the outside? Once you're done, what what does that look like? And most indices are done by rank, which is to say that you get a percentile result, but some are actually still use their absolute value and that matters because of how much variability you might see or how you interpret those results.
00;10;12;19 - 00;10;33;25
Meera Kotagal
In our paper, we sort of give an example of using one of those indices to understand how rank might show up and how that's different than absolute value. And using that index, which is the Material Community Deprivation Index. If you look at the-- if you divide it into percentiles and you take the top quintile, the top 20% of deprived communities, that actually includes a massive amount of the variation.
00;10;33;25 - 00;10;51;02
Meera Kotagal
It actually is, you know, way more variation in that top quintile than there is across the other 80% combined. So when you do it by percentile, sometimes you lose some degree of the of the variation. And that's important to understand with rank versus value.
00;10;51;19 - 00;11;34;22
Alan Weil
Okay. You've done a really excellent job of describing some of the reasons you might combine in different ways. I want to focus a little bit more on what's in and what's out, and then we'll return to the topic of how you define a community. We'll cover those topics after we take a short break. And we're back. I'm speaking with Dr. Meera Kotagal, who's talking to me about area-based socioeconomic deprivation indices.
00;11;34;22 - 00;12;30;14
Alan Weil
Before the break, we were discussing some of the methodological dimensions of combining different measures into an index. But I'd like to focus a little bit more now on the actual domains included or excluded. Earlier on, you mentioned the inclusion of race in the Social Vulnerability Index. Given the conversations we're having in the health literature about the role of racism and the appropriate uses of race as an explanatory variable for analysis, I'd really like to get your thoughts on first of all, in your review of these 15 indices, did you see race as a factor and what your understanding is of the reasons you might include or exclude it?
00;12;31;11 - 00;12;40;26
Meera Kotagal
So we did. There definitely are. Of the 15 indices we reviewed, there are four that include race or ethnicity in the makeup of their index. And so it does play a role.
00;12;41;09 - 00;13;00;20
Alan Weil
And I should probably just when we say it's included, what you mean here is that when you're capturing an index of the level of deprivation of a community, you're taking one of the. One of the dimensions is the racial and ethnic composition of the residents of that community. That's what it means for it to be in.
00;13;01;08 - 00;13;23;21
Meera Kotagal
That's correct. It controls for the percentage of people living in a community that are of a minority race or specifically that are of Black race, and that can be included in the indices. And there are four that that do that. I think it's important when we think about that to recognize that a neighborhood's racial composition is not in and of itself actually a marker of socioeconomic deprivation.
00;13;23;29 - 00;13;44;04
Meera Kotagal
And to acknowledge that race is a social construct and not a biologic one. And often when we're thinking about race in these contexts, what we're really measuring is structural racism and racism in the ways it shows up, not race itself. And so I think there are some instances we talked a little bit about SVI where it makes a lot of sense that it was included.
00;13;44;04 - 00;14;12;13
Meera Kotagal
SVI was developed to assess vulnerability after disaster and the intentional choice was made to include race because there are higher rates of vulnerability after disaster in communities with a higher percentage of Black residents even after you consider their socioeconomic status. But it's important to know when you're doing using that tool for other analyses how having race as a part of that tool might affect both your results and be implicated in the assumptions that you're making.
00;14;12;24 - 00;14;39;19
Meera Kotagal
It may be entirely appropriate to use an index that includes race, but if you're trying to assess for racial inequities and you control for individual-- you're trying to assess for individual-level racial inequities, and you control for area-level racial makeup and your deprivation index impacts your results. And so calling out those assumptions is really important and making sure that when we pick the indices and the tools we use, we understand the implications of doing so for our results.
00;14;40;15 - 00;14;51;14
Meera Kotagal
I think we really need to recognize and acknowledge that race is a social construct, and when we're looking at racial inequities, we need to understand what we're finding. Are the differences really about race or are they really about racism?
00;14;52;01 - 00;15;11;09
Alan Weil
So that's a really helpful way of thinking about their implications to this decision. It's not a right or wrong. It's a it affects your results. And and you need to know going in that it's going to affect your results. And then the question is, does it affect them in ways that are intentional and useful for what you're trying to understand?
00;15;12;06 - 00;15;36;19
Alan Weil
The same logic, I suppose, applies to lots of other domains, but probably less complex in terms of the what is it you're measuring? Can you describe some of the other sort of more substantive areas where some of the indices chose to include and others didn't, and maybe some of your thinking about those choices and why you would in certain instances prefer to use one over another?
00;15;37;05 - 00;16;04;25
Meera Kotagal
Sure. I think there are a couple of really important examples. I think one is around transportation. So some folks have included there are four or five indices that include transportation-- availability of transportation-- as a part of that metric and the likelihood the percentage of people who own cars in the neighborhood, things like that, and that can be useful because it can be a measure of understanding access to care, you know, poverty and deprivation and the ways it impacts people's access.
00;16;05;03 - 00;16;27;28
Meera Kotagal
But it also is very region-dependent. So if you're doing a study in Manhattan and you look at the percentage of people who own a car, that's probably not really reflective of community-level deprivation as it might be in other communities around the United States. So again, the assumptions we make are important and understanding them and I totally agree with your point that there isn't a right or wrong answer to these indices.
00;16;27;28 - 00;16;53;06
Meera Kotagal
There are just choices. And we need to understand with intentionality what those and choices are and how they impact both the assumptions we're making and the results that we find. I think the other example that can be included is thinking about language and primary language and understanding how that might impact. There are definitely barriers that come with access to health care and access to health interventions as a result of non-English primary language.
00;16;53;17 - 00;17;11;18
Meera Kotagal
But not all things are impacted by that and not all communities where English is not the primary language are are communities of deprivation or communities without resources. So again, thinking about when it’s a useful tool to include or when we would rather that we didn’t include it because we're controlling for something that isn't our intention.
00;17;12;04 - 00;17;44;13
Alan Weil
Well, you know, we're a health policy journal, and so we're often publishing papers that look at the effects of program or policy intervention. And many of the papers that we publish include some use of a area deprivation index. Can you talk a little bit about sort of based on what you found in your review and and some of these very important considerations and what are the right ones to use?
00;17;44;20 - 00;17;57;05
Alan Weil
Can you talk about sort of the the policy implications of having so many and the policy implications of the differences in the ways social deprivation is described?
00;17;57;18 - 00;18;19;18
Meera Kotagal
I think there's definitely a role for multiple indices because each has some advantages and some disadvantages. And I think acknowledging that is important. I don't think the goal should be to get to a single index that we use for everything because they're not all appropriate for all studies or all interventions. Some input variables are not always clearly linked to socioeconomic deprivation.
00;18;19;18 - 00;18;46;14
Meera Kotagal
And so like we've talked about, the assumptions are important. However, having multiple indices does generate challenges around interpretation and generalizability of those results. I think they can be used in really useful ways in policy, and some examples, and we cite some of these in the paper, Mass Health uses census tract neighborhood, an index score of sort of stress at the census tract level, to adjust payments according to a patient social risk.
00;18;46;14 - 00;19;14;04
Meera Kotagal
So if you live in communities of higher deprivation, those payments are different. CMS’ Accountable Care Organization, which is for realizing equity access and community health, or REACH, that model, adjusts reimbursement rates based on area deprivation and uses that and that really-- those implications and that use of these tools in policy is really a reflection of understanding the difference between equality and equity and recognizing that equity is around what resources are needed to get everyone to the same outcome.
00;19;14;04 - 00;19;31;23
Meera Kotagal
And for many of our communities that are increasingly deprived or that have lower access to resources, they may need additional resources in order to reach the same outcome. And so those policy choices use these indices to help make adjustments to provide those additional resources to communities in specific need.
00;19;32;17 - 00;20;14;17
Alan Weil
I am curious, you know, a lot of work around social policy has been trying to move us from a deprivation model to a strengths model. And even the term socioeconomic deprivation index sort of implies you're missing something. How do you feel about the term here and the reasoning for developing indices like this relative to this question of are we just sort of minimizing the strengths and resources that exist in communities that don't have them for for reasons of exclusion and racism, not because of something inherent in the community?
00;20;15;14 - 00;20;39;23
Meera Kotagal
I love that question, and I think it's really important so that, you know, social determinants of health, we often think of them as negative social determinants of health, right? You don't have this, you don't have that. That's how it impacts your access to care. But there are also very significant, positive social determinants of health and our role as as individuals trying to make a difference in this space is to both mitigate the negative ones and augment positive ones.
00;20;39;23 - 00;21;01;02
Meera Kotagal
And I think you are right that these tools really focus on deficits within communities. The advantage to that, I think, is that it allows us to identify gaps that we can then remedy in order to make a positive impact. So it does provide us with an opportunity for making a change and really making an intervention to address a specific gap that may be resulting in negative health outcomes.
00;21;01;02 - 00;21;26;27
Meera Kotagal
But you are correct that these do not measure the well-being of communities or social cohesion or the ways that communities are intersected and thrive. And there are other tools that can do that. These particular indices don't, but there are tools that can and I think important to understand social dynamics, modeling or other tools that can be used to demonstrate the ways in which the intersections and connections with communities provide strength to those communities.
00;21;27;18 - 00;21;52;21
Alan Weil
So I realize we got to almost the end of our conversation and I somehow managed to skip over a topic and Jeff will have to decide whether he throws it in here or tries to cut and paste it back earlier, but can you say a little bit about the dimension of what the area is? You mentioned the the choice of zip code or census tract, but just in as a general matter,
00;21;52;21 - 00;22;04;24
Alan Weil
how do we think about what the scale of community is that's relevant to understanding whether people's living circumstances have the resources they need to thrive or not?
00;22;05;07 - 00;22;26;08
Meera Kotagal
In an ideal world, we would go to the neighborhood and neighborhoods are hard to measure in big data. And so we often think about this at a couple of different levels. The zip code level, census tract level or census block group level. Most of the indices use the census tract level. Area deprivation index is available at the census block group level and some of the other indices,
00;22;26;08 - 00;22;47;24
Meera Kotagal
the Community Need Index, the Distressed Communities Index are only available at the zip code level. And the question of why it matters I think is really important. Zip codes can be quite variable in size, and they also may include multiple different types of communities. And so sometimes you can lose the signal at the zip code level. And so it's not truly a reflection of neighborhood context.
00;22;48;07 - 00;23;09;09
Meera Kotagal
I think there's a great example of this in the initial work around the lead issues in Flint. And as people started to measure lead levels in Flint and they used zip codes, they really weren't able to identify a signal. But when you look at a map of Flint, those zip codes actually extend beyond the city boundaries of Flint, where the lead levels were really quite problematic, into suburban areas.
00;23;09;09 - 00;23;28;12
Meera Kotagal
And so you're pulling a number of of individuals and communities into that assessment. And as a result, you lose your signal. When you're able to define it much more specifically at the census tract level, you actually can identify that there's a significant problem. So sometimes people use zip codes because the data can be more complete, there's no geocoding that's required,
00;23;28;20 - 00;23;38;13
Meera Kotagal
you're able to address concerns around privacy and confidentiality, but we have to recognize that there's significant heterogeneity within zip codes and they may be a questionable proxy for actual neighborhood.
00;23;39;04 - 00;24;02;17
Alan Weil
Well, Dr. Kotagal, thank you so much for really taking a quite complex topic and making it accessible, for helping me and hopefully our listeners understand the importance of these indices, but also the importance of picking the right one and the implications of maybe picking one that doesn't exactly line up with what we need. There's no question these are here to stay,
00;24;02;25 - 00;24;24;27
Alan Weil
and so I hope you'll continue to keep an eye on them and give us ongoing guidance on what the right way is to think about the the interaction between where someone lives and their health status. For the excellent conversation and the paper, I thank you, and I also thank you for being my guest today on “A Health Podyssey”.
00;24;25;28 - 00;24;45;09
Meera Kotagal
Thank you so much for having me. I really want to acknowledge that this work was the work of many and in particular, Dr. Stephen Trinidad, who is the first author on this work, who really led this effort and is a really doing incredible work. And I think we're grateful for the opportunity to have this in Health Affairs and to talk with you about it today.