Adriana Bermeo (Host): It is well known that individuals living with epilepsy are at risk of experiencing cognitive decline and psychiatric comorbidities. But did you know that the individual's home address plays a significant role in the cognitive and psychiatric outcomes of people living with temporal lobe epilepsy?

Welcome to episode number six of Epilepsy Currents podcast. Today, we will be talking about the role of environmental factors on cognition and mood of patients with temporal lobe epilepsy. I am your host, Adriana Bermeo. I am the Senior Podcast Editor for Epilepsy Currents, the official journal of the American Epilepsy Society.

I want to first welcome contributing editor, Dr. Samuel Terman, who wrote the commentary, Hiding in Plain Sight--Neighborhood vs. Individual Determinants of Psychological Outcomes in Patients with Epilepsy. This commentary was published on the March 2024 issue of Epilepsy Currents. Dr. Terman is an Assistant Professor at the University of Michigan. Dr. Terman, welcome to Epilepsy Currents Podcast.

Samuel W. Terman, MD, MS: Thanks so much. Happy to be here.

Host: It is also my pleasure to welcome Dr. Bruce Hermann, who is the Senior Author of the work that inspired this commentary, titled Association of Neighborhood Deprivation with Cognitive and Mood Outcomes in Adults with Pharmacoresistant Temporal Lobe Epilepsy. Their paper was published in Neurology in June 2023. Dr. Hermann is Emeritus professor of Neuropsychology at the University of Wisconsin School of Medicine and Public Health. Welcome, Dr. Hermann. Thank you for being with us.

Bruce Hermann: Thank you, and good afternoon.

Host: We are also joined by Dr. Jarrod Dalton, who is co-author on the original paper. Dr. Dalton is an Associate Professor and the Director of the Center for Population Health Research at the Cleveland Clinic's Lerner Research Institute. Dr. Dalton, thank you for being with us today.

Jerrod Dalton, PhD: And thank you.

Host: While neurologists and epileptologists are familiar with the effects of the biological characteristics of the lesions causing epilepsy, the neurophysiological findings of the treatments we use in the clinic, we are much less likely to consider their patient's immediate living environment, their access to healthy foods, exposure to violence, or their access to transportation as potential modifiable variables directly affecting someone's epilepsy outcomes.

Dr. Hermann. Would you mind sharing with us a snapshot of the findings from your study in a way our listeners can grasp?

Bruce Hermann: Dr. Busch and I have been collaborators for quite some time, and she was able to compile an amazing cohort of 800 patients with pharmacoresistant temporal lobe epilepsy who underwent neuropsychological assessment as part of their inpatient evaluations at the Cleveland Clinic. And then, she and Dr. Dalton extracted home addresses from the EMR and binned the patients into ADI quintiles. And from the neuropsychological battery that was quite extensive, as is commonly the case in preoperative evaluations, the data was reduced to seven mean cognitive domains, things such as intelligence, language, visual perception, verbal and visual memory, executive function, processing speed. And then, secondarily classified all the patients into four cognitive phenotypes that range from intact, essentially unaffected, to generalized cognitive impairment.

So, the core finding and most important finding was that examination of the relationship of the Area Deprivation index was closely aligned with the cognitive scores. So as deprivation increased, cognitive scores went down. This was seen across six of the seven cognitive domain scores, and greater deprivation was associated with the more abnormal cognitive phenotype. And increasing neighborhood deprivation was also associated with greater anxiety and depression. So, it really was quite a signal that we were able to observe. And as Dr. Terman pointed out in his commentary, you can just take a look at IQ. Quintile 1, the least deprived group, had a mean IQ score of 100, which is dead average. And the most deprived quintile, Quintile 5, their mean IQ was 86, which is a full standard deviation below average. So, very reliable and fairly strong signal throughout these results.

Host: Yeah. Really striking findings, which seem to point out that the deprivation that patients are exposed is almost as important as the biology of the condition. And I'm sure we will discuss that in length later. Dr. Dalton, the concept of the Area Deprivation Index, or ADI, plays a central role in the study. Can you please help us understand what does this measure? How do these quintiles work? And how is this measurement different from other measurements of socioeconomic status?

Jerrod Dalton, PhD: Sure. Yeah, the Area Deprivation Index is a tool that we use in our research in a lot of settings. We work with clinical specialties across Cleveland Clinic and use this as one tool for understanding place-based health disparities. Essentially, the Area Deprivation Index, it's a latent variable model or a factor model that produces an index that captures correlation among 17 neighborhood-level characteristics. The ADI was established at the National Cancer Institute in 2003 by a demographer and sociologist named Gopal Singh. It was designed as scale of neighborhood socioeconomic position as opposed to socioeconomic status, which I can distinguish a little bit between the two. And it incorporates aspects of social organization, income, housing, economic inequality, and opportunity structure.

So, the indicators that comprise the ADI are varied, but they all speak to overall socioeconomic characteristics. So, it'll include measures of education, proportions by level of education. It includes measures of income, such as median family income, income disparities. It includes measures of housing and housing affordability, median home value, median rent, monthly mortgage amounts, and percent of households

It also includes economic and employment measures such as labor force participation rates and poverty rates, family structure measures such as the percent of single-parent households among those with somebody less than 18 years of age, and a series of resources such as motor vehicle ownership, telephone ownership, and whether or not households have plumbing.

Host: And those variables are related to the neighborhood where the patient is living rather than to the individual patient. Is that correct?

Jerrod Dalton, PhD: That is absolutely correct. When we look at these quintiles that you and Dr. Hermann were referring to, we have to be very careful not to stratify our patient populations based on observed socioeconomic position associated with their community, but rather stratify them according to the communities themselves.

Host: Dr. Herrmann, it is certainly very intriguing to think that the environment can shape cognitive functions, potentially not only for patients with epilepsy, but particularly for patients with epilepsy. Can you elaborate on how factors like the person's neighborhood or community may influence their health, particularly their cognitive abilities and maybe if you can make a mechanistic idea of how does that work or what would be the factors that intervene in that relationship?

Bruce Hermann: Yeah. Well from a neuropsychological standpoint, we don't have much of an understanding of how these neighborhood factors affect cognition. That, of course, was the point of the study. And as you infer, in the field of neuropsychology of epilepsy, the lion's share of attention has been paid to the relationships between cognition and features of the epilepsy and its treatment.

So, there are innumerable studies looking at aspects of cognition like intelligence, memory, language, and the epilepsy syndrome that the patient might have, or their age of onset, the duration, seizure frequency or severity, or the number of medications that they take. This is a classic heritage. And in addition, there's a large literature examining diverse neuroimaging correlates of cognition and psychiatric comorbidities, including depression and anxiety.

But we have known for some time, just for example, through the CDC behavioral surveillance studies, that the economic, social, and lifestyle correlates of epilepsy can be punishing, and those have been documented. But how these factors relate to cognition and behavior have been less well understood.

Host: Dr. Terman, in your commentary, you highlight some strengths and also some limitations of the study. Would you like to share some of the specific issues that our readers may need to be aware when interpreting the findings?

Samuel W. Terman, MD, MS: Yes. The study has many strengths, and like all studies, there are limitations and future directions. Some of the strengths have already been mentioned through Dr. Dalton and Hermann's discussion of the work, including a large sample size with an impressively long period of time in which patients were recruited with detailed expert cognitive assessments. Measuring the outcome well is important for any study. And while sometimes having many different domains and submeasures can raise concern for multiple comparisons when results are inconsistent, or you're not sure whether significant findings are true positives. In this case, I think it was a strength given convergence of results across multiple domains and subdomains of cognitive and neuropsychological measures.

But, that said, there are numerous limitations that I see. For example, as I read it, it was a cross-sectional study, which is good for hypothesis generation, but also that lack of temporal correlation, what happens over time. You can't distinguish causation from reverse causation. We can theorize what direction the relationship is going, but you can't be sure in a cross-sectional study. In other words, do neighborhoods affect people? Which is what we think from the study. Or do people non-randomly sort themselves into neighborhoods? That would be reverse causation, which is also plausible, but we can't tell.

And there's some confusing relationships. For example, age was one factor in the multivariable models, and it actually was not associated with degree of cognitive impairment, which we know that aging is a very important feature of a person's cognitive health. So, future studies could address these things. For example, longitudinal data could surmount some of these questions about causation versus reverse causation. There are even specialized statistical models that seek to isolate when in a person's life, what exposure matters most. For example, if a person's exposed to a neighborhood with a higher degree of deprivation earlier in life versus later in life, when does it matter most?

It's also been mentioned that the mechanism by which some of the features of the ADI leads to outcome changes is unclear. And related to measurement of the ADI, a major strength is its ability to aggregate and summarize over the 17 different components that have been mentioned. But we're also left not knowing exactly quite which one is most critical when we lump rather than we split.

And the ADI also, while it measures lots of things well, it also lacks potentially key measures: pollution, healthy food, social cohesion, the way a city is laid out. Also, if a high ADI neighborhood is juxtaposed next to a low ADI neighborhood, there may be features that are complicated dynamics that we're not capturing here. So, we can't be sure which component of the ADI and also if there's a component in there that's just correlated with something that we haven't measured. We don't know exactly what's the driving force here.

And then, two other thoughts in terms of perhaps limitations and/or future directions. The census block is an important grouping, which is how the ADI is categorized, but it's also not the only way that people are grouped. Neighborhoods in the world in which we live, we live in an online communities and a globalized society. So, just the sheer geographic location is likely important for healthy food and parks and pollution, but there's also so much more as we're connected to each other than just that specific census block.

Host: Let me give you a follow up on that. Dr. Dalton, any thoughts on following these findings in a longitudinal way or including other measures of deprivation or other social determinants of health.

Jerrod Dalton, PhD: Yeah. I think, you know, Dr. Terman's comments are 100% right on the spot. You know, I think when I work with scientists, what I usually do is I will mention a lot of these limitations. And I will say that the ADI is just an analytic tool.

A great first step in understanding social and place based health disparities in a particular, in a particular disease context, such as epilepsy. There are many issues with it. And the field of neighborhood measurement is very rapidly evolving. In terms of longitudinality, I think it's helpful for us to recall the social context under which we have place-based disparities in the United States, which was largely driven 80, 90 years ago by racial residential segregation and federal housing policy that discriminated against racial and ethnic minority groups.

So, in large part, we have this entrenched issue of economic deprivation in communities that disproportionately impacts those of minority race and ethnic status. And these things are very difficult to reverse. So as it relates to the question of the stability of overall measures of socioeconomic position over time, unfortunately, it's too stable. It's more stable than we would like it to be, that I personally would like it to be. And as a scientific rationale for a cross-sectional evaluation of area deprivation and relationship with outcomes, generally speaking, we have pretty stable results over time. Now, that said, there are of course issues of gentrification and neighborhood change over time that should be taken into account. And these available measures currently cross-sectionally defined. They don't accommodate in measurement fashion neighborhood change.

I might respond to a couple of the other good points that Dr. Terman made. Regarding the age, the age in this sample was relatively homogenous on the earlier side of the life course, around 20 to 50 years of age. So, mechanisms of cognitive decline in older adults may or may not had a large influence on the observed measures of cognition. And of course, the area deprivation and other place-based socioeconomic measures lack key mechanistic exposures like the ones that he mentioned. And like I said, it's a very active area of sociological research, and we frankly need enriched models that are more expressive than a single domain, and we've done a little bit of methodological work in our lab exploring just that.

Host: Dr. Hermann, besides the cognitive outcomes, the study also highlights worsening measurements of depression and anxiety in the patients with the most deprivation or the highest ADI quintile. How do you interpret these findings. And particularly, can you share with our listeners how did these findings vary when you took race into account, race and ethnicity?

Bruce Hermann: Let me address this in the following way. First, we've been speaking in the abstract a little bit about deprivation and in the supplemental table of the neurology article, Drs. Busch and Dalton put together a very nice table that's informative about deprivation where people stand in these various quintiles.

Let me just give you a few examples. For example, mean income. So, the mean income in the neighborhood and please correct my language if this is imprecise, Jarrod, but the mean income in the least disadvantaged quintile in that neighborhood is $118,000. The mean income in the most disadvantaged neighborhood, Quintile 5, was $36,000. Huge difference. The percent of families below the poverty level in the least disadvantaged neighborhood, 1%; in the most disadvantaged neighborhood, 27%. Single parent households with youth under 18, 12% in the least disadvantaged, 68% in the most disadvantaged. And just one more, the percent of households without a vehicle, 1% in the least disadvantaged, 19% in the most disadvantaged.

So in regards to the psychiatric comorbidities, there has been a long history of debate about what's important for a long time. The debate centered around where the seizure started from. So for those who had temporal lobe epilepsy, psychiatric comorbidity was driven by volumetric abnormalities in the depths of the temporal lobe and mesial regions, and/or the degree of spiking that might be shooting through limbic systems.

And so, the way role played by disadvantaged in social circumstances was relegated to have little role, actually. But if you think about these sorts of effects that I just mentioned, how can they not have an effect on somebody's emotional adjustment? And if you look at the general population and the relation of psychiatric comorbidities to these sorts of trends, they're quite apparent. So the most disadvantaged, and I'll start this sentence, but then ask Dr. Dalton to finish it, but in the most disadvantaged quintiles of our particular study minoritized groups were overrepresented as expected. And to address the issue of neighborhood disadvantage versus minority status, Jarrod carried out a number of analyses. And let me turn that over to you, Jarrod.

Jerrod Dalton, PhD: When we look at race and ethnicity and its role in relationships with health disparities, it's important to consider the mechanisms by which you would hypothesize there to be racial and ethnic differences and to study those mechanisms. And we did not hypothesize all such mechanisms associated with racial and ethnic minority status in this study. We evaluated one mechanism, which was the one I referred to earlier around racial residential segregation and the fact that in societal terms, not necessarily for a specific individual, but in societal terms, there are many barriers to economic mobility, and we have this prevalent problem of racial and ethnic minority populations residing in communities with low resources.

So, there are a couple potential causal models that can be used to motivate analyses into these relationships. One would be a mediation analysis where you consider the extent to which selection into low-resource communities is responsible for observed racial and ethnic disparities in your outcome. And that's exactly what we looked at here. So, we looked at models with and without adjustment for ADI and studied how the relationship between racial and ethnic minority status changed after accounting for what in technical terms would be the mediating effect of living in low-resource communities.

So for our proportional odds logistic regression analysis looking at severity of cognitive phenotypes, what we found without adjustment was that minority race or ethnicity was associated with three times the risk of being in a higher or more severe cognition phenotype compared to non-minority patients. And then, after accounting for the neighborhood deprivation, that relationship was reduced to about 1.8. So in other words, we do seem to have some signal of explanatory effect of ADI in terms of the racial and ethnic disparities in cognition in this patient population.

Host: Thank you for that. It seems like there's so many layers to these. I would like to follow with Dr. Terman. When we study, and even from the clinic, when you see patients with medication-resistant epilepsy, how can we differentiate what is what

Samuel W. Terman, MD, MS: well, it's a bit different from a research standpoint than a clinical standpoint, because there are things that you can do in one setting but can't do in another. From a research standpoint, it's a matter of measuring what matters when it matters, and we don't always have the luxury of having everything that we would like at exactly the right time points over a long period of time. But if you're trying to distinguish the effect of neighborhood deprivation versus the condition of epilepsy itself, you need to measure as many things about both of those things as accurately as you can. So, variables like what anti-seizure medicines is the person on, at what doses, and what adherence, and what days did they have seizures, and were they convulsive, and how long were the seizures, and were they in status epilepticus? And you can very quickly see that, "Oh my goodness, there are so many." The world is so complicated and there's so many things that could influence the outcome. So, it's matter of what's feasible and being able capture those variables that you wish to distinguish and then you can decide, "Am I going to do a mediation or a structural equation model or some other type of model to try to tease out the effect of different potentially correlated covariates.

From a clinical standpoint, sometimes if you're not sure, ultimately what you do is you manipulate what you can in the direction that you think it should be manipulated. For example, if you think a person could be having cognitive side effects due to their anti-seizure medicine and their seizures are not as problematic, then you go down. Or if you feel that their seizures are a driver of their cognitive status, you go up and see if they can tolerate it better. But unfortunately, at this day and age, epilepsy care is a bit of a trial and error process.

Maybe to take the next step in terms of another research avenue that people have done in socioeconomic disparities, people have even done some sorts of quasi experiments to exploit natural variation. There have been a limited number of actual randomized experiments where people were randomized to healthcare insurance versus not, or vouchers versus not, or even refugees coming out of dangerous neighborhoods or communities or countries. People were sorted new locations and people have studied these things. There're all kinds of variation, distance from a highway in terms of pollution and access to healthy food. So, there are lots of creative ways to measure the things that you would like to and the things that you think are important and would like to distinguish between. Sometimes if you can't, there are some other tricks up our sleeves using clever quasi experiments or true experiments when it's ethically and practically feasible to do so.

Host: Thank you very much. I will stay with that question and I would love to ask Dr. Hermann and Dr. Dalton, what's in the future? In the research arena when we consider the realm of environmental influences on neurologic conditions. In the discussion in your paper, you hint that maybe looking into how these variables affect the anatomical studies in patients with temporal lobe epilepsy may be in the works. Anything else, any future directions on where these type of studies are going or should go?

Bruce Hermann: Yeah. I think an immediate applicability is with children with epilepsy, and even better, children who have new-onset epilepsies that addresses some of the points that have been raised in this discussion. So if you go and you start at the beginning with a cohort of children with epilepsy and controls in their homes of origin developing over time, does epilepsy occur more often in high deprivation areas? Maybe, maybe not. What's the impact of deprivation on cognitive development going forward prospectively? What's the impact on brain development going forward? And if deprivation has an effect, how long does it last? Does it last one year, two years, three years? I mean, the literature on lifetime outcomes of youth with epilepsy generally, but not invariably, points to a somewhat difficult course in many regards as they age and grow older. And if these deprivation factors have an influence on cognition, cognitive development, brain development, then it becomes even more critical to get serious about this. And I think Dr. Terman had a very interesting point at the end of his commentary, you know, if you want to intervene, do we intervene on or about? And we need to have sense, we don't have a relative ranking of the importance of deprivation versus type of epilepsy versus seizure control versus type of medication on memory, on intelligence, on depression. How much variance do these factors account for? And if you get a rank ordering, this is easy to say but hard to do, but if you can generate a rank ordering, the explanatory power correlation at the least gives you some ideas about where to go and what to do. And I think that's at least one important future for the direction, as well as getting a really good handle on exactly how is deprivation exerting its effect on a person. And one way is through imaging.

Host: Thank you for that. I want to ask Dr. Terman to follow up on that, because you have some interesting insight in your commentary about what can communities do besides the providers. You mentioned urban development and maybe food deserts come into mind. Any suggestions or any ideas of what can we do as providers or as a community to maybe mitigate some of these inequities.

Samuel W. Terman, MD, MS: Yeah. Neighborhoods and cities are not randomly evolving entities. They are the product of planning. And so when we develop communities and neighborhoods, ideally we plan for success. Now, I'm far out of the realm of civil engineering and city planning and all this is very complicated, interfacing with politics and economics and political will. But, as Dr. Hermann mentioned, we need to understand which of these things are most critical. And after understanding which of these things are most critical, then deciding, well, if the problem is pollution, then that's a very different path than if the problem is lack of healthy food, or if the problem is crime and violence, or if the problem is physical activity, or if it's truly a product of medical care. You know, these lead to very different types of solutions. Ultimately, it may be that as is often the case, that causes and effects are bidirectional and multifactorial, so it may be there is not one single solution, but a comprehensive suite of solutions to plan communities as well as we can.

Host: Thank you all. I want to especially thank you for these very thought-provoking conversations. I'm sure as we go home to our own neighborhood and our own community and can imagine the communities and neighborhoods our patients are going back to after visiting us in the clinic. We may be planning in the future how to learn more about it and how can we make a difference over time.

We are running out of time. I want to thank each of our guests. I want to thank our listeners. Special thanks also to the American Epilepsy Society in partnership with the CDC and their initiative Disrupting Disparities: Documenting and Addressing Gaps in Epilepsy Care Through Healthcare Provider Education and Training. Their collaboration is the sponsor of today's episode.

I want to thank Dr. Rohit Marawar and the SAGE podcast production team. We look forward to having all of you back on our next episode. Remember to subscribe to Epilepsy Currents podcast, wherever you get your podcast, and send us your feedback, suggestions, or questions through our website, epilepsycurrents.Org, and follow us on X, former Twitter, @aescurrents. Until next time, everybody.