Association of Neighborhood Deprivation With Cognitive and Mood Outcomes in Adults With Pharmacoresistant Temporal Lobe Epilepsy

Abstract

Background and Objectives

Temporal lobe epilepsy (TLE) is the most common adult form of epilepsy and is associated with a high risk of cognitive deficits and depressed mood. However, little is known about the role of environmental factors on cognition and mood in TLE. This cross-sectional study examined the relationship between neighborhood deprivation and neuropsychological function in adults with TLE.

Methods

Neuropsychological data were obtained from a clinical registry of patients with TLE and included measures of intelligence, attention, processing speed, language, executive function, visuospatial skills, verbal/visual memory, depression, and anxiety. Home addresses were used to calculate the Area Deprivation Index (ADI) for each individual, which were separated into quintiles (i.e., quintile 1 = least disadvantaged and quintile 5 = most disadvantaged). Kruskal-Wallis tests compared quintile groups on cognitive domain scores and mood and anxiety scores. Multivariable regression models, with and without ADI, were estimated for overall cognitive phenotype and for mood and anxiety scores.

Results

A total of 800 patients (median age 38 years; 58% female) met all inclusion criteria. Effects of disadvantage (increasing ADI) were observed across nearly all measured cognitive domains and with significant increases in symptoms of depression and anxiety. Furthermore, patients in more disadvantaged ADI quintiles had increased odds of a worse cognitive phenotype (p = 0.013). Patients who self-identified as members of minoritized groups were overrepresented in the most disadvantaged ADI quintiles and were 2.91 (95% CI 1.87–4.54) times more likely to be in a severe cognitive phenotype than non-Hispanic White individuals (p < 0.001). However, accounting for ADI attenuated this relationship, suggesting neighborhood deprivation may account for some of the relationship between race/ethnicity and cognitive phenotype (ADI-adjusted proportional odds ratio 1.82, 95% CI 1.37–2.42).

Discussion

These findings highlight the importance of environmental factors and regional characteristics in neuropsychological studies of epilepsy. There are many potential mechanisms by which neighborhood disadvantage can adversely affect cognition (e.g., fewer educational opportunities, limited access to health care, food insecurity/poor nutrition, and greater medical comorbidities). Future research will seek to investigate these potential mechanisms and determine whether structural and functional alterations in the brain moderate the relationship between ADI and cognition.

Temporal lobe epilepsy (TLE) is the most common type of focal epilepsy and is associated with a high risk of cognitive deficits and depressed mood, particularly in patients with pharmacoresistant seizures.^1,2 In fact, up to 80% of patients with intractable epilepsy demonstrate cognitive impairments on neuropsychological testing and up to 70% have a Diagnostic and Statistical Manual of Mental Disorders psychiatric diagnosis, with depression and anxiety disorders among the most common.^1-4 These comorbidities negatively affect daily functioning and quality of life⁵ and are reported by patients to be among the most concerning aspects of their disease.⁶ Over decades, a major focus in the neuropsychology of epilepsy has been the relationship between cognitive status and the taxonomy of epilepsy and its related clinical characteristics. In TLE, poor neuropsychological outcomes have been linked to early age at seizure onset, long duration of epilepsy, a history of secondarily generalized tonic-clonic seizures, dominant-sided seizures, and etiology (presence of mesial temporal sclerosis [MTS]).⁷ While the focus on disease-related factors is understandable, precious little is known about the impact of the social and environmental complications of epilepsy on cognition and mood in TLE. Epilepsy is more prevalent among individuals in lower (more disadvantaged) socioeconomic groups, independent of social drift and other known epilepsy risk factors.^8-10 Furthermore, individuals with epilepsy are more likely to live in households with the lowest annual incomes.¹¹

By contrast, over the past decade, research on the social determinants of health has grown exponentially and suggests that social factors are a fundamental cause of health and disease.¹² The neighborhood environments in which people reside or otherwise spend their time are important contextualizers of health and longevity. Neighborhood characteristics, such as socioeconomic status and racial/ethnic composition, are associated with health disparities and health outcomes in many chronic conditions, including epilepsy.^13-16 Furthermore, factors such as systemic and structural racism (e.g., housing discrimination, educational segregation, unfair lending practices, and environmental injustice) and systematic inequality often force minoritized racial and ethnic groups to reside in regions with greater deprivation.¹⁷ A number of studies over the past several years outside of epilepsy have shown that neighborhood deprivation is associated with brain morphology, volume, and connectivity, cognitive performance/decline and depressed mood.^18-22 However, research on the complex interactions among neighborhood deprivation, race/ethnicity, and brain health is only in its infancy.

A conceptual framework for social determinants of health in epilepsy has been proposed,²³ and a growing body of research has demonstrated the impact of socioeconomic and/or neighborhood deprivation on not only the incidence and prevalence of epilepsy and utilization and access to medical care but also on a number of important epilepsy outcomes including health literacy, food insecurity, stigma, pregnancy outcomes, and surgical outcomes.^23-25 However, despite the high prevalence of cognitive dysfunction and depressed mood in TLE and the growing awareness of the social determinant of health, very few studies have examined the relationship between socioeconomic status and neuropsychological functioning,^26-28 and we are not aware of any studies on the impact of neighborhood deprivation, more broadly, on cognitive and mood outcomes.

The Area Deprivation Index (ADI) was developed by Singh in 2003²⁹ and updated by Kind and Buckingham in 2018³⁰ to quantify neighborhood-level socioeconomic position and includes 17 data elements derived from US Census and American Community Survey data (e.g., education, employment, housing, and poverty).³⁰ High ADI scores indicate greater neighborhood deprivation. The objective of this study was to use the ADI to examine the relationship between neighborhood deprivation and neuropsychological outcomes in adults with pharmacoresistant TLE.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

This was a retrospective observational cohort study. All data for the study were obtained from an existing institutional review board (IRB)–approved data registry at Cleveland Clinic. The requirement for informed consent was waived.

Participants

Cases were selected from an IRB-approved neuropsychology registry for older adolescents and adults with pharmacoresistant epilepsy who were being evaluated for epilepsy surgery at Cleveland Clinic between 1986 and 2021. Individuals were included if they met the following criteria: (1) age 16 years or older, (2) a history of pharmacoresistant TLE, (3) completed a comprehensive neuropsychological evaluation as part of a presurgical work-up, (4) had no history of prior resective neurosurgery, and (5) available home address at or around the time of cognitive evaluation. For those individuals who may have been assessed multiple times, data from their first neuropsychological assessment (i.e., cognitive, mood, and anxiety) were used in these analyses.

Assessment of Neighborhood Deprivation

Patient addresses were extracted from the EPIC (Verona, WI) electronic health record. The R sociome package was used to estimate 2018 ADIs, nationally, at the census block group–level using 5-year American Community Survey data (2014–2018). The ADI is a standardized score ranging from 40 to 160 with a mean score of 100, with higher scores indicating greater neighborhood deprivation. ADI percentiles were categorized into groups according to their respective quintiles of the overall ADI distribution (i.e., quintile 1 = least disadvantaged and quintile 5 = most disadvantaged).

Neuropsychological Outcomes

All patients completed a comprehensive neuropsychological evaluation as part of a multidisciplinary workup for potential epilepsy surgery, including measures of intelligence, attention/working memory, visuomotor processing speed, language, executive function, visuospatial skills, and verbal and visual memory (eTable 1, links.lww.com/WNL/C733). Test scores within each cognitive domain were administered and scored according to the test manuals, and all scores were transformed into standard scores (mean = 100, SD = 15). Scores within each cognitive domain were averaged to generate a composite score for that domain. Delayed recall trials were used to generate composite scores for the verbal and visual memory domains.

Numerous studies in recent years have identified divergent cognitive and behavioral phenotypes among patients with seemingly homogenous epilepsy syndromes that relate to sociodemographic, clinical, and network characteristics.^31-34 In 2020, a task force was assembled by the International League Against Epilepsy and the International Neuropsychological Society to develop a consensus-based empirically driven approach to cognitive diagnostics in epilepsy research—International Classification of Cognitive Disorders in Epilepsy (IC-CoDE)—to enhance global collaboration and facilitate big data approaches to the neuropsychology of epilepsy.^35,36 The IC-CoDE was used to characterize overall cognitive phenotypes using a ≤1.5 SD cutoff to define cognitive impairment, across cognitive domains, for all study participants as intact, single-domain impairment, bi-domain impairment, or generalized impairment (i.e., impairment in 3 or more cognitive domains).

Finally, depression was assessed with the Beck Depression Inventory (BDI) and anxiety with the Beck Anxiety Inventory (BAI), self-report symptom inventories in which higher scores reflect greater symptoms of depression and anxiety, respectively. Raw scores were used to analyze these outcomes.

Other Demographic and Clinical Variables

Demographic variables included age, sex, race/ethnicity, and years of education. Information regarding each patient's race and ethnicity was obtained from self-report demographic data contained in their EPIC medical record. Available options for race were American Indian/Alaska Native, Asian, Black, Native Hawaiian/Pacific Islander, White, multiracial/multicultural, declined, or unavailable. Available options for ethnicity were Hispanic, not Hispanic, declined, or unavailable. Clinical variables included age at seizure onset, duration of epilepsy (i.e., age during neuropsychological assessment minus age at seizure onset), number of antiseizure medications (ASMs), side of seizure onset, and presence/absence of MTS on MRI.

Statistical Analyses

Descriptive statistics were calculated for the overall sample and then stratified by ADI quintile using standard univariable summary statistics. Categorical variables are presented as number (percentage) and continuous variables (including continuous baseline characteristics and cognitive, mood, and anxiety score outcomes) as median (25th and 75th percentiles). Nonparametric Kruskal-Wallis tests were used to compare ADI quintile groups on cognitive, mood, and anxiety scores.

Cognitive phenotype was modeled as an ordinal response variable using cumulative logit models. We estimated 2 multivariable models for cognitive phenotype. The first model (model 1) included age, sex, education, minority or other race, and ethnicity (vs non-Hispanic White), age at seizure onset, duration of epilepsy, number of ASMs, seizure side, and the presence of MTS on MRI. The second model (model 2) included these covariates and ADI quintile. This was performed to compare covariate relationships in the absence and presence (respectively) of ADI effects. BDI and BAI scores were modeled in a similar way, with the only difference being that we used multivariable ordinary least squares regression instead of cumulative logit regression. Effect estimates were presented as proportional odds ratio (POR) (for cognitive phenotype) or difference in mean values (for BDI and BAI scores) with Wald 95% CIs and p values. Probability distributions of cognitive phenotypes were summarized graphically by ADI, both on a univariable basis (sample proportions by group) and on a covariate-adjusted basis (adjusted mean values from model 2).

We conducted a sensitivity analysis for our model of cognitive phenotype in which ADI was treated a continuous predictor (as opposed to groups defined based on quintiles). For this model, we used a quadratic effect for the ADI. Analyses were implemented on a firewalled Unix server located at Cleveland Clinic, using the RStudio Integrated Development Environment,³⁷ Server Edition, version 1.3.1093, and R statistical software version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria).

Data Availability

Datasets analyzed in this study are not publicly available, but further information about the datasets is available from the corresponding author on reasonable request.

Results

Of 1,362 individuals with available data, a total of 800 patients met all inclusion criteria for the study. Most of those not included predated institution of the electronic health record at our hospital and, as such, did not have a home address available for calculation of the Area Deprivation Index. Of the final 800 patients included, neuropsychological testing was completed between 1990 and 2021, with most of the patients (98%) completing their evaluation between 2000 and 2021. Patients were a median of 38.3 years of age (interquartile range [IQR] 28.3–47.9 years) with 12 years (IQR 12–18 years) of education. A majority of the participants were female (n = 460; 58%) and most were White, non-Hispanic (n = 718; 90%). The median age at seizure onset was 16 years (IQR 8–27 years), and the median duration of epilepsy was 17 years (IQR 8–28 years). Of the 800 total patients included in the cohort, 763 had complete data on study covariates and were used in the models for cognitive phenotype. Of them, BDI scores were unavailable for 49, such that our models for that metric incorporated data on 714 patients. Similarly, our models for BAI included data on 360 patients for whom this outcome was assessed.

Demographic and disease characteristics of the sample are summarized, by ADI quintile, in Table 1, and neighborhood characteristics of each ADI quintile are summarized in eTable 2 (links.lww.com/WNL/C733). Patients from neighborhoods in higher ADI quintiles (higher neighborhood disadvantage) were more likely to be male; were more likely to be of non-Hispanic Black, Hispanic, or other race and ethnicity; had fewer years of education; and were slightly younger at seizure onset. Distributions of duration of epilepsy, number of ASMs, and seizure side were generally similar across ADI quintiles. Patients from neighborhoods in the lowest (least disadvantaged) ADI quintile had a lower prevalence of MTS (38%) than patients from neighborhoods in ADI quintiles 2–5 (55%).

Table 1.

Demographic and Clinical Information Stratified by ADI Quintile

Summary statistics for cognitive, mood, and anxiety variables are listed in Table 2. Based on the Kruskal-Wallis tests, we found significant declines in all these outcomes as a function of increasing ADI quintile (worsening neighborhood disadvantage), with the exception of executive function, for which the relationship was marginally insignificant.

Table 2.

Cognitive, Mood, and Anxiety Scores Stratified by ADI Quintile

Table 3 contains PORs for being in a more (vs less) severe cognitive phenotype as a function of the covariates specified earlier (model 1) and additionally ADI quintile (model 2). Before including ADI quintile, patients who were members of minoritized groups were nearly 3 times as likely to be in a more severe cognitive phenotype than non-Hispanic White patients (POR 2.91, 95% CI 1.87–4.54; p < 0.001). This relationship was attenuated after the inclusion of ADI quintile in the model (model 2 POR 1.82, 95% CI 1.37–2.42; p < 0.001), suggesting a potential mediating effect of neighborhood deprivation on the relationship between race and ethnicity and cognitive phenotype.

Table 3.

Proportional ORs for More Severe Cognitive Phenotypes

By contrast, ADI quintile seemed not to account for racial and ethnic differences in BDI scores (covariate-adjusted difference [95% CI] in the mean BDI score between the minoritized group and non-Hispanic White group before including ADI quintile in the model: 5.6 [3.1–8.1], p < 0.001; difference [95% CI] after including ADI quintile in the model: 5.7 [3.1–8.4], p < 0.001) (Table 4). Covariate-adjusted racial and ethnic differences in BAI scores were not statistically significant before or after including ADI quintile in the model (Table 5).

Table 4.

Linear Models of Depression Based on BDI Score

Table 5.

Linear Models of Anxiety Based on BAI Score

More disadvantaged ADI quintiles (quintiles 3–5) were generally associated with increased odds of more severe cognitive phenotypes in comparison with quintile 1 (POR estimates ranging from 1.31 to 1.42; see Figure 1 and “Model 2” results in Table 3), although the comparison of quintile 5 with quintile 1 was not statistically significant (p = 0.10). Sensitivity analysis suggested a significant relationship between ADI as a continuous predictor and cognitive phenotype (Wald χ² test p = 0.013 for the ADI effect as represented through a quadratic curve). eFigure 1 (links.lww.com/WNL/C733) displays covariate-adjusted phenotype probabilities as a function of (continuous) ADI from this model.

Figure 1. Phenotype Probabilities by ADI Quintile.

(A) Unadjusted model. (B) Covariate-adjusted model controlling for age, sex, education, race, age at seizure onset, duration of epilepsy, number of antiseizure medications, seizure side, MTS, and ADI quintile. ADI = Area Deprivation Index; MTS = mesial temporal sclerosis.

Our models suggest that higher education is associated with a reduced likelihood of having more severe disease, the use of additional antiseizure medications was associated with an increased likelihood of having more severe disease, and male sex was associated with lower BAI scores, although we note that the models were not primarily designed to independently evaluate these effects and may be subject to residual confounding.³⁸

Discussion

Our results demonstrate a significant association between neighborhood deprivation and neuropsychological outcomes in adults with pharmacoresistant epilepsy. Notably, we observed significant declines across nearly all measured cognitive domains (e.g., attention, memory, and language) and significant increases in symptoms of depression and anxiety as a function of increasing ADI. Furthermore, patients in higher ADI quintiles had increased odds of having a more severe cognitive phenotype.

Drivers of the relationships between ADI and neuropsychological function are undoubtedly multifactorial and extremely complex. Neighborhood disadvantage can adversely affect cognition through many different mechanisms (e.g., fewer and less adequate educational opportunities, limited access to health care, food insecurity/poor nutrition, increased exposure to environmental pollutants and toxins, chronic stress, physical inactivity, less socialization, and greater medical comorbidities).^22,39 Furthermore, recent studies outside the field of epilepsy have demonstrated strong relationships between neighborhood deprivation and neuroimaging findings, with greater abnormalities observed among individuals who live in disadvantaged areas.^18,19,21,40 We hypothesize that brain morphology, as assessed using various neuroimaging techniques, may mediate the relationship between ADI and cognition in patients with epilepsy and have additional research underway to test this hypothesis. It should also be noted that epilepsy can place a significant burden on the patient and their support structure (e.g., medical expenses, adverse treatment events, employment concerns, and inability to drive),⁴¹ potentially pushing them into areas with greater neighborhood deprivation. Furthermore, there is evidence to suggest that racialized inequalities in epilepsy result in greater burden among some racial/ethnic groups compared with non-Hispanic White individuals.⁴²

Consistent with known racial disparities in the United States and findings in prior ADI studies,^43,44 we observed an overrepresentation of individuals identifying as members of minoritized groups in the highest ADI quintiles. These individuals were also more likely to demonstrate a more severe cognitive phenotype than non-Hispanic White individuals. Such differences in cognitive performance are commonly observed on neuropsychological measures and known to be affected by test biases, normative issues, and societal inequities (e.g., education and adverse childhood experiences).^45,46 Our observation that the strength of the relationship between race and cognitive function was in fact attenuated by the inclusion of the ADI suggests that neighborhood deprivation may mediate, at least partly, the relationship between race/ethnicity and cognitive profile. Prior studies have also found neighborhood factors, such as socioeconomic position, explain more variability in cognitive test performance than race and ethnicity,⁴⁷ highlighting the importance of contextual factors in cognitive and behavioral functioning. Furthermore, it is important to remain cognizant that race is a social construct, and the residual disparity observed between racial and ethnic groups in this study is likely due to uncaptured differences in the experiences of these minoritized groups rather than to intrinsic differences in biology (e.g., brain volume/function). Of interest, recent research has suggested that social support may serve as a buffer against volume loss and cognitive decline in minoritized groups.^48,49 Certainly, future research in larger, more diverse samples will be needed to disentangle the complex relationships between race/ethnicity, neighborhood disadvantage, other contextual factors and contributors to cognitive function, and neuropsychological functioning.

The study results should be considered in light of several limitations. First, the ADI was measured at only 1 time point in adulthood (at a date close to the time of neuropsychological evaluation) and does not account for potential neighborhood deprivation during critical periods of early development or changes in neighborhood deprivation across the life span. Furthermore, we used the 2018 ADI to analyze our data, which were obtained over a large time span (1990–2021). ADI data generated from a limited date range may have affected our results to some extent. However, prior work has suggested that most neighborhoods are unlikely to change drastically, even over relatively long periods of time, and that the ADI methodology is fairly robust to temporal variation.^29,50 But more directly, the relationship between cognition and the ADI remained as described when the subset of patients seen from 2014 to 2018 was examined. Second, this was a cross-sectional study that examined cognition at only 1 point in time. Longitudinal studies will be required to examine the relationship between the ADI and changes in cognition over time. Third, all patients in this study had pharmacoresistant TLE, and results may not generalize to other epilepsy subtypes or to less severe forms of epilepsy. Fourth, most of the individuals in our sample (90%) self-identified as White, non-Hispanic, and most (67%) resided within the state of Ohio or surrounding states in the Midwest (20%). Furthermore, given the relatively small number of non-White and Hispanic patients in the study, we had to aggregate all patients from minoritized groups into a single “minority/other” category for purposes of our statistical analyses. Future studies, in larger cohorts with greater racial and ethnic diversity, will be required to examine potential differences between specific minoritized subgroups. Nevertheless, most of the patients (91%) in our minoritized group were Black or Hispanic, groups that have been systematically disadvantaged and forced to live in more deprived areas. Finally, the measures of mood and anxiety were self-report, and comparable investigations using formal psychiatric diagnoses is an important future direction.

Research in the neuropsychology of epilepsy has arguably ignored the potential impact of the social determinants of health on cognition—factors that have been hiding in plain sight for many years. Their impact on cognition seems to be pervasive and of clinical significance because ADI remains a significant predictor of neuropsychological functioning even after for controlling for the clinical and seizure features of epilepsy that have for so long been the dominant interest in the field. Clearly, a realignment of research focus seems warranted with a special eye to those social factors that may be modifiable and serve to improve cognition and prevent further worsening. Interest in modifiable risk factors to improve cognition is a major focus in other disorders such as Alzheimer disease (e.g., diet, exercise, cognitive activities, etc.) and would represent important advances in the neuropsychology of epilepsy.

Glossary

ADI

Area Deprivation Index

ASM

antiseizure medication

BAI

Beck Anxiety Inventory

BDI

Beck Depression Inventory

IC-CoDE

International Classification of Cognitive Disorders in Epilepsy

IQR

interquartile range

IRB

institutional review board

MTS

mesial temporal sclerosis

POR

proportional odds ratio

TLE

temporal lobe epilepsy

Appendix. Authors

Study Funding

Primary support for this study was provided by the Cleveland Clinic Epilepsy Center.

Disclosure

R.M. Busch receives support from the National Institute of Neurological Disorders and Stroke (NINDS) (R01NS120976, R01NS035140, R01NS097719, and R61AG069729). J.E. Dalton receives support from the NIA (R01G055480 and R01AG059979) and NHLBI (R01HL153175). L. Jehi receives support from NINDS (R01NS097719) and NCATS (UL1TR002548). L. Ferguson receives support from the NINDS (R01NS120976, R01NS035140, and R01NS109493). A.F. Struck receives funding from Ceribell. A.F. Struck and B.P. Hermann are supported by the NINDS (R01NS111022, R01NS120976, R01NS117568). None of these grants are directly related to the project reported in this manuscript. Go to Neurology.org/N for full disclosures.