Individual- and community-level social determinants of health are associated with cognition in older adults with focal epilepsy

Abstract

Objective:

Epilepsy is associated with significant health disparities, including access to specialized care and adverse outcomes that have been associated with several social determinants of health (SDOH). We sought to examine the relationship between individual- and community-level SDOH and cognitive outcomes in older adults with epilepsy.

Materials and methods:

We collected clinical, SDOH, and neuropsychological data in 57 older adults with epilepsy. Individual-level SDOH included patient factors (quality of education, income, insurance, marital status) and early-life environmental factors (parental education and occupation, childhood employment). Neighborhood deprivation was measured with the Area Deprivation Index (ADI). Stepwise regressions were conducted to examine the independent contribution of individual-level SDOH to cognitive performance, and Spearman rho correlations were conducted to examine the relationship between ADI and cognitive performance. The SDOH profiles of patients who met the criteria for cognitive impairment were examined.

Results:

After controlling for clinical variables, patient factors (public health insurance, poorer quality of education) and early-life environmental factors (lower mother’s education, lower father’s and mother’s occupational complexity, history of childhood employment) were significant predictors of lower performance on measures of global cognition, verbal learning and memory, processing speed, and executive function. Higher ADI values (greater disadvantage) were associated with lower scores on global cognitive measures, verbal learning and memory, and executive function. Patients who met criteria for cognitive impairment had, on average, a greater number of adverse SDOH, including lower household incomes and father’s education, and higher ADI values compared to those who were cognitively intact.

Conclusion:

We provide new evidence of the role of individual- and community-level SDOH on cognitive outcomes in older adults with epilepsy. This emerging literature highlights the need to examine SDOH beyond epilepsy-related clinical factors. These data could inform the development of interventions focused on increasing access to epilepsy care, education, and resources and promoting brain and cognitive health within the most at-risk communities.

1. Introduction

Older adults with epilepsy represent the largest and fastest growing segment of individuals diagnosed with the disease [1]. These patients present with multiple and complex medical-somatic comorbidities, including hypertension, diabetes, cerebrovascular disease, head injury, pharmacological complications (i.e., polypharmacy, side effects of antiseizure medications: ASM), and psychiatric comorbidities, that place them at increased risk for cognitive decline and dementia. [2] In fact, several community-based studies have reported up to a 3-fold increased risk for Alzheimer’s disease and related dementias (ADRD) among individuals with epilepsy [3]. The onset and co-occurrence of neurobehavioral comorbidities, such as cognitive impairment, among older adults with epilepsy places significant physical, social, and economic burdens on the patient, caregiver, and society [4]. This may exacerbate existing health disparities among historically underserved and under-resourced populations, particularly those with lower socioeconomic status (SES) [5].

Social determinants of health (SDOH) are defined as the social, economic, political, and environmental factors that contribute to health and disease [6,7]. These factors include structural determinants, defined as the economic and social policies and cultural norms that shape the distribution of resources across society, which in turn shape downstream social determinants (e.g., education, income, neighborhood characteristics) influencing inequities in health. Compared to the general population, individuals with epilepsy demonstrate lower education, income, and occupation attainment, and are more likely to experience discrimination and stigma, and report reduced social support [8,9]. These SDOH factors have been associated with disparities in diagnoses, treatment, and health outcomes in epilepsy [5,10,11]. Notably, most of the research in this area has focused on investigating the impact of SDOH on access to epilepsy care (e.g., medication, surgery), treatment adherence, and epilepsy outcomes (i.e., seizure outcomes), with a more recent literature examining the impact of SDOH on cognitive outcomes [12–15]. Specifically, individual-level SDOH including education, occupation, and income have been associated with poor cognitive outcomes [15,16]. Neighborhood deprivation, a community-level SDOH, has been of recent interest given the availability of metrics that capture several community-level SDOH domains (e.g., education and occupation opportunities, housing, transportation). Studies in epilepsy have demonstrated that neighborhood deprivation is associated with adverse cognitive outcomes in children [17] and young-to-middle-aged adults with epilepsy [13,15]. However, the impact of individual and community-level SDOH on the cognitive health of older adults with epilepsy has not been examined. This is particularly critical given the extensive literature linking SDOH with cognitive decline and dementia in the general population [18,19]. Further, there are no studies examining early-life environmental factors (e.g., parental education) known to impact late-life cognitive decline in older adults with epilepsy.

In this study, we 1) characterize the individual-level SDOH (i.e., patient and early-life environment factors) of older adults with epilepsy, 2) evaluate the relationship between individual and community-level SDOH and cognitive performance, and 3) examine the SDOH profiles of older adults with cognitive impairment. We define cognitive impairment (i.e., intact versus impaired) based on the International Classification of Cognitive Disorders in Epilepsy (IC-CoDE), a validated cognitive diagnostic taxonomy for epilepsy [20].

2. Methods

2.1. Standard protocol Approvals, Registrations, and patient consents

This study was approved by Institutional Review Boards at University of California, San Diego (UCSD), Cleveland Clinic (CC), and University of Wisconsin-Madison (UWM). Written informed consent was obtained from all patients. Patients were recruited as part of the BRain, Aging and Cognition in Epilepsy (BrACE) study–a prospective ongoing longitudinal study of cognitive and brain aging in older adults with epilepsy. All participants in BrACE currently have baseline data collected. Inclusion criteria were: age ≥ 55 years old, diagnosis of focal epilepsy by a board-certified neurologist with expertise in epileptology in accordance with the criteria defined by the International League Against Epilepsy (ILAE) [21], no history of prior therapeutic neurosurgery (e.g., resection, laser ablation, neuromodulation device), no history of stroke, and English as primary language. Exclusion criteria included a diagnosis of dementia or a primary neurological or psychiatric condition, including stroke, Parkinson’s disease, and history of psychosis requiring hospitalization.

2.2. Clinical and sociodemographic data

Demographic (e.g., age, sex, race, ethnicity) and epilepsy clinical history were obtained from patient self-report and electronic health records (EHR).

2.3. Neuropsychological and self-report measures

All patients completed a comprehensive neuropsychological assessment that included the Montreal Cognitive Assessment (MoCA, English 7.1), the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog), the Test of Practical Judgement (TOP-J), and standard neuropsychology measures of memory, language, executive function, and processing speed. Measures of memory included the Rey Auditory Verbal Learning Test Long Term Percent Retention (RAVLT-LTPR); Wechsler Memory Scale 4th Edition (WMS-4) Logical Memory Story B delayed recall (LMII); and the WMS-4 Visual Reproduction (VR) delayed recall (VRII). Measures of language included visual confrontation naming with the Multilingual Naming Test (MINT), auditory naming with the Auditory Naming Test (ANT), and semantic fluency with Animal Fluency. Measures of processing speed included the Trail-Making Test Condition A (TMT-A) and the Cancellation subtest from ADAS-Cog. Measures of executive function included letter fluency and TMT Condition B (TMT-B). MoCA Total score was corrected for education by adding one point to the total score for patients with less than 12 years of education (i.e., less than a high school diploma) [22]. Scores for the MoCA and TOP-J are reported in raw scores and total number of errors are reported for the ADAS-Cog. Raw scores for all other tests were corrected for age, education, sex, and race when appropriate based on published norms or tests manuals. Cancellation raw scores were converted into z-scores based on data from a sample of 370 cognitively normal older adults from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). All neuropsychological scores were converted into T-scores for ease of interpretability, and composite scores were created for each cognitive domain (i.e., learning, memory, language, processing speed, and executive function) by averaging the T-scores across individual tests within each domain.

2.4. Individual-level social determinants of health

Individual-level SDOH were collected from self-report and EHR during the baseline study visit and included patient and early-life environmental factors (Box 1). Patient factors included personal income, household income, education, marital status (married/cohabiting versus divorced/never married), and insurance coverage (Medicare/Medicaid/public health insurance versus private health insurance). Early-life environmental factors included parental education (i.e., father and mother’s education), parental occupation complexity (i.e., father and mother’s occupation), patient history of childhood employment (i. e., employment before 18 years old), and family class status during childhood (i.e., low/working, middle, upper). Occupation complexity was categorized into high (managerial and professional specialty, technical, sales, and administrative support) or low (service, precision production, repair, operators, fabricators, laborers, homemakers) occupational complexities based on the Dictionary of Occupational Titles [23] and categories previously used in epilepsy [16]. Personal and household incomes were divided into low and high based on each state’s (i.e., California, Ohio, Wisconsin) median individual and household income. Word reading with the American New Adult Reading Test (ANART) was used as a proxy for quality of education [24].

Box 1. Terms and definitions.

Social determinants of health (SDOH):

Refers to the conditions in which people are born, grow, live, work, and age, which impacts a wide range of health outcomes and risks. These determinants are shaped by the distribution of resources across societies and are influenced by economic and social policies and cultural and social norms.

Adverse SDOH:

Are the negative conditions or factors in a person’s social, economic, and physical environment that can negatively impact health outcomes and quality of life. Some examples of adverse SDOH include lower levels of educational or occupational attainment, lack of health insurance, low income, food insecurity, housing instability, low health literacy, discrimination or systemic racism, and lack of social support.

Individual-level SDOH:

Refers to personal factors and circumstances that influence a person’s health and well-being, including education, income, occupation, health behaviors, social support, housing, and access to healthcare.

Early-life environmental factors:

Refers to the conditions and influences an individual is exposed to early in childhood (i.e., conception through childhood) that can significantly impact development, health outcomes, and aging trajectories. Examples include maternal health, parental education and socioeconomic status, nutrition, home environment, and stimulation and enrichment.

Community-level SDOH:

Refers to the social, political, economic, and environmental conditions in which individuals and groups live, affecting health outcomes. These determinants influence the health and well-being of entire communities and can contribute to health disparities. Examples include neighborhood and built environment, economic and educational opportunities, access to healthcare, and public safety.

Neighborhood deprivation:

Refers to the relative socioeconomic conditions of communities or neighborhoods, with greater neighborhood disadvantage indicating a lack of resources, services, opportunities, and conditions necessary to support its residents’ well-being (including health outcomes).

2.5. Community-level SDOH: Area deprivation index

The area deprivation index (ADI) is a widely-used, validated measure of neighborhood-level socioeconomic disadvantage that is based on several SDOH factors including poverty, employment, housing, and education opportunities [25,26]. ADI data were obtained from the University of Wisconsin School of Medicine and Public Health Neighborhood Atlas. Patients’ addresses were geocoded and assigned an ADI decile and a national percentile according to their residential Census block for their state. ADI decile values range from 1 to 10 with higher ADI values indicating greater neighborhood disadvantage. Given that participants were recruited across three different states (i.e., California, Ohio, and Wisconsin), we used the ADI deciles as we were interested in exposure to neighborhood deprivation for each participant relative to their state. Based on previous studies that used binary characterization of ADI, [27–29] we created a binary (yes/no) ADI Disadvantage Score (Declines 9 or 10) to capture patients living in the most disadvantaged neighborhoods.

2.6. Application of IC-CoDE

The IC-CoDE was applied to classify patients into cognitively impaired and intact groups. The IC-CoDE taxonomy classifies patients into cognitive phenotypes based on the number of impaired domains with Single-Domain defined as impairment in one cognitive domain, Bi-Domain defined as impairment in two domains, Generalized defined as impairment in three or more domains, and Intact defined as no impairment in any cognitive domain [20]. Although the IC-CoDE consists of five cognitive domains including memory, language, attention/processing speed, executive function, and visuospatial abilities, given the lack of availability of visuospatial tests in our battery, we used the four-domain classification taxonomy that includes all the above domains but visuospatial. The four-domain IC-CoDE classification has been validated in a large sample of patients with temporal lobe epilepsy and the rates are reported in McDonald et al. [20]. We used < 1 standard deviation below the normative mean as the impairment cutoff as this cutoff maximizes sensitivity and specificity in older cohorts, and we have previously used this cutoff in older adults with focal epilepsy [16,30]. To be impaired in a domain, at least two tests per domain must be impaired. Given the sample size, patients with either Single, Bi-Domain, or Generalized impairments were combined into one group (i.e., IC-CoDE Impaired), whereas those with no impaired domains were classified as IC-CoDE Intact.

2.7. Statistical analyses

All statistical analyses were conducted using IBM SPSS Version 29.02 and figures were created on R version 4.1.2. Descriptive statistics were conducted to characterize the overall sample. To reduce the number of variables included in the model, we conducted stepwise regressions to examine the differential contribution of individual-level SDOH to cognitive performance after controlling for important epilepsy clinical factors [age of onset, number of antiseizure medications (ASM), lateralization, localization]. The following SDOH were entered into the stepwise regression model: insurance, personal and household income, father’s and mother’s education, father’s and mother’s occupation complexity, childhood employment, and family class status during childhood. Given that most cognitive measures were corrected for years of education, word reading was included in the model as a proxy for quality of education [24]. Power analyses revealed that to achieve power > 0.80 with 14 predictors, a sample of 55 was sufficient for a medium effect size (f² = 0.15) and a sample of 27 was sufficient for a for a large effect size (f² = 0.35). Spearman rho correlations (95 % confidence intervals method proposed by Fieller, Hartley, and Pearson [31]) were conducted to examine the relationship between ADI deciles and composite cognitive T-scores. Independent t-test, Fisher’s Exact, and Fisher–Freeman–Halton tests were used to test for differences in clinical, sociodemographic, and SDOH variables between the IC-CoDE Impaired and Intact groups. A stepwise discriminant function analysis (DFA) was conducted to determine which clinical, demographic, and SDOH predictors can correctly identify patients with cognitive impairment (IC-CoDE Impaired vs Intact). A SDOH composite score was calculated to reflect the number of adverse SDOH factors and included the following relevant SDOH variables: patient education 12 years or less, public health insurance, divorce/never married, personal and family income less than the median, father’s and mother’s education 12 years or less, low father’s and mother’s occupation complexity, lower or working class as a child, history of childhood employment, and ADI Disadvantage Score (Decile 9 or 10). All results are reported at an alpha of 0.05 and with a Benjamini-Hochberg false discovery rate (FDR) correction. Prior to conducting the statistical analyses, Shapiro-Wilk Tests were performed, indicating normal distribution for all cognitive tests T-scores across the entire sample except for the MINT and ANT; however, visual inspection of Normal Q-Q Plots suggested normal distribution without any significant outliers.

2.8. Data availability

Authors have full access to all study data and participant consent forms and take full responsibility for the data, the conduct of the research, the analyses and interpretation of the data, and the right to publish all data. The data supporting the findings of this study are available on request from the senior author (C.R.M.).

3. Results

3.1. Participants

Fifty-seven older adults with focal epilepsy were included in the study (Table 1). The cohort was mostly non-Hispanic White, mostly female, had on average 15 years of education, and were mostly married/cohabiting. Clinically, the cohort had an earlier age of epilepsy onset, mostly temporal and left-lateralized seizure focus, and more than half were taking two or more ASM. Over half of the cohort had Medicare or public health insurance, had a household income less than the median, had fathers with 12 years of education or less, and reported a lower- or working-class background during childhood. Approximately 70 % of the patients who reported a history of childhood employment also reported a lower/working class status. The majority of the cohort had mothers with 12 years of education or less and with lower occupational attainment. On average, the cohort had an ADI of 4.47 (range = 1–10), with 8.8 % having an ADI Disadvantage Score.

Table 1.

Sociodemographic and clinical variables of the cohort.

N	57
Site: CCF | UCSD | UWM	27 (47.4 %) | 14 (24.6 %) | 16 (28.1 %)
Age: Years	66.33 (6.65)
Education: Years	15.02 (2.63)
Sex: Female (%)	30 (52.6 %)
Race: White	49 (86 %)
Ethnicity: Hispanic (%)	2 (2.1 %)
Marital status: Married (%)	41 (74.5 %)
Number of SDOH+	4.56 (2.19)
Primary Insurance: Medicare/Public (%)	31 (54.4 %)
Personal income < Median	15 (31.3 %)
Household income < Median	22 (51.2 %)
Father’s education: 12 years or less	25 (54.3 %)
Mother’s education: 12 years or less	31 (66 %)
Father’s occupation: Low	21 (42 %)
Mother’s occupation: Low	30 (60 %)
Childhood class status: Lower or working class	28 (50.9 %)
Worked as a child	23 (41.8 %)
Area deprivation index	4.47 (2.72)
ADI Disadvantage Score (%)	5 (8.8 %)
Age of onset: Years	42.75 (21.13)
Late onset (>55 years)	24 (42.1 %)
Duration: Years	23.58 (19.29)
Number of ASM	1.80 (0.923)
Side
Left	22 (38.6 %)
Right	9 (15.8 %)
Bilateral	13 (22.8 %)
Unknown	13 (22.8 %)
Localization
Frontal	3 (5.3 %)
Temporal	28 (49.1 %)
Frontotemporal	14 (24.6 %)
Unknown	12 (21.1 %)

CCF: Cleveland Clinic Foundation; UCSD: University of California, San Diego; UWM: University of Wisconsin, Madison; ASM: antiseizure medication.

⁺SDOH Composite: Patient education 12 years or less, public health insurance, divorce/never married, personal and family income less than the median, father’s and mother’s education 12 years or less, low father’s and mother’s occupation complexity, lower or working class as a child, history of childhood employment, and ADI Disadvantage Score (Decline 9 or 10).

3.2. Independent contribution of individual-level SDOH to cognitive performance

Table 2 and Fig. 1 include individual-level SDOH factors that were significant predictors across cognitive measures after controlling for important clinical variables. For the MoCA [F (5, 24) = 3.74, p = 0.012), better word reading performance was associated higher MoCA Total scores (β = 0.148, p = 0.002). Although the overall model was not significant [F (6, 23) = 1.89, p = 0.126), higher father’s occupation attainment (β = −2.99, p = 0.011) and private health insurance (β = −1.99, p = 0.049) were associated with fewer errors on the ADAS-Cog. Higher Learning Composite T-scores [F (8, 21) = 4.34, p = 0.003) were associated with private health insurance (β = 8.45, p = 0.002), no history of childhood employment (β = 7.08, p = 0.008), and higher mother’s occupation complexity (β = 6.02, p = 0.034). Higher Memory Composite T-scores [F (6, 23) = 3.40, p = 0.015) were associated with higher mother’s education (β = 1.49, p = 0.003) and private health insurance (β = 4.39, p = 0.049). Higher Processing Speed T-scores [F (2, 23) = 3.57, p = 0.012) were associated with better word reading performance (β = 0.402, p = 0.004) and private health insurance (β = 6.44, p = 0.033). Higher Executive Function T-scores [F (7, 22) = 4.87, p = 0.002) were associated with private health insurance (β = 11.04, p < 0.001), no history of childhood employment (β = 8.04, p = 0.017), and better word reading performance (β = 0.321, p = 0.017).

Table 2.

Stepwise regression examining contribution of individual-level SDOH to cognitive performance.

	Predictors	R2 change	F Change	p-value
MoCA	Covariates	0.141	1.03	0.414
	Word reading	0.297	12.68	0.002
ADAS-Cog	Covariates	0.015	0.094	0.984
	Father’s occupation (Low vs High)	0.189	5.69	0.025
	Insurance (Public vs Private)	0.126	4.33	0.049
Learning	Covariates	0.093	0.641	0.638
	Mother’s education (Years)	0.164	5.31	0.030
	Insurance (Public vs Private)	0.144	5.56	0.027
	Childhood employment	0.129	6.04	0.022
	Mother’s occupation (Low vs High)	0.093	5.16	0.034
Memory	Covariates	0.155	1.15	0.357
	Mother’s education (Years)	0.215	8.20	0.009
	Insurance (Public vs Private)	0.100	4.33	0.049
Language	Covariates	0.514	6.67	<0.001
Processing Speed	Covariates	0.100	0.697	0.601
	Word reading	0.266	10.08	0.004
	Insurance	0.116	5.12	0.033
Executive Function	Covariates	0.074	0.498	0.738
	Insurance	0.306	11.84	0.002
	Childhood employment	0.110	4.93	0.036
	Word reading	0.118	6.63	0.017

Covariates: Duration of epilepsy, number of antiseizure medications, side of seizure onset, and localization

MoCA: Montreal Cognitive Assessment; ADAS-Cog: Alzheimer’s Disease Assessment Scale-Cognitive Subscale.

Fig. 1.

Variance explained by clinical and individual-level social determinants of health factors across cognitive measures A) MoCA Total Score, B) Number of errors on the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog), C) Learning Composite, D) Memory Composite, E) Processing Speed Composite, and F) Executive Function Composite. Word reading is a proxy for quality of education. Clinical covariates include duration of epilepsy, number of antiseizure medications, side of seizure onset, and localization.

3.3. Relationship between the area deprivation index and cognitive measures

There were no differences in ADI values across sites [F (2, 53) = 0.169, p = 0.845; averages: CCF = 4.61, SD = 2.64, range = 1–9, UCSD = 4.07, SD = 2.56, range = 1–10; UWM = 4.38, SD = 3.00, range = 1–9]. Fig. 2 shows correlations between cognitive measures and ADI deciles. Higher ADI deciles were associated with lower MoCA Total Score, TOP-J, Learning, Memory, and Executive Function composite T-scores, and with more errors on the ADAS-Cog.

Fig. 2.

Associations between area deprivation index (ADI) and cognitive measures: A) Montreal Cognitive Assessment (MoCA) Total Score, B) Number of errors on the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog, C) Performance on the Test of Practical Judgement, D) Language Composite, E) Learning Composite, F) Memory Composite, G) Processing Speed Composite, and F) Executive Function Composite. Comparisons corrected with an FDR of 0.0375.

3.4. Social determinants of health in cognitively impaired older adults

Of the 57 participants, 54 had comprehensive cognitive data to apply the IC-CoDE taxonomy (Table 3). Approximately 33 % of the cohort had a cognitively impaired profile (i.e., IC-CoDE Impaired). These patients were more likely to have a higher SDOH composite score, a household income less than the median, lower father’s education, and higher ADI deciles. They were also taking on average more ASM. The DFA revealed that mother’s education and household income correctly classified 74 % of the cases (Wilks’s Λ = 0.575, p = 0.001).

Table 3.

Demographic, clinical, and SDOH factors between IC-CoDE Impaired and Intact Groups.

	IC-CoDE Impaired	IC-CoDE Intact	Statistics	p-value
N	18 (33.3 %)	36 (66.7 %)
Age: years	66.50 (7.14)	65.50 (6.08)	0.650	0.519
Sex: Female (%)	9 (50 %)	19 (52.8 %)	– -	1.00
Race: White	15 (83.3 %)	33 (88.9 %)	– -	0.674
Ethnicity: Hispanic (%)	0 (0 %)	2 (5.9 %)	– -	1.00
Age of onset: Years	42.06 (20.9)	43.97 (20.6)	0.321	0.749
Duration: Years	23.44 (18.6)	22.8 (19.1)	0.117	0.908
Number of ASM	2.12 (0.928)	1.58 (0.806)	2.146	0.037
Education: years	14.72 (2.61)	15.08 (2.72)	0.466	0.643
Word reading (Quality of education)	49.44 (12.9)	54.08 (13.2)	1.229	0.225
Number of SDOH +	5.78 (2.16)	3.81 (1.94)	3.394	0.001
Insurance: Medicare/Medicaid/Public (%)	12 (66.7 %)	18 (50 %)	– -	0.384
Marital status: Divorced/Never married (%)	5 (27.8 %)	8 (23.5 %)	– -	0.747
Personal income < Median	6 (46.2 %)	7 (21.2 %)	– -	0.145
Household income < Median	11 (78.6 %)	10 (35.7 %)	– -	0.020
Father’s education: 12 years or less	11 (78.6 %)	12 (41.4 %)	– -	0.027
Mother’s education: 12 years or less	12 (80 %)	16 (55.2 %)	– -	0.185
Father’s occupation: Low	8 (50 %)	11 (35.5 %)	– -	0.366
Mother’s occupation: Low	12 (70.6 %)	15 (50 %)	– -	0.226
Childhood class status: Lower or working class	12 (66.7 %)	14 (41.2 %)	– -	0.144
Childhood employment	7 (38.9 %)	15 (44.1 %)	– -	0.775
Area deprivation index	5.56 (3.05)	3.94 (2.43)	2.108	0.040

CC: Cleveland Clinic; UCSD: University of California, San Diego; UWM: University of Wisconsin, Madison; SDOH: social determinants of health.

Bold: significant at alpha 0.05.

⁺SDOH Composite: Patient education 12 years or less, public health insurance, divorce/never married, personal and family income less than the median, father’s and mother’s education 12 years or less, low father’s and mother’s occupation complexity, lower or working class as a child, history of childhood employment, and Area deprivation Index Disadvantage Score (ADI of 9 or 10).

4. Discussion

Given the increasing number of older adults living with epilepsy, alongside an aging global population, there is a need to identify factors associated with cognitive decline in this population. We add to this emerging literature by demonstrating the role of individual and community-level SDOH in the cognitive profiles of older adults with epilepsy. First, we show that a large proportion of the cohort had several SDOH factors that have been previously associated with cognitive decline and dementia, including ADRD. Second, we demonstrate that several individual-level SDOH factors and neighborhood disadvantage were associated with lower scores on measures of global cognition, learning and memory, executive function, and processing speed. Specifically, measures of global cognition (e.g., MoCA), learning, memory, and executive function were associated with both individual-level SDOH and neighborhood disadvantage, whereas processing speed was only associated with individual-level SDOH. Lastly, we show that older adults who meet criteria for cognitive impairment based on a validated diagnostic taxonomy for epilepsy show a greater number of SDOH factors previously shown to be associated with adverse health outcomes.

4.1. Individual-level SDOH

There is a compelling literature linking SDOH with disparities in neurological disorders and cognitive health [32–34]. In a study of young-to-middle aged adults with epilepsy, Hohmann et al. [15] demonstrated that individual-level SDOH, including unemployment, lower income, and lower education, were associated with greater cognitive impairments in the areas of verbal learning and psychomotor speed and increased levels of mental distress. We extend this literature by demonstrating that individual-level SDOH (e.g., insurance coverage, word reading), including early-life environment factors (e.g., parental education and occupation), are associated with cognitive performance in older adults with epilepsy, explaining 33 %-62 % of the variance in cognitive scores after controlling for epilepsy clinical factors.

Insurance coverage was the most common predictor of poor cognitive performance in our sample, with Medicare/Medicaid and other public health insurance associated with worse performance on a global cognitive measure and on tests of verbal learning, memory, executive function, and processing speed. Studies have shown that individuals with epilepsy who are either uninsured or have Medicare/Medicaid have gaps in epilepsy care, including access to diagnostic tools and treatments [10,11]. This gap in access to care is concerning given that older adults with chronic medical conditions, such as epilepsy, require comprehensive management and interdisciplinary care in order to maximize health outcomes, particularly cognitive health. Whether insurance coverage is capturing other factors not included in our study that may better explain these cognitive outcomes, our findings highlight the critical need to create programs to educate Medicare/Medicaid beneficiaries with epilepsy about brain and cognitive health.

Despite most of our measures being corrected for years of education, we found that word reading, a commonly used proxy for quality of education, was associated with worse performance on the MoCA (global cognitive measure) and on measures of processing speed and executive function. Given that quality of education differs significantly across ethnoracial and socioeconomic statuses [35], years of education do not fully capture the educational experiences impacting cognitive development and late-life cognitive decline. Our findings highlight the importance of including proxies for quality of education, such as word reading or reading level, in studies examining cognitive outcomes in epilepsy.

4.1.1. Early-life environmental factors and cognitive outcomes

Given the vast literature demonstrating the impact of early-life environmental factors (e.g., parental education) on late-life cognition and risk of dementia [36–38], we were interested in examining the relationship between cognition and early-life factors. Mother’s education was a significant predictor of performance on measures of learning and memory, whereas father’s occupation was a significant predictor of a global cognitive measure (i.e., ADAS-Cog). Large population studies have revealed a direct relationship between parental education and occupation and late-life cognition [37,39], with several proposed explanations for this relationship, including the effect of stimulating and enriched environments on cognitive development, quality of parent–child interactions, access to higher educational and health promoting resources, and the impact of parental education on SES. In a group of young-to-middle-aged adults with temporal lobe epilepsy, we previously demonstrated that patients with a worse cognitive profile had lower parental education [40]. Oyegbile-Chidi et al. [14] showed that lower parental education (as part of a sociodemographic disadvantage composite) was associated with poorer cognitive and academic trajectories in children with newly diagnosed epilepsy. Here, we also found a relationship between a history of childhood employment and lower scores on measures of verbal learning and executive function. Given that the majority (70 %) of our patients who reported a history of childhood employment also reported a low/working-class family status, childhood employment in our cohort may reflect childhood SES. Several mechanisms have been proposed to explain the effects of childhood SES on late-life cognition, including risk mechanisms associated with low SES, such as exposure to environmental toxins and material deprivation, and protective mechanisms associated with high SES, such as better educational opportunities, social networks, and financial resources [38,41]. However, longitudinal studies with large samples of older adults with epilepsy and comprehensive SDOH data are needed to better understand the pathways through which early-life environmental factors impact late-life cognition in this population.

4.1.2. Community-level SDOH

There has been an increased interest in examining the role of neighborhood-level indicators of economic deprivation in disparities in neurological disorders and cognitive outcomes [19,42,43]. In epilepsy, measures of neighborhood deprivation have been associated with a higher incidence of epilepsy, status epilepticus, disparities in access to specialized care, intellectual disability, poorer quality of life [5,44,45] and adverse cognitive outcomes in children and young-to-middle-age adults with epilepsy. [12,13,15,17] We provide further evidence on the role of neighborhood deprivation on cognitive outcomes by showing that greater neighborhood disadvantage is associated with poorer performance on global cognitive measures, including the MoCA, and on measures of learning, memory, and executive function in older adults with epilepsy. Given that epilepsy is a chronic medical condition with onset throughout the life-course, longitudinal studies examining neighborhood deprivation at multiple life-course stages can elucidate the causal relationships between deprivation and epilepsy and its impact on health outcomes. Notably, there was lack of variability in ADI deciles across our cohort (only 8.8 % of the cohort lived within the most disadvantaged neighborhoods). Given that participants in our study were recruited from tertiary care centers, recruitment of community-based cohorts may better capture variability in neighborhood composition, including the most disadvantaged neighborhoods. Nonetheless, this literature can inform epilepsy-related policy aimed at developing system-level interventions focused on increasing access to epilepsy care, education, and resources within the most at-risk communities. Examples of community-level interventions that can address SDOH include food banks and pantries, housing programs, income enhancements and supplements, mobile health clinics, urban green spaces, local health-promoting programs, and opportunities for social engagement among older adults. [46–48].

4.1.3. SDOH Profiles in Cognitively Impaired Older Adults

We also examined SDOH factors in patients with cognitively impaired profiles based on a diagnostic taxonomy developed and validated for epilepsy (i.e., IC-CoDE). Patients with cognitive impairment had a higher number of SDOH factors previously associated with adverse cognitive outcomes, including lower household incomes, lower father’s education, higher ADI values, and were taking more ASM. We extend our previous work, where we showed that lower educational and occupational attainment was associated with a cognitively impaired profile in a community-based cohort of older adults with late-onset epilepsy [16], by demonstrating that early-life environmental factors (lower father’s education) and income are also associated with a cognitively impaired profile. Notably, the association between income and cognitive impairment suggests that older adults with epilepsy may benefit from supplemental income programs, particularly given the research demonstrating that supplemental income programs improve well-being and health, including cognitive health, in older adults.[49] Older adults with epilepsy can also benefit from programs that promote brain health, including cognitive training, physical activity, and social engagement programs and programs that aim to improve health behaviors surrounding sleep hygiene, nutrition, and mood. Further research is needed to better understand the social risks and needs of older adults with epilepsy, including food insecurity, transportation needs, house instability, and medication affordability, among others. To facilitate these clinical and research efforts, epilepsy clinics can integrate screening tools for social risk and needs, such as the ICD-10-CM Z-code health-related social needs or the Epic SDOH Toolkit, both clinical tools that can be used to track a patient’s social needs and risks over time, enabling providers to refer patients to local community resources, which may improve access to these services.

5. Limitations

Despite the novelty of our findings, there are limitations to the current study that we plan to address in the future. These include a modest sample size, recruitment of participants from tertiary care centers, limited diversity in ethnoracial representation, lack of longitudinal data, and potential recall biases in patient self-report. Importantly, there are many confounding factors associated with early versus late-life exposures not captured in our study that may impact cognitive trajectories and require further investigation. Larger samples will allow for the examination of ADI using quintiles [13] or other relevant categorizations and to detect more subtle associations between cognitive outcomes and different SDOH measures (e.g., processing speed and neighborhood disadvantage). Studies that include community-based cohorts of diverse patients with epilepsy who receive their care within their communities are needed, given that these populations are significantly impacted by a lack of access to epilepsy care. Further, community-based cohorts are more representative of the diverse U.S. populations which will improve the generalizability of our findings. Longitudinal data will elucidate the mechanistic pathways through which SDOH impacts epilepsy, cognitive, and overall health outcomes and, in older adults, will help identify factors associated with the risk of developing dementia. Importantly, longitudinal data can address the challenge of causality in cognitive decline (i.e., social causation versus social selection, also known as social drift) [50]. Specifically, studies can examine whether SDOH (e.g., SES, education) increases the risk of cognitive decline (social causation) or whether having epilepsy results in individuals experiencing a downward drift in SES leading to exposures to risk factors known to increase the risk of cognitive decline (social selection/drift). Further, longitudinal studies that include participants with new and recent onset epilepsy can track the long-term impact of seizures on social outcomes (e.g., employment, income, education) and how these outcomes then impact cognitive trajectories. We also did not have information on where patients lived as children to examine the role of life-course residential disadvantage, which has been shown to be associated with adverse outcomes [51]. This is particularly important for older adults with childhood (i.e., age 1–14) onset epilepsy (12.3 % of our cohort) who may harbor decades of community-level risk factors for accelerated brain and cognitive aging. Lastly, although our assessment of SDOH was based on the literature, we did not use a standardized SDOH screening tool. Inclusion of standardized screening tools, such as the PRAPARE, [52] which assesses patients’ assets, risks, and experiences, into epilepsy studies will be useful in systematically collecting data on this population.

6. Conclusions

Although the role of SDOH in cognition has been investigated in the general population and to some extent in other neurological disorders, just recently has there been an interest in examining the impact of SDOH on cognitive outcomes in epilepsy. Importantly, no prior studies have focused on older adults and as such we add to this important emerging literature by demonstrating the role of individual- and community-level SDOH in cognitive outcomes in older adults with epilepsy. The recently published SDOH framework from the National Institute of Neurological Disorders and Stroke (NINDS) [34] provides an opportunity to guide research questions in this area and inform the collection of rich SDOH data across the lifespan to better understand the social forces that impact patient-centered outcomes in epilepsy. Notably, given the impact of policies (i.e., economic, health, social) on the care of patients with epilepsy, there is a need to understand the structural determinants associated with the documented health inequities in individuals living with epilepsy.