Disparate dementia risk factors associate with cognitive impairment and rates of decline in African Americans

Abstract

Objective:

Evaluate the frequency of modifiable dementia risk factors and association with cognitive impairment and rate of decline in diverse participants engaged in studies of memory and aging.

Methods:

Modifiable dementia risk factors and their associations with cognitive impairment and cognitive decline were determined in community-dwelling African American (AA, n=261) and non-Hispanic White participants (nHW, n=193) who completed ≥2 visits at the Mayo Clinic Alzheimer Disease Research Center in Jacksonville, Florida. Risk factors and their associations with cognitive impairment (global Clinical Dementia Rating^® [CDR] ≥0.5) and rates of decline (CDR Sum-of-Boxes) in impaired participants were compared in AA and nHW participants, controlling for demographics, APOE ε4 status, and Area Deprivation Index.

Results:

Hypertension, hypercholesterolemia, obesity, and diabetes were overrepresented in AA participants, but were not associated with cognitive impairment. Depression was associated with increased odds of cognitive impairment in AA (OR: 4.30; 95%CI: 2.13-8.67) and nHW participants (OR: 2.79; 95%CI: 1.21-6.44) but uniquely associated with faster decline in AA participants (β: 1.71; 95%CI: 0.69-2.73; p=0.001). Fewer AA participants reported antidepressant use (9/49, 18%) than nHW counterparts (57/78, 73%; p<0.001). Vitamin B12 deficiency was also associated with an increased rate of cognitive decline in AA (β: 2.65; 95%CI: 0.38-4.91; p=0.023).

Interpretation:

Modifiable dementia risk factors are common in AA and nHW participants, representing important risk mitigation targets. Depression was associated with dementia in AA and nHW participants, and with accelerated declines in cognitive function in AA participants. Optimizing depression screening and treatment may improve cognitive trajectories and outcomes in AA participants.

Introduction

Advancing age is the most important non-modifiable risk factor for dementia. In the United States, 73 million people (~20% of the population) are projected to be 65-years or older by 2030.¹ Accordingly, shifts in population demographics are expected to increase the number of Americans living with Alzheimer disease (AD) from 6.5 in 2022 to 11.2 million in 2040.² The upward trajectory coupled with the practical and financial burden imposed by AD and related causes of dementia emphasizes the need for population-wide strategies to mitigate modifiable dementia risk factors.^3–5 This need is particularly urgent amongst Black/African Americans (AA) who account for 12.4% of the United States population (12% of Americans ≥65-years⁶), yet 19% of patients with dementia,^{2, 4, 7–9} and experience a higher incidence of dementia compared to other racial and ethnic groups.¹⁰ Inequalities in structural (e.g., social processes, policy, systems) and social (e.g., physical, social environment) determinants of health contribute to disproportionate dementia risk in AA individuals,¹¹ and include disparities in the quantity and quality of education, health care, and physical health.¹² Left unchecked, disparities in dementia burden are likely to increase in lockstep with an aging AA population.¹³

Systematic reviews and meta-analyses suggest that up to 40% of dementia worldwide is attributable to “modifiable risk factors”, including low early-life educational attainment (<10 years); midlife hearing loss, traumatic brain injury, hypertension, and obesity; and later-life smoking, depression, social isolation, physical inactivity, air pollution, and diabetes.^{3, 14} Several of these risk factors are overrepresented in AA individuals (e.g., hypertension, diabetes, obesity, smoking and air pollution),^{4, 5} raising the possibility that these factors may contribute to disparate dementia risk. If true, these would present efficient targets for risk reduction strategies—coupled with efforts to improve disparities in structural and social determinants of health (e.g., socioeconomic status, education, health literacy and access to care)—that may improve brain health, modify cognitive trajectories, reduce economic costs, and prevent or minimize the burden of AD and related causes of dementia in AA communities. Recognizing this, we evaluated the frequency of dementia risk factors in community-dwelling AA and non-Hispanic White (nHW) participants enrolled within the Mayo Clinic Alzheimer Disease Research Center (ADRC) and assessed the association between modifiable risk factors and cognitive impairment, and the rate of functional decline in cognitively impaired individuals.

Methods

Protocol approvals and patient consent

Community-dwelling older adults were recruited from the Jacksonville metro area (2008-2022) and longitudinally assessed via the Mayo Clinic ADRC in Jacksonville, Florida. Participants or their legally appointed representatives provided written informed consent. Study procedures were approved by the Mayo Clinic Institutional Review Board.

Clinical Research Assessments

Participants completed core ADRC procedures (National Alzheimer’s Coordinating Center Unified Data Set, version 3.0¹⁵), including annual clinical assessments, detailed neurological examination performed by expert clinicians (neurologists or psychiatrists), and neuropsychological testing performed by trained psychometricians. Race and ethnicity were self-reported at study entry. Medical history was recorded at study entry and updated annually, including past medical history, medication use, and past/present depression. Weight and height were measured at each visit and body mass index calculated (BMI = weight in kg / height in m²). When multiple assessments were completed, variables measured at the last clinical assessment were used in the analyses. Cognitively impaired participants were identified by study clinicians at the conclusion of each assessment referencing National Institute on Aging–Alzheimer’s Association Work Group recommendations.^16–20 Cognitive status (impaired or unimpaired) and specific clinical diagnoses were reviewed and ratified at monthly consensus conferences, integrating clinical and neuropsychological assessments by behavioral neurologists and neuropsychologists. Etiologic diagnoses were assigned referencing established diagnostic criteria.^16–20 The severity of cognitive impairment was staged using the global Clinical Dementia Rating^® (CDR), assigned by qualified clinicians in accordance with established scoring rules, where CDR 0 identifies cognitively normal participants and CDR ≥0.5 identifies participants with cognitive impairment.²¹ The CDR has established high inter-rater reliability and stability across time in ADRCs.²² Annualized change in CDR Sum-of-Box (CDR-SB) scores were used as the primary measure of cognitive decline, similar to industry-standard clinical trials.²³

Data from AA and nHW participants with two or more research visits were included in this nested cohort study. Participants with fewer than two visits were excluded given this study’s focus on rates of cognitive decline. Demographic information, cognitive status, and the presence or absence of potentially modifiable dementia risk factors were prospectively reported by the participant or coparticipant at each visit, including hypertension, hypercholesterolemia, obesity, hearing loss, smoking (>100 cigarettes lifetime), depression (15-item Geriatric Depression Scale score ≥5), diabetes, traumatic brain injury, thyroid disease, vitamin B12 deficiency, heart disease, and stroke/transient ischemic attack. Hypertension was deemed present in participants who reported active hypertension or current use of anti-hypertensive medications, or in whom elevated blood pressure was documented at the research visit (systolic blood pressure >140 mmHg or diastolic blood pressure >90 mmHg). Hypercholesterolemia was deemed present in participants who reported active hypercholesterolemia or current use of cholesterol/lipid-lowering medications. Diabetes was defined by self-reported history alone as injectable medications (including insulin) were inconsistently recorded. The accuracy of self-reported history of diabetes has been established in similar ADRC cohorts.²⁴ Depression was defined in accordance with the 15-item Geriatric Depression Scale—a well validated depression scale for use in older individuals, where a cutoff score of ≥5 suggests the presence of depression,²⁵ or the active use of antidepressant medications. Area Deprivation Index (ADI) was obtained according to zip code for each participant. The ADI is a public, United States Census-based geographic disadvantage metric including years of education, housing-quality, poverty/income and employment domain measures that allow for characterization and ranking of neighborhood socioeconomic conditions.²⁶ For this study, ADI was used as a proxy of social-economic-environmental disadvantage at the neighborhood, census level (higher numbers reflect greater deprivation). APOE ε4 genotype, a non-modifiable dementia risk factor, was measured in blood using Taqman genotyping assays and incorporated in analyses.²⁷

Community recruitment strategies were supported by the Mayo Clinic Florida ADRC Outreach, Recruitment and Engagement (ORE) core, and included educational talks during health fairs, church groups, and other activities (e.g., health screenings) suggested by Community Ambassadors. As previously reported, ADRC datasets may present selection-related factors as potential limitations.^{28, 29} This study is no exception. AA participants were enrolled regardless of cognitive status, whereas nHW participants were predominantly recruited from behavioral neurology clinics, favoring participants with cognitive impairment. Systematic biases in recruitment necessitated stratification by race for all analyses.

Statistical analyses

The frequency of modifiable dementia risk factors was compared in AA and nHW participants with and without cognitive impairment. The relationship between modifiable risk factors and rates of cognitive decline were also considered in participants with cognitive impairment, where cognitive decline corresponded to the annualized rate of change in the CDR-SB scores, derived by dividing the final CDR-SB by years from onset of cognitive impairment. Reported use of medications to treat hypertension, hypercholesterolemia, and depression were compared between AA and nHW participants as a proxy measure of disease treatment.

Continuous variables were summarized as mean (SD). Categorical variables were reported as frequency (percentage). Demographics and risk factors at first and last visit were compared between AA and nHW participants using a Wilcoxon rank sum test (continuous and ordinal variables) or Fisher’s exact test (categorical variables). Associations between cognitive impairment (dependent variable) and potentially reversible risk factors (independent variables) were assessed using multivariable logistic regression, including modifiable dementia risk factors (i.e., hypertension, hypercholesterolemia, BMI, obesity, hearing loss, smoking, depression, diabetes, traumatic brain injury, thyroid disease, vitamin B12 deficiency, heart disease, and stroke/transient ischemic attack). Multivariable linear regression was used to determine the relationships between modifiable risk factors and rates of cognitive decline. Multivariable analyses were adjusted for age at first visit, sex, years of education, ADI, and APOE ε4 genotype. Forest plots were created for visual comparisons of adjusted ORs across groups.

All tests were two-sided with p-value <0.05 considered statistically significant. Statistical analyses were performed using R Statistical Software (version 4.1.2; R Foundation for Statistical Computing, Vienna, Austria).

Results

Demographics and Modifiable Risk Factors

Three-hundred-thirteen AA and 253 nHW participants met inclusion criteria. Fifty-two AA and 60 nHW participants completed only 1 visit and were excluded, leaving 261 AA and 193 nHW participants for analyses. Age, sex, education, and ADI were similar in included vs excluded participants. Unique to AA participants, individuals with greater impairment (CDR ≥1) and those with history of stroke or TIA were less likely to return for follow-up visits, and therefore more likely to be excluded (Supplemental Table 1). There were no differences in level of impairment or modifiable risk factors in excluded vs included nHW participants (Supplemental Table 2). Average (SD) ages were similar between AA (77.3 ± 7.1 years) and nHW participants (75.6 ± 9.0 years; p=0.12) at study entry (Table 1). Females were overrepresented in the AA cohort (197/261, 75% vs 85/193, 44%; p<0.001); nHW had more years of education, on average (mean difference, 95%CI: 2.4 years, 1.8-3.0; p<0.001). The ADI was higher in AA participants, on average (mean difference, 3.8 points, 95%CI: 3.4-4.2; p<0.001), indicating greater social-economic-environmental deprivation. On average, AA participants completed more research visits (6.0 ± 3.2) than nHW participants (4.2 ± 2.3; p<0.001), with longer follow-up (mean difference, 95%CI: 2.1 years, 1.5-2.6; p<0.001). Cognitive impairment (global CDR ≥0.5) was less frequent in AA compared to nHW participants (72/261, 28% vs. 127/193, 66%; p<0.001), as was the frequency of APOE ε4/ε4 carriers (AA: 13/260, 5%; nHW: 19/175, 11%; p=0.025)—findings that reflect differences in recruitment strategies and retention. Symptomatic AD was the most common cause of impairment in both cohorts but was more common in AA participants (AA: 53/86, 62%; nHW: 55/134, 41%; p=0.004). Symptomatic Lewy body disease was less common in AA participants (AA: 6/86, 7%; nHW: 51/134, 38%; p<0.001). Other causes of cognitive impairment (e.g., vascular cognitive impairment, frontotemporal dementia, unclassified mild cognitive impairment) were diagnosed in similar proportions of participants.

Table 1.

Demographics and risk factors stratified by race

Variables	N	African American	N	non-Hispanic White	P-value
Participants (N)		261		193	--
Demographics
Age (years) at initial visit	261	77.3 (7.1)	193	75.6 (9.0)	0.12
Males	261	64 (25%)	193	108 (56%)	< 0.001
Years of education	260	13.2 (3.5)	193	15.6 (2.7)	< 0.001
ADI State Rank	261	7.8 (2.3)	193	4.0 (2.1)	< 0.001
Number of visits	261	6.0 (3.2)	193	4.2 (2.3)	< 0.001
Global CDR at last visit	261		193
0		189 (72%)		66 (34%)	< 0.001
0.5		34 (13%)		41 (21%)	0.022
≥1		38 (15%)		86 (45%)	< 0.0001
ApoE ε4 genotype	260		175
None		162 (62%)		103 (59%)	0.48
One copy		85 (33%)		53 (30%)	0.67
Two copies		13 (5%)		19 (11%)	0.025
Risk Factors
Smoking at initial visit	258	111 (43%)	171	70 (41%)	0.69
History of TBI at first visit	257	3 (1%)	187	20 (11%)	< 0.001
Hypertension	252	217 (86%)	184	133 (72%)	< 0.001
Hypercholesterolemia	253	162 (64%)	186	101 (54%)	0.049
Body mass index	210	28.1 (5.8)	116	25.8 (4.3)	0.001
Obesity	210	67 (32%)	116	19 (16%)	0.002
Diabetes	254	90 (35%)	185	21 (11%)	< 0.001
Hearing loss	261	136 (52%)	193	83 (43%)	0.058
Depression	257	49 (19%)	187	78 (42%)	< 0.001
Thyroid disease	260	47 (18%)	190	35 (18%)	1.00
B12 deficiency	244	17 (7%)	164	16 (10%)	0.36
Heart disease	245	74 (30%)	170	43 (25%)	0.32
Stroke or TIA	244	29 (12%)	168	12 (7%)	0.13

Values represent mean (sd), or n (%). P-values result from Wilcoxon rank sum test (continuous variables) or Fisher’s exact test (categorical variables).

Area Deprivation Index (ADI) state ranking, Clinical Dementia Rating (CDR), Traumatic Brain Injury (TBI), Transient Ischemic Attack (TIA).

Modifiable dementia risk factors were common in both groups but more common in AA participants (Table 1). The disparate distribution of risk factors was most apparent when considering hypertension, hypercholesterolemia, obesity, and diabetes. Hearing loss was also more common in AA participants (p=0.058), although differences did not reach the predetermined threshold of statistical significance. Depression was less frequent in AA compared to nHW participants (49/257, 19% vs. 78/187, 42%; p<0.001). Other modifiable risk factors were similarly represented in AA and nHW participants, including reported history of coronary artery disease and stroke/transient ischemic attack.

Dementia Risk Factors and Cognitive Impairment

Next, we considered the association between modifiable dementia risk factors and cognitive impairment (Table 2). AA participants with cognitive impairment had lower BMI (OR, 0.90; 95%CI: 0.84-0.96), and higher frequency of depression (OR, 3.88; 95%CI: 2.02-7.45) and stroke/transient ischemic attack (OR, 2.46; 95%CI: 1.10-5.50) compared to AA participants without cognitive impairment. Depression was overrepresented in nHW participants with cognitive impairment (OR, 4.02; 95%CI: 2.02-8.01) compared to nHW participants without cognitive impairment. History of traumatic brain injury (OR, 0.32; 95%CI: 0.12-0.83), hypertension (OR, 0.36; 95%CI: 0.16-0.77), hearing loss (OR, 0.49, 95%CI: 0.27, 0.90), and thyroid disease (OR, 0.36; 95%CI: 0.17-0.75) were all less likely in nHW participants with cognitive impairment. Unique to nHW participants, participants with cognitive impairment were younger at enrollment (p<0.001), and males were more likely to receive a diagnosis of cognitive impairment than females (p<0.001). APOE ε4 genotype was associated with cognitive impairment in AA (OR, 1.92; 95%CI: 1.22-3.00) and nHW participants (OR, 2.50; 95%CI: 1.44-4.32), with a similar magnitude of association.

Table 2.

Association between risk factors and cognitive impairment

	African American Participants					Non-Hispanic White Participants
	N	No Cognitive Impairment	N	Cognitive Impairment	P-value	N	No Cognitive Impairment	N	Cognitive Impairment	P-value
Number of participants		189		72			66		127
Age (years) at initial visit	189	77.0 (6.9)	72	78.2 (7.5)	0.19	66	78.5 (9.0)	127	74.0 (8.6)	< 0.001
Males	189	41 (22%)	72	23 (32%)	0.11	66	17 (26%)	127	91 (72%)	< 0.001
Years of education	188	12.8 (3.4)	72	14.1 (3.5)	0.006	66	15.5 (2.5)	127	15.6 (2.8)	0.54
ADI State Rank	189	7.9 (2.1)	72	7.4 (2.6)	0.29	66	3.5 (2.0)	127	4.2 (2.2)	0.053
ApoE ε4 genotype	188		72			63		112
None		126 (67%)		36 (50%)	0.015		47 (75%)		56 (50%)	0.002
One copy		56 (30%)		29 (40%)	0.14		14 (22%)		39 (35%)	0.089
Two copies		6 (3%)		7 (10%)	0.051		2 (3%)		17 (15%)	0.020
Smoking at initial visit	188	87 (46%)	70	24 (34%)	0.091	66	30 (45%)	105	40 (38%)	0.42
History of TBI	186	2 (1%)	71	1 (1%)	1.00	66	12 (18%)	121	8 (7%)	0.024
Hypertension	188	162 (86%)	64	55 (86%)	1.00	64	54 (84%)	120	79 (66%)	0.009
Hypercholesterolemia	188	123 (65%)	65	39 (60%)	0.46	65	37 (57%)	121	64 (53%)	0.64
Body mass index	160	28.8 (5.8)	50	25.9 (5.0)	0.002	50	25.4 (4.6)	66	26.1 (4.1)	0.25
Obesity	160	56 (35%)	50	11 (22%)	0.12	50	8 (16%)	66	11 (17%)	1.00
Diabetes	188	68 (36%)	66	22 (33%)	0.76	64	6 (9%)	121	15 (12%)	0.63
Hearing loss	189	103 (54%)	72	33 (46%)	0.22	66	36 (55%)	127	47 (37%)	0.022
Depression	188	24 (13%)	69	25 (36%)	< 0.001	65	14 (22%)	122	64 (52%)	< 0.001
Thyroid disease	189	37 (20%)	71	10 (14%)	0.37	65	19 (29%)	125	16 (13%)	0.010
B12 deficiency	184	14 (8%)	60	3 (5%)	0.77	63	8 (13%)	101	8 (8%)	0.42
Heart disease	184	55 (30%)	61	19 (31%)	0.87	63	15 (24%)	107	28 (26%)	0.86
Stroke or TIA	184	17 (9%)	60	12 (20%)	0.037	63	1 (2%)	105	11 (10%)	0.032

Values represent mean (sd), or n (%). P-values result from Wilcoxon rank sum test (continuous variables) or Fisher’s exact test (categorical variables).

Area Deprivation Index (ADI) state ranking, Traumatic Brain Injury (TBI), Transient Ischemic Attack (TIA).

The relationship between modifiable risk factors and cognitive impairment was further evaluated through multivariable logistic regression controlling for age, sex, years of education, ADI, and APOE ε4 genotype (Table 3). Side-by-side comparisons of these multivariable analyses establish similarities in the direction and magnitude of association between modifiable brain health risk factors and odds of cognitive impairment in AA and nHW participants (Figure 1), including depression, which was associated with greater odds of cognitive impairment in AA (OR, 4.30; 95%CI: 2.13-8.67) and nHW participants (OR, 2.79; 95%CI: 1.21-6.44). History of stroke/transient ischemic attack was associated with cognitive impairment in AA participants (OR, 2.52; 95%CI: 1.08-5.83); a clinically meaningful association could not be excluded in nHW participants (OR, 8.70; 95%CI: 0.80-94.83). Uniquely, higher BMI was associated with lower odds of cognitive impairment in AA but not nHW participants (OR, 0.91; 95%CI: 0.85-0.98), whereas past medical history of traumatic brain injury (OR, 95%CI: 0.11, 0.03-0.43) was associated with lower odds of cognitive impairment in nHW but not AA participants.

Table 3.

Relationship between risk factors and cognitive impairment- stratified by race

Multivariable Analysis
Variable	N	Odds Ratio (95% CI)	P-value	Variable	N	Odds Ratio (95% CI)	P-value
African American Participants				Non-Hispanic White Participants
Age at initial visit (every 10-years)	259	1.40 (0.93, 2.10)	0.11	Age at initial visit (every 10-years)	175	0.48 (0.30, 0.77)	0.002
Males	259	1.77 (0.94, 3.33)	0.078	Males	175	7.47 (3.35, 16.62)	<0.001
Years of education	259	1.12 (1.03, 1.22)	0.009	Years of Education	175	0.89 (0.76, 1.03)	0.12
ADI State Rank	259	0.95 (0.84, 1.08)	0.44	ADI State Rank	175	1.22 (1.02, 1.47)	0.033
APOE ε4 genotype	259	1.95 (1.22, 3.13)	0.005	APOE ε4 genotype	175	2.32 (1.22, 4.40)	0.010
Smoking	256	0.56 (0.30, 1.02)	0.059	Smoking	154	0.68 (0.31, 1.53)	0.35
TBI	255	0.86 (0.07, 10.58)	0.91	TBI	171	0.11 (0.03, 0.43)	0.001
Hypertension	250	1.01 (0.43, 2.37)	0.98	Hypertension	167	0.51 (0.20, 1.32)	0.17
Hypercholesterolemia	251	0.81 (0.44, 1.49)	0.50	Hypercholesterolemia	169	0.48 (0.21, 1.08)	0.074
Body mass index	210	0.91 (0.85, 0.98)	0.017	Body mass index	108	0.97 (0.86, 1.09)	0.60
Obesity	210	0.68 (0.30, 1.56)	0.37	Obesity	108	0.69 (0.19, 2.50)	0.57
Hearing loss	259	0.58 (0.31, 1.11)	0.099	Hearing loss	175	0.58 (0.26, 1.28)	0.18
Depression	255	4.30 (2.13, 8.67)	< 0.001	Depression	170	2.79 (1.21, 6.44)	0.016
Diabetes	252	1.01 (0.54, 1.87)	0.99	Diabetes	168	2.25 (0.68, 7.44)	0.18
Thyroid disease	258	0.58 (0.25, 1.32)	0.19	Thyroid disease	172	0.69 (0.28, 1.73)	0.43
B12 deficiency	243	0.62 (0.16, 2.36)	0.48	B12 deficiency	147	0.31 (0.08, 1.18)	0.087
Heart disease	244	1.10 (0.57, 2.14)	0.77	Heart disease	153	0.87 (0.35, 2.18)	0.77
Stroke or TIA	243	2.52 (1.08, 5.83)	0.032	Stroke or TIA	151	8.70 (0.80, 94.83)	0.076

The reference category for sex was female. Multivariable models were adjusted for age at first visit, sex, education, Area Deprivation Index (ADI) state ranking, and ApoE ɛ4 genotype.

Traumatic Brain Injury (TBI), Transient Ischemic Attack (TIA).

Figure 1.

Relationship between risk factors and cognitive impairment. The relationship between modifiable dementia risk factors and cognitive impairment is shown, stratified by race (closed circle = Black/African American participants, open circle = non-Hispanic White participants). Models were adjusted for age at first visit, sex, education, Area Deprivation Index (ADI) state ranking, and APOE ε4 genotype.

Traumatic Brain Injury (TBI), Body Mass Index (BMI), Transient Ischemic Attack (TIA).

Dementia Risk Factors and Rate of Cognitive Decline

The relationship between modifiable dementia risk factors and annualized rates of decline (maximum CDR-SB / years from symptomatic onset) was considered in participants with cognitive impairment using multivariate linear regression, controlling for age at first visit, sex, years of education, ADI State Rank, and APOE ε4 genotype (Table 4; Figure 2). Unique to AA participants, smoking was associated with a slower annualized decline in CDR-SB (β = −1.09, 95%CI: −2.19-0.00; p=0.049), while depression (β = 1.71, 95%CI: 0.69-2.73; p=0.001) and vitamin B12 deficiency (β = 2.65, 95%CI: 0.38-4.91; p=0.023) were associated with increased rates of decline. APOE ε4 genotype was associated with an increased rate of decline in nHW participants only (β = 0.42, 95%CI: 0.01-0.83; p=0.046).

Table 4.

Relationship between risk factors and rate of cognitive decline- stratified by race

Multivariable Analysis
Variable	N	β (95% CI)	P-value	Variable	N	β (95% CI)	P-value
African American Participants				Non-Hispanic White Participants
Age at initial visit	72	−0.29 (−0.98, 0.39)	0.40	Age at initial visit	109	−0.31 (−0.69, 0.07)	0.11
Men	72	−0.28 (−1.35, 0.79)	0.60	Men	109	−0.08 (−0.76, 0.60)	0.81
Years of education	72	0.05 (−0.10, 0.20)	0.48	Years of education	109	−0.09 (−0.21, 0.02)	0.11
ADI State Rank	72	−0.12 (−0.32, 0.09)	0.27	ADI State Rank	109	0.07 (−0.08, 0.21)	0.37
ApoE ε4 genotype	72	0.29 (−0.51, 1.08)	0.47	ApoE e4 genotype	109	0.42 (0.01, 0.83)	0.046
Smoking (initial visit)	70	−1.09 (−2.19, 0.00)	0.049	Smoking (initial visit)	91	0.12 (−0.57, 0.81)	0.72
TBI (initial visit)	71	−2.05 (−6.36, 2.26)	0.35	TBI (initial visit)	108	0.43 (−0.85, 1.71)	0.50
Hypertension	64	0.30 (−1.05, 1.65)	0.66	Hypertension	103	−0.21 (−0.87, 0.44)	0.52
Hypercholesterolemia	65	−0.19 (−1.20, 0.81)	0.70	Hypercholesterolemia	104	−0.17 (−0.77, 0.44)	0.58
Body mass index	50	0.02 (−0.10, 0.15)	0.71	Body mass index	59	−0.06 (−0.17, 0.05)	0.30
Obesity	50	0.35 (−1.25, 1.95)	0.66	Obesity	59	−0.05 (−1.35, 1.25)	0.94
Hearing loss	72	−0.66 (−1.78, 0.45)	0.24	Hearing loss	109	−0.16 (−0.82, 0.50)	0.63
Depression	69	1.71 (0.69, 2.73)	0.001	Depression	105	0.43 (−0.15, 1.02)	0.15
Diabetes	66	0.83 (−0.24, 1.89)	0.13	Diabetes	104	−0.14 (−1.06, 0.78)	0.76
Thyroid disease	71	−0.91 (−2.41, 0.59)	0.23	Thyroid disease	107	0.42 (−0.54, 1.39)	0.39
B12 deficiency	60	2.65 (0.38, 4.91)	0.023	B12 deficiency	85	−0.71 (−1.95, 0.53)	0.26
Heart disease	61	−0.29 (−1.40, 0.81)	0.60	Heart disease	91	−0.46 (−1.22, 0.30)	0.23
Stroke or TIA	60	−0.14 (−1.41, 1.13)	0.82	Stroke or TIA	89	0.23 (−0.81, 1.27)	0.66

The reference category for sex was female. Multivariable models were adjusted for age at first visit, sex, education, Area Deprivation Index (ADI) state ranking, and ApoE ε4 genotype.

Traumatic Brain Injury (TBI), Transient Ischemic Attack (TIA).

Figure 2.

Relationship between risk factors and rate of cognitive decline. The relationship between modifiable dementia risk factors and rate of cognitive decline is shown, stratified by race (closed circle = Black/African American participants, open circle = non-Hispanic White participants). Models were adjusted for age at first visit, sex, education, Area Deprivation Index (ADI) state ranking, and APOE ε4 genotype.

Traumatic Brain Injury (TBI), Body Mass Index (BMI), Transient Ischemic Attack (TIA).

Treatment of Dementia Risk Factors in AA and nHW Participants

AA participants with a history of hypertension more frequently reported using medications to manage blood pressure (172/217, 79%) compared to nHW participants (90/133, 68%; p=0.016). The inverse was observed concerning antidepressant medications: AA participants with depression were considerably less likely to report taking medications to manage depression (9/49, 18% vs nHW participants: 57/78, 73%; p<0.001). No differences were reported in the use of medications to manage hypercholesterolemia (Table 5). Differences in prescribing may contribute to the divergent associations between risk factors, cognitive impairment, and rates of decline in AA and nHW participants.

Table 5.

Disease subgroup analyses of treatment at last visit by race

Treatment	African American	Non-Hispanic White	P-value
Hypertension
No	45 (21%)	43 (32%)	0.016
Yes	172 (79%)	90 (68%)
Hypercholesterolemia
No	46 (28%)	22 (22%)	0.25
Yes	116 (72%)	79 (78%)
Depression
No	40 (82%)	21 (27%)	< 0.001
Yes	9 (18%)	57 (73%)
B12 deficiency
No	10 (59%)	10 (62%)	1.00
Yes	7 (41%)	6 (38%)

Values represent n (%). P-values from Fisher’s exact test.

Discussion

Modifiable dementia risk factors were common in community-dwelling AA and nHW participants in this series, while selected vascular risk factors—hypertension, hyperlipidemia, obesity, and diabetes—were overrepresented in AA participants. Despite the high frequency of risk factors, only depression (nHW and AA participants) and history of stroke/transient ischemic attack (AA participants only) were associated with an increased odds of cognitive impairment after controlling for relevant covariates. Depression and vitamin B12 deficiency were also associated with accelerated declines in cognitive function, although these associations were unique to AA participants, who were less likely than their nHW counterparts to report treatment with antidepressant medications.

Vascular risk factors are associated with increased markers of cerebrovascular disease, cerebral amyloid-β deposition, and impaired cognition.³⁰ However, in our cohort, vascular risk factors were not associated with cognitive impairment. Furthermore, hypertension was more frequent in cognitively normal nHW participants compared to impaired individuals. Secondary prevention could account for this finding, as most participants with cognitive impairment were treated for hypertension and hyperlipidemia. The high frequency of obesity, diabetes, heart disease, hypertension, and hypercholesterolemia in our cohort and others emphasize the need for active screening and treatment of prevalent risk factors in community-dwelling adults.^31–35 This need is particularly acute in AA communities where overrepresented vascular risk factors may contribute to disparate dementia risk.^{4, 5} Vascular risk factors can be objectively measured and effectively managed in clinical practice, providing fertile targets for population-level interventions that may decrease dementia risk and improve quality of life.^{36, 37} However, effective design and deployment of these programs requires substantial investment in programs to address existing gaps in structural and social determinants of health that exacerbate dementia risk and compromise access to care in AA communities, including disparities in access to quality education, health literacy, socioeconomic status, food security, and exposure to adversity and discrimination.³⁸

Gaps in structural and social determinants of health between AA and nHW cohorts were apparent across cohorts in this study. Average ADI was higher (a proxy measure of social-economic-environmental deprivation) and years of education were lower in AA participants. After adjusting for these covariates, the magnitude and direction of association between modifiable risk factors and cognitive impairment were largely similar (Figure 1), emphasizing the contributions of these factors to disparate dementia risk, and underscoring the importance of systematically measuring and adjusting for structural and social determinants of health in studies of memory and aging. In this study, a persistent difference was limited to history of traumatic brain injury, which was associated with a lower odds of cognitive impairment in nHW but not AA participants. The factors that drive findings in participants reporting traumatic brain injury are unclear but should not be overinterpreted given the small number of participants who reported a history of traumatic brain injury; further research in this area is warranted. Body mass index was also inversely associated with odds of cognitive impairment (higher body mass index, lower odds) in AA participants. Being overweight (BMI ≥25 kg/m²) in mid-life is associated with an increased risk of dementia later in life,³⁹ although an increasing number of studies report that decreasing BMI may herald the symptomatic onset of dementia in older individuals.^{40, 41} The reasons underlying this paradoxical relationship are unclear but may reflect declines in intake or changes in nutrition attributed to the progression of neurodegenerative diseases that precede the onset of symptomatic impairment.^{42, 43} Few studies to date have considered the association between weight loss, cognitive impairment, and race. Differences in association across race may point to the effects of disparate structural and social determinants of health, which may be amplified in older individuals who are more likely to be living alone and aging within communities that lack sufficient elder care resources.

Our findings are similar to other series that report a lower prevalence of depression in AA compared to nHW participants, despite higher levels of adversity amongst AA individuals (i.e. the “Black-White depression paradox”).^44–46 We engaged community-dwelling volunteers with high levels of education who may have higher health literacy and improved access to healthcare resources, evidenced by high frequency of antihypertensive treatment. In this context, the low frequency of self-reported medications to treat depression in AA participants is even more surprising. Greater stigma associated with mental health issues in AA communities or lower awareness of the benefits of treatment and risks of undertreatment of depression may contribute to these findings. We also directly assessed participants for depressive symptoms using 15 validated questions, with established thresholds signaling active depression (i.e., geriatric depression scale ≥5; sensitivity 86%, specificity 91%),⁴⁷ rather than relying on self-report or review of medical records. This approach likely minimized the potential for under-reporting attributable to lower mental health literacy, strengthening our findings. Further studies are required to improve understanding of the complex relationship between depression and cognitive impairment.

Depression in late-life was included as a modifiable risk factor by the Lancet Commission on dementia prevention,³ yet depression may represent a prodrome or symptom of Alzheimer disease when emerging in late-life.⁴⁸ In line with this, depression was more frequent in cognitively impaired individuals in our cohort, regardless of race. Depression was also uniquely associated with accelerated functional declines in AA participants with cognitive impairment. This finding may reflect selective undertreatment of depression. AA participants were less likely than nHW to be prescribed antidepressant medications but just as likely (or more likely) to report use of medications to manage hypertension and hypercholesterolemia. Disparities in mental health awareness and management are well recognized in AA individuals.^{46, 49} Undertreatment of depression in AA participants is likely multifactorial and may reflect disparities in structural and social determinants of health such as lower socioeconomic status and reduced access to care, inadequate health literacy, or underreporting of depression due to mental health stigma, health provider distrust, or lack of confidence in treatment leading to decreased compliance. Differences in presenting symptoms and signs in AA and nHW participants, and differences in providers’ perceptions and attitudes towards the treatment of depression may also contribute to treatment disparities.^{46, 50} Regardless of the causes and contributors, our findings suggest that depression may represent a modifiable dementia risk factor that warrants increased attention. Initiatives designed to raise awareness and strengthen access to mental health services in AA communities may improve quality of life, and potentially lower risks of cognitive impairment and rates of cognitive decline.

Vitamin B12 deficiency was also associated with accelerated decline in cognitively impaired AA participants. This relationship may reflect undertreatment of vitamin B12, although inconsistent reporting of non-prescription medications (including vitamins and supplements) precluded formal exploration of this possibility. Intriguingly, although the deleterious effects of smoking on dementia risk are well recognized, smoking history in our cohort was associated with a slower rate of cognitive decline in AA participants. Differences in characterization of “smoking history” across studies may account for this unexpected finding, acknowledging that the threshold for cigarette exposure used in this study (>100 lifetime cigarettes) may be overly inclusive.¹⁵ Alternatively, the inverse relationship may reflect benefits of screening and treatment of vascular risk factors in individuals with a history of smoking, independent of or in combination with cognitive benefits of smoking cessation.

Our findings are subject to limitations. This was not an epidemiologic study and results should not be extended to inform the incidence and prevalence of modifiable dementia risk factors in the community. Additionally, differences in recruitment patterns precluded direct comparisons of the frequency of cognitive impairment in AA and nHW participants, while reliance on self-reported medical history and medication use almost certainly contributed to underreporting of risk factors. Prospective measurement of modifiable risk factors—e.g., lipid/cholesterol profiles, hemoglobin A1C, thyroid, vitamin B12 levels, neuroimaging for cerebrovascular disease—is needed to accurately determine the contributions of disparate risk factors to disparate cognitive impairment in AA participants. At the same time, there is a need to develop and deploy strategies to promote retention of participants (and caregivers) with greater impairment (CDR ≥1) or prior history of stroke/transient ischemic attack who may experience greater burden associated with longitudinal participation. Higher quality evidence (e.g., randomized control trials) is also needed to establish whether successful mitigation of these modifiable dementia risk factors leads to reduction in incident cognitive impairment or rates of cognitive decline. Although cognitive impairment was defined in accordance with standard definitions,^16–20 levels of impairment were staged and tracked using the global CDR and CDR-SB scores. The CDR was developed in patients with AD and is uniformly applied across ADRC participants. We acknowledge, however, that this measure may underestimate rates of change in patients with non-AD dementia, potentially underestimating the strength of associations between risk factors and rates of decline. Additionally, the sample size of our study is not overly large, and therefore the possibility of a type II error (i.e., a false-negative finding) is important to consider, particularly for rarer risk factors; emphasis should be placed on 95% confidence limits when interpreting the results of our association analyses. Finally, our study engaged a diverse population in Northeast Florida that increasingly mirrors the racial, ethnic, and financial disparity of the US population.⁹ Although diversity in the sample is a strength, the constrained geographic focus and reliance on volunteer participants who may have higher education and motivation than the general public may limit generalizability of results to other communities in the United States and beyond. Replication in other cohorts recruited from other communities is needed to support and extend findings.

Modifiable dementia risk factors were frequent in AA and nHW participants engaged in longitudinal studies of memory and aging. Depression was associated with cognitive impairment and faster cognitive decline in AA participants who were less likely to report use of medications to treat depression. Disparities in the prevalence and treatment of modifiable dementia risk factors exemplify the potential to leverage primary and secondary prevention strategies to promote brain health and mitigate disparities in cognitive impairment in at-risk communities. There is a particular need to develop and reinforce bidirectional, equitable partnerships between researchers, clinicians, and community members to ensure that programs are responsive to the needs of the community. This may be even more important when attempting to address disparities in screening and treatment of frequently stigmatized mental health syndromes, including depression.

Data Availability

Anonymized data will be shared with qualified researchers upon reasonable requests to the corresponding author.