Associations between comorbid conditions with mild cognitive impairment and Alzheimer's disease by race‐ethnicity and sex based on NACC data

Abstract

INTRODUCTION

We investigated the associations between diabetes (type 2), hypertension and hypercholesterolemia with mild cognitive impairment (MCI) and Alzheimer's disease (AD) diagnoses by race‐ethnicity and sex.

METHODS

Data (n = 22,950) were derived via the National Alzheimer's Coordinating Center. Logistic regression was used to assess the relationship between each comorbid condition and MCI and AD.

RESULTS

For non‐Hispanic White (NHW) and non‐Hispanic Black (NHB) males, diabetes and hypertension had a significant positive association with MCI. For NHW females, diabetes, hypertension, and hypercholesterolemia were significantly associated with increased MCI. Diabetes was significantly associated with AD in Asian and Pacific Islander (API) males and NHW females.

DISCUSSION

Targeted interventions aimed at curbing diabetes and hypertension onset for NHW and NHB males, and potentially reducing all comorbid conditions for NHW females may reduce MCI outcomes. Heterogeneity in the Hispanic and API diasporas should be considered when delineating comorbid conditions with MCI and/or AD.

Highlights

• For non‐Hispanic White (NHW) and non‐Hispanic Black (NHB) males, diabetes and hypertension were associated with MCI.

• Diabetes, hypertension, and hypercholesterolemia were associated with mild cognitive impairment (MCI) for NHW females.

• Only hypertension was associated with MCI in NHB females.

• Diabetes was associated with Alzheimer's disease (AD) in NHW females and Asian and Pacific Islander (API) males.

1. BACKGROUND

The number of Americans living with Alzheimer's disease (AD) is projected to double to 13.8 million by 2060. ¹ However, AD burden is not distributed equally across racial and ethnic groups ¹ , ² or between sexes. ³ The prevalence of mild cognitive impairment (MCI) and AD and related dementias (ADRD) is higher among Hispanic (14%) and non‐Hispanic Black (NHB) (19%) older adults compared to non‐Hispanic White (NHW) older adults (10%), and Hispanic and NHB older adults face higher risks of developing dementia. ¹ , ² In addition to the higher prevalence of MCI and AD in women, ⁴ women also face a higher lifetime risk with a hazard ratio (HR) of 1.12 (95% confidence interval [CI]: 1.02, 1.23) for all‐cause dementia. ¹ , ⁵ Due to increased life expectancy, higher life expectancy for women, and population diversification, ⁶ it is imperative to investigate the intersectionality between race‐ethnicity, sex and other factors potentially associated with MCI and AD.

Hypertension, diabetes, and hypercholesterolemia have shown to be risk factors for dementia incidence. ⁷ , ⁸ , ⁹ In addition, risk for dementia with these comorbid conditions vary between different racial‐ethnic groups. A study showed that population‐attributable risks associated with ADRD were higher in NHB participants (40%), as well as Hispanic participants (34%) compared to NHW participants (29%). ¹⁰ The increased risk for and prevalence of dementia among NHB and Hispanic older adults may be explained by higher prevalence of these comorbid conditions. ¹⁰ , ¹¹ , ¹² Moreover, in a study of Hispanic and NHW participants matched on age, education, sex, cognitive status, and apolipoprotein E (APOE) ε4 status, hypertension and higher body mass index (BMI) were associated with decreased executive performance in Hispanic participants. ¹³ Another study found that persistent hypertension decreased cognitive performance, and both incident and persistent hypertension increased the odds for MCI in Hispanic participants. ¹⁴ The higher likelihood for dementia among NHB older adults may be attributed to higher prevalence of vascular risk factors, as well as other socioeconomic conditions, such as reduced access to healthcare. ¹⁰ , ¹¹ , ¹⁵ , ¹⁶ More notably, NHB men are underrepresented in research studies. ¹⁵ , ¹⁶ Similarly, Asian and Pacific Islanders (API) and particularly Native Hawaiians and Pacific Islanders (NHPI) are understudied, and a recent community‐based study showed increased odds of MCI in those reporting high blood pressure, high cholesterol, and pre‐diabetes or borderline diabetes compared with those not reporting these respective conditions. ¹⁷

Sex differences can also impact risks for ADRD ¹ , ¹⁸ and vascular comorbidities. ¹¹ , ¹⁹ , ²⁰ The lifetime risk of women developing AD is higher than it is for men. ¹ A population‐based study with 105,796 participants found that, after adjusting for age, women had an increased risk (17%) of developing late‐onset AD compared to men across NHB, Hispanic, API, and NHW cohorts, with varying risk ranges for each ethnic group. ¹⁸ Cardiometabolic conditions attenuated ADRD risk across race‐ethnic groups, but only exploratory mediation analyses were conducted. ¹⁸ A review on the intersectionality of sex and race on AD found that vascular risk factors for AD were more prevalent in NHB women than any other race and sex. ²¹ However, few studies have examined the influence of sex and comorbid conditions on MCI and AD across multiple racial and ethnic cohorts.

Therefore, the aim of the present study is to investigate the relationship between comorbid conditions, including diabetes (type 2), hypertension, hypercholesterolemia, with MCI and AD within and between NHW, Hispanic, NHB, and API cohorts stratified by sex in the National Alzheimer's Coordinating Center (NACC) dataset. The use of a single large‐scale dataset ensures the uniformity of data collected via a standardized protocol. We hypothesize that these comorbid conditions are associated with MCI and AD, and that these associations vary by race, ethnicity, and sex.

RESEARCH IN CONTEXT

1. Systematic review: The authors found studies investigating the relationship between diabetes (type 2), hypertension and hypercholesterolemia, primarily for non‐Hispanic whites (NHW).

2. Interpretation: Race‐ethnicity and sex‐specific data show that associations between diabetes and mild cognitive impairment (MCI), and hypertension and MCI are more pronounced for NHW and non‐Hispanic Black (NHB) males. For NHW females, those with diabetes, hypertension and hypercholesterolemia had significant increased odds for MCI, while only hypertension was significantly associated with MCI in NHB females. Diabetes had a significant association with AD in NHW females and in Asian and Pacific Islander (API) males, but an inverse association for Hispanic males.

3. Future directions: Potentially, race‐ethnic and sex‐specific targeted interventions aimed at reducing comorbid conditions which may impact cognitive outcomes should be considered. In addition, clinical trials and large standardized studies are encouraged to recruit larger, diverse populations to further delineate the role of comorbid conditions in diverse race‐ethnic and sex groups.

2. METHODS

2.1. Study population and data ascertainment

The study population and data were derived via the NACC data repository. NACC data collection procedures and protocols have been described previously. ²² , ²³ Briefly, NACC data are collected by a consortium of the National Institute of Aging's Alzheimer's Disease Research Centers (ADRCs) found across the United States. Participants recruited by ADRCs complete a standardized clinical interview and examination and are followed longitudinally, including the Uniform Data Set (UDSv3). ²⁴ Data were scrutinized prior to release with a centralized multi‐stage quality control and quality assurance protocol that culminates in a certification process by NACC. ²²

Data from 42 ADRCs were utilized in the present study. The source data includes all participants in the NACC database through NACC's June 2023 data freeze with data collection beginning in September 2005. Data were obtained from NACC through an approved data use request. For the present study (due to retaining sample size and power), only the first observations were utilized via NACC Uniform Dataset version 3 (UDSv3). Data for 48,605 participants were available before exclusions were applied. Participants younger than 55 or older than 85 were excluded (n = 6100). Of those remaining, 62 reporting type 1 diabetes were excluded. Additional exclusions include participants without distinct diagnostic status (n = 4799, see §2.3) or those without APOE genotype (n = 9631). Among the remaining 28,013 participants, 1002 were excluded due to missing demographic data (e.g., ethnicity, education). Therefore, those with complete diagnostic, demographic and other covariate data were included in the present study (n = 27,011); although sample sizes varied for each respective model (see statistical analyses) depending on availability of participant comorbid condition and covariate data). Utilization of NACC data and analyses were reviewed and approved by the California State University, Fullerton IRB (HSR‐19‐20‐527).

2.2. Measures

2.2.1. AD, MCI, and comorbidities

NACC evaluation and diagnostic criteria have been described previously. ²³ For this analysis, we categorized diagnostic status as either AD, MCI, normal cognition, or other status. Those who were issued a clinical diagnosis of normal cognition by the ADRC (this includes being coded as normal cognition on both AD and cognitive impairment variables) were categorized as having normal cognition. Participants with either amnestic or non‐amnestic MCI and without AD diagnoses were categorized as MCI. Participants with non‐Alzheimer's dementia or who were cognitively impaired but did not meet the criteria for MCI were recorded as other status. All participants who were issued a presumptive etiologic diagnosis of AD were categorized as AD.

Comorbid conditions were evaluated primarily by participants' and co‐participants’ self‐reports. Participants reporting any history of diabetes, hypertension, and hypercholesterolemia were recorded as having each of the respective diagnoses. Participants reporting a history of diabetes were also queried on the type (type 1, type 2, or other).

2.2.2. Race and ethnicity

NACC ascertains race‐ethnicity via the clinical interview. The question prompts are unstructured and for race‐ethnicity data, the interviewer codes the following questions: “Does the subject report being of Hispanic/Latino ethnicity (i.e., having origins from a mainly Spanish‐speaking Latin American country) regardless of race?” and “What does the subject report as his or her race?”

2.2.3. Covariates

Sex, age, education, and APOE genotype were also utilized in the present study. Sex was coded by the NACC interviewer. Age was a derived variable. Interviewers recorded the number of years of school that a participant had completed. APOE genotyping was (predominantly) conducted by the ADRCs, and the results are then reported to NACC using standardized forms.

2.3. Statistical analyses

Variable distributions were measured via summary statistics and univariate visualizations. Stacked bar plots representing the frequency of observations by sex and race‐ethnicity were plotted for each of the categorical variables. Prior to analyses, age was centered and rescaled to units of decades. Evaluation and reporting of summary statistics conformed with the STROBE guidelines for epidemiologic research. ²⁵

We conducted multiple logistic regression analyses, adjusting for co‐variates [race‐ethnicity (NHW, ref), age (years), education (years), sex (female, ref.), and presence of APOEε4 alleles (no allele, ref), to investigate the relationship between each respective comorbid condition (diabetes, hypertension, or hypercholesterolemia) with MCI and AD. Separate models were conducted for each respective comorbid condition and for MCI and AD; interaction terms were also included for the comorbid conditions by race‐ethnicity and by sex, separately. We also conducted sets of stratified analyses by race‐ethnicity and sex, to investigate the relationship between each comorbid condition with MCI and AD, adjusting for covariates. Outcome variables were binary and are as follows: AD versus normal cognition (ref.) or MCI versus normal cognitive status (ref.). Models to calculate OR were estimated using R 4.4.0 ²⁶ and the stats package. ²⁷ P‐values were used in statistical inference as a continuous measure of evidence, and CIs were used to gauge the plausibility of parameter values. ²⁸ , ²⁹ , ³⁰

3. RESULTS

Of the 27,011 participants included in this analysis, 20,858 were NHW (77.2%), 3379 were NHB (12.5%), 2092 were Hispanic (7.7%), and 682 were API (2.5%). There were 15,431 female participants (57.1%) and 11,580 male participants (42.9%), with mean ages 71.7 years and 72.0 years, respectively. There was little variation in age among subgroups; the mean age among sex and racial‐ethnic groups was lowest for Hispanic women (71.0 years) and highest for Hispanic men (72.2 years). The group with the highest levels of education were NHW men (M = 16.4 years, SD = 2.9 years) with overall mean years of education being 15.4 (SD = 3.3).

Sample comorbid conditions are shown in Figure 1. Diabetes was the least common comorbidity. Men had a slightly higher rate (n = 1646, 14.2%) of diabetes than women (n = 1817, 11.8%), and NHB men reported the highest rate of diabetes (n = 264, 28.3%), which contrasted with the much lower rates among NHW women (n = 848, 7.6%). Hypertension was slightly more common among men in the sample, with 5936 men (51.3%) and 7462 women (48.4%) reporting hypertension. The highest rate of hypertension was among NHB women (n = 1836, 75.0%) and lowest for NHW women (n = 4691, 41.8%). Similarly, hypercholesterolemia was lower in women (n = 7709, 50.0%) compared with men (n = 6507, 56.2%), and hypercholesterolemia variation among sex and racial‐ethnic groups was modest, with the highest among API men (n = 169, 60.1%) and lowest among NHW women (n = 5456, 48.6%).

FIGURE 1.

Descriptive statistics, APOEε4 alleles, and diagnoses status by sex and racial‐ethnic groups. APOE, apolipoprotein E.

Figure 1 also shows sample descriptives for clinical diagnoses and APOEε4 status. Briefly, just under half of the sample had at least one APOEε4 allele (Figure 1). Among men, 4086 had one APOEε4 allele (35.3%), and 947 were homozygous for APOEε4 (8.2%); similarly, among women, 5494 had one APOEε4 allele (35.6%), and 1103 were homozygous for APOEε4 (7.1%). API had the lowest rates of heterozygous or homozygous APOEε4 genotypes (114 women, 28.4%; 84 men, 29.9%). NHB and NHW had similar genotype frequencies with 1113 NHB and 4910 NHW women having at least one APOEε4 allele (45.5% and 43.8%, respectively); in contrast, 466 NHB and 4233 NHW men had at least one APOEε4 allele (50.0% and 43.9%, respectively).

In our overall separate models (Table 1) investigating the relationship between comorbid conditions (diabetes, hypertension, and hypercholesterolemia) with MCI and AD, we found that previously established variables, including age [OR (95% CI) shown for diabetes and MCI model): OR (95% CI): 1.18 (1.11, 1.24), p < 0.001] and APOEε4 status [one allele: [OR (95% CI):1.33 (1.22, 1.45), p < 0.001; two alleles: 2.27 (1.89, 2.73), p < 0.001]) were significantly associated with MCI and AD odds, respectively, while education [OR (95% CI): 0.91 (0.9, 0.92), p < 0.001] was significantly and inversely associated with MCI and AD odds. We also found that comorbid conditions, including diabetes [OR (95% CI): 1.50 (1.23, 1.82), p < 0.001], hypertension [OR (95% CI): 1.21 (1.07, 1.37), p = 0.002], and hypercholesterolemia [OR (95% CI): 1.21 (1.07, 1.37), p = 0.002] were positively and significantly associated with MCI. Only diabetes was significantly associated with AD [OR (95% CI): 1.24 (1.07, 1.43), p = 0.004].

TABLE 1.

The association (OR, 95% CI, and p‐value) between comorbid conditions, race/ethnicity, and sex with MCI and AD, including interaction by race/ethnicity and sex with comorbid conditions.

	MCI			AD
Parameter	Diabetes n = 15,790	Hypertension n = 15,666	Hypercholesterolemia n = 23,403	Diabetes n = 15,806	Hypertension n = 15,586	Hypercholesterolemia n = 23,412
Comorbid conditions	1.50 (1.23, 1.82), p < 0.001	1.21 (1.07, 1.37), p = 0.002	1.17 (1.03, 1.32), p = 0.012	1.24 (1.07, 1.43), p = 0.004	1.03 (0.95, 1.12), p = 0.495	0.99 (0.91, 1.07), p = 0.812
Race/ethnicity (ref: NHW)
HISP	1.26 (1.07, 1.49), p = 0.006	1.37 (1.1, 1.7), p = 0.004	1.23 (0.98, 1.55), p = 0.074	0.96 (0.84, 1.09), p = 0.532	0.91 (0.77, 1.08), p = 0.269	0.99 (0.83, 1.17), p = 0.892
NHB	0.99 (0.87, 1.13), p = 0.858	0.77 (0.61, 0.97), p = 0.028	1.01 (0.85, 1.19), p = 0.938	0.55 (0.49, 0.61), p < 0.001	0.52 (0.44, 0.62), p < 0.001	0.57 (0.5, 0.64), p < 0.001
API	1.37 (1.06, 1.75), p = 0.015	1.10 (0.77, 1.53), p = 0.597	1.67 (1.20, 2.30), p = 0.002	0.94 (0.77, 1.15), p = 0.554	0.89 (0.69, 1.15), p = 0.387	0.88 (0.66, 1.15), p = 0.345
Race x comorbid conditions
Comorbid conditions x HISP	0.78 (0.56, 1.08), p = 0.136	0.83 (0.63, 1.10), p = 0.203	1.06 (0.79, 1.41), p = 0.706	0.66 (0.51, 0.86), p = 0.002	0.95 (0.77, 1.19), p = 0.666	0.86 (0.69, 1.07), p = 0.178
Comorbid conditions x AA	0.87 (0.66, 1.14), p = 0.318	1.32 (1.01, 1.72), p = 0.042	1.00 (0.80, 1.25), p = 0.997	0.87 (0.70, 1.08), p = 0.214	1.04 (0.85, 1.26), p = 0.726	0.92 (0.77, 1.09), p = 0.336
Comorbid conditions x API	0.47 (0.22, 0.94), p = 0.04	1.27 (0.79, 2.04), p = 0.327	0.58 (0.36, 0.93), p = 0.024	1.39 (0.88, 2.21), p = 0.16	1.26 (0.88, 1.81), p = 0.206	1.31 (0.91, 1.9), p = 0.143
Age (years)	1.18 (1.11, 1.24), p < 0.001	1.15 (1.09, 1.21), p < 0.001	1.16 (1.1, 1.23), p < 0.001	1.49 (1.43, 1.55), p < 0.001	1.49 (1.43, 1.55), p < 0.001	1.49 (1.44, 1.55), p < 0.001
Education (years)	0.91 (0.9, 0.92), p < 0.001	0.91 (0.9, 0.93), p < 0.001	0.91 (0.9, 0.92), p < 0.001	0.88 (0.88, 0.89), p < 0.001	0.89 (0.88, 0.89), p < 0.001	0.88 (0.88, 0.89), p < 0.001
Sex (female)	2.19 (2.01, 2.38), p < 0.001	2.23 (1.99, 2.5), p < 0.001	2.17 (1.93, 2.44), p < 0.001	1.94 (1.83, 2.07), p < 0.001	1.97 (1.82, 2.13), p < 0.001	1.92 (1.76, 2.08), p < 0.001
Sex x comorbid	0.84 (0.68, 1.05), p = 0.128	0.92 (0.79, 1.08), p = 0.323	0.98 (0.83, 1.14), p = 0.775	0.85 (0.72, 1.01), p = 0.072	0.95 (0.85, 1.06), p = 0.363	1.00 (0.89, 1.12), p = 0.998
APOEε4 (ref = no alleles)
1 allele	1.33 (1.22, 1.45), p < 0.001	1.33 (1.22, 1.45), p < 0.001	1.32 (1.21, 1.43), p < 0.001	2.67 (2.52, 2.84), p < 0.001	2.68 (2.52, 2.84), p < 0.001	2.69 (2.54, 2.86), p < 0.001
2 alleles	2.27 (1.89, 2.73), p < 0.001	2.27 (1.89, 2.72), p < 0.001	2.21 (1.83, 2.65), p < 0.001	8.39 (7.42, 9.5), p < 0.001	8.36 (7.4, 9.47), p < 0.001	8.33 (7.37, 9.43), p < 0.001

Abbreviations: AA, African American; AD, Alzheimer's disease; API, Asian and Pacific Islander; APOE, apolipoprotein E; CI, confidence interval; HISP, Hispanic; MCI, mild cognitive impairment; NHB, non‐Hispanic Black; NHW, non‐Hispanic White; OR, odds ratio.

For our sample, race‐ethnicity data (Table 1) showed that Hispanic individuals had a significant to borderline significant increased odds of MCI compared with NHW [OR (95% CI): 1.26 (1.07, 1.49), p = 0.006 (diabetes model); OR (95% CI): 1.37 (1.1, 1.7), p = 0.004 (hypertension model); OR (95% CI): 1.23 (0.98, 1.55), p = 0.074 (hypercholesterolemia model)]. Similarly, API also had a significant increased odds of MCI compared with NHW [OR (95% CI): 1.37 (1.06, 1.75), p = 0.015; OR (95% CI): 1.10 (0.77, 1.53), p = 0.597; OR (95% CI):1.67 (1.20, 2.30), p = 0.002, respective comorbid models]. For AD, NHB had a significantly reduced odds of AD compared with NHW in the diabetes, hypertension and hypercholesterolemia models, [OR (95% CI): 0.55 (0.49, 0.61), p < 0.001; 0.52 (0.44, 0.62), p < 0.001; 0.57 (0.5, 0.64), respectively]. We also observed significant three‐way interactions among sex, comorbidity, and racial‐ethnic groups. The interaction models for MCI included an omnibus three‐way interaction test that was significant for all three comorbidities [χ²(df = 10) > 22.3, p < 0.014]. The omnibus interaction test for the three‐way interaction among sex, comorbidity, and racial‐ethnic group in the AD models was significant only for diabetes [χ²(df = 10) > 19.8, p = 0.031].

Table 2 presents model estimates for the associations between each comorbidity and MCI by each race‐ethnicity and sex, adjusted for covariates. Analyses by race‐ethnicity and sex showed that, for both NHW and NHB males, that those with diabetes and hypertension had significant increased odds for MCI [for NHW: OR (95% CI): 1.20 (0.99, 1.45), p = 0.057; OR (95% CI): 1.15 (1.01, 1.30), p = 0.031; for NHB: OR (95% CI): 1.65 (1.11, 2.45), p = 0.013; OR (95% CI): 1.58 (1.03, 2.47), p = 0.04, respectively]. For NHW females, in separate models, those with diabetes, hypertension and hypercholesterolemia, compared to those without these respective conditions, had significant increased odds for MCI [OR (95% CI): 1.58 (1.26, 1.97), p = 0.001; OR (95% CI): 1.16; (1.02, 1.33), p = 0.029; OR (95% CI): 1.18 (1.03, 1.34), p = 0.016, respectively]. In addition, a positive significant association between hypertension and MCI was observed for NHB females [OR (95% CI): 1.43 (1.06, 1.95), p = 0.023]. The only significant association for API females was an inverse association between hypercholesterolemia and MCI [OR (95% CI): 0.41 (0.20, 0.82), p = 0.014].

TABLE 2.

Association (OR, 95% CI, and p‐value) between comorbid conditions and MCI for NHW, NHB, HISP, and API males and females

Males
	Diabetes				Hypertension				Hypercholesteremia
Variable	Model (NHW) n = 5061	Model (HISP) n = 382	Model (NHB) n = 576	Model (API) n = 163	Model (NHW) n = 5050	Model (HISP) n = 383	Model (NHB) n = 576	Model (API) n = 162	Model (NHW) n = 5026	Model (HISP) n = 373	Model (NHB) n = 570	Model (API) n = 161
Comorbid conditions	1.20 (0.99, 1.45), 0.057	1.09 (0.69, 1.71), 0.724	1.65 (1.11, 2.45), 0.013	0.85 (0.33, 2.05), 0.734	1.15 (1.01, 1.30), 0.031	1.01 (0.65, 1.58), 0.949	1.58 (1.03, 2.47), 0.04	1.78 (0.87, 3.74), 0.119	1.12 (0.99, 1.27), 0.084	1.38 (0.89, 2.17), 0.157	1.24 (0.85, 1.81), 0.269	0.95 (0.48, 1.91), 0.887
Age (years)	1.12 (1.03, 1.21), 0.010	1.12 (0.83, 1.51), 0.471	1.02 (0.78, 1.33), 0.896	1.16 (0.74, 1.83), 0.518	1.10 (1.01, 1.20), 0.024	1.11 (0.82, 1.50), 0.513	0.95 (0.73, 1.24), 0.719	1.05 (0.65, 1.70), 0.828	1.11 (1.02, 1.21), 0.015	1.13 (0.83, 1.54), 0.428	1.00 (0.76, 1.30), 0.974	1.18 (0.75, 1.87), 0.464
Education (years)	0.91 (0.89, 0.93), 0.001	0.94 (0.90, 0.99), 0.013	0.91 (0.86, 0.97), 0.003	1.01 (0.89, 1.14), 0.900	0.91 (0.89, 0.93), 0.001	0.94 (0.90, 0.98), 0.007	0.91 (0.85, 0.96), 0.001	1.02 (0.90, 1.16), 0.788	0.91 (0.89, 0.93), 0.001	0.94 (0.89, 0.98), 0.005	0.91 (0.85, 0.96), 0.001	1.01 (0.89, 1.15), 0.864
APOEε4 (ref = no alleles)
1 allele	1.37 (1.20, 1.57), 0.001	0.93 (0.55, 1.54), 0.772	1.50 (1.02, 2.20), 0.040	2.24 (0.96, 5.25), 0.061	1.37 (1.20, 1.57), 0.001	0.94 (0.56, 1.56), 0.824	1.49 (1.02, 2.19), 0.041	2.25 (0.93, 5.39), 0.068	1.37 (1.20, 1.57), 0.001	0.84 (0.50, 1.42), 0.526	1.45 (0.98, 2.12), 0.06	2.25 (0.96, 5.22), 0.058
2 alleles	2.13 (1.60, 2.83), 0.001	1.28 (0.16, 7.95), 0.793	1.77 (0.71, 4.14), 0.199	2.56 (0.10, 67.08), 0.513	2.15 (1.62, 2.85), 0.001	1.31 (0.17, 8.13), 0.771	1.80 (0.73, 4.21), 0.185	2.77 (0.10, 73.68), 0.484	2.13 (1.60, 2.82), 0.001	1.16 (0.15, 7.32), 0.875	1.72 (0.69, 4.01), 0.223	2.59 (0.10, 67.76), 0.508

Females
	Diabetes				Hypertension				Hypercholesteremia
Variable	Model (NHW) n = 6854	Model (HISP) n = 789	Model (NHB) n = 1707	Model (API) n = 258	Model (NHW) n = 6849	Model (HISP) n = 801	Model (NHB) n = 1721	Model (API) n = 264	Model (NHW) n = 6793	Model (HISP) n = 781	Model (NHB) n = 1701	Model (API) n = 261
Comorbid conditions	1.58 (1.26, 1.97), 0.001	1.16 (0.79, 1.67), 0.45	1.08 (0.82, 1.42), 0.573	0.42 (0.10, 1.31), 0.183	1.16 (1.02, 1.33), 0.029	1.02 (0.73, 1.42), 0.912	1.43 (1.06, 1.95), 0.023	1.31 (0.66, 2.57), 0.436	1.18 (1.03, 1.34), 0.016	1.25 (0.89, 1.76), 0.202	1.10 (0.86, 1.41), 0.444	0.41 (0.20, 0.82), 0.014
Age (years)	1.19 (1.09, 1.30), 0.001	1.24 (0.98, 1.59), 0.075	1.38 (1.16, 1.66), 0.001	1.46 (0.94, 2.28), 0.092	1.16 (1.06, 1.27), 0.001	1.21 (0.95, 1.54), 0.123	1.34 (1.12, 1.61), 0.001	1.40 (0.90, 2.21), 0.137	1.17 (1.07, 1.28), 0.001	1.17 (0.92, 1.49), 0.199	1.39 (1.16, 1.66), 0.001	1.62 (1.05, 2.55), 0.032
Education (years)	0.90 (0.88, 0.92), 0.001	0.93 (0.90, 0.96), 0.001	0.89 (0.85, 0.93), 0.001	0.92 (0.83, 1.03), 0.132	0.90 (0.88, 0.92), 0.001	0.93 (0.90, 0.96), 0.001	0.89 (0.86, 0.93), 0.001	0.95 (0.86, 1.06), 0.361	0.90 (0.88, 0.92), 0.001	0.93 (0.90, 0.96), 0.001	0.89 (0.85, 0.93), 0.001	0.92 (0.83, 1.02), 0.109
APOEε4 (ref = no alleles)
1 allele	1.44 (1.25, 1.66), 0.001	0.90 (0.61, 1.31), 0.572	1.10 (0.85, 1.43), 0.446	1.62 (0.71, 3.49), 0.23	1.44 (1.25, 1.66), 0.001	0.89 (0.60, 1.29), 0.536	1.10 (0.85, 1.42), 0.473	1.70 (0.77, 3.58), 0.174	1.41 (1.22, 1.63), 0.001	0.88 (0.59, 1.28), 0.498	1.10 (0.85, 1.42), 0.478	1.93 (0.87, 4.16), 0.097
2 alleles	2.79 (2.07, 3.71), 0.001	0.75 (0.11, 3.00), 0.716	2.23 (1.24, 3.88), 0.006	0.99 (0.05, 6.50), 0.989	2.76 (2.05, 3.68), 0.001	0.77 (0.12, 3.06), 0.737	2.16 (1.20, 3.76), 0.008	1.12 (0.06, 7.34), 0.921	2.63 (1.95, 3.51), 0.001	0.70 (0.10, 2.79), 0.648	2.20 (1.23, 3.84), 0.006	1.14 (0.06, 7.83), 0.911

Abbreviations: API, Asian and Pacific Islander; APOE, apolipoprotein E; CI, confidence interval; HISP, Hispanic; MCI, mild cognitive impairment; NHB, non‐Hispanic Black; NHW, non‐Hispanic White; OR, odds ratio.

Table 3 presents model estimates for the association between each comorbidity and AD by race‐ethnicity and sex, adjusted for covariates. In these models, associations were only identified between diabetes and AD. For API males, diabetes had a significant [OR (95% CI): 2.02 (1.03, 4.04), p = 0.044] increased association with AD. Further, a significant positive [OR (95% CI): 1.19 (1.01, 1.40), p = 0.041] association was observed between diabetes and AD for NHW females. However, we observed that for Hispanic males, those with diabetes had a significant inverse odds for AD [OR (95% CI): 0.65 (0.44, 0.96), p = 0.032].

TABLE 3.

Association (OR, 95% CI and p‐value) between comorbid conditions and AD for NHW, NHB, HISP, and API males and females

Males
	Diabetes				Hypertension				Hypercholesteremia
Variable	Model (NHW) n = 8155	Model (HISP) n = 580	Model (NHB) n = 768	Model (API) n = 226	Model (NHW) n = 8145	Model (HISP) n = 580	Model (NHB) n = 769	Model (API) n = 225	Model (NHW) n = 8103	Model (HISP) n = 569	Model (NHB) n = 761	Model (API) n = 225
Comorbid conditions	1.10 (0.95, 1.27), 0.209	0.65 (0.44, 0.96), 0.032	1.02 (0.72, 1.44), 0.903	2.02 (1.03, 4.04), 0.044	1.01 (0.92, 1.10), 0.912	0.92 (0.64, 1.33), 0.664	1.03 (0.74, 1.45), 0.844	1.25 (0.69, 2.27), 0.458	0.99 (0.90, 1.09), 0.842	1.06 (0.73, 1.51), 0.771	0.97 (0.72, 1.32), 0.865	1.12 (0.62, 2.03), 0.71
Age (years)	1.34 (1.26, 1.42), 0.001	1.66 (1.30, 2.13), 0.001	2.10 (1.70, 2.62), 0.001	2.01 (1.36, 3.04), 0.001	1.34 (1.26, 1.43), 0.001	1.64 (1.29, 2.11), 0.001	2.06 (1.66, 2.57), 0.001	1.85 (1.25, 2.79), 0.003	1.34 (1.26, 1.43), 0.001	1.63 (1.28, 2.09), 0.001	2.07 (1.67, 2.58), 0.001	1.94 (1.32, 2.90), 0.001
Education (years)	0.91 (0.89, 0.92), 0.001	0.96 (0.93, 0.99), 0.014	0.95 (0.91, 0.99), 0.020	0.96 (0.87, 1.05), 0.381	0.91 (0.89, 0.92), 0.001	0.96 (0.93, 1.00), 0.029	0.95 (0.90, 0.99), 0.016	0.95 (0.86, 1.04), 0.248	0.91 (0.89, 0.92), 0.001	0.96 (0.93, 0.99), 0.017	0.95 (0.90, 0.99), 0.014	0.95 (0.86, 1.04), 0.265
APOEε4 (ref = no alleles)
1 allele	2.47 (2.24, 2.72), 0.001	2.46 (1.68, 3.66), 0.001	2.11 (1.54, 2.91), 0.001	6.21 (3.09, 13.25), 0.001	2.46 (2.23, 2.71), 0.001	2.48 (1.69, 3.68), 0.001	2.10 (1.53, 2.89), 0.001	5.67 (2.86, 11.88), 0.001	2.47 (2.24, 2.73), 0.001	2.39 (1.62, 3.56), 0.001	2.14 (1.56, 2.95), 0.001	5.78 (2.92, 12.11), 0.001
2 alleles	6.48 (5.30, 7.99), 0.001	15.49 (5.28, 66.43), 0.001	7.27 (3.94, 14.07), 0.001	16.06 (2.65, 309.79), 0.011	6.45 (5.27, 7.95), 0.001	15.00 (5.15, 63.97), 0.001	7.28 (3.94, 14.12), 0.001	13.89 (2.32, 266.34), 0.016	6.45 (5.27, 7.96), 0.001	14.48 (4.96, 61.85), 0.001	7.30 (3.95, 14.12), 0.001	13.66 (2.28, 261.81), 0.017

Females
	Diabetes				Hypertension				Hypercholesteremia
Variable	Model (NHW) n = 10,102	Model (HISP) n = 1144	Model (NHB) n = 2080	Model (API) n = 348	Model (NHW) n = 10,092	Model (HISP) n = 1154	Model (NHB) n = 2094	Model (API) n = 353	Model (NHW) n = 10032	Model (HISP) n = 1128	Model (NHB) n = 2071	Model (API) n = 349
Comorbid conditions	1.19 (1.01, 1.40), 0.041	0.90 (0.66, 1.22), 0.489	1.05 (0.83, 1.32), 0.692	1.37 (0.69, 2.69), 0.364	1.01 (0.93, 1.11), 0.801	0.97 (0.74, 1.26), 0.813	0.94 (0.75, 1.19), 0.602	1.08 (0.65, 1.79), 0.76	0.99 (0.91, 1.08), 0.874	0.73 (0.56, 0.96), 0.022	0.84 (0.69, 1.03), 0.1	1.28 (0.78, 2.10), 0.329
Age (years)	1.43 (1.35, 1.51), 0.001	1.83 (1.53, 2.19), 0.001	1.91 (1.66, 2.22), 0.001	2.52 (1.81, 3.56), 0.001	1.43 (1.35, 1.51), 0.001	1.86 (1.55, 2.23), 0.001	1.92 (1.66, 2.22), 0.001	2.51 (1.80, 3.55), 0.001	1.43 (1.35, 1.52), 0.001	1.87 (1.56, 2.25), 0.001	1.94 (1.68, 2.25), 0.001	2.53 (1.82, 3.58), 0.001
Education (years)	0.85 (0.84, 0.86), 0.001	0.88 (0.86, 0.91), 0.001	0.87 (0.84, 0.90), 0.001	0.87 (0.81, 0.93), 0.001	0.85 (0.84, 0.86), 0.001	0.88 (0.86, 0.91), 0.001	0.87 (0.84, 0.90), 0.001	0.88 (0.82, 0.94), 0.001	0.85 (0.83, 0.86), 0.001	0.88 (0.86, 0.91), 0.001	0.87 (0.84, 0.90), 0.001	0.88 (0.82, 0.94), 0.001
APOEε4 (ref = no alleles)
1 allele	3.00 (2.74, 3.28), 0.001	2.28 (1.73, 3.01), 0.001	2.41 (1.96, 2.96), 0.001	3.06 (1.73, 5.51), 0.001	3.01 (2.75, 3.29), 0.001	2.36 (1.79, 3.12), 0.001	2.38 (1.94, 2.92), 0.001	3.03 (1.72, 5.41), 0.001	3.01 (2.75, 3.29), 0.001	2.36 (1.78, 3.13), 0.001	2.48 (2.01, 3.05), 0.001	3.11 (1.75, 5.61), 0.001
2 alleles	9.53 (7.96, 11.47), 0.001	8.62 (4.12, 19.84), 0.001	11.25 (7.56, 17.02), 0.001	12.33 (4.31, 41.40), 0.001	9.49 (7.93, 11.42), 0.001	8.91 (4.27, 20.47), 0.001	11.22 (7.55, 16.97), 0.001	12.50 (4.38, 41.87), 0.001	9.43 (7.87, 11.35), 0.001	9.36 (4.46, 21.63), 0.001	11.31 (7.60, 17.12), 0.001	12.21 (4.26, 41.04), 0.001

Abbreviations: AD, Alzheimer's disease; API, Asian and Pacific Islander; APOE, apolipoprotein E; CI, confidence interval; HISP, Hispanic; NHB, non‐Hispanic Black; NHW, non‐Hispanic White; OR, odds ratio.

4. DISCUSSION

Via a nationally distributed sample and uniform data collection procedures, we examined the association between comorbid conditions, including diabetes, hypertension, and hypercholesterolemia, on MCI and AD across race‐ethnic and sex groups. We found that previously established covariates such as education was inversely associated with MCI and AD, while age and increasing numbers of APOEε4 alleles were associated with increased odds of MCI and AD. Associations between comorbid conditions and MCI and/or AD by race‐ethnicity and sex showed that, for both NHW and NHB males, diabetes and hypertension had 20%– 60% (respectively) increased odds for MCI compared to those without the respective condition. We also found that NHW females with diabetes had 60% increased odds for MCI, and NHW women with hypertension and hypercholesterolemia had nearly 20% increased MCI odds compared to those without the condition. NHB females with hypertension had nearly a 45% increased odds of MCI. For AD, diabetes was associated with increased odds of AD in NHW females, as well as in API males, yet we observed an inverse relationship with diabetes and AD in Hispanic males.

We found in our overall model that diabetes was associated with increased odds of MCI and AD (50% and 24%, respectively). Studies have shown a relationship between diabetes, particularly midlife, and cognitive decline. ³¹ , ³² , ³³ , ³⁴ A meta‐analysis of 17 longitudinal cohort studies (predominantly in NHW populations) amounting to about 1.7 million people concluded that diabetes mellitus increased the risk of developing AD with a relative risk (RR) of ≈ 1.5. ³¹ Recent findings from the Atherosclerosis Risk in Communities (ARIC) study showed that midlife diabetes was associated with rapid dementia biomarker changes, specifically, 0.11‐SD faster for neurofilament light (NfL) and 0.15‐SD faster for glial fibrillary acidic protein (GFAP). ³² Studies show that type 2 diabetes induces vascular and cognitive decline (memory, mood, and behavior), and therefore key pathophysiological changes due to diabetes onset may subsequently contribute to MCI and/or AD. ³³ , ³⁴

In addition, our race‐ethnic and sex‐specific data showed that NHW and NHB males with diabetes (compared with those without) had increased odds (20%–65%) of MCI, and that only NHW females with diabetes showed increased odds for both MCI and for AD. Two earlier studies in African Americans, one a cohort study and the other cross‐sectional, showed that diabetes was associated with cognition; the latter showing an inverse association between diabetes and a battery of cognitive functioning tests and that lower poverty levels exacerbated this association. ³⁵ , ³⁶ Nonetheless, few studies have reported on the relationship between diabetes with clinically diagnosed/assessed MCI and AD in NHB and to our knowledge no studies have shown sex‐specific outcomes. Our findings of the relationship between diabetes and MCI in NHB males, but not in NHB females may be due to the differential mediating effects of other comorbid conditions (i.e., stroke) and/or unknown factors for NHB males compared with NHB females, which needs further investigation.

On the contrary, we found an inverse association between diabetes and AD for Hispanic males, and a strong positive association for API males. Several earlier studies have investigated the relationship between diabetes and cognitive decline in Hispanic/Latino individuals; however, results have been mixed. ³⁷ , ³⁸ , ³⁹ Data on Mexican Americans from two Texas studies showed no association between diabetes and MCI, after adjusting for established covariates. ³⁸ Studies in Caribbean Latino individuals suggest a modest 11% MCI attributable risk to diabetes, and that those with diabetes had reduced language and visuospatial abilities, but no differences in rate of cognitive change between those with and without diabetes mellitus was observed over a 6‐year follow‐up. ³⁹ Recent studies suggest a clearer association between diabetes and steeper cognitive decline, MCI and A1C for middle‐aged Latino inividuals. ⁴⁰ , ⁴¹ In terms of API, few to no data are available on diabetes and AD; Hayes‐Larson and colleagues conducted a study investigating diverse Asian American populations in California and showed heterogeneous associations between type 2 diabetes and incident dementia, with South Asian individuals having the highest hazards ratio (HR = 1.32; 95% CI, 1.08–1.62) and Filipino individuals the lowest (HR = 1.31; 95% CI, 1.08–1.60). ⁴² A community‐based study in NHPI showed an association between prediabetes, but not diabetes with MCI, however small sample sizes may limit the findings. ¹⁷ Heterogeneity in both the Hispanic/Latino/Caribbean, as well as API diaspora in terms of disease severity, age at diabetes onset, and divergent cardiovascular factors in these sub‐groups warrant further investigation to more clearly delineate the association between type 2 diabetes and cognitive decline.

We also found (in our overall model) a strong association between hypertension with MCI. Race‐ethnic and sex‐specific data also showed, similar to diabetes, that hypertension increased the odds of MCI in both NHW (15%) and NHB (nearly 60%) males and females. The association between hypertension and cognitive decline has been reported in other racial/ethnic groups. ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ The Whitehall II cohort study, in predominantly UK non‐Hispanic white individuals, showed that participants with longer exposure to hypertension (systolic blood pressure [SBP] ≥ 130 mmHg), during midlife, had an increased risk of dementia compared to those with shorter exposure to hypertension (HR 1.29, 95% CI 1.00, 1.66). ⁴³ A recent pooled analysis from five cohorts (n = 34,349) on the relationship between hypertension and various cognitive domains showed that higher blood pressure levels in Black individuals were related to a steeper decline compared with White individuals for global cognition and memory, but not for executive function. ⁴⁶ Although research shows a link between hypertension and cognitive impairment – awareness and management of these comorbid conditions is still lacking for vulnerable, at‐risk groups.

Although hypercholesterolemia was associated with increased MCI odds for a majority of the race‐ethnic and sex groups, we only found a significant association for increased MCI odds for NHW females, and an inverse association for API females. Studies on the relationship between elevated cholesterol and cognitive decline are equivocal. Sáiz‐Vazquez and colleagues conducted a meta‐meta‐analysis on 100 primary studies and five meta‐analyses showing an association between only low‐density lipoprotein (LDL) with AD, and no effects of high‐density lipoproteins, total cholesterol, or triglycerides on AD. ⁴⁷ Biological plausibility of the influence of cholesterol on cognitive decline shows that increased LDL has vascular and neurotoxic effects. ⁴⁸ , ⁴⁹ , ⁵⁰ However, for the present study, only total cholesterol was assessed and therefore, potentially the role of cholesterol subtypes should be considered for future studies. In addition, API is an expansive disparate diaspora and therefore, further studies are needed to more clearly delineate the relationship between hypercholesterolemia and cognitive decline in API subgroups, such as South Asian, Japanese, and NHPI populations.

With multiple, stratified models, resulting in small sample sizes, associations may be difficult to detect in subgroup analysis; however, even in a majority of these models, established variables, including age, education, and APOEε4 status remained significantly associated in the established direction. And we were able to detect race‐ethnic and sex‐specific associations between comorbid conditions and MCI and AD. Further, although studies have examined race‐ethnic and sex‐specific ADRD differences in select race‐ethnic groups, few studies have ascertained, analyzed and reported on cognitive outcomes consistently for and across multiple race‐ethnic and sex‐specific groups in one study. In addition, ADRD diagnoses in the NACC dataset are based on standardized measures, including clinical assessments and a battery of neuropsychological tests across multiple race‐ethnic and sex groups, resulting in comprehensive and uniform dementia outcome measurements, which may contribute to more clearly delineating ADRD etiology and respective factors associated with brain health outcomes for the aforementioned groups.

Our findings suggest that those reporting comorbid conditions, including diabetes, hypertension and hypercholesterolemia had an increased odds of MCI, and that those with diabetes had increased odds of AD compared to those without the respective condition. We further found that diabetes and hypertension increased MCI odds for both NHW and NHB males, while all three conditions were associated with increased MCI odds in NHW females and that only hypertension was related to increased MCI in NHB females. Overall, these data may potentially suggest that preventive strategies to reduce diabetes onset, as well as high blood pressure and cholesterol may reduce cognitive impairment; and that potentially more targeted interventions, specifically reducing diabetes and hypertension for NHW and NHB males (who had nearly a 65% increased odds), as well as lower all these conditions for NHW females, while specifically targeting hypertension in NHB females may curb cognitive declines in these aforementioned populations. Nonetheless, further studies are warranted to further investigate the complex role of comorbid conditions in heterogeneous populations, including API and Hispanic populations.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. Author disclosures are available in the supporting information.

CONSENT STATEMENT

The present study utilized secondary data; therefore, human subjects consent was not necessary. Utilization of NACC data and secondary analyses were reviewed and approved by the California State University, Fullerton IRB (HSR‐19‐20‐527).

Supporting information

Supporting information

McEligot AJ, Xiao C, Do TQ. Associations between comorbid conditions with mild cognitive impairment and Alzheimer's disease by race‐ethnicity and sex based on NACC data. Alzheimer's Dement. 2025;21:e70650. 10.1002/alz.70650