Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2022 Sep 1;77(12):e234–e246. doi: 10.1093/geronb/gbac128

Socioeconomic Status, Race/Ethnicity, and Unexpected Variation in Dementia Classification in Longitudinal Survey Data

Elizabeth A Luth 1,✉, Holly G Prigerson 2
Editor: Jessica Kelley
PMCID: PMC9799200  PMID: 36048568

Abstract

Objectives

As dementia affects a growing number of older adults, it is important to understand its detection and progression. We identified patterns in dementia classification over time using a longitudinal, nationally representative sample of older adults. We examined the relationship between socioeconomic status and race/ethnicity, and patterns in dementia classification.

Methods

Data for 7,218 Medicare beneficiaries from the 2011–2017 National Health and Aging Trends Study (NHATS) were classified into five categories: consistently no dementia, consistently cognitive impairment, “typical” dementia progression, “expected” variation, and “unexpected” variation. Multivariable multinomial logistic regression assessed relative risk of dementia classification by sociodemographic and health factors.

Results

Among NHATS respondents, 59.5% consistently were recorded as having no dementia, 7% consistently cognitively impaired, 13% as having typical progression, 15% as having expected variation, and 5.5% as having unexpected variation. In multivariable models, compared with consistent dementia classification, less education, Medicare–Medicaid-dual enrollment, and identifying as non-Hispanic Black were associated with increased likelihood of unexpected variation (e.g., non-Hispanic Black adjusted risk ratio: 2.12, 95% CI: 1.61–2.78, p < .0001).

Discussion

A significant minority of individuals have unexpected patterns of dementia classification over time, particularly individuals with low socioeconomic status and identifying as non-Hispanic Black. Dementia classification uncertainty may make it challenging to activate resources (e.g., health care, caregiving) for effective disease management, underscoring the need to support persons from at-risk groups and to carefully evaluate cognitive assessment tools to ensure they are equally reliable across groups to avoid magnifying disparities.

Keywords: Alzheimer’s disease, Health disparities, Minority aging (race/ethnicity), Socioeconomic status

An estimated 6.2 million older Americans have Alzheimer’s disease and over 11% of the population age 65 and older have dementia, a number that is expected to grow to 13.8 million by 2060 (Alzheimer’s Association, 2021; Rajan et al., 2021). Socioeconomic and racial/ethnic disparities in dementia are well documented. Dementia is more common among those with less education and non-Hispanic Black and Hispanic individuals (Alzheimer’s Association, 2021; Lines, 2014; Power et al., 2021; Vega et al., 2017).

In terms of socioeconomic status (SES), years of formal education build cognitive reserve, which may help mitigate cognitive impairment as changes to the brain occur (Meng & D’Arcy, 2012; Roe et al., 2007). Additionally, formal education is associated with mentally stimulating, higher paying jobs that may build cognitive reserve (Alzheimer’s Association, 2021; Pool et al., 2016). Higher paying jobs also increase access to health-promoting resources such as one’s ability to afford health care to prevent and manage diseases associated with dementia (e.g. heart disease, diabetes) and to access physician specialists who can diagnose dementia at earlier stages and prescribe patients medications that may slow cognitive decline (Gudala et al., 2013; Majoka & Schimming, 2021; Samieri et al., 2018). SES, and the financial resources it confers, also affects one’s ability to address the long-term effects of dementia, specifically, having flexibility in managing increased caregiving and support needs as dementia becomes more advanced.

Older Black and Hispanic adults are more likely to have missed or delayed dementia diagnoses and are less likely to be receive medication for dementia and participate in clinical trials on dementia (Alzheimer’s Association, 2021; Lines, 2014; Sayegh & Knight, 2013). Given the link between race/ethnicity and SES in the United States, SES factors related to dementia are replicated within racial/ethnic minoritized groups. Compared with non-Hispanic Whites, Hispanic older adults have less formal education and Hispanic and Black persons have lower median household income, and higher poverty rates, which may account for some of their increased risk for dementia (DeNavas-Walt & Proctor, 2015; Kohler & Lazarín, 2007). More proximate risk factors for dementia, such as heart disease and diabetes, disproportionately affect Black and Hispanic persons (Glymour & Manly, 2008; Lines, 2014). However, race/ethnicity in the U.S. carries with it characteristics independent of SES that affect access to the health-promoting resources (e.g., high-quality health care). In the United States, individuals from racial/ethnic minoritized groups have experienced individual discrimination and systemic racism—the most famous example being the Tuskegee Syphilis Study. Perceived discrimination is a potential barrier to accessing diagnoses and treatment for dementia. Higher proportions of Black and Hispanic persons believe their race/ethnicity make it more difficult to get high-quality care, including for dementia (Alzheimer’s Association, 2021). Perceived discrimination has also been associated with poorer cognitive functioning (Barnes et al., 2012). Heterogeneity among Black and Hispanic populations in the United States is often ignored in how data is captured, particularly in large surveys, obscuring potential within-group differences in dementia prevalence and risk (McDonough et al., 2022; Williams et al., 2016). However, membership in racial/ethnic groups carries shared social consequences, such as racial/ethnic minoritized groups’ experience of being universally subjected to the effects of systemic racism, which has been associated with poorer cognitive functioning among African American women (Coogan et al., 2020) and affects access to high-quality health care and disease management (Bailey et al., 2021; Stites et al., 2021; Williams et al., 2016).

The observed relationship between socioeconomic status (SES), race, ethnicity, and health outcomes, including dementia, is the result of complex, overlapping, intersecting social and structural processes that affect individuals’ experiences of the world and how those experiences are reflected in their health and cognition. There are several potential explanations for these associations. Fundamental causes theory postulates that SES contributes to health disparities because it confers access to resources that help individuals avoid disease (Link & Phelan, 1995; Phelan & Link, 2005). In the United States, structural racism creates inequalities in power, prestige, neighborhood context, and health care, which operate through and independently of SES to create and perpetuate racial inequalities in health (Phelan and Link 2015). Intersectionality theory acknowledges and engages with the collective and compounded effects of simultaneously occupying multiple aspects of identity (race, ethnicity, gender, SES) on individuals’ experiences (Crenshaw, 1989), including health disparities (Weber, 2006), dementia (Hulko, 2004), and dementia caregiving (Liu et al., 2022). Differences in health literacy and suboptimal communication can further compound challenges in accurately measuring dementia.

Population-based studies, such as the National Health and Aging Trends Study (NHATS) and the Health and Retirement Study (HRS), rely on cognitive tests that can be easily and reliably administered by a trained data collector. In longitudinal studies such as these, repeated assessment of cognitive function provides a signal about the potential development and progression of dementia at the population level, allowing researchers and policy makers to map current and plan for future disease, disability burden, and caregiving needs. This is particularly important in capturing the trajectories of individuals who are less likely to seek medical care—early or ever—regarding changes to their cognitive functioning, such as persons with lower SES and from racial/ethnic minoritized groups. Uncertainty about cognitive status can lead to a delay or failure to activate resources (e.g., specialist physician consultation, planning for disability and caregiving needs), potentially compounding disparities in dementia management and outcomes.

As individuals age, they may experience any number of trajectories with respect to their cognition, with a predominant understanding that cognition will remain relatively stable or proceed along a path of overall decline. Any observed improvements in cognitive function are expected to be modest, such as moving from mild cognitive impairment to normal cognition (Abner et al., 2012; Mungas et al., 2010), and able to be explained by classification error, misinformation from informants, or underlying medical conditions (Abner et al., 2012). Although cognitive trajectories in order adults are heterogeneous and should therefore be analyzed with care (Steinerman et al., 2010), researchers do not usually attempt to identify or understand larger improvements in cognitive functioning over time, such as moving from “dementia” to “normal” cognition. Recently, epidemiologists have used two points of dementia classification, including predicting incidence and prevalence rates in large population studies (Chen et al., 2020; Freedman et al., 2018; Zissimopoulos et al., 2018). Researchers using more time points to examine cognitive trajectories assume stability or a linear decline in their analytic approaches (e.g. Zahodne et al., 2011). Alternatively, researchers acknowledge inconsistencies in dementia classification over time, but use a workaround such as dementia classification in year before death (Sullivan et al., 2022), rather than attempting to understand inconsistencies in classification. However, overall, there is a lack of attention to identifying specific patterns in heterogeneity of trajectories of cognition over longer periods of time in large samples, particularly trajectories that fall outside the range of expected improvements in cognition.

This study has two aims: First, it adds to our understanding of heterogeneous cognitive trajectories in older adults by presenting a novel way of attending to trajectories that fall outside of expected fluctuations in cognition. Understanding differences in dementia classification over time can allow for planning and resource activation at individual and systemic levels to meet the changing, dementia-related needs of the aging U.S. population. Second, rather than dismissing these trajectories as uncredible or devising an analytic plan to sidestep them, this analysis presents an initial step toward engaging with these unexpected, heterogeneous classifications by identifying factors associated with them. We pay particular attention to the role of sociodemographic factors related to dementia: SES and race/ethnicity. Consistent with sociological and intersectional perspectives (e.g., Weber, 2006), we take the approach that measures of SES and racial and ethnic categories are not intrinsic traits that inherently lead to disparities in health or cognitive outcomes. Rather, they are proxy measures for social constructs, dynamics, and structures that impose restrictions (such as lack of access to health care) or subject individuals to experiences (individual experiences of discrimination and structural racism) that negatively affect health and cognition—and reliable assessment of patient’s cognitive status and competence. We propose reasons for the existence of different dementia classifications, identify possible explanatory mechanisms for variation in dementia classification by SES and race/ethnicity (e.g., bias in assessment tools, health literacy, challenges to reliable assessment associated with reduced contact with the high-quality health care), and discuss implications for research, policy, and clinical care.

Method

Data

This cohort study analyzed seven waves of pooled data from the 2011–2017 National Health and Aging Trends Study (NHATS). NHATS follows a nationally representative sample of 12,427 Medicare beneficiaries aged 65 and older. Participants who entered NHATS in 2011 have up to seven waves of data. A replenishment sample was added in 2015; there are up to three waves of data for this cohort. We analyzed data for 7,218 respondents who were community dwelling the year they entered the study and had at least three waves of dementia classification data (n = 7,642). Due to small sample numbers of respondents, we excluded individuals who identified as “Asian/Other race” (n = 200) or “do not know/refused” for race/ethnicity (n = 115). We excluded 109 individuals with missing data on covariates. In the final analytic sample, 99% of respondents had no missing data and were still in the study, had died, or were lost to attrition; 1% (n = 80) were missing dementia classification(s) mid-participation in the study.

Outcome: Dementia Classification Over Time

NHATS provides researchers with syntax to classify respondents as having none, possible, or probable dementia for each wave of data collection. NHATS dementia classification uses three types of information to identify persons with cognitive impairment: (a) self or proxy report that a physician told the NHATS participant they had dementia or Alzheimer’s disease; (b) score indicating probable dementia on the AD8 Dementia Screening Interview (AD8) assessing memory, temporal orientation, judgment, and function (administered to proxy respondents when NHATS participant could not respond); and (c) cognitive tests evaluating participants’ memory (immediate and delayed 10-word recall), orientation (date, month, year, day of the week; naming President and Vice President), and executive function (clock draw; Kasper et al., 2013). Participants with a reported physician diagnosis of dementia or proxy responses to the AD8 indicating likely dementia (score of 2+) were classified as “probable” dementia. For cognitive tests, scores 1.5 SD at or below the mean in 2 or 3 cognitive domains (memory, orientation, executive functioning) were classified as “probable dementia.” Scores 1.5 SD at or below the mean in one domain were classified as “possible” dementia (Kasper et al., 2013). NHATS’ dementia classification has been validated against the Aging, Demographics, and Memory Study (Kasper et al., 2013).

To allow for multiple possible changes in dementia classification across years of data collection and to maximize inclusion of the 2015 replenishment sample, we examined patterns in dementia classifications for respondents with at last three dementia classification data points. Initially, we identified eleven categories of dementia classification patterns into which all respondents fell (Table 1). Based on data distribution and conceptual groupings, we collapsed the 11 categories into five (Table 1). Four categories reflect “expected” patterns in dementia classification from year to year. Of these four, two categories capture respondents with the same dementia classification every year data is reported: no dementia and some cognitive impairment (probable or possible dementia). A third captures “typical” dementia progression (moving from no to possible or probable dementia, or possible to probable dementia). The fourth captures “expected” variation in dementia classification that could be explained by disease progression or slight variations in performances on cognitive assessments from year to year. For example, as mild cognitive impairment sets in, an individual may oscillate between no and possible dementia on cognitive assessments over several years. Alternatively, as an individual progresses from mild cognitive impairment through more advanced stages of dementia, they may oscillate between possible and probable dementia. The fifth category captures “unexpected” variation in dementia classification, jumping across classification categories in unintuitive patterns that are not easily explained by disease progression or small fluctuations in cognitive assessment scores. In this case, individuals move directly from probable dementia one year to no dementia (reflecting a substantial (1.5 SD) improvement in two or three cognitive assessment domains) in subsequent years or may jump across all three classification categories—from possible to no to probable dementia—in as many years (reflecting substantial [1.5 SD] fluctuations in at least one cognitive assessment domain). As dementia is a progressive disease in which sufferers do not make significant cognitive improvements, particularly at later stages, these substantial and inconsistent fluctuations are unusual. In sensitivity analysis keeping “probable to no dementia” and “possible to no to probable dementia” as separate categories, overall results did not substantively change and the separate categories were similarly in how they compared with “consistently no dementia.”

Table 1.

Description of Dementia Classifications Over Time

Original Category n (%) Final category n (%) Description Examples Notes
Expected outcomes
 Always no dementia 4,293 (59.5%) Consistently no dementia 4,293 (59.5%) Individual recorded as no dementia in all waves 3333333
3333…
….333		 Signifies individuals who have consistent scores on cognitive tests over time
 Always probable dementia 443 (6.1%) Consistently cognitive impairment 494 (6.8%) Individual recorded as possible dementia or probable dementia in all waves 1111111
1111…
 Always possible dementia 51 (0.7%) 2222222
….222
 No to possible dementia 386 (5.3%) Dementia progression 958 (13.3%) Individual progresses through dementia classifications (e.g. no to possible to probable) 3322222
332….		 Signals individuals who experience gradual cognitive decline over time
 No to possible to probable dementia 117 (1.6%) 33322211
321….
 No to probable dementia 297 (4.1%) 3331111
….311
 Possible to probable dementia 158 (2.2%) 2222111
….221
 Move between possible and probable dementia 131 (1.8%) Anomalous, expected variation 1,074 (14.9%) Individual oscillates between two contiguous (probable and possible; possible and no) dementia classifications. 1221111
2122111
121….
….212
3333121		 Signifies individuals who may experience small fluctuations in cognitive assessments, likely close to cutoff points. Individuals starting with “no” dementia and then moving between “probable” and “possible” dementia are included here, provided they are not classified as “no” dementia in waves following a “probable” dementia classification.
 Move between no and possible dementia 943 (13.1%) 3323233
232….
….323
Unexpected outcomes
 Move from probable to no dementia 215 (3.0%) Anomalous, unexpected variation 399 (5.5%) Individual moves across all three dementia classifications, (probable to no; possible to no to probable dementia) 1131111
1333211
131….
….313		 Signifies individuals who experience large fluctuations on cognitive assessments from year to year, including having “probable” dementia, followed by “no dementia” in subsequent year(s).
 Move from possible to no to probable dementia 184 (2.5%) 2123321
3222132
32321..
….231
Open in a new tab

Notes: 1 = probable dementia, 2 = possible dementia, 3 = no dementia. “.” indicates there is no data collection during this wave. In all but 80 cases (1%), this is because the participant entered the sample in 2015 (“….” followed by 3 waves of data as represented by ….333, ….111, etc. above), died, or was lost to attrition (latter two represented by a series of “.” after several waves of data, as represented by 333…, 1111…, etc. above).

Sociodemographic and Health Risk Factors for Dementia

We examined the relationship between multiple sociodemographic and health risk factors for dementia and variation in dementia classification over time. Race/Ethnicity was categorized as non-Hispanic White, non-Hispanic Black, Hispanic. We measured socioeconomic status using dual Medicare–Medicaid enrollment in any year of the study (yes = 1, no = 0) and education level, measured as less than high school (0–11 years), high school (12 years), some college (13–15 years), and Bachelor’s degree or higher (16+ years). We controlled for health factors associated with increased dementia risk the year the NHATS participant entered the sample (2011 or 2015), including self or proxy reports of a physician telling the NHATS participant they had heart disease, hypertension, or diabetes (yes = 1, no = 0). Depression, a modifiable risk factor for dementia, was measured using the PHQ-4 and categorized based on established cutoff points (normal = 0–2, mild = 3–5, moderate = 6–8, severe = 9–12; Livingston et al., 2020).

Covariates

As age is the largest risk factor for dementia (Alzheimer’s Association, 2021; De Reuck et al., 2018; James et al., 2012), we control for age in 5-year intervals. Dementia is more prevalent among women, likely due to their longer life span, but also possibly due to their historically lower levels of education (male = 1, female = 0; Alzheimer’s Association, 2021; Chêne et al., 2015). Immigrant status has also been associated with increased prevalence of dementia and delayed diagnosis (Franco & Choi, 2020; Moon et al., 2019), and so we control for being born in the United States (yes = 1, no = 0). We also control whether the respondent had a proxy respondent in any study year (yes = 1, no = 0).

Analysis

We calculated descriptive statistics for the total sample and by the five dementia classification categories. Chi-square tests identified significant differences in sociodemographic and health risk factors within dementia classification categories. To model the association between dementia classification categories and risk factors, we used bivariate and multivariable multinomial logistic regression. We ran postestimation analyses to assess for potential collinearity among variables. All analyses were conducted using Stata MP 15.1.

Results

Sixty percent (59.5%) of the 7,218 NHATS participants consistently reported no dementia at every wave of data collection, 7% consistently reported some cognitive impairment at every wave (possible or probable dementia), 13% indicated dementia progression (e.g., no to probable dementia), 15% indicated an expected variation in dementia classification (e.g., oscillating between no and possible dementia), and 6% (5.5%) indicated an unexpected variation in dementia classification (e.g., probable dementia one year and no dementia the following year; Table 2). The sample was predominantly non-Hispanic White (72%), followed by non-Hispanic Black (22%), and Hispanic (6%), and were relatively evenly split by education level: less than high school (23%), high school (27%), some college (26%), bachelor’s degree, or higher (24%). One-quarter (25%) received Medicaid at some point during the study period.

Table 2.

Descriptive Statistics for 7,218 Medicare Beneficiaries Aged 65 and Older, 2011–2017 NHATS

Dementia classification over time
Sample Consistent no dementia Consistent cognitive impairment Dementia progression Anomalous-expected variation Anomalous-unexpected variation
n % n % n % n % n % n % p value
Dementia classification 7,218 1.00 4,293 59.48 494 6.84 958 13.27 1,074 14.88 399 5.53
Sociodemographic factors
 Race/ethnicity ***
  Non-Hispanic White 5,210 72.18 3,427 79.83 284 57.49 626 65.34 648 60.34 225 56.39
  Non-Hispanic Black 1,580 21.89 713 16.61 155 31.38 244 25.47 331 30.82 137 34.34
  Hispanic 428 5.93 153 3.56 55 11.13 88 9.19 95 8.85 37 9.27
 Education ***
  Less than high school 1,651 22.87 559 13.02 224 45.34 321 33.51 372 34.64 175 43.86
  High school 1,937 26.84 1,162 27.07 128 25.91 256 26.72 295 27.47 96 24.06
  Some college 1,866 25.85 1,258 29.30 83 16.80 211 22.03 245 22.81 69 17.29
  Bachelor’s degree or higher 1,764 24.44 1,314 30.61 59 11.94 170 17.75 162 15.08 59 14.79
  Received Medicaid (any wave) 1,799 24.92 662 15.42 232 46.96 332 34.66 371 34.54 202 50.63 ***
Health factors (year entered NHATS sample)
  Heart disease diagnosis 1,211 16.78 663 15.44 122 24.70 172 17.95 190 17.69 64 16.04 ***
  Hypertension diagnosis 4,817 66.74 2,777 64.69 342 69.23 659 68.79 758 70.58 281 70.43 ***
  Diabetes diagnosis 1,813 25.12 968 22.55 149 30.16 245 25.57 323 30.07 128 32.08 ***
Depression/anxiety (PHQ-4) ***
  Normal (0–2) 5,134 71.13 3,309 77.08 243 49.19 609 63.57 718 66.85 255 63.91
  Mild (3-5) 1,474 20.42 773 18.01 137 27.73 246 25.68 235 21.88 83 20.80
  Moderate (6–8) 443 6.14 163 3.80 74 14.98 72 7.52 94 8.75 40 10.03
  Severe (9–12) 167 2.31 48 1.12 40 8.10 31 3.24 27 2.51 21 5.26
Covariates
 Age (year entered NHATS sample) ***
  65–69 1,624 22.50 1,318 30.70 27 5.47 86 8.98 156 14.53 37 9.27
  70–74 1,584 21.95 1,115 25.97 41 8.30 151 15.76 215 20.02 62 15.54
  75–79 1,476 20.45 880 20.50 83 16.80 183 19.10 243 22.63 87 21.80
  80–84 1,328 18.40 626 14.58 120 24.29 230 24.01 250 23.28 102 25.56
  85–89 749 10.38 251 5.85 124 25.10 179 18.68 135 12.57 60 15.04
  90+ 457 6.33 103 2.40 99 20.04 129 13.47 75 6.98 51 12.78
Born outside the United States 625 8.66 256 5.96 55 11.13 117 12.21 137 12.76 60 15.04 ***
Male 3,000 41.56 1,774 41.32 185 37.45 379 39.56 495 46.09 167 41.85 **
Proxy respondent (any wave) 1,456 20.17 329 7.66 357 72.27 367 38.31 211 19.65 192 48.12 ***
Open in a new tab

Notes: NHATS = National Health and Aging Trends Study. p Value based on chi-square tests.

*p < .05, **p < 0.01, ***p < .0001,

In bivariate multinomial logistic regression models (Table 3), all sociodemographic and health factors were significantly associated with having any dementia classification versus consistently no dementia. The only exception is heart disease, which was significantly associated with consistently having some cognitive impairment versus consistently no dementia, but was not significant for other categories. Compared with non-Hispanic Whites, non-Hispanic Black and Hispanic, persons had 146% and 228% higher relative risk of having unexpected variation in dementia classification versus consistently no (non-Hispanic Black relative risk ratio [RRR]: 2.46, 95% CI: 2.10–2.87, p < .0001; Hispanic RRR: 3.28, 95% CI: 2.51–4.30, p < .0001). Compared with persons with less than a high school education, having a high school diploma or more education was associated with 74%–86% lower risk of unexpected variation versus consistently no dementia (high school RRR: 0.26, 95% CI: 0.20–0.35, p < .0001; Bachelor’s or more RRR: 0.14, 95% CI: 0.11–0.20, p < .0001). Being Medicare–Medicaid dual enrolled was associated with 462% higher risk of unexpected variation versus consistently no dementia (Medicaid RRR: 5.62, 95% CI: 4.55–6.96, p < .0001). Among health factors, having hypertension or diabetes were associated with 30% and 62% higher risk of unexpected variation versus consistently no dementia (hypertension RRR: 1.30, 95% CI: 1.04–1.65; p < .0001; diabetes RRR: 1.62, 95% CI: 1.30–2.03, p < .0001). Elevated levels of depression and anxiety, compared with normal, were associated with 39%-468% higher risk of unexpected variation versus consistently no dementia (mild RRR: 1.39, 95% CI: 1.97–1.81, p = .012; severe RRR: 5.68, 95% CI: 3.35–9.63, p < .0001). Results were similar for having consistent cognitive impairment, progressive dementia, and expected variation in dementia classification versus consistently no dementia.

Table 3.

Relative Risk Ratios and 95% Confidence Intervals for Bivariate Multinomial Logistic Regression Models of Individual Sociodemographic Factors on Dementia Classification Over Time in 7,218 Medicare Beneficiaries Age 65 and Older, 2011–2017 NHATS

Consistent no dementia (ref) Consistent cognitive impairment Dementia progression Anomalous-expected variation Anomalous-unexpected variation
RRR RRR 95% CI p value RRR 95% CI p value RRR 95% CI p value RRR 95% CI p value
M1: Race/ethnicity (ref = NHW)
 NHB 1.00 2.62 2.12 3.24 *** 1.87 1.58 2.22 *** 2.46 2.10 2.87 *** 2.46 2.10 2.87 ***
 Hispanic 1.00 4.34 3.12 6.04 *** 3.15 2.39 4.15 *** 3.28 2.51 4.30 *** 3.28 2.51 4.30 ***
M2: Education (ref = less than HS)
 HS 1.00 0.27 0.22 0.35 *** 0.38 0.32 0.47 *** 0.38 0.32 0.46 *** 0.26 0.20 0.35 ***
 Some coll 1.00 0.16 0.13 0.22 *** 0.29 0.24 0.36 *** 0.29 0.24 0.35 *** 0.18 0.13 0.24 ***
 BA/more 1.00 0.11 0.08 0.15 *** 0.23 0.18 0.28 *** 0.19 0.15 0.23 *** 0.14 0.11 0.20 ***
M3: Received Medicaid (any wave)
1.00 4.86 4.00 5.90 *** 2.91 2.49 3.40 *** 2.89 2.49 3.37 *** 5.62 4.55 6.96 ***
M4: Heart disease diagnosis
1.00 1.80 1.44 2.24 *** 1.20 1.00 1.44 n.s. 1.18 0.99 1.40 n.s. 1.05 0.79 1.38 n.s.
M5: Hypertension diagnosis
1.00 1.23 1.00 1.50 * 1.20 1.04 1.40 * 1.31 1.13 1.51 *** 1.30 1.04 1.63 *
M6: diabetes diagnosis
1.00 1.48 1.21 1.82 *** 1.18 1.00 1.39 * 1.48 1.27 1.71 *** 1.62 1.30 2.03 ***
M7: Depression/anxiety (PHQ-4; ref = normal)
 Mild 1.00 2.41 1.93 3.02 *** 1.73 1.46 2.04 *** 1.40 1.19 1.66 *** 1.39 1.07 1.81 *
 Moderate 1.00 6.18 4.56 8.38 *** 2.40 1.79 3.21 *** 2.66 2.04 3.47 *** 3.18 2.20 4.60 ***
 Severe 1.00 11.35 7.31 17.61 *** 3.51 2.22 5.56 *** 2.59 1.61 4.18 *** 5.68 3.35 9.63 ***
M8: Age (year entered NHATS sample; ref = 65–69)
 70–74 1.00 1.79 1.10 2.94 * 2.08 1.57 2.74 *** 1.63 1.31 2.03 *** 1.98 1.31 3.00 ***
 75–79 1.00 4.60 2.96 7.17 *** 3.19 2.43 4.18 *** 2.33 1.88 2.90 *** 3.52 2.37 5.22 ***
 80–84 1.00 9.36 6.10 14.36 *** 5.63 4.32 7.34 *** 3.37 2.70 4.21 *** 5.80 3.94 8.56 ***
 85–89 1.00 24.12 15.57 37.35 *** 10.93 8.17 14.61 *** 4.54 3.48 5.93 *** 8.52 5.53 13.11 ***
 90+ 1.00 46.92 29.31 75.10 *** 19.19 13.68 26.93 *** 6.15 4.38 8.65 *** 17.64 11.04 28.17 ***
M9: Born outside the United States
1.00 1.98 1.45 2.69 *** 2.19 1.74 2.76 *** 2.31 1.85 2.87 *** 2.79 2.06 3.78 ***
M10: Male
1.00 0.85 0.70 1.03 n.s. 0.93 0.81 1.07 n.s. 1.21 1.06 1.39 ** 1.02 0.83 1.26 n.s.
M11: Proxy respondent (any wave)
1.00 31.40 25.03 39.39 *** 7.48 6.30 8.89 *** 2.95 2.44 3.55 *** 11.18 8.91 14.01 ***
Open in a new tab

Notes: BA/more = Bachelor’s degree or higher; CI = confidence interval; HS = high school; NHATS = National Health and Aging Trends Study; NHB = non-Hispanic Black; NHW = non-Hispanic White; RRR = relative risk ratio.

*p < .05, **p < 0.01, ***p < .0001, n.s.= not significant.

In multivariable multinomial logistic regression models (Table 4), identifying as non-Hispanic Black and measures of lower socioeconomic status remained significantly associated with unexpected variation in dementia classification (vs. consistently no dementia). Non-Hispanic Black persons continued to have higher risk of unexpected variation versus consistently no dementia classification compared to non-Hispanic whites (adjusted risk ratio [ARR]: 2.12, 95% CI: 1.61–2.78, p < .0001). Compared with having less than a high school education, having a high school diploma or more education as associated with 52–63% lower risk of unexpected variation versus consistently no dementia (high school ARR: 0.48, 95% CI: 0.335–0.64, p < .0001; Bachelor’s or more ARR: 0.37, 95% CI: 0.26–0.53, p < .0001). Being Medicare–Medicaid dual enrolled was associated with 161% higher risk of unexpected variation versus consistently no dementia (Medicaid ARR: 2.6104, 95% CI: 2.01–3.39, p < .0001). Heart disease was associated with 30% lower risk of unexpected variation versus consistently no dementia (ARR: 0.70, 95% CI: 0.51–0.95, p = .021). Diabetes was associated with 35% higher risk of unexpected variation versus consistently no dementia (ARR: 1.35, 95% CI: 1.05–1.73, p = .019). Compared with normal depression, moderate and severe anxiety were associated with 145% and 229% higher risk of unexpected variation versus consistently no dementia (moderate ARR: 2.45, 95% CI 1.63–3.68, p < .0001; severe ARR: 3.29, 95% CI: 1.80–6.00, p < .0001).

Table 4.

Adjusted Risk Ratios and 95% Confidence Intervals for Multivariable Multinomial Logistic Regression of Sociodemographic Factors on Dementia Classification Over Time for 11,218 Medicare Beneficiaries Age 65 and Older, 2011–2017 NHATS

Consistent no dementia (ref) Consistent some cognitive impairment Dementia progression Anomalous-expected variation Anomalous-unexpected variation
ARR ARR 95% CI p value ARR 95% CI p value ARR 95% CI p value ARR 95% CI p value
Race/ethnicity (ref = NHW)
 NHB 1.00 2.59 1.97 3.40 *** 1.77 1.45 2.17 *** 1.99 1.66 2.38 *** 2.12 1.61 2.78 ***
 Hispanic 1.00 3.73 2.29 6.07 *** 2.15 1.49 3.10 * 1.66 1.17 2.35 ** 1.47 0.88 2.45 n.s.
Education (ref=less than HS)
 HS 1.00 0.52 0.39 0.69 *** 0.59 0.48 0.74 n.s. 0.58 0.47 0.71 *** 0.48 0.35 0.64 ***
 Some college 1.00 0.42 0.30 0.58 *** 0.55 0.44 0.69 n.s. 0.51 0.41 0.63 *** 0.41 0.29 0.57 ***
 BA/more 1.00 0.31 0.22 0.45 *** 0.44 0.34 0.56 n.s. 0.32 0.26 0.41 *** 0.37 0.26 0.53 ***
Received Medicaid (any wave)
1.00 1.84 1.42 2.38 *** 1.59 1.31 1.93 *** 1.52 1.27 1.82 *** 2.61 2.01 3.39 ***
Heart disease diagnosis (first wave)
1.00 1.07 0.82 1.41 n.s. 0.87 0.70 1.07 *** 0.92 0.76 1.12 n.s. 0.70 0.51 0.95 *
Hypertension diagnosis (first wave)
1.00 0.77 0.60 0.98 * 0.92 0.78 1.09 *** 0.98 0.83 1.15 n.s. 0.90 0.70 1.16 n.s.
Diabetes diagnosis (first wave)
1.00 1.24 0.96 1.59 n.s. 1.07 0.89 1.29 *** 1.23 1.04 1.45 * 1.35 1.05 1.73 *
Depression/ Anxiety (PHQ-4; first wave; ref = normal)
 Mild 1.00 1.89 1.45 2.45 *** 1.48 1.23 1.79 *** 1.21 1.02 1.45 * 1.10 0.83 1.45 n.s.
 Moderate 1.00 5.23 3.60 7.59 *** 2.09 1.52 2.88 *** 2.15 1.61 2.88 ** 2.45 1.63 3.68 ***
 Severe 1.00 6.47 3.71 11.26 *** 2.54 1.52 4.25 *** 2.04 1.22 3.41 *** 3.29 1.80 6.00 ***
Age (year entered NHATS sample; ref = 65–69)
 70–74 1.00 1.75 1.04 2.96 * 2.03 1.52 2.70 *** 1.60 1.27 2.02 *** 1.88 1.22 2.89 **
 75–79 1.00 3.96 2.45 6.40 *** 2.95 2.23 3.92 *** 2.29 1.82 2.88 *** 3.08 2.03 4.65 ***
 80–84 1.00 7.73 4.84 12.36 *** 5.28 3.99 6.98 *** 3.63 2.87 4.59 *** 5.47 3.62 8.25 ***
 85–89 1.00 17.05 10.48 27.76 *** 9.43 6.92 12.86 *** 4.92 3.70 6.53 *** 7.03 4.43 11.16 ***
 90+ 1.00 25.68 15.05 43.80 *** 14.12 9.77 20.43 *** 6.65 4.62 9.57 *** 12.54 7.49 20.99 ***
Born outside the United States
1.00 0.83 0.54 1.29 n.s. 1.40 1.04 1.90 *** 1.61 1.22 2.13 ** 1.77 1.19 2.62 **
Male
1.00 1.27 1.01 1.60 * 1.21 1.03 1.43 * 1.53 1.32 1.78 *** 1.36 1.08 1.72 **
Proxy respondent (any wave)
1.00 16.02 12.50 20.53 *** 4.43 3.67 5.35 *** 1.96 1.60 2.40 *** 6.58 5.14 8.43 ***
Model fit statistics
Log likelihood −7,200.64 McFadden’s R2 .172 AIC/BIC 14,569.28/15,147.57
Open in a new tab

Notes: AIC = Akaike information criterion; ARR = adjusted risk ratio; BA/more = Bachelor’s degree or higher; BIC = Bayesian information criterion; CI = confidence interval; HS = high school; NHATS = National Health and Aging Trends Study; NHB = non-Hispanic Black; NHW = non-Hispanic White.

*p < .05, **p < 0.01, ***p < .0001, n.s.= not significant.

Results for expected variation versus consistently no dementia were similar to unexpected variation, except Hispanic persons and those with mild depression had significantly higher risk of expected variation, and heart disease was not significant. All sociodemographic and health factors except education were significantly associated with dementia progression versus consistently no dementia. All sociodemographic and health factors were significantly associated with consistently some cognitive impairment versus consistently no dementia except heart disease, and diabetes.

Discussion

We identified five dementia classification trajectories over seven years (2011–2017) in nationally representative sample of non-Hispanic White, non-Hispanic Black, and Hispanic older adults. Four categories represented expected patterns in dementia classification. Two-thirds of respondents consistently reported the same classification for each wave of data: 59.5% always had no dementia, 7% always had the same degree of cognitive impairment (probable [6%] or possible dementia [1%]). Another 13% reported “typical” dementia progression, moving along the continuum from no to probable dementia. About 15% had anomalous, but unremarkable, patterns in their dementia classification, moving between “no and possible” or “possible and probable” dementia at each wave of data collection. The remaining 5.5% of NHATS participants reported an unexpected variation in dementia progression over time, moving from probable dementia in one wave to no dementia in subsequent years (3%) or from possible to no to probable dementia (2.5%). Significantly, unexpected variation in dementia classification was not equally distributed throughout the older adult population.

Sociodemographic factors, including identifying as non-Hispanic Black and markers of lower SES (having less education, being Medicare–Medicaid dual enrolled), were associated with higher risk of unexpected variation in dementia classification. We observed this association independent of health and other risk factors for dementia such as heart disease, diabetes, elevated depression, and older age. Our findings have important implications for how dementia classification is measured in survey data, what that means for our understanding of dementia progression in the population over time, and how current measurement efforts may both reflect, and potentially exacerbate, dementia-related health disparities.

To a certain extent, the observed “expected” variation in dementia classification reflects the nature of cognitive decline. Dementia is progressive, but not necessarily linear, with individuals experiencing dips and peaks in cognitive function that may be suggestive of, but not definitively, dementia. These dips and peaks are more common in earlier stages of dementia, when it is more difficult to diagnose the condition, and may be reflected in differing classifications over time, particularly for individuals for whom dementia varied between two adjacent classifications, such as when an individual moved from possible dementia 1 year to no dementia a subsequent year or moved from probable dementia 1 year to possible dementia another year. Alternatively, the variation could be the result of slight fluctuations in cognitive assessment scores for individuals who hovered near the assessment tests’ cut points.

Although some fluctuation in dementia classification may be expected (Gao et al., 2014; Roberts et al., 2014), it is less clear why some individuals reported more substantial variation in dementia classification from year to year than others. For example, individuals in the “unexpected” variation category were classified as having probable dementia 1 year and no dementia the next. Observed patterns in changes in dementia classification for these individuals are not easily explained by disease progression or small fluctuations in cognitive assessment scores. Moving from probable to no dementia requires a substantial (1.5 SD) improvement in two or three cognitive assessment domains of memory, orientation, and executive functioning. Moving across all three classification categories—possible to no to probable dementia—in as many years requires similar substantial changes in at least one cognitive assessment domain. Given dementia is a progressive disease in which cognitive impairment rarely, if ever, improves, these substantial and inconsistent fluctuations are unexpected.

This susceptibility to unexpected variation may be heightened among adults aged 70 and older who are at increased risk of dementia (De Reuck et al., 2018; James et al., 2012), many of whom may begin to experience declines in cognitive function but are not yet diagnosed as having dementia. Alternatively, unexpected variation may be attributable to transient events that could temporarily affect performance on cognitive assessments, such as changes in medication regimen or infections (e.g., urinary tract infections), which may be more poorly managed in populations with less contact with the health care system due to structural factors that hinder or discourage access (e.g., lack of insurance coverage, experiences with racial discrimination in health care settings), such as persons with fewer economic resources and from racial/ethnic minoritized groups (Shiota et al., 2014). Increased unexpected variation may also be capturing initial changes in cognitive functioning that later present as greater prevalence of dementia among non-Hispanic Black persons and those with less education (Alzheimer’s Association, 2021). Increased risk of unexpected variation classification among non-Hispanic Black and Hispanic individuals supports prior findings of cultural bias in measurement by health care professionals and in the measurement tools used, depending on test and administration method, that may also apply to the assessment tools used in NHATS (Manly et al., 2011; Ramirez et al., 2006; Sloan & Wang, 2005; Zsembik & Peek, 2001). For example, differences in health literacy may present obstacles to accurate cognitive assessment and assessor bias may result in different ratings among individuals with lower SES.

In NHATS, individuals who report being told by their physician that they have dementia are automatically flagged as having “probable” dementia in that and all subsequent years of data collection. As such, the cognitive assessment tests provide a picture of cognitive functioning for individuals who are not diagnosed by a physician. Systematically reduced access to primary care among Black and Hispanic individuals and those with lower levels of SES may lead to even greater dependence on cognitive assessments among these segments of the NHATS sample (Arnett et al., 2016; Kangovi et al., 2013; Kleinman et al., 1981; Ryvicker & Russell, 2018). Our finding that 1 in 20 older adults experiences an unexpected variation in dementia classification over time, with greater risk among non-Hispanic Black and Hispanic individuals and those with lower SES, exposes potential uncertainty in dementia trajectories for these individuals. This uncertainty may affect individuals’ ability to plan for and activate support and other resources that will be needed as their cognitive impairment emerges and progresses. The fact that these individuals are not diagnosed by a physician may be a reflection of structural barriers to consistent access to health care, potentially putting them at increased risk of poor dementia management and insufficient caregiving support, with higher risk for negative consequences among non-Hispanic Black and Hispanic individuals and those with lower SES.

Additionally, the existence of this “unexpected variation” category suggests the cognitive assessment tests used in NHATS may not accurately capture cognitive functioning over time in a national sample, particularly for non-Hispanic black individuals and those with less than a high school education. Unexpected variation in dementia classification may be attributable to measurement error in the tools used to assess cognitive functioning, as there can be variation in sensitivity and specificity of cognitive assessment tools (Snyder et al., 2011). Although the NHATS dementia classification is comprehensive and cognitive assessment tools are often tested with older adults (Feeney et al., 2016; Hunter et al., 2021; Kasper et al., 2013; Martin-Khan et al., 2010), it may be harder to reliably assess older adults for cognitive impairment and functioning. Our finding that non-Hispanic Black individuals and those with lower SES experience more unexpected variation highlights a need to better understand potential additional implicit biases in assessment tools that may result in greater challenges in accurate, reliable dementia classification and diagnosis in these groups. Cognitive assessment tools that are validated in socioeconomically, culturally, and linguistically diverse samples should be developed and incorporated into large surveys. Future research should examine the progression of dementia among individuals in the “unexpected variation” category to determine how their caregiving needs emerge and are addressed. Policy makers should consider resources to support the needs of this group of older adults for whom dementia emergence, progression, and accompanying needs may be particularly difficult to predict.

Limitations

This study has limitations. Responses to cognitive assessments relied on survey participant and/or proxy reports and had not been verified with information from medical records. However, the assessment tools used in NHATS are multidimensional and consistent with those in other large-scale surveys. Second, we do not have information about respondents’ stage of dementia to determine whether these stages align with NHATS’ dementia classification categories. Third, the NHATS sample includes Medicare beneficiaries only, precluding us from examining the relationship between dementia classification and receiving Medicaid without Medicare among older adults. Older adults receiving Medicaid without Medicare may differ along key social factors (Freedman et al., 2016). However, 96% of individuals aged 65 and older in the United States are enrolled in Medicare, making the potential proportion of individuals over age 65 who may receive Medicaid without Medicare is relatively small (Freedman et al., 2016). Nonetheless, additional research is needed to understand how dementia classification may vary among older adults who receive Medicaid, but not Medicare. Our use of the number of years of education and the three racial/ethnic categories provide by NHATS does not allow us to account for the heterogeneous lived experiences of the individuals who fall within each category as sociodemographic factors are imperfect proxies for social experience. This is an unfortunate methodological limitation of many surveys and quantitative analyses of survey data. Small sample sizes of groups within the Hispanic category preclude a more granular analysis. NHATS does not collect information about education quality or experience with perceived discrimination which may help explain variation in dementia classification. Future studies should assess health literacy, quality of education in addition to amount, perceived discrimination, and more granular racial and ethnic groups to allow for heterogeneity within these broad categories and their potential differential associations with dementia classification to be identified.

Conclusion

Analyzing dementia classification over time in NHATS offers a new way of understanding dementia-related outcomes that focuses on the emergence and progression of dementia in a national, population-based sample of older adults. Unexpected variation in dementia classification cannot be explained as typical of dementia progression or small degrees of measurement error and may be due to undetected bias in assessment tests. Given the uncertainty and unpredictability in unexpected variation in dementia classification, non-Hispanic Black persons and those with lower levels of SES, who have less access to health-promoting resources that facilitate disease management, may have additional difficulties planning for and activating resources to effectively manage support and caregiving needs that emerge as dementia progresses. Future research should examine the relationship between SES, race/ethnicity, dementia classification, and subsequent caregiving needs and burden, particularly among persons with unexpected variation in dementia classification. Policy makers may want to consider the needs of individuals with unexpected variation in dementia classification when planning for future disease, disability burden, and caregiving needs. Our findings suggest that clinicians should take extra care in assessing dementia in patient populations at increased risk of unexpected dementia classification, as there may be fluctuations in cognitive status or bias in assessment tools.

Contributor Information

Elizabeth A Luth, Institute for Health, Healthcare Policy and Aging Research, Department of Family Medicine and Community Health, Rutgers University, New Brunswick, New Jersy, USA.

Holly G Prigerson, Division of Geriatrics and Palliative Medicine, Weill Cornell Medicine, New York, New York, USA.

Funding

This work was supported by grants from the National Institute on Aging (AG065624; E. A. Luth) and the National Cancer Institute (CA197730; H. G. Prigerson).

Author Contributions

Dr. Luth planned the study, conducted data analysis, and drafted the manuscript. Dr. Prigerson advised on study framing and provided input into manuscript.

Conflict of Interest

None declared.

References

  1. Abner, E. L., Kryscio, R. J., Cooper, G. E., Fardo, D. W., Jicha, G. A., Mendiondo, M. S., Nelson, P. T., Smith, C. D., Van Eldik, L. J., Wan, L., & Schmitt, F. A. (2012). Mild cognitive impairment: Statistical models of transition using longitudinal clinical data. International Journal of Alzheimer’s Disease, 2012, 291920. doi: 10.1155/2012/291920 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Alzheimer’s Association. (2021). 2021 Alzheimer’s disease facts and figures. Alzheimer’s & Dementia, 17(3), 327–406. doi: 10.1002/alz.12328 [DOI] [PubMed] [Google Scholar]
  3. Arnett, M. J., Thorpe, R. J., Gaskin, D. J., Bowie, J. V., & LaVeist, T. A. (2016). Race, medical mistrust, and segregation in primary care as usual source of care: Findings from the exploring health disparities in integrated communities study. Journal of Urban Health, 93(3), 456–467. doi: 10.1007/s11524-016-0054-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bailey, Z. D., Feldman, J. M., & Bassett, M. T. (2021). How structural racism works—Racist policies as a root cause of U.S. racial health inequities. New England Journal of Medicine, 384(8), 768–773. doi: 10.1056/NEJMms2025396 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barnes, L. L., Lewis, T. T., Begeny, C. T., Yu, L., Bennett, D. A., & Wilson, R. S. (2012). Perceived discrimination and cognition in older African Americans. Journal of the International Neuropsychological Society, 18(5), 856–865. doi: 10.1017/S1355617712000628 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chen, C., Thunell, J., & Zissimopoulos, J. (2020). Changes in physical and mental health of Black, Hispanic, and White caregivers and non-caregivers associated with onset of spousal dementia. Alzheimer’s & Dementia, 6(1), e12082. doi: 10.1002/trc2.12082 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chêne, G., Beiser, A., Au, R., Preis, S. R., Wolf, P. A., Dufouil, C., & Seshadri, S. (2015). Gender and incidence of dementia in the Framingham Heart Study from mid-adult life. Alzheimer’s & Dementia, 11(3), 310–320. doi: 10.1016/j.jalz.2013.10.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Coogan, P., Schon, K., Li, S., Cozier, Y., Bethea, T., & Rosenberg, L. (2020). Experiences of racism and subjective cognitive function in African American women. Alzheimer’s & Dementia, 12(1), e12067. doi: 10.1002/dad2.12067 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Crenshaw, K. (1989). Demarginalizing the intersection of race and sex: A black feminist critique of antidiscrimination doctrine, feminist theory and antiracist politics. University of Chicago of Legal Forum, 139, 139–167. [Google Scholar]
  10. DeNavas-Walt, C., & Proctor, B. D. (2015). Income and poverty in the United States: 2014 (Current Population Reports No. P60-252). United States Census Bureau. https://www.census.gov/content/dam/Census/library/publications/2015/demo/p60-252.pdf [Google Scholar]
  11. De Reuck, J., Maurage, C. -A., Deramecourt, V., Pasquier, F., Cordonnier, C., Leys, D., & Bordet, R. (2018). Aging and cerebrovascular lesions in pure and in mixed neurodegenerative and vascular dementia brains: A neuropathological study. Folia Neuropathologica, 56(2), 81–87. doi: 10.5114/fn.2018.76610 [DOI] [PubMed] [Google Scholar]
  12. Feeney, J., Savva, G. M., O’Regan, C., King-Kallimanis, B., Cronin, H., & Kenny, R. A. (2016). Measurement error, reliability, and minimum detectable change in the mini-mental state examination, montreal cognitive assessment, and color trails test among community living middle-aged and older adults. Journal of Alzheimer’s Disease, 53(3), 1107–1114. doi: 10.3233/JAD-160248 [DOI] [PubMed] [Google Scholar]
  13. Franco, Y., & Choi, E. Y. (2020). The relationship between immigrant status and undiagnosed dementia: The role of limited English proficiency. Journal of Immigrant and Minority Health, 22(5), 914–922. doi: 10.1007/s10903-019-00963-w [DOI] [PubMed] [Google Scholar]
  14. Freedman, V. A., Kasper, J. D., Spillman, B. C., & Plassman, B. L. (2018). Short-term changes in the prevalence of probable dementia: An analysis of the 2011–2015 national health and aging trends study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 73(Suppl 1), S48–S56. doi: 10.1093/geronb/gbx144 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Freedman, V. A., Spillman, B. C., & Kasper, J. D. (2016). Making National Estimates with the National Health and Aging Trends Study (NHATS Technical Paper #17). Johns Hopkins University School of Public Health. https://nhats.org/sites/default/files/inline-files/Making_National_Estimates_in_NHATS_Technical_Paper_2018_0.pdf [Google Scholar]
  16. Gao, S., Unverzagt, F. W., Hall, K. S., Lane, K. A., Murrell, J. R., Hake, A. M., Smith-Gamble, V., & Hendrie, H. C. (2014). Mild cognitive impairment, incidence, progression, and reversion: Findings from a community-based cohort of elderly African Americans. Aging, Mental Health, and Ethnicity, 22(7), 670–681. doi: 10.1016/j.jagp.2013.02.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Glymour, M. M., & Manly, J. J. (2008). Lifecourse social conditions and racial and ethnic patterns of cognitive aging. Neuropsychology Review, 18(3), 223–254. doi: 10.1007/s11065-008-9064-z [DOI] [PubMed] [Google Scholar]
  18. Gudala, K., Bansal, D., Schifano, F., & Bhansali, A. (2013). Diabetes mellitus and risk of dementia: A meta-analysis of prospective observational studies. Journal of Diabetes Investigation, 4(6), 640–50. doi: 10.1111/jdi.12087 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hulko, W. (2004). Social science perspectives on dementia research: Intersectionality. In A. Innes, C. Archibald & C. Murphy (Eds.), Dementia and Social Inclusion: Marginalised Groups and Marginalised Areas of Dementia Research, Care, and Practice (pp. 237–254). [Google Scholar]
  20. Hunter, M. B., Jenkins, N., Dolan, C., Pullen, H., Ritchie, C., & Muniz-Terrera, G. (2021). Reliability of telephone and videoconference methods of cognitive assessment in older adults with and without dementia. Journal of Alzheimer’s Disease, 81(4), 1625–1647. doi: 10.3233/JAD-210088 [DOI] [PubMed] [Google Scholar]
  21. James, B. D., Bennett, D. A., Boyle, P. A., Leurgans, S., & Schneider, J. A. (2012). Dementia from Alzheimer disease and mixed pathologies in the oldest old. JAMA, 307(17), 1798–1800. doi: 10.1001/jama.2012.3556 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kangovi, S., Barg, F. K., Carter, T., Long, J. A., Shannon, R., & Grande, D. (2013). Understanding why patients of low socioeconomic status prefer hospitals over ambulatory care. Health Affairs, 32(7), 1196–1203. doi: 10.1377/hlthaff.2012.0825 [DOI] [PubMed] [Google Scholar]
  23. Kasper, J. D., Freedman, V. A., Spillman, B. C., & Skehan, M. E. (2013). Classification of persons by dementia status in the national health and aging trends study (Technical paper #5). Johns Hopkins University School of Public Health. [Google Scholar]
  24. Kleinman, J. C., Gold, M., & Makuc, D. (1981). Use of ambulatory medical care by the poor: Another look at equity. Medical Care, 19(10), 1011–1029. https://www.jstor.org/stable/3764367 [DOI] [PubMed] [Google Scholar]
  25. Kohler, A. D., & Lazarín, M. (2007). Hispanic education in the United States. National Council of La Raza, Statistical Brief, 8(1), 7–8. http://hdl.handle.net/10919/83669 [Google Scholar]
  26. Lines, L. W., & Wiener, J. M. (2014). Racial and ethnic disparities in Alzheimer’s disease: A literature review (A. and L.-T. C. P. Office of Disability, Trans.; RR-0024-1412). https://aspe.hhs.gov/report/racial-and-ethnic-disparities-alzheimers-disease-literature-review
  27. Link, B. G., & Phelan, J. (1995). Social conditions as fundamental causes of disease. Journal of Health and Social Behavior, 35, 80–94. doi: 10.2307/2626958 [DOI] [PubMed] [Google Scholar]
  28. Liu, R., Chi, I., & Wu, S. (2022). Caregiving Burden Among Caregivers of People With Dementia Through the Lens of Intersectionality. The Gerontologist, 62(5), 650–661. doi: 10.1093/geront/gnab146 [DOI] [PubMed] [Google Scholar]
  29. Livingston, G., Huntley, J., Sommerlad, A., Ames, D., Ballard, C., Banerjee, S., Brayne, C., Burns, A., Cohen-Mansfield, J., Cooper, C., Costafreda, S. G., Dias, A., Fox, N., Gitlin, L. N., Howard, R., Kales, H. C., Kivimäki, M., Larson, E. B., Ogunniyi, A., & Mukadam, N. (2020). Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. The Lancet, 396(10248), 413–446. doi: 10.1016/S0140-6736(20)30367-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Majoka, M. A., & Schimming, C. (2021). Effect of social determinants of health on cognition and risk of Alzheimer disease and related dementias. Clinical Therapeutics, 43(6), 922–929. doi: 10.1016/j.clinthera.2021.05.005 [DOI] [PubMed] [Google Scholar]
  31. Manly, J. J., Schupf, N., Stern, Y., Brickman, A. M., Tang, M.-X., & Mayeux, R. (2011). Telephone-based identification of mild cognitive impairment and dementia in a multicultural cohort. Archives of Neurology, 68(5), 607–614. doi: 10.1001/archneurol.2011.88 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Martin-Khan, M., Wootton, R., & Gray, L. (2010). A systematic review of the reliability of screening for cognitive impairment in older adults by use of standardised assessment tools administered via the telephone. Journal of Telemedicine and Telecare, 16(8), 422–428. doi: 10.1258/jtt.2010.100209 [DOI] [PubMed] [Google Scholar]
  33. McDonough, I. M., Harrell, E. R., Black, S. R., Allen, R. S., & Parmelee, P. A. (2022). Sources of nonreplicability in aging ethnoracial health disparities research. Psychology and Aging, 37(1), 60–71. doi: 10.1037/pag0000606 [DOI] [PubMed] [Google Scholar]
  34. Meng, X., & D’Arcy, C. (2012). Education and dementia in the context of the cognitive reserve hypothesis: A systematic review with meta-analyses and qualitative analyses. PLoS One, 7(6), e38268. doi: 10.1371/journal.pone.0038268 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Moon, H., Badana, A. N. S., Hwang, S. -Y., Sears, J. S., & Haley, W. E. (2019). Dementia prevalence in older adults: Variation by race/ethnicity and immigrant status. The American Journal of Geriatric Psychiatry, 27(3), 241–250. doi: 10.1016/j.jagp.2018.11.003 [DOI] [PubMed] [Google Scholar]
  36. Mungas, D., Beckett, L., Harvey, D., Farias, S. T., Reed, B., Carmichael, O., Olichney, J., Miller, J., & DeCarli, C. (2010). Heterogeneity of cognitive trajectories in diverse older persons. Psychology and Aging, 25(3), 606–619. PubMed. doi: 10.1037/a0019502 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Phelan, J. C., & Link, B. G. (2005). Controlling disease and creating disparities: A fundamental cause perspective. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 60(Special Issue 2), S27–S33. doi: 10.1093/geronb/60.Special_Issue_2.S27 [DOI] [PubMed] [Google Scholar]
  38. Phelan, J. C., & Link, B. G. (2015). Is Racism a Fundamental Cause of Inequalities in Health? Annual Review of Sociology, 41(1), 311–330. doi: 10.1093/geronb/60.Special_Issue_2.S27 [DOI] [Google Scholar]
  39. Pool, L. R., Weuve, J., Wilson, R. S., Bültmann, U., Evans, D. A., & Mendes de Leon, C. F. (2016). Occupational cognitive requirements and late-life cognitive aging. Neurology, 86(15), 1386–1392. doi: 10.1212/WNL.0000000000002569 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Power, M. C., Bennett, E. E., Turner, R. W., Dowling, N. M., Ciarleglio, A., Glymour, M. M., & Gianattasio, K. Z. (2021). Trends in relative incidence and prevalence of dementia across non-Hispanic black and white individuals in the United States, 2000–2016. JAMA Neurology, 78(3), 275–284. doi: 10.1001/jamaneurol.2020.4471 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Rajan, K. B., Weuve, J., Barnes, L. L., McAninch, E. A., Wilson, R. S., & Evans, D. A. (2021). Population estimate of people with clinical Alzheimer’s disease and mild cognitive impairment in the United States (2020–2060). Alzheimer’s & Dementia, 17(12), 1966–1975. doi: 10.1002/alz.12362 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ramirez, M., Teresi, J. A., Holmes, D., Gurland, B., & Lantigua, R. (2006). Differential Item Functioning (DIF) and the Mini-Mental State Examination (MMSE): Overview, Sample, and Issues of Translation. Medical Care, 44(11), S95–S106. [DOI] [PubMed] [Google Scholar]
  43. Roberts, R. O., Knopman, D. S., Mielke, M. M., Cha, R. H., Pankratz, V. S., Christianson, T. J. H., Geda, Y. E., Boeve, B. F., Ivnik, R. J., Tangalos, E. G., Rocca, W. A., & Petersen, R. C. (2014). Higher risk of progression to dementia in mild cognitive impairment cases who revert to normal. Neurology, 82(4), 317–325. doi: 10.1212/WNL.0000000000000055 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Roe, C. M., Xiong, C., Miller, J. P., & Morris, J. C. (2007). Education and Alzheimer disease without dementia. Neurology, 68(3), 223. doi: 10.1212/01.wnl.0000251303.50459.8a [DOI] [PubMed] [Google Scholar]
  45. Ryvicker, M., & Russell, D. (2018). Individual and environmental determinants of provider continuity among Urban older adults with heart failure: A retrospective cohort study. Gerontology and Geriatric Medicine, 4. doi: 10.1177/2333721418801027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Samieri, C., Perier, M. -C., Gaye, B., Proust-Lima, C., Helmer, C., Dartigues, J.-F., Berr, C., Tzourio, C., & Empana, J.-P. (2018). Association of cardiovascular health level in older age with cognitive decline and incident dementia. JAMA, 320(7), 657–664. doi: 10.1001/jama.2018.11499 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sayegh, P., & Knight, B. G. (2013). Cross-cultural differences in dementia: The sociocultural health belief model. International Psychogeriatrics, 25(4), 517–530. doi: 10.1017/S104161021200213X [DOI] [PubMed] [Google Scholar]
  48. Shiota, T., Torimoto, K., Momose, H., Nakamuro, T., Mochizuki, H., Kumamoto, H., Hirayama, A., & Fujimoto, K. (2014). Temporary cognitive impairment related to administration of newly developed anticholinergic medicines for overactive bladder: Two case reports. BMC Research Notes, 7(1), 672. doi: 10.1186/1756-0500-7-672 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Sloan, F. A., & Wang, J. (2005). Disparities among older adults in measures of cognitive function by race or ethnicity. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 60(5), P242–P250. doi: 10.1093/geronb/60.5.P242 [DOI] [PubMed] [Google Scholar]
  50. Snyder, P. J., Jackson, C. E., Petersen, R. C., Khachaturian, A. S., Kaye, J., Albert, M. S., & Weintraub, S. (2011). Assessment of cognition in mild cognitive impairment: A comparative study. Alzheimer’s & Dementia, 7(3), 338–355. doi: 10.1016/j.jalz.2011.03.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Steinerman, J. R., Hall, C. B., Sliwinski, M. J., & Lipton, R. B. (2010). Modeling cognitive trajectories within longitudinal studies: a focus on older adults. Journal of the American Geriatrics Society, 58(Suppl. 2), S313–S318. doi: 10.1111/j.1532-5415.2010.02982.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Stites, S. D., Midgett, S., Mechanic-Hamilton, D., Zuelsdorff, M., Glover, C. M., Marquez, D. X., Balls-Berry, J. E., Streitz, M. L., Babulal, G., Trani, J. -F., Henderson, J. N., Barnes, L. L., Karlawish, J., & Wolk, D. A. (2021). Establishing a framework for gathering structural and social determinants of health in Alzheimer’s Disease research centers. The Gerontologist, 62(5), 694–703. doi: 10.1093/geront/gnab182 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Sullivan, S. S., Li, C.-S., de Rosa, C., & Chang, Y.-P. (2022). Development of a longitudinal dataset of persons with dementia and their caregivers through end-of-life: A statistical analysis system algorithm for joining national health and aging trends study/national study of caregiving. American Journal of Hospice and Palliative Medicine, 39(9), 1052–1060. doi: 10.1177/10499091211057291 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Vega, I. E., Cabrera, L. Y., Wygant, C. M., Velez-Ortiz, D., & Counts, S. E. (2017). Alzheimer’s disease in the Latino community: Intersection of genetics and social determinants of health. Journal of Alzheimer’s Disease, 58(4), 979–992. doi: 10.3233/JAD-161261 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Weber, L. (2006). Reconstructing the landscape of health disparities research: Promoting dialogue and collaboration between feminist intersectional and positivist biomedical traditions. In A. J. Schulz & L. Mullings (Eds.), Gender, Race, Class, and Health: Intersectional Approaches (pp. 21–59). [Google Scholar]
  56. Williams, D. R., Priest, N., & Anderson, N. B. (2016). Understanding associations among race, socioeconomic status, and health: Patterns and prospects. Health Psychology, 35(4), 407–411. doi: 10.1037/hea0000242 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Zahodne, L. B., Glymour, M. M., Sparks, C., Bontempo, D., Dixon, R. A., MacDonald, S. W. S., & Manly, J. J. (2011). Education does not slow cognitive decline with aging: 12-year evidence from the victoria longitudinal study. Journal of the International Neuropsychological Society, 17(6), 1039–1046. doi: 10.1017/S1355617711001044 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Zissimopoulos, J. M., Tysinger, B. C., St.Clair, P. A., & Crimmins, E. M. (2018). The impact of changes in population health and mortality on future prevalence of Alzheimer’s disease and other dementias in the United States. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 73(Suppl 1), S38–S47. doi: 10.1093/geronb/gbx147 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Zsembik, B. A., & Peek, M. K. (2001). Race differences in cognitive functioning among older adults. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 56(5), S266–S274. doi: 10.1093/geronb/56.5.S266 [DOI] [PubMed] [Google Scholar]

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES