The ACTIVE Study: Association of Race and Social Determinants of Health (SDoH) with Long-Term Outcomes and Cognitive Training Effects

Background of ACTIVE Project

The Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) study is the largest randomized clinical trial of the efficacy of cognitive training in improving both cognitive abilities and everyday functioning in normal older adults (Ball et al., 2002; Jobe et al., 2001; Rebok et al., 2014; Willis et al., 2006). It is unique in 1) focusing on cognitively normal older adults “at risk of loss of independence” and thus most vulnerable to health, functional, and cognitive deficits; 2) including participants in six geographical locations in the United States (US); 3) comparing three distinct cognitive training programs with a control group; and 4) oversampling African Americans (AA) who have twice the risk of Alzheimer’s disease (AD) as non-Hispanic Whites. The Alzheimer’s Association calls this disparity a ‘silent epidemic’ and projects that it will increase over the next 30 years (Alzheimer’s Association, 2010).

The ACTIVE Trial

Major Phases in the ACTIVE Trial and Conceptual Model

ACTIVE began in September 1996 at six field sites: University of Alabama at Birmingham, Boston Hebrew Rehabilitation Center for Aged (now Hebrew SeniorLife), Indiana University School of Medicine, Johns Hopkins University, Pennsylvania State University, and Wayne State University, with a data-coordinating center at the New England Research Institutes. Eligible community-dwelling individuals aged 65 and older were recruited from community centers, senior housing, churches, hospitals and clinics, and various registries and rosters (e.g., state driver’s license and identification card registry rosters of assistance/service programs for low-income older adults) between March 1998 and October 1999. The ACTIVE sample was not intended to be representative of the US population. Compared to the US population at the time of recruitment, the ACTIVE participants were slightly younger than the US population aged 65 and above and were more likely to be female and not married. As a result of the targeted efforts to oversample African Americans (AA), to our knowledge, this is the largest AA sample in any cognitive training study and follow-up (see Tzuang et al., 2018). At baseline the randomized ACTIVE sample (N = 2802) included 728 AA participants (26%). Eighty-seven percent (N = 473) of AA participants completed the training intervention (i.e., > 8 of 10 training sessions), 73% (N = 529) were assessed at 2-year follow-up; 61% (N = 441) were assessed at 5-year follow-up, and 53% (N = 383) at 10-year follow-up.

Figure 1 depicts the conceptual model that guided the design of the ACTIVE study. If the cognitive training program had positive effects on the mental ability that it targeted, then this effect should transfer or carry over to a positive effect on daily function. While it was assumed that each training program would affect only the mental ability that it focused on, it was expected that the effect on daily function would be less specific. That is, daily function could be improved by any or all three training interventions. It was further hypothesized that improvements in mental abilities and daily function would ultimately impact longer-term health-related quality of life (HRQoL), health service use, and mobility. Participant characteristics such as baseline cognitive status, sensory ability, disease, and demographics were included in the model as potential moderators of training effectiveness. In the ACTIVE 20-year follow-up, we are using administrative data in addition to ACTIVE data to expand our definition of participant characteristics and explore the five domains of SDoH as presented in Healthy People 2030 (Office of Disease Prevention and Health Promotion (n.d.)).OPEN IN VIEWER

Race and Social Determinants of Health in Relation to Cognition and Cognitive Training

One of the innovative aspects of ACTIVE is the multiple measures of SDoH (Mehta et al., 2004). Primary data from ACTIVE participants merged with their data from administrative sources were used to assess the five domains of SDoH described in the Healthy People 2030 Report (Office of Disease Prevention and Health Promotion (n.d.)): economic stability, education access and quality, health care access and quality, neighborhood and built environment, and social and community context (Clay et al., 2023). The following descriptions include measures assessed in the Clay et al. manuscript and additional measures available for analysis. Economic stability is measured by (1) occupational prestige, gathered from self-reported occupation data available in the ACTIVE study and coded according to the Bureau of Labor Statistics; and (2) annual credit score ratings provided by TransUnion. Education access and quality is measured by (1) self-reported years of education and, additionally by (2) self-reported cognitively stimulating activities, both of which come from the first phase of ACTIVE. Health care access and quality are measured by variables from the ACTIVE Phase 1 data including (1) self-reported health care visits, (2) health insurance, and (3) health care utilization. Neighborhood and built environment characteristics are measured by indicators of (1) neighborhood socioeconomic position (e.g., % of community employed), (2) median income, and (3) % below the poverty level) that come from the US Census. Finally, social and community context is measured by (1) marital status and (2) presence of a spouse, (3) number of co-residents and relations among members of a household, (4) preferred methods of mobility (e.g., self-driving, public transportation), and by (5) census tract measures including population density as provided by the US Census.

Contents of Two ACTIVE Special Issues

Manuscripts in the 2023 Special Issue

In this Special Issue, we report on baseline and longitudinal follow-up data related to race and SDoH through up to 20-years of follow-up. Although cognitive training has been shown to improve cognitive and everyday function in older adults, there is a large knowledge gap regarding the role of multiple domains of SDoH in understanding racial disparities in the efficacy of cognitive training. Across the papers in this special issue, the analyses focused on race or some/all of the five domains of SDoH proposed in Healthy People 2030. We explored whether participants’ levels of advantage across these five domains yielded differential training gains using ACTIVE survey data, supplemental neighborhood level Census data, North American Industry Classification System (NAICS) data, and Medicare claims data. We also addressed the association between race and SDoH regarding participant outcomes, comparing the trajectories in cognition when B/AA and White groups are demographically similar. Collectively, results from the papers in this special issue can help inform future generations of cognitive intervention research and the role of intervention regarding the significantly higher incidence of dementia in African American older adults.

Evaluating Social Determinants of Health Domains and Their Predictive Validity Within Black/African American and White Older Adults From the ACTIVE Trial

The paper by Clay and colleagues aimed to assess the five domains of SDoH described above and to evaluate their associations with cognition and HRQoL. They hypothesized that higher scores on SDoH measures are associated with higher levels of cognitive function and HRQoL. Baseline data from the ACTIVE cognitive training trial were used for this investigation. These data included speed of processing, memory, and reasoning composites. US Census data, NAICS data, and occupational status coding were merged with ACTIVE data and used as additional measures of SDoH. A principal components analysis with varimax rotation was used to examine the factor structure of the SDoH domains. The resulting uncorrelated factor scores were subsequently used in covariate-adjusted multiple regression models to assess the associations of SDoH with baseline cognitive function and HRQoL. Their results showed that data collected from the ACTIVE study and administrative sources can be used to evaluate the five SDoH domains as presented in Healthy People 2030 and the ability of higher scores on the composites to predict higher levels of baseline performance on measures of cognition and self-reported HRQoL within a sample of older adults. Further, higher SDoH domain scores were associated with better functioning on composite measures of cognition and on mental and physical HRQoL with the domain of Access to Health Care associated with all outcomes, a result highlighting the importance of access to health care within older B/AA and White older adults.

Indicators of Crash Risk in Older Adults: A Longitudinal Analysis From the ACTIVE Study

The Ball et al. paper examined cognitive function as a predictor of crashes in ACTIVE study control participants. The analyses included demographic variables (age, sex, race, education) and covariates as appropriate (depression, visual acuity, turn 360, MMSE, site location, and replicate), in addition to the cognitive composites as predictors of time to at-fault crashes. Racial differences were further explored with the SDoH variables. Older age, male sex, and site location were each predictive of higher crash risk. In addition, worse scores on the speed of processing composite were associated with higher crash risk. These results support previous findings that both older age and male sex predict higher crash risk.

Multiple cross-sectional studies have assessed the relationship between cognition and driving mobility within healthy older adults. Results have consistently shown a positive relationship between cognition and driving mobility, and these findings have been similar across multiple cognitive domains (Clay et al., 2005; Edwards et al., 2008; Pope et al., 2016). The ACTIVE study allows a unique opportunity to assess these relationships over time and in the context of racial differences and SDoH.

Social and Neighborhood Context Moderates the Associations Between Processing Speed and Driving Mobility: A 10-Year Analysis of the ACTIVE Study

The paper by Pope and colleagues examined measures of cognitive function at baseline, and over time, as predictors of driving mobility outcomes over 10 years within ACTIVE study participants in the control group (no training received). The results showed that decline in processing speed measures was associated with increased self-reported driving difficulty in older adults with below-average scores for neighborhood and built environments and social community context SDoH domains. These findings suggest individuals who are in neighborhood and built environment and social and community context settings that are below-average may be at a higher risk of the negative outcomes associated with declines in processing speed.

Effects of Cognitive Training on Alzheimer’s Disease and Related Dementias (ADRD): The Moderating Role of Social Determinants of Health

In addition to the utility of SDoH to predict cognitive scores, the inclusion of an intervention component in the ACTIVE study allowed examination of the impact of SDoH on cognitive training gains as measured by cognitive ability measures. As described above, participants assigned to the ACTIVE intervention groups demonstrated cognitive gains as measured by testing and by reduced rates of diagnosed Alzheimer’s and related dementias. However, prior work has not examined the extent to which training gains from ACTIVE depend on the SDoH that affect the study’s participants.

Rebok and his colleagues explored whether participants’ levels of functioning across the five SDoH domains yield differential training gains using ACTIVE survey data, supplemental Census data, and Medicare claims data. They hypothesized that a greater level of advantage as measured by SDoH factor scores would be associated with greater training gains and a lower rate of dementia diagnosis. They reported that higher scores on health care access and education access and quality were associated with higher ADRD risk. Trained participants in all three intervention conditions (memory, reasoning, speed of processing) obtained a greater degree of protection from ADRD when they had higher scores for the SDoH domain associated with health care access and quality.

Disparities in First IADL Difficulty Between Older Black and White Adults

Feger and colleagues studied disparities in first self-reported IADL difficulty between older Black and White adults. They analyzed data from ACTIVE participants who entered free of any self-reported IADL difficulties at baseline. Black older adults were significantly more likely to report travel as the first difficult IADL compared to White older adults (14% vs. 8%). This difference was explained by differences in current driving status by race at baseline and greater reliance on and less access to public transportation among Black older adults. Although many studies have reported disparities in the ability to perform IADLs in older Black and White adults, this is one of the first studies to examine differences by race in the ordering of incident IADL self-reported difficulty and may help explain why travel is more likely to become difficult for older Black adults.

Associations Between Body Mass Index (BMI) and Cognitive Change in the ACTIVE Study: Variations by Race and Social Determinants of Health

While African American adults have higher rates of both overweight/obesity and cognitive decline, little research has examined these relationships within this population. Results from the few longitudinal studies with diverse samples have been mixed; one study found lower BMI was related to faster cognitive decline (Arvanitakis et al., 2018) and higher dementia risk (Gao et al., 2011), while another study (Sturman et al., 2008) found that overweight and obese BMI was related to better cognitive performance, but for B/AA participants only.

The manuscript by Aiken-Morgan and her colleagues examined the relationships between BMI and cognitive performance/decline and race among the ACTIVE study participants. The aims were: 1) to examine how associations between BMI (baseline and change) and cognitive change (memory, reasoning, and speed) vary as a function of race, over the 10-year period in ACTIVE, and 2) to determine the extent to which age, sex, education, and other SDoH mediate differences by race. They found that increases in BMI were associated with less cognitive decline across domains, with significant BMI by race interactions for processing speed (both level and slope) and memory (slope). There was a significant positive relationship between BMI slope and processing speed level in African American participants but not in Whites. Further, there was a significant positive relationship between BMI slope and memory level in White participants, but not in African Americans. Finally, contrary to their hypothesis, differences by race were not mediated by SDoH; however, relationships between BMI change and cognitive change for the full sample were mediated by Economic Stability for processing speed and reasoning.

Do Associations Between Vascular Risk and Mild Cognitive Impairment Vary by Race?

The article by Rotblatt et al. is another example of the utility of the ACTIVE data beyond intervention effects. They investigated whether the association between vascular risk and mild cognitive impairment (MCI) varied across B/AA and White participants. The main findings demonstrated that increasing vascular risk burden, high cholesterol, and obesity were associated with greater odds of non-amnestic MCI (naMCI) for Black participants but not White participants. The pattern of findings for amnestic MCI (aMCI) was less consistent, with the effects of vascular risk factors more similar/overlapping by race. These findings are consistent with previous studies showing increasing health problems associated with greater cognitive declines, particularly in non-amnestic domains of perceptual/processing speed (Byrd et al., 2018; Carmasin et al., 2014). They also support findings of obesity and high cholesterol having a greater association with cognitive impairment in older Black than older White adults (Sturman et al., 2008). As such, the findings may reflect the influence of race-related variations in SDoH on vascular health, and ultimately on cognitive health. Across participants, diabetes and hypertension were associated with increased odds of aMCI and naMCI, respectively, consistent with previous findings (Luchsinger et al., 2007).

Does Consumer Credit Precede or Follow Changes in Cognitive Impairment Among Older Adults? An Investigation in the ACTIVE Trial

There is a growing body of evidence linking consumer credit to cognitive function, which is particularly important to examine for older adults. The paper by Dean and her colleagues addresses the questions of whether credit scores or changes in credit scores predict future MCI, and whether MCI is associated with subsequent poor credit scores. Using ACTIVE survey data from the 1-, 2-, 3-, and 5-year follow-up visits linked to TransUnion consumer credit data, they assessed the relationships between consumer credit history (individual credit scores, debts unpaid, or debts in collections excluding medical debt) and pre- and post-classification of algorithmically defined MCI. They reported that poorer credit predicts increased risk of future MCI up to 3 years later, though the amount of change in credit itself was not predictive. They found limited evidence that non-medical debt collections predicted future MCI, suggesting that the relationships found could be due to increased medical debt as health worsens in the lead-up year to MCI. It was concluded that monitoring credit among older adults may help identify potential future MCI risk. After a diagnosis, older adults may need assistance managing changes in credit or other financial challenges.

The Effects of Occupational Complexity on Late-Life Cognition in ACTIVE: Examining the Mediating And Moderating Effects of Race

Owens and his colleagues studied occupational complexity (OC) in ACTIVE participants and its association with cognitive performance. They examined whether: 1) OC explains individual differences in cognition at baseline, 2) this relationship is differentially related to cognition by Black/White race, and 3) OC mediates some or all the Black/White race-related variance in late life cognition. They reported that multiple dimensions of OC are related to cognition, but there were relatively few Black/White differences in these associations. Across all cognitive dimensions except for useful field of view, a history of having jobs lower in substantive complexity and fine motor skills and higher in physical demands may explain some of the Black/White race differences in older adults’ cognition. The authors conclude that occupations can be a target to reduce social disparities in late-life occupations.

Rural-Urban Differences in Cognition: Findings From the ACTIVE Trial

There has been increasing interest in studying the relationship between the geographic region in which one resides and subsequent cognitive decline. Using longitudinal data from the ACTIVE trial, the paper by Steinberg et al. examines associations between three geographic areas (urban, suburban, rural) and cognition (memory, reasoning, processing speed) over a 10-year period, after adjusting for SDoH. They report that compared to urban and suburban participants, rural participants fared worse on all cognitive measures across the 10-year trajectory. These findings are consistent with prior work suggesting poorer cognitive performance of rural older adults when compared to urban older adults (Cassarino et al., 2018; Lorenzo-Lopez et al., 2017; Saenz et al., 2018). Across geographic areas, greater economic stability, health care access and quality, and neighborhood resources were associated with better cognition over time. The authors conclude that further work to understand the role of geographic location and SDoH domains on cognition is needed to make the greatest impact on geographically diverse communities.

The Relationship Between Cognition and Mortality Among Older Black and White Men in Advanced Cognitive Training for Independent and Vital Elderly

There has been sparse attention paid to cognition and mortality among older men and potential race differences. Men’s health disparities are likely to continue to increase over the next few decades. This is largely because of the aging baby boom cohort, increases in racial diversity in the US, and increases in life expectancy among Black and White men. Thorpe and his colleagues examined the relation between cognition and mortality among older Black and older White men in the ACTIVE study. They found that among White men, higher performance on the memory composite measure was associated with a decreased risk of all-cause mortality (HR: .93; 95% CI: .89–.98), whereas the other cognitive measures were not associated with all-cause mortality risk. Among Black men, none of the cognitive measures was associated with all-cause mortality risk. However, the mean effects of the cognitive scores were larger than that of older White men. These findings underscore the need for additional work to retain and recruit a larger sample of older Black men and to advance our understanding of the cognition-mortality relationship.

Conclusions

The ACTIVE study is the largest and longest running cognitive intervention study to-date involving cognitively normal older adults at baseline, and it continues to be an important study to understand how factors such as race and SDoH relate to long-term participant outcomes and moderate intervention effects. In addition to primary training effects, the ACTIVE study’s community-based recruitment efforts allowed for inclusion of a diverse sample, which has enabled the ACTIVE study to address questions related to sociodemographic group and race differences in cognitive performance.