RACIAL AND ETHNIC DISPARITIES IN COGNITIVE DIFFICULTY AMONG OLDER ADULTS: EVIDENCE FROM NEW YORK CITY

Abstract This study examined racial and ethnic disparities in cognitive difficulty among older adults in New York City (NYC). Also, we tested whether physical health, family structure, individual socioeconomic status (SES), and neighborhood SES explained the disparities. Based on community districts, individual-level data from the 2019 American Community Survey were merged with neighborhood data from NYC Community District Profile. A sample of 5,622 NYC residents aged 60 or older was included across 55 community districts. The outcome variable, cognitive difficulty, was measured by a binary variable in which respondents’ self-reported challenges with cognitive health (1=having challenge, 0=no). Racial and ethnic groups included Whites, Blacks, Latinos/Hispanics, and Asians. We used multilevel logistic regressions for analysis. Results show that Latinos/Hispanics had the highest odds of reporting cognitive difficulty across groups. Physical health, marital status, individual SES, and access to parks were significantly associated with cognitive difficulty. Physical health, family structure, and multilevel SES partially explained or influenced the racial and ethnic disparity in cognitive difficulty. However, such influence varied by race and ethnicity. Physical health and individual SES contributed to the disparities for Latinos/Hispanics and Blacks, compared to Whites. Neighborhood SES attenuated the disparity in cognitive difficulty between Latinos/Hispanics and Whites. Also, family structure uniquely explained the disparity for Blacks. No significant disparity was identified between Asians and Whites. This study shed light on the important roles of multilevel factors in predicting racial and ethnic disparities in cognitive difficulty. Findings provide direction for interventions to reduce racial and ethnic disparities in cognitive difficulty.


UPDATING ESTIMATES OF DRIVING LIFE EXPECTANCY Jonathon Vivoda, Miami University, Oxford, Ohio, United States
Information is limited about driving life expectancy, and the amount of time between loss of driving ability and death. The most often cited study (by Foley and colleagues, 2002) analyzed data collected in the mid-1990s and used life expectancy estimates to determine survival probabilities. Although that study was well designed, longitudinal data from the Health and Retirement Study (HRS) are now available to use participants' actual date of death to assess issues related to driving life expectancy. HRS data from 1996-2018 were assessed; only participants who had answered the driving ability question for at least one wave, and had a reported death date were included. The percentages of participants were determined who never reported the ability to drive, always reported the ability to drive, and those who transitioned from driving to non-driving. Time between loss of driving ability and death was calculated among former drivers by subtracting the interview date when respondents reported an inability to drive (after having previously reported being able to drive) from the participants' date of death. Only about 3% reported never driving, with a nearly even split between those who stopped driving (49%) and those who always drove (48%). Women were more likely to report being a never/former driver (65%) compared to men (36%). Among former drivers, the average time between inability to drive and death was 1312 days (3.59 years), and was significantly longer for women (437 days). Among former drivers, the average age for reporting inability to drive was about 83 years old. We previously used an evidence-based mathematical model to evaluate the cost-effectiveness of psychosocial interventions that reduce the risk of a nursing home admission for people with dementia from a healthcare payer perspective. We found the incremental cost-effectiveness of MIND, an in-home intervention for people with mildmoderate dementia, compared to usual care was $271,456 per quality-adjusted life-year (QALY). The incremental cost-effectiveness of NYU Caregiver Intervention (NYUCI), which is for people with moderate dementia, compared to usual care was $3,964/QALY. Here we quantify the uncertainty around our cost-effectiveness estimates. First, we calculated the expected value of partial perfect information (EVPPI), which is the value of eliminating uncertainty around the treatment effect (i.e., risk of entering a nursing home) of MIND and NYUCI, and represents the maximum willingness-to-pay for a study to inform these estimates. Given a willingness-to-pay of $110,000/QALY, population EVPPI for MIND and the NYUCI were $81,000,000 and $395,000,000, respectively. Second, we calculated the expected value of sample information (EVSI), the expected net benefit of sampling (ENBS) and the optimal sample size (OSS). EVSI is the amount of uncertainty reduced from a pragmatic trial evaluating the risk of entering a nursing home for people in the intervention compared to usual care. ENBS is the return of a pragmatic trial with a fixed ($1,050,000) and per person ($2,000) cost to conduct the study. The OSS is the sample size that maximizes ENBS and was 3,571 for MIND and 5,357 for NYUCI. There is value in conducting pragmatic trials on MIND and NYUCI.

RACIAL AND ETHNIC DISPARITIES IN COGNITIVE DIFFICULTY AMONG OLDER ADULTS: EVIDENCE FROM NEW YORK CITY
Ethan Siu Leung Cheung 1 , and Jinyu Liu 2 , 1. Columbia University,New York,New York,United States,2. Columbia University,New York City,New York,United States This study examined racial and ethnic disparities in cognitive difficulty among older adults in New York City (NYC). Also, we tested whether physical health, family structure, individual socioeconomic status (SES), and neighborhood SES explained the disparities. Based on community districts, individual-level data from the 2019 American Community Survey were merged with neighborhood data from NYC Community District Profile. A sample of 5,622 NYC residents aged 60 or older was included across 55 community districts. The outcome variable, cognitive difficulty, was measured by a binary variable in which respondents' self-reported challenges with cognitive health (1=having challenge, 0=no). Racial and ethnic groups included Whites, Blacks, Latinos/ Hispanics, and Asians. We used multilevel logistic regressions for analysis. Results show that Latinos/Hispanics had the highest odds of reporting cognitive difficulty across groups.
Physical health, marital status, individual SES, and access to parks were significantly associated with cognitive difficulty. Physical health, family structure, and multilevel SES partially explained or influenced the racial and ethnic disparity in cognitive difficulty. However, such influence varied by race and ethnicity. Physical health and individual SES contributed to the disparities for Latinos/Hispanics and Blacks, compared to Whites. Neighborhood SES attenuated the disparity in cognitive difficulty between Latinos/Hispanics and Whites. Also, family structure uniquely explained the disparity for Blacks. No significant disparity was identified between Asians and Whites. This study shed light on the important roles of multilevel factors in predicting racial and ethnic disparities in cognitive difficulty. Findings provide direction for interventions to reduce racial and ethnic disparities in cognitive difficulty.

AGING IN COMMUNITY: LESSONS LEARNED IN THE FIRST YEAR OF DEVELOPING A GRASSROOTS, RURAL PROGRAM Emily Kinkade, Melisa Hajdar, and Heather Fuller, North Dakota State University, Fargo, North Dakota, United States
The North Dakota Aging in Community project has an overarching goal of creating programs that improve the quality of life of rural older adults by increasing communitylevel support to help them age in place in two rural communities. The progress of program development and implementation was assessed across the project's first year through program data and steering committee surveys. Program data, including steering committee meeting minutes and monthly activity reports completed by program staff, were qualitatively analyzed for successes, challenges, and lessons learned by finding common themes within these data. Successes included the use of an apprenticeship/mentorship model utilizing consultants from similar grassroots, rural organizations, recruitment of an optimistic, enthusiastic, and collaborative steering committee, and employing patience to develop community-led efforts based on unique community needs. Challenges included hiring qualified staff in rural communities, limitations in infrastructure opportunities, and need for creative marketing strategies. Survey results aligned with these conclusions, yet demonstrated the strengths and challenges related to incorporating the ideas and needs of diverse stakeholders. These findings highlight the unique challenges in developing programs to support rural agingin-place, yet also highlight unique strategies that leverage rural strengths. Based on the findings of this first-year evaluation, future directions include conducting needs assessments to identify new avenues for program development, ensuring local program ownership through continued development of local steering committees, and continuing the mentorship/apprenticeship model while creating strategies to successfully foster independence. By sharing lessons learned, this program may serve as a replicable model to foster rural aging-in-place.

FOOD INSECURITY AMONG OLDER ADULTS IN NEW YORK CITY: DOES LOCATION MATTER? Ethan Siu Leung Cheung, Columbia University, New York, New York, United States
Having access to adequate and appropriate food sources is essential to addressing food insecurity among older adults. However, the role of locational characteristics in explaining food insecurity remains unclear, especially in urban areas. This study investigated the association of distance to grocery stores, neighborhood disadvantage, and social cohesion with food insecurity among older adults in New York City. Individual-level data were drawn from a 2-year Poverty Tracker Study. The sample included New York City residents aged 65 or older (baseline N = 710). Based on the respondents' residential address and neighborhood ZIP codes, the individual-level data were merged with two spatial datasets: American Community Survey and ReferenceUSA. ArcGIS 10 (near analysis) was used to manage spatial data and calculate the distance to grocery stores. Hierarchical logistic regression models were employed for analyses. Descriptive results show that more older adults in neighborhoods with economic disadvantage and lower level of social cohesion reported more food insecurity. Logistic regressions suggested that after controlling for individual-level characteristics (e.g., age, gender, race and ethnicity, and education), living farther (0.26-0.50 miles and 0.51-0.70 miles) from the nearest grocery store was positively associated with food insecurity. Residing in economically disadvantaged neighborhoods also increased the odds of food insecurity. Community social cohesion was a marginally significant protective factor against food insecurity. Findings suggest that locational characteristics play a significant role in predicting food insecurity in New York City, suggesting that community outreach and grocery delivery programs are needed to mitigate the risk. Older adults living in rural areas have particular challenges to accessing critical supportive services such as home modifications to promote functioning and safety. Conducting remote home assessments through telehealth has the potential to reduce time spent and overall cost that occur in conducting in-person assessments. During the pandemic, providers turned to telehealth to preserve continuity of assessment services with few research-based practices to guide them. With support from a NIDLRR SBIR grant, Thrive for Life LLC in partnership with the USC Leonard Davis School of Gerontology conducted research to develop a remote home assessment that aims to connect health and home modification providers with rural older adults (65+) with disabilities, a population that may not receive home modifications otherwise. Research included a literature review, key informant interviews with five experts in the field, and individual phone interviews with 30 rural older adults who have disabilities. The literature was analyzed and used to inform the interview questions. Key informant interview responses were analyzed for models, potential challenges, lessons learned, and opportunities to impact priority needs. Consumer interview responses were analyzed for needs, preferences, concerns, and challenges related to technology use. Findings demonstrate common barriers such as lack of access to broadband and smart technology; circumstances in which remote assessments are, and are not, likely to be successful; and the