Harmonization of late-life participation in cognitively stimulating activities across four cohort studies of cognitive aging.

OBJECTIVE

We harmonized self-reported survey data measuring late-life participation in cognitively stimulating activities across four cohort studies of cognitive aging.

METHODS

Data came from the Rush Memory and Aging Project (MAP), the Minority Aging Research Study (MARS), the Rush ADRC Clinical Core (Clinical Core), and the Religious Orders Study (ROS) in individuals free of dementia at baseline. We used an item-banking approach that leverages all available items assessing cognitively stimulating activities, including items common and unique across studies. We generated harmonized factors to summarize an individual's cognitive activity. We then validated the harmonized factor by testing its association with incident AD dementia, cognitive decline, and cognitive resilience i.e. cognitive decline after adjusting for common ADRD pathologic indices.

RESULTS

We included 4550 participants (MAP = 1977; MARS = 825; Clinical Core = 366; ROS = 1382); 75% were female, mean baseline age was 76.9 years (SD = 7.6) and mean education was 16 years (SD = 3.7). Confirmatory factor analysis models fit the data well in each study (comparative fit index ≥ 0.90, root mean square error of approximation ≤ 0.08, and standardized root mean residual ≤ 0.08). The harmonized factor score was associated with incident AD dementia (HR = 0.80, 95%CI = 0.74-0.86, p < 0.001), cognitive decline (est. = 0.005, SE = 0.002, p = 0.02), and cognitive resilience (est. = 0.011, SE = 0.004, p = 0.003).

CONCLUSION

We statistically harmonized survey data measuring frequency of participation in cognitive activities across four studies of aging and subsequently validated the harmonized latent trait by linking it to ADRD outcomes. Findings demonstrate the utility of combining self-reported psychosocial survey data collected across multiple studies to thoroughly evaluate the impact of modifiable risk factors on later-life cognitive outcomes in heterogeneous populations.