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

Confirmatory factor analysis models fit the harmonized cognitive activity data well across all four cohort studies.

• Fit indices met acceptable thresholds in each study: comparative fit index ≥ 0.90, root mean square error of approximation ≤ 0.08, and standardized root mean residual ≤ 0.08.
• Data came from MAP (n=1977), MARS (n=825), Clinical Core (n=366), and ROS (n=1382).
• Total sample included 4550 participants free of dementia at baseline.
• An item-banking approach was used that leveraged both common and unique items across studies.

Higher harmonized cognitive activity factor scores were associated with lower risk of incident Alzheimer's disease dementia.

• Hazard ratio = 0.80 (95% CI = 0.74–0.86, p < 0.001).
• This indicates each unit increase in the harmonized factor score was associated with a 20% reduction in the hazard of incident AD dementia.
• Analysis was conducted in participants free of dementia at baseline across four cohort studies.

Higher harmonized cognitive activity factor scores were associated with slower cognitive decline.

• Estimate = 0.005, SE = 0.002, p = 0.02.
• Cognitive decline was assessed as a longitudinal outcome across the four cohort studies.
• The association was statistically significant at the conventional alpha level of 0.05.

Higher harmonized cognitive activity factor scores were associated with greater cognitive resilience, defined as cognitive decline after adjusting for common ADRD pathologic indices.

• Estimate = 0.011, SE = 0.004, p = 0.003.
• Cognitive resilience was operationalized as cognitive decline after adjusting for common ADRD pathologic indices.
• This finding suggests cognitive activity may confer resilience to AD-related neuropathology beyond its direct association with cognitive decline.

The study harmonized self-reported cognitive activity data from four cohort studies of cognitive aging comprising a heterogeneous older adult sample.

• Total sample: 4550 participants (MAP = 1977; MARS = 825; Clinical Core = 366; ROS = 1382).
• 75% of participants were female; mean baseline age was 76.9 years (SD = 7.6); mean education was 16 years (SD = 3.7).
• All participants were free of dementia at baseline.
• Harmonization used an item-banking approach leveraging all available items, including those common and unique across studies, to generate harmonized factor scores.