Abstract
BACKGROUND AND PURPOSE: The complex aetiology of Alzheimer's disease suggests prevention potential. Risk scores have potential as risk stratification tools and surrogate outcomes in multimodal interventions targeting specific at-risk populations. The Australian National University Alzheimer's Disease Risk Index (ANU-ADRI) was tested in relation to cognition and its suitability as a surrogate outcome in a multidomain lifestyle randomized controlled trial, in older adults at risk of dementia. METHODS: In this post hoc analysis of the Finnish Intervention Study to Prevent Cognitive Impairment and Disability (FINGER), ANU-ADRI was calculated at baseline, 12, and 24 months (n = 1174). The association between ANU-ADRI and cognition (at baseline and over time), the intervention effect on changes in ANU-ADRI, and the potential impact of baseline ANU-ADRI on the intervention effect on changes in cognition were assessed using linear mixed models with maximum likelihood estimation. RESULTS: A higher ANU-ADRI was significantly related to worse cognition, at baseline (e.g., estimate for global cognition [95% confidence interval] was -0.028 [-0.032 to -0.025]) and over the 2-year study (e.g., estimate for 2-year changes in ANU-ADRI and per-year changes in global cognition [95% confidence interval] was -0.068 [-0.026 to -0.108]). No significant beneficial intervention effect was reported for ANU-ADRI, and baseline ANU-ADRI did not significantly affect the response to the intervention on changes in cognition. CONCLUSIONS: The ANU-ADRI was effective for the risk prediction of cognitive decline. Risk scores may be crucial for the success of novel dementia prevention strategies, but their algorithm, the target population, and the intervention design should be carefully considered when choosing the appropriate tool for each context.