Do changed behaviors predict cognitive decline in a community sample?

Abstract

Objective: To examine whether later-life development of neuropsychiatric symptoms (NPS) or a modified diagnosis of mild behavioral impairment (MBI) are associated with future cognitive decline.

Design: Prospective Cohort Study SETTING: Community PARTICIPANTS: 823 individuals without dementia aged 70-90 years from the Sydney Memory and Ageing Study, followed over six years.

Measurements: Biennially, cognition was assessed through neuropsychological testing, and clinical diagnoses of mild cognitive impairment (MCI) and dementia were made by expert consensus. NPS was evaluated using the Bayer Activities of Daily Living scale and Neuropsychiatric Inventory (NPI). Based on published algorithms, modified diagnoses of MBI (MBI-Lite) were obtained. The relationship between behavior at baseline and neuropsychological test scores six years later was examined using linear regression. Cox regression was performed to evaluate associations between behavior and both incident dementia and incident categorical progression. Apolipoprotein E ε4 allele carrier status and cardiovascular disease risk were controlled for in all analysis.

Results: Higher total NPI scores were associated with worse global cognition scores at six years (β = -0.03, p = .007) and increased risk of incident dementia over six years (HR=1.06, p = .003). Presence of NPS of clinically significant severity and frequency was associated with worse global cognition scores for those with normal cognition (β = -0.44, p = .001), but not those with MCI at baseline (β = 0.22, p = .389). Diagnosis with MBI-Lite, was also associated with worse global cognition scores (β = -0.28, p = .028).

Conclusions: Certain measures of NPS in cognitively normal older adults may herald future cognitive decline and be useful for early diagnosis and dementia research.