Loneliness, cerebrovascular and Alzheimer's disease pathology, and cognition.

INTRODUCTION: Loneliness has a rising public health impact, but research involving neuropathology and representative cohorts has been limited.

METHODS: Inverse odds of selection weights were generalized from the autopsy sample of Rush Alzheimer's Disease Center cohorts (N = 680; 89 ± 9 years old; 25% dementia) to the US-representative Health and Retirement Study (N = 8469; 76 ± 7 years old; 5% dementia) to extend external validity. Regressions tested cross-sectional associations between loneliness and (1) Alzheimer's disease (AD) and cerebrovascular pathology; (2) five cognitive domains; and (3) relationships between pathology and cognition, adjusting for depression.

RESULTS: In weighted models, greater loneliness was associated with microinfarcts, lower episodic and working memory in the absence of AD pathology, lower working memory in the absence of infarcts, a stronger association of infarcts with lower episodic memory, and a stronger association of microinfarcts with lower working and semantic memory.

DISCUSSION: Loneliness may relate to AD through multiple pathways involving cerebrovascular pathology and cognitive reserve.

HIGHLIGHTS: Loneliness was associated with worse cognition in five domains. Loneliness was associated with the presence of microinfarcts. Loneliness moderated cognition-neuropathology associations. Transportability methods can provide insight into selection bias.