Aging is a weak but relentless determinant of dementia severity.

Structural Equation Models (SEM) can explicitly distinguish "dementia-relevant" variance in cognitive task performance (i.e., "δ" for dementia). In prior work, δ appears to uniquely account for dementia severity regardless of the cognitive measures used to construct it. In this study, we test δ as a mediator of age's prospective association with future cognitive performance and dementia severity in a large, ethnically diverse longitudinal cohort, the Texas Alzheimer's Research and Care Consortium (TARCC). Age had adverse effects on future cognition, and these were largely mediated through δ, independently of education, ethnicity, gender, depression ratings, serum homo-cysteine levels, hemoglobin A1c, and apolipoprotein e4 status. Age explained 4% of variance in δ, and through it, 11-18% of variance in future cognitive performance. Our findings suggest that normative aging is a dementing condition (i.e., a "senility"). While the majority of variance in dementia severity must be independent of age, age's specific effect is likely to accumulate over the lifespan. Our findings also constrain age's dementing effects on cognition to the age-related fraction of "general intelligence" (Spearman's "g"). That has broad biological and pathophysiological implications.