Later-Life Food Insecurity Raises Probable Dementia Probability to 34 Percent in PSID Cohort

University of Michigan researchers tracked food insecurity across two windows in the same families using the longest-running U.S. household panel survey. Exposure in later life (2015-2019) drove the association with probable dementia; midlife exposure alone showed no significant link. The design captured within-family changes over decades, isolating timing effects that shorter studies miss. Absolute probabilities reached 40 percent for persistent exposure and 34 percent for later-life exposure only. The observational cohort cannot separate causation from confounding by income volatility, chronic disease, or social isolation that co-occur with food insecurity. Absolute risk differences of 23-29 percentage points exceed many reported associations for single nutritional factors, yet residual confounding remains plausible. The study adds to evidence that material hardship in older age affects cognition through combined nutritional and psychosocial pathways rather than midlife deficits alone. Clinicians already screen for food insecurity at visits but lack referral pathways; expanding SNAP outreach and medically tailored meals for older adults offers testable interventions. Future work must link administrative food-assistance records to cognitive endpoints in larger cohorts to test whether reducing later-life insecurity alters incidence within five years.