Among 1,491 dementia-free Norwegians aged 70-77, directly measured peak oxygen uptake predicted Alzheimer disease over a median 8.7 years, with 77 incident cases. Moderately fit (HR 0.54, 95% CI 0.29-0.98) and fit participants (HR 0.52, 0.29-0.93) had about half the risk of the unfit group. Each additional 1 mL/kg/min of VO2peak carried a 4% lower risk. Risk rose sharply below roughly 25.1 mL/kg/min for men and 20.9 for women, while one-year fitness changes showed only weak associations.
Measuring VO2peak on a treadmill rather than estimating it from questionnaires is a real strength, since self-reported fitness is notoriously noisy. The threshold finding is the most practical takeaway: dementia risk seems to track where you stand entering old age, which fits the broader literature linking midlife and late-life fitness to lower dementia incidence. But caution is warranted. With only 77 events, the confidence intervals are wide and the upper bound of the per-unit estimate touches 1.00. AD ascertainment relied on registries and prescription records, which likely undercounts mild or undiagnosed disease. Reverse causation is also plausible, because preclinical Alzheimer pathology can erode fitness and motivation years before diagnosis. The null result for one-year changes should not be read as proof that late-life training is pointless; a single year is a short window, and trial data on exercise and cognition remain mixed. This is a confirmatory, hypothesis-supporting cohort result rather than a paradigm shift. It strengthens the case for building fitness early and treating VO2 max as a trackable health marker.