To determine if superior health at old age protects against cognitive impai
rment (CI) and Alzheimer's disease (AD), we prospectively studied 100 optim
ally healthy oldest-old (greater than or equal to 85 years) individuals. In
itially, subjects represented the top 3% of the oldest old for health. Duri
ng 5.6 +/- 0.3 years of follow-up, 34 subjects developed CI, and 23 progres
sed to AD. By age 100, probability of CI and AD were. 65 +/- .09 and .49 +/
- .10. Median onset age was 97 years for CI and 100 for AD. Clearly, superi
or health at old age does not guarantee protection against cognitive declin
e. Lifetime risks were similar to the general population but onset ages wer
e later, suggesting factors that delay onset are key to improving cognitive
health in the elderly. In this population, absence of apolipoprotein E-is
an element of4 and male gender were associated with delayed onset, whereas
estrogen use and education had no detectable effect on cognitive outcome.