Correlation is not causation. Even if exercise were associated with a 99% risk reduction, it could remain the case that people who are prone to Alzheimer's are less motivated to exercise early in life rather than the exercise preventing Alzheimer's later in life. There are plenty of other good reasons to exercise, we don't have to stretch this one beyond what the data says.
