Clarity is what's important, and the notion of absolute percentage isn't necessarily universal.
The statement "No. of people with dementia dropped by 2.8% between 2000 and 2012" is wrong given a standard interpretation of the base of a percentage. In an average population, where 4 people once would have gotten dementia, now only 3 will. That is absolutely a drop of ~25%. Your statement would imply that out of a 100 people who would gotten dementia, now only ~97 will, which understates the effect.
Again though, this is best solved by a clear explanation in the first place. I didn't think the summary was unclear, but it appears to have caused some confusion.