As far as I know, the underlying causes of dementia are not very well understood. Besides the fact that 6% is not very rare, there's an issue of the Texas sharpshooter fallacy (http://en.wikipedia.org/wiki/Texas_sharpshooter_fallacy): taking whatever problem exists and claiming that may be linked to whatever one disapprove of.
"She has Parkinson's, which X in 100 have by age 70. Clearly there are some unusual things going on with those unlucky X..."
"She has Huntington's, which X in 100 have by age 70. Clearly there are some unusual things going on with those unlucky X..."
"She has Alzheimer's, which X in 100 have by age 70. Clearly there are some unusual things going on with those unlucky X..."
"She has MLS, which X in 100 have by age 70. Clearly there are some unusual things going on with those unlucky X..."
"She has brain cancer, which X in 100 have by age 70. Clearly there are some unusual things going on with those unlucky X..."
"She had a crippling stroke, which X in 100 have by age 70. Clearly there are some unusual things going on with those unlucky X..."
You see my point. There are an awful lot of things that can go wrong with a brain over time, the rate at which things go wrong goes up with time (we call it "aging"), and sooner or later something will go wrong (and you will "die"). It is inevitable that some short sleepers will suffer awful brain problems. Just like everyone else.
I saw another example yesterday in a discussion of famous drug chemist & inventor Shulgin; apparently, upon getting cancer at age 80 and asking for help, his friends had to deny to people that it was related to any of the drugs he took! I was like, "are you serious? A large minority or a majority of the entire population will have cancer at some point in their lifetime, many will not even make it to their 80s at all much less only develop cancer at that advanced age, and people seriously think that this might be more than-microscopically-small evidence that some of the drugs cause cancer, when if anything it may be evidence against?"
The first, a survey, seems to show that too little sleep increases all-cause mortality (i.e. that "aging" you mention), the second suggests that sleep disturbances (unknown, admittedly how that relates to sleep duration) may be linked (cause or effect unknown) with Alzheimer's.