Don't worry about the dementia thing too much. While it's a very strong correlation, it only increases the risk of developing dementia to 120% of normal for nondiabetics and 140% of normal for diabetics, which is still only about 1-2% of the people in their study.
About a quarter of the participants (524 of 2067) developed dementia over the course of the study. I'm not sure where you got the 1-2% figure. Perhaps you are referring to the p-values, but p indicates how likely it would be to observe the data if the null hypothesis were true.