I wouldn't jump to this conclusion too quickly, because there are a lot of possible confounding factors and possible reverse causality at play here.
For example, it's entirely plausible that being at high risk of any of these diseases would cause people to cut back on their sweetener intake into the lowest consumption bracket, leading to unhealthy individuals showing a bimodal distribution (those who cut back dramatically on advice from doctors and those who have very very bad habits) while healthy people have a more typical curve.
For example, it's entirely plausible that being at high risk of any of these diseases would cause people to cut back on their sweetener intake into the lowest consumption bracket, leading to unhealthy individuals showing a bimodal distribution (those who cut back dramatically on advice from doctors and those who have very very bad habits) while healthy people have a more typical curve.