This is technically correct, but it hardly matters, because the author clearly wants you do draw casual conclusions from this correlation.

Even though looks a lot like a classic example of collider bias, where obesity looks like it protects you from dysglycemia.