My very crude understanding would be (if I read the abstract right): Too low total cholesterol is bad, too high is bad (that U shaped relationship between total cholesterol and mortality, also holds for age-sex groups). With the link between high and bad weaker than between low and good.