Per your article, "below the range 22.5-25 kg/m2, BMI was associated inversely with overall mortality, mainly because of strong inverse associations with respiratory disease and lung cancer" (italics mine) i.e. if your BMI is low, average mortality is comparatively high, but only because of smokers. (Study recruitment year looks to be 1979 when there were a lot more smokers).
Moreover, your study recruited people at age 46 (mean) and followed them through their death. Meaning, a whole lot of people got older, got sick, subsequently lost weight and died. Unless proper adjustments were made, the low BMI-high mortality connection is rather unproven.
The main concern regarding the newer CDC analysis is that it did not adequately account for weight loss from serious illnesses such as cancer and heart disease. Including such individuals in the analysis created the false appearance that being overweight protected against death during the follow up.
and
The newest CDC analysis also failed to account adequately for the effect of smoking on weight. Smokers tend to be a little lighter than nonsmokers, although the negative health impact of smoking far outweighs that of a few extra pounds. As a result, the Flegal study underestimated the risks from obesity and overestimated the risks of leanness.
Regardless, I don't think we are on different pages since you acknowledged that no one really knows which direction the causality goes.
Moreover, your study recruited people at age 46 (mean) and followed them through their death. Meaning, a whole lot of people got older, got sick, subsequently lost weight and died. Unless proper adjustments were made, the low BMI-high mortality connection is rather unproven.
For a similar example, see http://onlinelibrary.wiley.com/doi/10.3322/canjclin.55.5.268...:
The main concern regarding the newer CDC analysis is that it did not adequately account for weight loss from serious illnesses such as cancer and heart disease. Including such individuals in the analysis created the false appearance that being overweight protected against death during the follow up.
and
The newest CDC analysis also failed to account adequately for the effect of smoking on weight. Smokers tend to be a little lighter than nonsmokers, although the negative health impact of smoking far outweighs that of a few extra pounds. As a result, the Flegal study underestimated the risks from obesity and overestimated the risks of leanness.
Regardless, I don't think we are on different pages since you acknowledged that no one really knows which direction the causality goes.