In the May 7th issue of Science,  Stanton Glantz, argues that the evidence on tobacco is well tested, but " the new numbers on obesity are weak--or as one critic in CDC says, "loosey-goosey."" Glantz and others grumble that the CDC authors use inconsistent methods for calculating relative risks associated with tobacco and bad diet. Apparently, this study exhibits an age bias whereas the study for tobacco takes out the age bias. Glantz argues, the paper should be "withdrawn."

The article continues, "Several epidemiologists at CDC and the National Institutes of Health (NIH) echoed Glantz's concerns but declined to speak on the record. "I don't want to lose my job," said one CDC staffer who does research in this area.... A CDC scientist says internal discussions on these issues got "very contentious" months before publication and left some feeling that the conclusions were not debatable."

This is the article (need a subscription): http://www.sciencemag.org/cgi/content/full/304/5672/804

This is the May 7th issue: http://www.sciencemag.org/content/vol304/issue5672/index.shtml#editorial