The question of the risks of smoking is actually exceedingly complicated and in the end it is statistical as is everything else in life. Yet the truth is that mainstream tobacco research is junk science of the worst sort with biases and method problems that are severe that the field simply can't be taken seriously by anyone interested empirical study. If you want to learn about the issues at stake a good place to start would the numerous links i've posted in the past and http://www.lcolby.com/
has a bunch of great stuff on issues at stake. As to insurance companies we should be realistic and understand that could care less about the soundness of any sort of research and will instead accept without question any mainstream study that will justify higher rates.