“Furthermore, the belief that multiple regression analyses will accurately apportion the contributions of smoking, increasing age, and other colinear factors has been shown to be unjustified.”