In my "other" job, I got involved in producing medical statistics for the insurance industry - if they're insuring your life, they want to have some idea of how long it's going to be. I make it clear that I'm not a doctor, I'm a mathematician. Whilst all the above comments are extremely good and valid, there are two further aspects which we found are equally important.

The first is weight. It's not just what you eat, but how much - people with a high Body-Mass Index (BMI) don't generally live as long as those with a perfect one, regardless of diet or exercise. The second is stress. There are many types of stress, so this is hard to quantify, but long term stressful situations can be shown to be linked statistically to reduced life expectancy.

Both these observations are based on empirical data. Incidentally, using empirics we knew that certain conditions were hereditary long before the medical profession discovered why and confirmed our findings.

Just trying to be helpful.