Hacker News Reader

Comment by xer0 | Hacker News Reader

Unfortunately you have to live your whole life to discover whether you're an exception or a statistic. I'll go with the statistics for my personal choices.

View on HN

But most smokers will be exceptions. Most smokers will not get cancer. They get cancer significantly more often than non-smokers, but that's not scary sounding enough, I guess.

To give specific numbers, according to one study (http://www.ncbi.nlm.nih.gov/pubmed/7895211), the lifetime risk of developing lung cancer is:

   male smokers: 17%
   female smokers: 12%
   nonsmokers: 1%

Which backs up what I said. It's really unusual to see anyone talk about the actual lifetime risk. It's not as scary as saying "your chances are X times greater"... greater than what?

If those numbers are accurate, and entirely due to the effects of smoking (not just other lifestyle behaviors more common with smokers, like drinking) that's easily reason enough to quit. But even in the worst case, the fact remains, most smokers will not get cancer.

I'm not sure where your argument leads. Are you justifying smoking as a choice, based on the numbers? Are you attempting to explain why people choose to smoke, based on the likelihood of cancer? Something else?

Regardless, smoking increases your overall risk pressure along with all of the other risky choices you can make. It's cumulative with the rest of life's choices.

"X times greater risk" refers to the likelihood ratio over the Bayesian prior, assuming that this bit of evidence (smoking/not) is independent of other known evidence. Usually the prior is "all people" or "people of <X> ethnotype" or something like that.

You can compute the lifetime risk if you have a prior for the lifetime risk of lung cancer (1% in above example), just by multiplying.

What about heart disease?

Smoking causes problems other than cancer.

Yes, but I'm talking about cancer. The fact still remains that the statement "Most smokers will not get lung cancer" is a controversial statement for some reason, even though it is demonstrably true.

Reply on news.ycombinator.com