Well, keep in mind that statistics are "the past" (what has actually happened) while odds are "the future" (what might happen); in some cases (like traffic accidents and deaths) the odds are based on statistics while in other cases (such as the lottery) the odds are pre-set.
With a lottery, it is (supposed to be) totally random who wins, and prior results have no impact on future results.
The odds of being in a fatal accident is more like the odds of getting cancer. There are things one can do to improve the odds (wear seat belt / helmet, stay off cell phone, keep the vehicle maintained, etc, etc) but despite doing all that folks still die in accidents.
How would one assign odds to an individual? I have no idea. Common sense says that if you live in an area that has less traffic, wear your seat belt, and have many years of driving experience, your odds of surviving improve. But nevertheless some freak accident can put you in the morgue even if previous stats never hinted at that freak combination ever happening. Then you would become a statistic upon which future odds could be based...
(Not implying that's what I would want to happen... just that prior nationwide statistics aren't necessarily a good predictor of what might happen to individuals)
Edit: Wanted to add link to the FARS system, which allows you to slice & dice the fatality data many different ways... might be a way to better estimate odds according to your lifestyle and location:
http://www-fars.nhtsa.dot.gov/Main/index.aspx. Unfortunately no similar system exists for accidents in general, apparently because of the different reporting approaches by state. Would be interesting to look at severity of injuries, including subsequent deaths in the hospital (which I don't believe count in the NHTSA fatality rate) similar to the way the report I linked above does (but for all states and all vehicle types).