I seem to remember reading something like this in the papers a while ago. On that occasion the statistics were based on the likelyhood to be killed if involved in an accident at work. It wasn't necessarily a reflection of how likely you were to have an accident in the first place?