I think it may be counter-productive to mention this, because it is difficult to prove. It is trivial to show that there are dead-weight costs incurred when computer systems need to deal with DST across political boundaries. Those costs alone should be sufficient to convince anyone to abandon DST (or just as well, adopt it full-time, as Russia recently did). The fact that "normal" people do not know these costs exist is, I think, the main reason they are sympathetic to emotional arguments such as "DST will save children's lives on Halloween" or equally your "DST kills drivers once or twice a year."
For starters, because it may not be true. Another comment here claimed that the increase in traffic accidents (note: not deaths or even injuries, we're just counting accidents) is offset almost perfectly when we gain an hour of sleep. See http://www.nejm.org/doi/full/10.1056/NEJM199604043341416
I am completely in favor of abandoning DST, but am not convinced that it helps the cause to argue in terms of deaths. The opposition may claim with even greater efficacy among the general public that stopping DST will endanger children on Halloween (as has been claimed before).
I am under the impression that traffic accidents increase briefly when the clocks go forward, and decrease briefly when they go back: there's no net difference overall.