The statistic of deaths / mile is irrelevant. Of course more drunk walkers die per mile walked than drunk drivers die per mile driven, because it takes them > 10 times as long to get anywhere. I think it's more likely that accident rates are based more on time traveled than on distance traveled.