Comprehensive study of pedestrian traffic fatalities and severe injuries in New York City:
This is a pretty thorough analysis of accidents involving death or severe injury. While it focuses much of its energy on pedestrians, it does analyze the rate of overall traffic deaths among all street users and determines that New York City is the safest city in the U.S. as far as traffic deaths are concerned. It adjusts for the surge in daytime population in Manhattan to stress that although the raw number of deaths and injuries is high, the rate per 100,000 population is quite low.
I believe that it alludes to NYC being even safer for pedestrians than the rate of death and severe injury per 100,000 population would imply, because there is actually a much higher rate of walking among the population there, among commuters, residents, and tourists. I've always wanted to see a study that attempts to compare pedestrian safety among U.S. cities by estimating something like miles walked per 100,000 population and using that to determine the actual likelihood of being killed or injured in a collision. My hunch has been that such a study would find that older, more dense cities, where walking is more common, are many times safer to walk in than our newer, lower density cities. It might also be interesting to have a study that factored in death or injury due to crime. If anyone is aware of such a study, please provide a link or other direction.