Regarding the biking injuries, I'd be willing to bet that biking (as a mode of transportation) injuries are skewed as well by those involved in riskier forms of leisure activity - racing or race training, single-track mountain biking, etc. It would be somewhat akin to reporting car accident statistics by including motocross or drifting, if those were a significant percentage of all driving. Also, another factor in the "well, I can just get my exercise somewhere else" argument is that many of the activities being chosen may have just as high an injury rate as biking. Beyond that, there's the fact that while one person may choose to get their exercise elsewhere, statistically people are just not doing that. 70% overweight in the US, remember?
Deaths are probably more fairly representative of the bike commuting risk, though.