If you want to quantify the health effect of touchback-related injury prevention, you need to know the rate of injury for coverage players (not returners and blockers, we don't control that end), and set the expected loss for coverage players, and set that against the field position benefit. If injuries are more common than an individual play but still infrequent, and the coverage players are chiefly reserves, touchbacks for health reasons may not move the needle from a game-impact point of view (humanitarian interests may be different, but wouldn't favor playing a violent game like football in the first place).

 The article you linked stated that injury rates were highest on kickoff plays (seems plausible), but did not break it down by role. And although it said there were 40% less concussion injuries from kickoffs (with TBs rising from 16.4% to 43.5%), the number of concussion injuries on kickoffs is not stated.

I found another article with numbers here:

http://usatoday30.usatoday.com/sports/football/nfl/story/2012-02-01/kickoff-returns-concussions/52926056/1

Concussions on all plays dropped by 12.5% from 0.679 concussions per game to 0.594 concussions per game.  There were 28 less total concussions and 658 less kickoff returns, so if the drop was solely from the kickoff drop, that represents 23.5 kickoffs per concussion.  However, the article also noted that the concussion rate was lower three years ago and a different article noted that non-concussion kickoff injuries hadn't gone anywhere, so this may be a bit of a high estimate.

Even taking the rate at face value, and cutting it in half because we only care about our team (and assuming, perhaps wrongly, that the rate is the same for coverage team as return team), that's 47 kickoffs per prevented concussion.  If we're able to hold the opposition an average of at least one yard below the 25 yard line (adjusted for turnovers induced), I think the value of 47 non-touchback returns would be larger than the point impact of a concussion to a random coverage team member.  The health benefit doesn't seem to be large enough to me to justify forgoing a possible competitive advantage.

If the injury rate is greater on the receiving team than the kicking team (my memory tells me this is the case, but it could be wrong), kickoff returns would give us a net competitive health benefit within the context of a single game, though carry-over to following games would still be a negative.