I remember reading an IEEE Spectrum article about death rates over 100, and while obviously very high, apparently they didn't increase much. I can't seem to find it unfortunately as it was a good article.