Link. Overall, not a bad result. It would be very interesting to randomize wealthy people to insurance/no insurance and compare results.