Using Large Data Sets and Statistical Tools to Define “Failure” and Demonstrate the Safety of DPR