What pitfalls would you watch for when analyzing an experiment with healthcare claims data, such as interference, peeking, or sample ratio mismatch?
