Per-Protocol, Intention-to-Treat, and Complier Average Causal Effects Analyses in Randomized Controlled Trials: Linking Evidence to Practice | Synapse