Should multiple imputation be the method of choice for handling missing data in randomized trials? | Synapse