Handling drop‐out in longitudinal clinical trials: a comparison of the LOCF and MMRM approaches