ABSTRACT
The term estimand is simply “the target of estimation of the treatment effect.” As discussed above, there may be several estimands of interest in a trial. When the “effect” is on a population level, as often is for the interest of the regulatory authority, drug maker, or insurance payer, we usually refer to the effect as population average effect. In a perfect situation, which seldom exists in clinical trials, all patients fully adhere to the data collection process—taking the randomized medicine as prescribed to the full course and never missing a visit during the trial. Principal stratum strategies. In this strategy, ICE means the potential ICE. The target population might be taken to be the “principal stratum” in which a potential ICE would occur or would occur. Proper imputation remains to be the most useful method to deal with the missing data problem for the estimands targeted by the treatment policy, hypothetical, and while-on-treatment strategies.
