Estimands: improving inference in randomized controlled trials in clinical nutrition in the presence of missing values

For randomized controlled trials, the impact of the amount and handling of missing data on the interpretation of the treatment effect has been unclear. The current use of intention to treat, per protocol, and complete-case analysis has shortcomings. The use of estimands may lead to improved estimation of treatment effects through more precise characterizations of the fate of treatments after dropout or other post-randomization events. A perspective on current and future developments with a view toward clinical nutrition is provided.