Abstract
Missing data in clinical trials can be handled by appropriate statistical methods, when the data are assumed to be missing at random (MAR). However, whether the missing data are MAR or missing not at random (MNAR) cannot be determined by the nature of the data. Therefore, clinical trials should be designed not only to gather the usual clinical data but to follow incomplete abnormal data from an early stage. Careful attention should be paid to informed consent and the balance between the burden on patients and the collection of precise clinical data.
