Statistical Primer: How to Deal With Missing Data in Scientific Research?

Papageorgiou and colleagues discuss dealing with missing data, a common challenge in clinical research. Although the best approach is to minimize the amount of missing data through good study design and data collection protocols, missing data cannot always be avoided and they must be treated appropriately to maintain the validity of the statistical inferences from a study. The authors outline the different reasons that data may be missing and they discuss the methods for handling these types of missing data, including limitations of such methods. Finally, they provide an example using the scenario of a study of congenital heart disease patients receiving an aortic allograft.