TL17: Implementations of tipping point analysis in assessing impact of missing data
*Susan Wang, Boehringer-Ingelheim 

Keywords: missing data, tipping point analysis, survival data, binary, repeated measurements

Missing data continues to be a significant impediment to the interpretation of clinical trial results. With careful planning and intensive efforts, missing data from clinical trials can be reduced, but missing data cannot be avoided in human clinical trials. With the recent development of more reasoned statistical approaches to the handling of missing data, the use of traditional methods to impute missing data, such as using LOCF (last-observation-carry-forward), or performing completer analysis, etc., have been discouraged for many legitimate reasons. Despite this, unverified assumptions are required to utilize these new methods. Perhaps as a result, more attention has been given to sensitivity analyses that assess the robustness of the study results through an enhanced understanding of the potential impact of missing data on the study conclusions. In recent years, the ‘tipping point’ analysis method has found its use in many clinical trials to evaluate the impact of missing data. In this session, we would like to discuss the applications of the tipping point analysis method for various types of data, such as binary, survival, and continuous data in single measurement or repeated measurements setting; and to discuss the extent of this implementation in assisting study conclusions. The questions to be discussed at the table will include, but not limit to, (1)Experiences using the tipping point analysis (2)Suggestions on further improvement of this method (3)Help from statistics to inform on whether or not the parameters of the tipping point are clinically plausible.