164 – Subgroup Analyses: How Friendly Are They?
Subgroup Analysis Findings in Safety with No Statistical Power: What Do These Really Mean?
Melvin S. Munsaka
Takeda Development Center Americas, Inc.
Various regulatory guidelines require safety data to be presented by subgroups of interest, including gender, age, race, and other factors of interest. For the most part, subgroup analyses of safety data tend to be more appropriately performed in the context of integrated analyses of safety for the obvious reason of the availability of large amounts of data which are useful, for example, in getting better estimates of incidences of adverse events and also allow for assessment of trends in subgroups which may not be possible with study level data. The subgroup analyses are typically a repetition of the tabular presentations done for the entire data. It is often not clear as to the purpose of these subgroups analyses, but could include addressing questions such as, are adverse events the same across subgroups and within and between treatments, is the time of occurrence of adverse events the same across subgroups, and so on. In this discussion, we will consider the question of subgroup analyses within the context of safety data and review some of the approaches used. Challenges related to the interpretation of findings from subgroup analyses of safety data will also be discussed.