When to adjust for multiplicity in cancer clinical trials

Interpreting cancer clinical trial results often depends on addressing issues of multiplicity. When testing multiple hypotheses, unreliable findings can occur by chance due to the inflation of the type I error rate, the probability of mistakenly rejecting the null hypothesis when the null hypothesis is true. In this setting, researchers may often set the type I error rate (or the alpha level) low to limit false positive findings and the interpretation of a causal relationship where none exists. Conversely, overly conservative type I error control may result in declaring findings, that do not meet multiplicity-adjusted alpha levels, as false when they are actually true, reducing opportunities for new discovery. This presentation focuses on multiplicity adjustment in the context of clinical trials conducted within the NCI's Community Oncology Research Program (NCORP). Because federally sponsored trials often require long-term participation from patients and represent a substantial investment by taxpayers, striking the right balance between optimizing what is learned from these trials, while avoiding false positive results, should be a priority.