Every clinical trial has a pre-specified statistical analysis plan that establishes the necessary statistical threshold for the primary endpoint to claim success. This statistical analysis plan also includes a hierarchical plan for how data will be analyzed for the secondary and other pre-specified analyses. Often, the first analysis in the hierarchical order to fail the significance threshold will mean that subsequent analyses are no longer valid and only exploratory.

Statistical significance thresholds are somewhat arbitrary; in reality not much different between p=0.04, 0.05, 0.06. But if p=0.05 is the necessary stated threshold for statistical significance, then unfortunately p=0.06 is interpreted as not significant. That's the reality. EMPA-REG OUTCOME achieved statistical significance for  superiority with p=0.04. ASSURE failed to achieve statistical significance with p=0.08 for the primary endpoint. Hopefully, BETonMACE achieves a p-value on the primary endpoint well below 0.05 so there is no question.

BearDownAZ