This is a textbook case of misinterpretation of Type II error due to heterogeneity and endpoint misalignment as rejection of the null hypothesis. It seems to have been designed to fail.