Even though they had two primary endpoints (D and PD), they analyze these data in terms of three binary outcome variables (D, PD, and D or PD) and find that the p -value for D is not significant (0.1); for PD, it is highly significant (0.002), and for D or PD, it is at a borderline level of significance (0.048).
2
—
—
The sentences
The observed difference in mortality is thus not statistically significant when considered by itself but becomes barely significant when it is combined with PD.