To satisfy the proportional hazard assumption, we relaxed the criterion in the last step by allowing non-demographic variables of borderline significance (i.e. p value between 0.05 and 0.10) to be individually added into the Cox-PH regression model.
1
—
—