No. Eliminate as many sources of bias as possible.
The effect however is to reduce the patient population from which events may be drawn.
For example, I'm sure many people think that any death would count as a progression event for the primary analysis of the primary endpoint. That's not true.