I have compared dE/dx distributions for AuAu 19 and pp500 using whole dE/dx sample and sample  with rejected runs taken from Yan Yang's bad run list.

I see that for AuAu 19 rejection helps to remove extra band for low dE/dx. 

For pp500 the rejection does not help. This means that there are still  some problematic tpcAnodeHVavg tables for runs which are not marked as "bad".

The plots :

1. pp500:

2. AuAu19