Some applications sometimes want to just read the secondary index (without the DB the index belongs to). In this case it is usually faster if you have a PCB referring directly to the secondary index. This is why you will sometimes find a PCB representing a secondary index.
However, when an application needs to access the data from the main database, but wants to use the secondary index for searching purpose, you code the PROCSEQ=secindex.
I guess that's clear, but just to make sure: Whenever you update your main database, the indexes will be updates automatically. There's no need to specify the secondary indexes in the PCB in this case.
I think IMS would allow you to perform updates on a secondary index. I believe this is usually a bad idea.