If you’re going to go the DBCS route then ‘O’ is what you will want 99.99% of the time. O allows both DBCS and SBCS to exist in the same field concurrently. E allows both DBCS and SBCS to exist in the same field, but NOT concurrently. That is E means field A of record 1 can be DBCS while field A of record 2 can be SBCS, but you cannot have both (DBCS and SBCS) in the same field, at the same time, for either record. O is just a whole lot more flexible.
You may also consider using G and a Unicode CCSID. DBCS is nice, but you cannot combine different DBCS encodings in the same file. So if you go DBCS (O or E) you can support Japanese in a file, or Korean in a file, or Simplified Chinese in a file, or Traditional Chinese in a file, but you cannot have ONE file with both Japanese and Simplified Chinese. You have to have multiple files in this situation. If you go the Unicode route you can have one file that contains the characters of all the DBCS languages, not to mention the SBCS languages, concurrently. Going Unicode is more work, but you also get a whole lot more flexibility!