I would expect no issues if you concentrate solely on IN_ROW_DATA. As I mentioned before, the only option for indexing BLOB type data is with full text indexing, so that won't show up in sys.dm_db_index_physical_stats. Re-organizing BLOB type data would be similar to re-orgnizing a heap, and involve a full export and re-import of the data, and hoping the data lands in the datafile in a contiguous manner. If BLOB fragmentation is a problem, I would move to a Filestream setup, and let the OS handle the fragmentation.