Hi,
As told by our friend,the need for the index really depends on the data access pattern used by your application.If your data is evenly distributed and has high cardinality and the application is experiencing heavy DML activity then create a b*tree index .Also ensure that the overhead of accessing the index and then reteriving the data is less when compared to perform a full table scan.As the main purpose of index is to provide faster data access in the table.And any index look up should result in fewer index block read and result in faster data reterival from the table.

hope this helps u

regards
anandkl