The line in there that would worry me is this one:
Scalar UDFs are generally accepted as one of the worst things you can do performance-wise in T-SQL. However, they may have earned a reputation that they simply don't deserve because a common method of measuring performance (SET STATISTICS TIME ON) injects a major performance problem of its own when used to measure the performance of Scalar UDFs.
Would it improve performance if this were converted to a view with some good indexing?