>>I think it works OK in SQL Server, but better to use field names. I got confused myself - thanks to Tracy for setting me straight.
>
>Very well, so it all comes down to the mention in Sergey's message about the high traffic in key changes which could have corrupted the index. So, I will have to check that performance on occasional basis.

So, what plans did you get when you tried both suggestions? Can you post them here?