Yeah, it's best to steer clear of core database alterations without having solid knowledge of how the core is designed. db_replace() takes into consideration moodle's idiosyncrasies. As you probably saw, it skips some tables and runs a function in blocks (if they implement it).
It usually works flawlessly, but technically still in beta (hence why it's not in the admin tool tree).
Afterwards, you can search the database for any remaining URLs, but exercise caution manipulating it.
Glad to have helped
Matt