There are a dozen different things you can do to help determine if a database is being used, but none of them are entirely reliable as a standalone solution.
There is one guaranteed way to determine if a database is being used... do the BSOFH thing. Take a full backup followed by a "Tail Log Backup" and then wait for the instant messages to start and trouble tickets to start popping up. 😉 If none of that happens for a year, drop the database and put the final backup files in long term storage.
And this is what I call "User Driven Documentation" (UDD). When all indications suggest the database is not in use, I turn to UDD. 😉