Were there any changes to application code or stored procedures?

Is it possible that the database was altered to set read_committed_snapshot on?

Is is possible that the users just stopped using the part of the application that was causing the deadlocks?