How do you optimize a slow-performing SQL query that is running against a database containing millions of clinical records?
