How to think like an epidemiologist