The stats they quote are based on clinical studies. Many studies have been done on chemo, herceptin and aramatose inhibitors and clinical trials and follow up over years is what determines it. The trouble with collecting all data is the sheer volume of people, compliance, if treatment was finished etc etc which is a lot of info to take on everyone I guess. Studies are snap shops...so for every 1000 women...this happens for instance. And then repeated with other studies. After collecting multiple studies and analysing (meta-ananlysis) is what leads to those recurrence tools like predict in UK. Studies are occurring all the time and across the world.