In these days of evidence-based medicine, you can’t do much better than a well-designed, double-blind, randomised controlled trial (RCT).

It is the staple ingredient of the systematic reviews and meta-analyses that we use to make sense of the burgeoning medical literature, hopefully distilling data into digestible bites that can inform clinical practice.

But what happens when researchers throw something new into the mix and end up with a better cake?

Because evidence evolves, Canadian and US researchers set out to determine how often RCTs were overturned by later research.