Getting real about evidence in surgery
When evidence based medicine (EBM) began in the 1990s, great emphasis was placed on the “right” methodology for studying different questions. For comparing treatments, the arguments for randomised controlled trials (RCTs) were irrefutable. Randomisation was the only reliable way of minimising bias in treatment allocation.
The way this was interpreted, however, led to difficulties. As RCTs were the only “good” design, non-RCTs were regarded as having little value. Subjects like surgery, where RCTs were rare and often small, were criticised,1 and surgeons were portrayed as lacking scientific rigour. In other areas of EBM, however, practical barriers to using the best design were already accepted. For instance, to evaluate a new diagnostic test, the best design is a parallel independent comparison with a “gold standard:” but where no gold standard exists, a reference standard is substituted.
The criticism of surgeons proved to be creative, as we began to think seriously about