Viewpoint

Introduction

There is a crisis in ‘evidence-based medicine’ (EBM). Reviewers and meta-analysts have become aware that much of their work may be futile because some investigators only publish selected results that favour their own point of view. Such reviews can only be reliable if there is access to all the raw data or if everyone is compelled to publish ‘negative’ results too. This is proving difficult and controversial. The latest example is dabigatran, which was hailed as a safe advance that needs no coagulation monitoring. It seems that the reduced frequency of bleeding if there is coagulation monitoring was not reported. There is also widespread uncertainty about the thresholds for treatment. This has been highlighted in the ‘over-diagnosis campaign’ to address a concern that many patients are subjected to harm without much benefit. Many see the problem as one of vested interest e.g. where those who gain from supplying a treatment will tend to advocate a low threshold for treating whereas those who wish to reduce costs or effort prefer to set high thresholds. The evidence needs to support what is best for the patient and gathered by those who care for them e.g. acute medicine physicians.