Evidence of mechanism in the evaluation of streptomycin and thalidomide.

This paper considers what evidence is needed to establish the effectiveness and safety of a drug therapy. The claim that A cures D is a particular case of a causal claim in medicine. So the paper begins with a general analysis of the evidence for causal claims in medicine. Such evidence is divided into two types: statistical evidence and evidence of mechanism. These are further divided into observational and interventional, producing a 2x2 classification. It is shown that historically there have different assessments of the importance of these different types of evidence. Evidence-based medicine (EBM) puts forward the thesis that claims of the form 'A cures D without harming the patient' can be established using only randomized controlled trials or RCTs. This thesis of EBM is criticized by considering two historical examples: streptomycin and thalidomide. Generalizing from these, it is claimed that the effectiveness and safety of a drug therapy can only be established by using both statistical evidence and evidence of mechanism. This is a specific instance of the Russo-Williamson thesis.