[Some reflexions about double blind]

Randomized, double blind, placebo-controlled clinical trials are currently the best means of demonstrating the clinical effectiveness of drugs. The double blind procedure, when ethically and technically feasible, is a necessary condition for validating results and for causal attribution of the observed difference between the two groups to the tested drug's pharmacological effect.

In practice, however, it appears that patients and independent investigators can guess who receives the drug and who receives a placebo through side effects, which are usually more frequent in patients receiving the drug. This phenomenon effectively "breaks the blind" and represents as such a major methodological bias, which cannot be avoided as it is inseparable of the drug's effect.

The impact of this "unavoidable" double blind breach nevertheless remains unclear. While it can reasonably be assumed that it may modify subjective symptoms such as anxiety or pain through suggestion, its influence on objective criteria remains to be demonstrated.