[Reflections on some methodological errors in the evaluation of drugs. Ten years experience at the National Marketing...

A ten year experience at the national french marketing authorization committee has permitted to notice the most commonly methodological errors in the field of clinical research and, particularly, in the dossiers for drug approval: a frequent insufficient sample size resulting in lack of statistical power, an unsatisfactory optimal dosing research, a misuse of the so-called surrogate markers, an erroneous opinion about the meaning of the p value, an abusive claim for equivalence in non significant superiority trials, a misuse of unadjusted multiple comparisons and too much confidence in subgroup analysis results.