[Facing the unreliability of clinical trials literature].

Journal publications of randomized controlled trials ("literature") have so far formed the basis for evidence of the effects of pharmaceuticals and biologicals. In the last decade, progressively accumulating evidence has shown that literature is affected by reporting bias with evident implications for the reliability of any decision based on literature or its derivatives such as research synthesis. Another important factor is the growing body of evidence of the fragility of editorial quality control mechanisms in biomedicine ande their easy exploitation for marketing purposes in the symbiosis between publishing and the pharmaceutical industry. Regulatory documents are probably more reliable than currently accessible other sources but there are many severe limitations to the long-term use of regulatory documents for research synthesis and decision-making. Instead of trying to reform the fields of research, industry, government, regulation and publishing, I propose basing public health decisions and reimbursement of any important interventions on independent trials and studies following the model pioneered by the Mario Negri Institute of Pharmacological Research.