Correspondence (letter to the editor): Search Bias

Clinical studies are intended to confirm characteristics of medical drugs. For this purpose, a study design is defined, which always means immediate sponsor bias. This is independent of whether the sponsor comes from pharmaceutical, academic, or independent institutions.

There is no legal requirement to publish. In spite of this, many pharmaceutical companies—additionally to Good Clinical Practice (GCP) and the Declaration of Helsinki—have subscribed to an ethical code according to which collected data have to be published.

Publication, however, presents a central problem. Positive studies are often published, and with enthusiasm; negative studies are rarely published or merely in review articles (publication bias). This is only very partly the pharmaceutical industry’s responsibility, because the potential for publication is actively orchestrated by the media, by insisting on exclusivity and selective publication of studies that meet medical-therapeutic or financial interests.

In describing the methods of the article under discussion, the authors say that suitable studies from a system literature search of PubMed were selected. This, of course, means selection bias: Studies with negative results or drug studies of lesser medical interest will rarely be indexed in PubMed. For drugs in the special therapeutic fields (such as phytotherapeutics, homeopathic, or anthroposophic drugs), hardly anything is listed in PubMed.

I have to express my objections to the underlying implication that pharmaceutical companies can do on the drug market as they wish. There are laws and authorities that regulate the drug market. All studies that have ever been conducted into a medical drug are known to the authorities. Without their approval, no drug will be licensed. If after approval of a drug negative results do indeed indicate a problem for the health of the public, then the authorities will step up to the mark—entirely independently of any publication.