Maintaining the integrity of the scientific record

Editor—The average scientific standard of what pharmaceutical sponsors present to regulators is far superior to that observed by medical journals.¹ Despite laudable efforts recently by various editors in employing statistical reviewers, much still finds its way into print that is, essentially, nonsense. If the Medicines Control Agency did its gate keeping job as badly I would be alarmed.

Despite Smith's comments, ethical standards are often not superior outside the pharmaceutical industry. During my work for the industry I came across the following behaviour from the sort of external investigator that Smith would like the industry to use: the faking of data; the changing of procedures in the interests of personal research without approval being sought from either the company or the ethics committee; and refusal to agree to abide by the prespecified (and mutually agreed) analysis because this would jeopardise the possibility of publication.

One example of refusal to abide by the prespecified analysis was particularly illustrative of the arrogance towards the industry that may be encountered if one works in it. I was informed that my source of employment meant that my opinion regarding statistical analysis carried no weight. (This was a crossover trial, a topic on which I had written a monograph.)² A third and academic opinion was insisted on, and when this person, although chosen by the investigator, agreed with me the investigator then disagreed with us both. To my then employer's credit, it refused to be party to the publication that resulted, even though the results would have been more positive to it than the prespecified analysis was.

I believe that a much greater improvement in our evidence base would be obtained if the Medicines Control Agency influenced the standards observed by medical journals than if the editors of these journals influenced the regulatory process. My hope for the future is that sponsors and regulators will move towards publication of regulatory dossiers on the web. This development should not, of course, be seen as an attack on journal editors.