Richard Smith: Amateurism still flourishing in scientific journals

richard_smith_2014In 1995 Stephen Lock, once editor of The BMJ and effectively the first person in Britain to be seriously concerned about research misconduct, called for an end of amateurism in the editing of scientific journals. He made his plea after reviewing a report on the gross failures of the editors of the British Journal of Obstetrics and Gynaecology in publishing two fraudulent papers which had the editor of the journal as an author on one and an assistant editor as an author on both. But more than 20 years later a report in the Toronto Star appears to show that amateur editing is flourishing as ever.

The report in the Toronto Star describes how a review of more than 1400 papers by Gideon Koren, one of Canada’s most prolific scientific authors, “reveals the inability—and unwillingness—of journals and research institutions to preserve the integrity of the scientific record.” Many journals had been notified of problems in the papers by Koren, but most of them had done nothing. The article quotes Ivan Oransky, founder of Retraction Watch, describe ‘“the vaunted self-correction mechanism of science” as one that is “held together by spit and bubble gum.”’ Retraction Watch, which reports almost every day on malfeasance in science, has brought to global prominence what Lock knew a quarter of a century ago–that research misconduct is common, poorly acknowledged, and inadequately managed.

Journals are only part of the weak system for preventing and responding to research misconduct, but they are the place that new scientific articles are sent. Peer review is, we know, poor at detecting such misconduct—because it works on trust: if authors say that there are 200 patients in a study then that is simply accepted, nobody asks for signatures or consent forms (which could anyway be forged). Similarly if authors say they have no conflicts of interest their statements are just accepted.

But sometimes fraud is so obvious that it is detected before publication, in which case journals have the unenviable task of acting. Many do nothing but simply reject the article, which is a failure of duty of care. Journals have neither the legal legitimacy nor the due process needed to conduct a proper inquiry into fraud, but they should not do nothing. They have to notify the employing organisation, usually a university, who do have legal legitimacy to investigate and should have due process. One common complication is that authors come from multiple institutions, making it easy for them to pass the buck. In the old days, as Lock explained in his editorial, it was common for everything to be hushed up, but in the age of transparency and social media that’s become a high risk response. Nevertheless, many institutions do little or act very slowly. There is then an onus on editors to chase the employers, but many think that they have discharged their duty once they’ve notified the employers. Editors do have the sanction of publishing their concerns and reporting that the institutions they have notified have failed to act. They should use that sanction when all else fails.

What is much more straightforward for editors is to take action when notified by employers that they have concluded that there has been misconduct. The editors should go ahead and retract the article, explaining exactly why they have done so. But many do not do this. Sometimes they may not do so for fear that it will reflect badly on their journal, but they are completely wrong: all publications make mistakes (and, as I’ve said, fraud is hard to detect), and good ones publish corrections and retractions, whereas poor ones publish few of either. It is not a problem to make a mistake, but it is to fail to correct it.

One of many odd things about scientific journals, which are at the heart of the scientific process, is that most of them are edited by amateurs. One day you are a cardiologist and professor of cardiology, the next day you’re an editor as well. It would never work the other way round, with an editor becoming a cardiologist overnight.

Dealing effectively with the many forms of misconduct is one of the jobs of an editor. Employing professional rather than amateur editors may not mean an end to journals failing to deal effectively with fraud, but it would be a step in the right direction. And the editors don’t need to be grand figures: indeed, humble editors with training and experience who follow protocols would be much more effective than the current grand figures with impressive academic backgrounds, but no editorial experience.

Unfortunately Lock, who is now 89, will have to wait longer for an end to amateurism in scientific publishing.

Richard Smith was the editor of The BMJ until 2004.