Discussion

Over the past 20 years, psychiatry has had a relatively low number of papers retracted compared with many other fields (in medicine, the specialties of oncology, cardiology and pharmacology were the highest) ^{Reference Okyay, Koçyiğit, Qumar, Yessirkepov and Sumbul16} and does not seem to have been affected to any major extent by ‘paper mills’, the profit-making journals producing low-quality or even fraudulent research (although there were some examples). The numbers of retractions, however, as in every other specialty, are generally on the increase. Most retracted papers in the Retraction Watch database were original articles, followed by reviews, as has been found in other specialties. The stated reasons for retraction were often vague – some had multiple reasons recorded, others had none. Most common was the nebulous reason ‘data error’. Some were identified as being errors by the journal. There were some cases of clearly stated research misconduct such as falsifying data or plagiarism.

We found that the UK, USA and China had the most psychiatry retractions, in contrast to Fanelli et al, ^{Reference Fanelli, Costas and Larivière17} who argue that generally nations with smaller scientific communities appear to have a bigger problem with retractions. Fanelli et al suggest that scientific misconduct is more likely in countries that lack research integrity policies, where individual publication performance is rewarded with cash, in cultures and situations where mutual criticism is hampered.

The psychiatry topics most commonly the subjects of retraction were mood disorders, service delivery and psychosis. Unlike Fang et al, we did not find Alzheimer’s disease or old-age psychiatry to be a common subject, ^{Reference Fang, Steen and Casadevall7} but it seemed that patterns of retraction were related more to individual journals and authors, or a particular conference, or even one investigator, leading to clusters by subject, rather than a subject itself being especially contentious.

Most authors only had one retraction, but there were some that were responsible for many, reported by Retraction Watch to have been subject to disciplinary procedures at their institutions because of problems such as plagiarism or data falsification. The one individual author with the most psychiatry entries in the database was Eysenck. Psychiatry may have relatively small numbers of retractions compared with some specialties, but it does have the problem of Eysenck’s work – there were seven in the database, although most are still expressions of concern rather than full retractions, prompted by the findings of a 2019 investigation by King’s College London, where Eysenck worked until 1983. That inquiry concluded that:

Some papers have crossed over to be covered in the non-psychiatric media. For example, there was a deal of controversy regarding the topic of ‘terminal anorexia’, leading to the retraction of a letter in the Journal of Eating Disorders in 2023. ^{Reference Yager, Gaudiani and Treem19} Papers related to sexuality have been retracted after criticism on social media or, for example, when the message of one was taken up by an alleged Neo-Nazi group.

Calls for retraction can appear on social media, the mainstream non-medical press or in academic journals. One of the most well-known controversies surrounds ‘paper 329’ on the antidepressant paroxetine, initially described as safe and effective for teenagers in 2001, ^{Reference Keller, Ryan, Strober, Klein, Kutcher and Birmaher20} but re-analysed in 2015 in The BMJ after years of pressure. ^{Reference Le Noury, Nardo, Healy, Jureidini, Raven and Tufanaru21} It has still never been formally retracted or even been subject to an expression of concern. The journal’s editors say the negative findings are included in a table, and that therefore there are no grounds to withdraw the article. This is not the only example of a paper that has, despite academic concerns being repeatedly raised, not attracted an expression of concern.

Rosenhan’s ‘On being sane in insane places’, published in 1973, launched a revolution in American psychiatry, leading professionals and the public to question psychiatric diagnoses, and is cited hundreds of times a year. ^{Reference Rosenhan22} Nearly five decades later, an investigative reporter attempted to track down the pseudo-patients used in Rosenhan’s study, but she was only able to find three of them (one of whom was Rosenhan himself and the other of whom was excluded from the publication). ^{Reference Cahalan23} Rosenhan’s study has been described as ‘one of the greatest scientific frauds of the past 75 years’, ^{Reference Scull24} but to this day it remains unretracted.

Our findings should be interpreted in light of several limitations. First, the data need to be viewed with caution because of the nature of the topic; how retractions and expressions of concern are recorded is not a uniform process. We explored one source because of its neutral and charitable status and its comprehensive exploration of practices against standards. Papers with the subject matter ‘psychiatry’ in Retraction Watch were extracted from the database, but on inspection, some were clearly not correctly categorised; it is possible that in the same way, relevant papers were missed if they had been misclassified. The judgement as to whether some papers could be considered to be psychiatry was subjective, but was decided by two psychiatrists separately, as is described as good practice. ^{Reference Kohl and Faggion25} Retraction Watch was chosen because it aggregates retractions across publishers, journals and databases, including cases where retraction metadata are missing, delayed or inconsistently applied. No other databases were examined, but Sebo and Sebo show that Retraction Watch Database (RWD) identified 90% of the retracted publications, which compared favourably with PubMed (84%) and Web of Science (86%). ^{Reference Sebo and Sebo26} They note that, although supplementing a RWD search with PubMed and Web of Science may be valuable:

The overwhelming majority of publications in our analysis pertained to original research, clinical studies and/or reviews and meta-analyses.

A major limitation is that this study focuses exclusively on formally retracted publications and EOCs. Although we identify that the presence of problematic and unretracted papers is an issue, systematically identifying such papers is beyond the scope of this analysis and arguably is not possible within the current publication governance structures. Better standards are required around recording and reporting at journal owner and editorial levels, perhaps on a public data base, to encourage public trust, interest and scrutiny.

In conclusion, it is difficult to know whether the increase in the number of retractions over the years is to be viewed as a sign of positive, improved editorial and publishing governance or of reduced standards because of the rise of unethical publications. It is likely to be a combination of both. Bhui et al have stated that ‘we need better guidance to deter legal threats designed to intimidate and influence editorial decisions; for example, legal threats could be posted on the journal website to ensure such threats are in the public domain.’ ^{Reference Bhui, O’Brien, Upthegrove, Tsai, Soomro and Newton–Howes5} Psychiatry might be regarded as one of the most political of medical specialties, given the multifactorial nature of mental illness and the influence of social, commercial and political determinants; debates about evidence are often imbued with allegations of human rights violations or of coercive practices, where evidence may be lacking or insufficiently developed in terms of mechanisms. Science, and arguable especially psychiatric science, is especially vulnerable to political interference in these troubled times. ^{Reference Aggarwal, Bhui, Jarvis, Raikhel, Rousseau and Kirmayer27}

Therefore, psychiatry inevitably includes research matters that are politically sensitive. Ideology and evidence frequently collide. So, for psychiatry and for all subjects, there needs to be a clear, accessible and accountable system if we are to maintain public trust and ethically advance science.