The rising network of scientific fraud

Scientific fraud has existed for as long as modern science has; documented cases date back centuries. With wrongdoers profiting off deceptive practices, cases have been sharply rising. A new study has uncovered a series of publications fabricating research, contributing towards the pervasive network of paper mills. 

The study titled ‘The entities enabling scientific fraud at scale are large, resilient, and growing rapidly’, published in the journal PNAS, identified footprints of activities connected to scientific fraud that have managed to escape traditional peer-review standards.  

Labiotech spoke to Jennifer A. Byrne, cancer researcher and co-author of the study, about the findings of her research and the growing problem of paper mills and disinformation. 

Scientific fraud and research paper mills: what’s the connection? 

Scientific fraud is defined as an act of deception or misrepresentation of one’s work that violates ethical standards, according to a research paper published in PubMed. Byrne pointed out that this misrepresentation is deliberate. 

“Fraud is different from people just making mistakes, which can happen by accident. They might be doing the best research that they can, and they might not know that what they have done is incorrect because of lack of training or knowledge and things like that,” said Byrne. “With fraud, generally, people on some level know that they’re not doing the right thing.” 

However, because fraud is about intent, arguably the only person who really knows their intent is that person, so it can be difficult to spot fraud. Still, there are factors that are regarded as a big no-no in the scientific community that point towards fraud. Byrne said that this includes stealing somebody’s ideas, publishing the same data more than once, and deliberately making up data. 

“Most people know that’s not the right thing to do,” said Byrne. “You can kind of prove beyond reasonable doubt that this data looks as if it must have been invented. It could not have happened in real life. Then, people start to think, well, that’s fraud.” 

And research paper mills are a clear indicator of this fraud. These are “undeclared sources” for churning out fake or low-quality publications. While there’s nothing wrong with utilizing an editorial service or even getting data from other people if this is revealed, paper mills are often written under the names of authors who have purchased that service. 

“They may have done a small part towards the paper. They may have done nothing towards the paper, but there’s no indication that the data have come from anywhere other than the authors,” Byrne said. 

Simply put, paper mills write the paper, invent data, and sell it to people who pay for it. The researchers of the study identified specific paper mills that have been running these operations. 

Case studies: fishy practices, journal hopping, and deceptive editors 

The authors investigated the Academic Research and Development Association (ARDA), based in India, which claims to provide “research and article writing services” for “research professionals who are desperate to have their research works known to the globe,” according to its website. 

Reese A. K. Richardson, first author of the study, along with Byrne and other scientists, found out that ARDA’s list of journals has expanded over many years. This was followed by databases delisting ARDA on suspicious grounds. 

It had listed 188 journals on its website in 2018, out of which only 56.4% had been indexed. Indexing is when journals are listed on a database, such as BioOne or PubMed. It’s often a sign of credibility and quality of a journal, and in turn raises the chances of a journal’s visibility. ARDA has been involved with more than 4,565 publications, and it charges between $250 and $500 per publication. 

What’s more striking is that 33.3% of the journals listed by ARDA that had been indexed were later de-indexed. this A journal is de-indexed when it no longer meets the quality and ethical standards of an indexing database, leading to its removal from the database. 

Moreover, every time a journal was de-indexed, it would disappear from ARDA’s website and be replaced with a new journal; an unethical practice known as ‘journal hopping.’ 

“To our knowledge, this is the first reported case of an entity engaging in fraudulent publishing that itself engages in journal hopping,” the study led by Richardson stated. 

Many articles that were published in the journals listed by ARDA were also found to be well outside the journal’s stated scope. For example, an article about roasting hazelnuts was in a journal about HIV/AIDS care. The researcher also discovered that another journal focused on special education had published an article about malware detection. 

This phenomenon of publishing unrelated articles in a journal was also observed in another journal called Aging, which is what set off the hunt for these kinds of journals. The journal claimed to specialize in the aging process, but one of Byrne’s colleagues noticed that this journal was generating several articles on cancer research, and rather basic cancer research. 

“They thought this was quite strange because there’s no obvious connection to this kind of cancer research and aging,” said Byrne. “We also realized that journals were publishing large numbers of papers that seemed a bit unusual. But then these journals would suddenly stop publishing these papers, and another journal would start publishing them.” 

Then, all of a sudden, these journals would get de-indexed, losing their impact factor, which is assigned to journals and reflects how often the articles are cited by either the authors or other people – a measure of recognition.  

“Once a journal is not indexed by databases and it doesn’t have an impact factor, it’s not a very attractive journal to publish in,” said Byrne. “So, what we thought was that these papers might come from paper mills, and the paper mills might be hopping between different journals. As one journal gets de-indexed, they’ll move to another one.” 

As the title of the research paper exposing these paper mills suggests, these fraudulent organizations are “resilient.” “If one thing stops working, they’ll move to another. They’ll change because they can do that very easily.” 

If these bad actors are just writing papers and inventing everything, it’s much easier to jump around journals, which sets them apart from genuine researchers, who take several years to change their research. 

“They have to finish doing the research that they were doing and start doing something else, and then finally publish that. And that takes a long time,” said Byrne. “But paper mills aren’t actually conducting any research at all, so they can switch their areas of focus very quickly, which is not a good thing.” 

It was also revealed that editors were colluding with paper mills to produce these fake papers in journals. One editor was involved with 79 papers at the journal PLOS One, out of which 49 were retracted. Editors, who had been flagged, edited 1.3% of the papers published in PLOS One as of last year, however, around a third of these papers were retracted. 

Richardson and team discovered a network; one where these editors consistently worked with specific authors – who were also editors at California-based PLOS One – and mostly only amongst themselves. Similar patterns with such editors were noticed in articles published by the controversial journal publisher Cairo-based Hindawi, which has had several research papers flagged in the past. 

RNA research targeted by paper mills 

Retraction rates are particularly high in the field of RNA research, the study pointed out.  

“There has been quite a lot of discussion around the possibility of paper mills producing papers in RNA research. That awareness has possibly given rise to higher retraction rates, simply because journals understand that these kinds of papers are being submitted,” said Byrne, adding that paper mill-linked retractions have occurred because of manipulated peer reviews too.  

Besides, in RNA research, different fields have had different retraction rates linked to paper mills, a lot of this coming down to awareness of the topic. 

So, for instance, CRISPR-Cas9, being a much more popular and studied field, had only about a 0.1% retraction rate. This is in stark contrast to microRNA and long non-coding RNA research, which saw retraction rates at 4%. 

“I just think it’s greater awareness amongst the journals and the journal editors. They are aware that these papers can be problematic,” said Byrne. 

Many of these papers contained blatant disinformation where reagents and polymerase chain reaction (PCR) primers – DNA sequences involved in the process of DNA synthesis – mentioned in the papers were wrong, as were cell lines that did not exist and manipulated images. 

“People are inventing actual cell lines. If you’re inventing a cell line, it’s very difficult to believe that you’ve actually done those experiments,” said Byrne. 

Europe’s largest paper mill exposed 

Meanwhile, what might be Europe’s largest paper mill was exposed by a team of investigators earlier this month. Around 1,500 research articles were linked to a Ukrainian network. These suspect papers have been published in 380 journals from authors based in Ukraine, Russia, and Kazakhstan since 2017. 

The paper mill was discovered by the research-integrity sleuth and scientist Anna Abalkina at the Free University of Berlin in 2022 when she noticed research papers with author e-mail IDs that had domains that did not match the geographical locations of academic institutions. 

Tracing more than 60 suspicious e-mail domains, she was able to link them to a paper mill – an enormous one. Like with Richardson, Byrne and team’s research, Abalkina spotted fake publications from manipulated images, plagiarism, fake peer reviews and fake reviewers, paid-for citations, and unusual patterns of co-authors. 

While many of these bogus papers have been flagged, the numbers are only growing. And Byrne believes that with the advancement of artificial intelligence (AI), this has become a burgeoning problem too widespread to tackle. 

“Peer reviewers have seen very large numbers of certain kinds of manuscripts, and that’s only been quite recent. What has changed? The big change has been the availability of generative AI,” said Byrne.  

Experts fear AI to worsen scientific fraud 

The advent of generative AI not only means that content can be falsified but also that fake photos can be generated that go undetected. While research sleuths hunt for fake images, new image detection systems need to be incorporated to identify AI-generated ones. As these tools need to keep up with the pace, it is challenging. 

Byrne said: “We are worried that AI is going to make the existing problem significantly worse because it’s much easier to produce papers with AI, and it might be harder to detect some of these papers because some of the features that we’ve perhaps been studying in the past might be disappearing from these papers.”  

And it’s not just the tools that need to keep up with AI. As it takes a long time to train new peer reviewers, Byrne explained that there are not enough of them to intercept fraudulent research from being published. In fact, reviewing fake manuscripts should not even be part of their job in the first place, according to Byrne, as this only leads to delays in reviewing genuine papers. 

“Some journals describe being overwhelmed by these sorts of scaled submissions. They can’t cope. It’s a bit like during the COVID pandemic, all of a sudden, we have hospitals overwhelmed with COVID patients, and you can’t treat people that are having heart attacks or cancer,” she said.  

But it’s a catch-22 situation considering there are not enough trained peer reviewers to review genuine papers, as they are busy tracking down the fake ones, but if they don’t chase after the fake ones, other researchers as well as the public are at risk of being misled. 

“I think the whole publication process risks slowing down,” said Byrne. “It’s certainly a very big problem. 

Spreading awareness, the only way ahead 

With scientists likening how paper mills operate to a cartel, Byrne thinks that the way forward is to create awareness on the matter, especially as it is understudied in the scientific community. 

“I think awareness is the first step… Awareness amongst peer reviewers, awareness amongst journal editors, and awareness amongst readers. If readers are aware of paper mill papers, they can avoid them and not read them. That’s a very easy thing that scientists can do,” she said.  

This can spur journals to take more aggressive steps to not publish fraudulent papers. Experts propose that a higher rate of corrections and retractions will help curb the numbers. 

“We can do that, but the journals, at the moment, don’t tend to put many resources into correction. They would rather publish papers because that’s what earns money,” she said. “If we don’t get on top of it, the problem is going to be much bigger quite soon.”