Rising Retraction Rates: A Symptom of a Strained System

Publishing research is a key part of a scientist’s career. Most researchers are excited to share their findings with their peers and the wider scientific community, and the news of an article being accepted for publication is a reason for celebration.

However, sometimes a study later turns out to be significantly flawed, either due to innocent error or more egregious breaches of research integrity, and either the authors or the publisher retracts the work. In 2015, the website Retraction Watch launched a database to track retractions across research fields. Today, it contains more than 64,000 articles, though some of these are categorized as having expressions of concern, corrections made, or been reinstated.

The data reveal that the number of retractions has almost steadily increased over time, yet this trend may not be the harbinger of failing science integrity that it seems. Some researchers say it may reflect an increased focus on correcting published literature. Others argue it has little effect on the community overall. Regardless, they all agree that a much bigger issue is at play, one which may actually drive the behaviors reflected in retractions: the emphasis on publication metrics in scientific careers.

Rising Retractions Could Have Many Causes

Jennifer Byrne, a cancer researcher-turned cancer research literature investigator at the University of Sydney, became interested in the question of integrity in her field’s publications when she came across several suspect articles. The publications, many of which allegedly came from entirely different authors and institutions, all touted the role of the same gene in several different cancers and had incredibly similar article structure. Today, Byrne and her group are investigating the impacts of these types of studies on the field, currently focusing on papers referencing cell line models that don’t exist.

Byrne said that the problem with these studies is that they lead to scientists wasting their time and resources. For example, she said that if researchers come across methods or topics that they think may be relevant for their group, they may try to replicate them. “If that experiment was not described honestly and actually never took place, then clearly reproducing the results is going to be very challenging,” she said. “That's the kind of damage that these papers do.”

Looking at the trend of retractions from health science and basic life science articles from the Retraction Watch database, Byrne said that she sees the increase in number of retractions over the years as positive. She said it may show a growing commitment to remove faulty science from the literature.

Ivan Oransky, the cofounder of Retraction Watch and executive director of its parent nonprofit organization, the Center for Scientific Integrity, echoed this idea. “I think it’s a good thing,” he said. “It means that the record is being cleaned up.”

However, the increase may also indicate a growing trend of academic research misconduct. Meta-analyses from 2009 and 2021 showed an average increase in falsification, fabrication, or plagiarism of 1.97 percent to 2.9 percent.^1,2 As of April 8, 2026, scientific misconduct, including these examples and other ethical violations, accounted for 60 percent of retractions. This is similar to rates found in previous reports.³

Despite this, Oransky said that the reason for the increase in retractions likely has more to do with a greater number of people looking for examples of unethical research, likening it to cancer screenings. “If you don't screen for it, it's going to seem as though it rarely happens. You might see a very late-stage case show up at a doctor's office, tragically, but it's going to seem like a much lower prevalence than it actually is. And so, you essentially have lots and lots of screening tests now,” he said. Oransky added that it is also easier to find fraud now, since more papers are available online and there are more tools to compare image and text discrepancies.

Even though retractions have increased overall, compared to the number of articles published in the same year, retractions represent less than one percent of the literature in that timeframe. In fact, both Oransky and Byrne said that more retractions are probably needed to remove the majority of the unreliable studies from the literature, but the estimated increase would still make flawed research a rare proportion of science articles. “Most people are honest and trustworthy and really try their best. And that's why I think we need the literature to be as trustworthy as most scientists are,” Byrne said. However, even when contacted about suspected research integrity concerns, Oransky and Byrne said that publishers don’t always respond to requests to flag or remove the articles.

Continue reading below...

Article

Steps to Improve Sustainable Laboratory Practices

Laboratory workers can shape sustainable practices without affecting scientific quality.

Read More

Byrne said that it is important to address these unreliable studies. “Without any flags, all of the literature looks clean, and that's just simply not true in some parts of the literature. So, post-publication corrections of any kind are really, really important.”

While immunologist and bibliometrics researcher Mark Hanson, at the University of Exeter, agreed that indicating an article is low quality can benefit newcomers to a field, he’s not concerned about some flawed articles being in the literature. “I think that the community just moves on from bad studies and bad information,” he said.

Although he said that fraud should be made evident, “You don't need to retract the papers. You should absolutely put an editorial note of concern on them.” Hanson added, “I don't think retractions do all that much.”

Instead, Hanson said that a much more pressing problem is the dramatic overall increase in science articles that has occurred in recent years, which is overwhelming publishers and researchers alike.⁴ Since 2011, more than one million biomedical papers have been added to PubMed every year, with this number exceeding 1.5 million additions per year in 2020.

“Scientists are increasingly feeling overwhelmed with the requests to review, the number of papers to publish, and also the drain of resources of funds towards science,” Hanson said.

One study estimated that these efforts cost scientists one billion dollars a year just in their time spent on peer review.⁵ Hanson said that, in terms of research advancements, many of these papers didn’t even need to be published. As an example, he pointed to many institutions’ requirement of a publication for degree conferral.

The root of this issue, Hanson said, is the motivation around publishing in academia. “If you incentivize people to publish papers in order to advance their careers, they will publish papers however they can,” he said. Sometimes, this leads to submitting lower-quality studies for publication, which is increasingly easy with many publishers willing to accept papers after a minimal peer review so that they can collect the publication fee.

Oransky said that the publish-or-perish mentality also promotes unethical behavior that leads to retractions; previous studies on the topic support this observation.^6,7 Models have also shown that the incentives to publish positive results and in high-impact journals lowers the trustworthiness of science literature overall.⁸

Additionally, this pressure drives the existence of paper mills, which take advantage of researchers’ need for publications by offering authorship on manuscripts with fabricated or stolen data; these fraudulent networks are growing rapidly.^9,10 Paper mill activity accounted for 20 percent of the health science and basic life science article retractions between 2016 and 2025, which was when this reason was added to the Retraction Watch database. A large spike in retractions in 2023 occurred when the former publisher Hindawi retracted thousands of articles that were discovered to have originated from a paper mill.

Now with generative AI, researchers anticipate that the problem will get worse before it gets better. “Publish-or-perish is kind of the kindling. Paper mills were gasoline. AI is an accelerant, so you’ve got a raging fire now that it's unclear if the firefighter is going to be able to put it out,” Oransky said.

Byrne agreed. “I feel that it highlights even more the need to value papers beyond numbers because numbers are now more easily achieved than they probably ever have been in the history of science.”

Improving Scientific Literature Requires Less Emphasis on Publishing

To address the career incentives to publish more papers, Oransky said, “We need to sort of really slow down and eventually abandon the use of [publication] metrics in research assessment.” One group, the Declaration on Research Assessment, proposes and advocates for new measurements of academic success.

Doing so could have other benefits beyond curbing research misconduct and publishing strain, too. Researchers demonstrated that given the changes to the publication landscape over the years, a researcher’s number of citations and published articles as well as the impact factor of the journals in which they publish are no longer useful measurements in academia.¹¹

In the meantime, though, Oransky and Byrne hope to see a continued interest in identifying and removing fraudulent and problematic articles from the literature. Byrne said that incentivizing corrections could be one way to make this a more routine part of the publication process.

Hanson agreed that more work should be done to remove incentives in publishing to address the deluge of scientific literature and strain on the system. He said that funding bodies can refuse to pay for certain publications, such as special issues that are large exploiters of low-quality articles. Alternatively, he said that moving grants to a lottery system or incorporating narrative curriculum vitae, in which researchers select and share only what they consider their top publications for funding bodies to review, could reduce the emphasis on one’s quantity of publications as opposed to the quality of their research.

Although he respects the work of people rooting out examples of scientific fraud, Hanson said that focusing on retractions is a losing game. “Hunting down all the bad actors and retracting all the individual articles doesn't help because it's like trying to play whack-a-mole except the molehill is the size of Mount Everest,” he said. “Addressing the fact that there's an entire subterranean culture underneath that hill is going to be more challenging, but also more permanent as a solution.”