Retracted papers about COVID-19 are more highly cited than they should be

Retraction, in theory, is supposed to work like a giant fluorescent warning sign. It tells the scientific community, “Careful, this paper is no longer reliable.” In a perfect world, that warning would stop the paper from being treated like solid evidence. In the real world, especially during the COVID-19 era, that sign often functioned more like a sticky note on a runaway shopping cart.

That is the uncomfortable truth behind retracted COVID-19 papers: many of them were cited far more often than they deserved, and some kept collecting citations long after they had been pulled back. In other words, the paper was out, but the idea kept wandering around the neighborhood like it still had a mailbox key.

This matters because citations are not just academic confetti. They shape future research, influence clinical discussions, feed systematic reviews, and sometimes leak into news coverage, policy debates, and social media arguments with the confidence of a person who read half a headline and decided they were now an epidemiologist. When a retracted paper keeps being cited as if it were trustworthy, the correction system is not really correcting much.

Retraction should be a brake, not a booster rocket

The clearest lesson from the pandemic is that retraction does not automatically erase influence. Several analyses of COVID-19 literature found that flawed or withdrawn papers kept circulating well after the scientific record had been corrected. One especially striking study found that retracted COVID-19 articles were cited, on average, far more often than the typical article in the journals that published them. That is a weird and alarming achievement. These were not obscure papers quietly fading into the wallpaper. They were loud, visible, and sticky.

Another analysis found that retracted COVID-19 publications continued gaining citations even after databases and search results began marking them as “retracted” or “withdrawn.” That means the problem is not simply whether a warning exists. The problem is whether researchers notice it, understand it, and change their behavior because of it. Too often, they do not.

The pandemic magnified this issue because COVID-19 papers were moving through the system at absurd speed. Journals were racing, preprints were exploding, public demand for answers was relentless, and every new result felt urgent. The scientific enterprise was trying to build the plane while flying it through a thunderstorm while millions of people watched from the runway and live-tweeted the bolts coming loose.

What made COVID-19 retractions so unusually visible?

Fast publishing created fast mistakes

COVID-19 research was published at a pace modern science had rarely seen. Review timelines compressed. Preprints spread before peer review. High-impact journals rushed to publish clinically relevant findings because doctors, policymakers, and the public were desperate for direction. That speed produced useful work, but it also created more room for weak methods, shaky data, and overconfident conclusions.

Some reviews of COVID-19 pharmacotherapy papers found that even high-profile journals published retracted studies after extremely rapid turnaround times. That is not proof that speed always ruins quality, but it is a reminder that velocity is not a substitute for rigor. A paper can arrive early and still be wrong. In fact, some of the earliest, splashiest, most heavily discussed COVID-19 papers turned out to be exactly that.

Big claims travel farther than quiet corrections

There is also a simple media problem: dramatic claims get attention, while retraction notices get homework. A paper that appears to answer a huge question about treatment, risk, or public health can spread through journals, blogs, group chats, conference talks, and news reports in a matter of days. The later retraction notice, by contrast, often arrives in a smaller package, with less fanfare, fewer reposts, and no headline writer trying to make it trend before lunch.

The best-known example is the pair of notorious COVID-19 papers tied to the Surgisphere scandal, including one in The New England Journal of Medicine and another in The Lancet. Both were retracted after concerns about unverifiable underlying data. Yet the influence of those papers did not vanish on cue. Researchers kept citing them, and many citations failed to clearly note that the papers had been retracted.

This is how zombie citations happen. The original claim enters the ecosystem wearing a tuxedo. The correction arrives later wearing sensible shoes and carrying a clipboard. Guess which one gets invited back into the building.

Retraction notices are not always clear enough

Retraction is not one standardized universal experience. Some notices are prompt, detailed, and impossible to miss. Others are vague, delayed, incomplete, or inconsistently linked across databases and publisher pages. During the COVID-19 period, several researchers pointed out that retraction practices remained uneven, which made it harder for readers to understand exactly what had gone wrong and whether a paper should still be discussed only as a cautionary example.

That inconsistency matters. If a notice does not clearly identify the article, explain the reason, and stay tightly linked to the original record, then a reader working quickly may miss the warning entirely. Search engines, downloaded PDFs, reference managers, screenshots, old news stories, and lecture slides can all preserve the paper in its pre-retraction form. Once that happens, the flawed paper starts living multiple afterlives.

Citations are often inherited, not freshly checked

Another reason retracted COVID-19 papers keep being cited is painfully ordinary: some authors do not verify every reference at the last minute. Citations get copied from earlier papers. Drafts sit for months. Literature reviews are assembled from citation chains that feel trustworthy because they have already been used elsewhere. By the time a paper gets submitted or accepted, a retraction may already have happened, but the reference list is still carrying yesterday’s version of reality.

That is not always malicious. Sometimes it is just bad workflow. But bad workflow at scale becomes a real integrity problem. One analysis of ongoing citations to a retracted COVID-19 study found that most verified citations appeared months after the retraction, and only a small minority explicitly noted that the paper had been withdrawn. That is not a little oversight. That is a systems failure wearing glasses.

What the numbers actually tell us

The broad pattern is hard to ignore:

1. Retracted COVID-19 papers were cited at rates that often exceeded what would normally be expected for papers in the same journals.
2. They continued collecting new citations after retraction rather than falling off a cliff.
3. Warnings in search results or metadata did not reliably stop future citations.
4. Many citing papers failed to state that the source had already been retracted.

That combination is the heart of the problem. A retraction is supposed to lower the evidentiary volume of a paper. Instead, many COVID-19 retractions merely changed the paper’s legal status while leaving its social life weirdly active.

There is another twist here: not all retractions are about fraud. Some are due to major error, flawed analysis, data problems, duplication, ethics issues, or problems that make the findings unreliable for different reasons. During COVID-19, several studies suggested that retracted COVID-related papers were often pulled for reasons other than classic misconduct. That is important, because it means the pandemic did not simply reveal more bad actors. It also revealed what happens when scientific systems are pushed into emergency mode.

Why this matters outside journals and university offices

It would be tempting to treat this as a niche problem for librarians, editors, and people who enjoy arguing about reference formatting. It is not. Retracted COVID-19 papers can influence patient care discussions, public understanding, and downstream evidence synthesis. When they are cited in reviews, commentaries, or summaries without a clear warning, they can keep shaping decisions long after their findings should have been quarantined.

Systematic reviews and meta-analyses are especially vulnerable if retracted material slips into the evidence base. Once a flawed paper is absorbed into a higher-level summary, the error becomes harder to spot and more prestigious in appearance. It no longer looks like one bad paper. It looks like part of the literature. That is a serious problem in any field, but during a global health emergency it can become a public-facing problem very quickly.

There is also a trust issue. Critics of science often point to retractions as proof that the entire research enterprise is broken. That is too simplistic. Retraction is actually a sign that self-correction exists. The real embarrassment is not that bad papers appear. The embarrassment is when the correction machinery works on paper but fails in practice, allowing invalid claims to keep collecting authority points.

How to stop retracted COVID-19 papers from becoming immortal

Authors need a final citation check

Every manuscript should undergo a final reference audit shortly before submission and again before publication. Not glamorous. Not thrilling. Extremely useful. Authors should verify whether key cited studies have been corrected, retracted, or replaced. If a retracted paper must be discussed for historical or methodological reasons, the citation should clearly say so and, ideally, point readers to the retraction notice.

Publishers need clearer warning systems

Retraction notices should be prominent, specific, free to read, and tightly linked to the original article. PDF watermarks, metadata updates, consistent labeling, and visible banners all help. A notice hidden behind polite formatting is not doing the job. The message should be unmistakable: this article remains part of the record, but its findings should not be treated as reliable evidence.

Databases and discovery tools need to be louder

PubMed and related systems have long marked retracted publications, which is good. But the wider discovery ecosystem is fragmented. People find papers through search engines, saved PDFs, citation managers, social platforms, institutional repositories, and plain old forwarded links. Warnings need to travel with the article, not stay confined to one neat database page that only the most careful reader checks.

Editors and reviewers should treat retracted citations as red flags

If a manuscript cites a retracted COVID-19 paper without explanation, that should trigger a query during peer review and editorial screening. It is one of the easiest quality-control catches available. No statistical wizardry required. Just a basic question: “Do the authors realize this source has been retracted?” Science deserves at least that much housekeeping.

Experience from the COVID-19 retraction era

One of the most revealing parts of this story is not just the data, but the experience around it. During the pandemic, many researchers, clinicians, librarians, and science journalists lived through a bizarre information environment in which papers could become influential before anyone had time to digest them properly. A study would appear in the morning, circulate in newsletters by noon, show up in social posts by dinner, and become part of a debate before the methods section had been read by more than six people and one stressed editor.

For clinicians trying to stay current, this created a daily whiplash effect. They were not browsing the literature as a leisurely hobby. They were searching for guidance while hospitals were overwhelmed, treatment protocols were evolving, and patients needed decisions in real time. In that environment, a high-profile paper in a major journal could carry enormous weight. If that paper was later retracted, the correction did not magically erase the earlier impression. It had already shaped conversations, influenced summaries, and possibly colored clinical judgment. The retraction might fix the archive, but it could not fully rewind the week.

Librarians and evidence specialists had their own version of the problem. They were often the people quietly chasing down whether a paper had been corrected, whether a preprint had changed, whether a database record was updated, and whether a review draft still contained sources that no longer belonged there. That work is not flashy, but during COVID-19 it became essential. The pandemic reminded everyone that literature management is not clerical busywork. It is part of research integrity. When retracted articles remain easy to download and hard to identify, the burden shifts onto the most careful readers to clean up the mess for everyone else.

Science journalists also ran into the same trap from another angle. A dramatic COVID-19 result could become news immediately, especially when it involved treatments, risk factors, or policy implications. But the later correction rarely received matching attention. A retraction is usually more technical, less emotionally satisfying, and harder to summarize in a punchy headline. The result is an information imbalance: the public hears the exciting claim loudly and the correction quietly. That pattern is not unique to COVID-19, but the pandemic made it impossible to ignore.

Even for ordinary readers, the experience was disorienting. People were told to “follow the science,” but science during a crisis was not a neat marching band. It was a crowded construction site. Papers were posted, questioned, revised, criticized, corrected, and sometimes withdrawn, all in full public view. That visibility was not itself a failure. In some ways it was healthy. The real problem was that retraction did not reliably function as a stop sign. Too often, it acted like a footnote to a rumor that had already gone viral.

That is why the lesson from COVID-19 should not be “never trust science.” It should be “build better correction systems.” Science earns trust not by pretending errors never happen, but by making sure that when they do happen, the warning travels as fast and as far as the original claim. During the pandemic, that did not happen nearly often enough. And the citation record still shows the bill.

Conclusion

Retracted papers about COVID-19 were more highly cited than they should have been because the pandemic rewarded speed, visibility, and dramatic claims, while the correction system lagged behind. Retraction notices existed, but they did not always travel through the citation ecosystem with enough force to prevent further misuse. As a result, flawed COVID-19 studies often continued to shape the literature even after they were officially withdrawn.

The good news is that this is a fixable problem. Better metadata, stronger retraction labeling, stricter editorial checks, and smarter citation habits can all reduce the spread of zombie papers. Science does not need to become error-free to be trustworthy. It needs to become better at making errors unmistakable once they are known. A retracted paper should be part of the historical record, not part of tomorrow’s evidence base.