Once a journal publishes a research paper, other researchers can keep citing that work and build on it to produce more research. Sometimes, this continues even after, say, the journal withdraws the original paper because it contains a mistake or, as is increasingly the case, it was produced by fraudulent means. So simply withdrawing a faulty paper does not correct the research record: the work based on it also needs to be checked and, if necessary, undone.
More research papers are being retracted than ever before. This is because researchers are also publishing more papers today — and because stakeholders, like journal editors and independent experts, are scrutinising papers more carefully.
According to Sabuj Bhattacharyya, an expert on research ethics at the Institute for Stem-cell Science and Regenerative Medicine, Bengaluru, retractions have increased substantially over the last four decades. In 2025, his team analysed dataof life-sciences papers retractedbetween 1981 and 2023 — and found that the reasons for retraction included non-compliance with ethical standards, plagiarism, data-related concerns, and publication misconduct.
In a 2026 study, research scholar P.R. Devanath and associate professor Rupesh Kumar of Tumkur University, Karnataka, analysed more than 12,000 citations of papers funded by the Indian government and later retracted. And found that 30% of the citations had been incurred after a paper was retracted. They also identified a “cascading retraction effect” in which some retracted papers continued to be cited by studies that were themselves later retracted. And nearly 64% of those second-generation papers were authored by the same groups, i.e. they had self-cited.
While not all post-retraction citations necessarily endorsed the retracted findings, many also did not acknowledge the paper’s retracted status, and some retracted papers also continued to appear in patents. Dr. Devanath said one important reason is a lack of awareness, particularly among early-career researchers with limited training in responsible citation practices.
Spreading the message
Retraction is part of how science corrects itself — but the correction often comes long after a paper has entered the scientific literature, according to Partha Pratim Majumder, former National Science Chair with the Science and Engineering Research Board, Government of India. By the time a study is withdrawn, other researchers may already have built on its findings. He said investigations often take time and there is no foolproof way to ensure researchers are quickly informed when a paper is retracted.
Dr. Bhattacharyya said that clearer, standardised retraction notices and consistent implementation of the National Research-Integrity Guidelines could reduce ambiguity and improve accountability.
Whether that communication reaches all the relevant research groups is a separate problem. Bhaskar Mukherjee, professor and head, Department of Library and Information Science, Banaras Hindu University, also said retractions often receive little institutional attention in India. So journals rarely notify an author’s institution while researchers themselves often avoid discussing retracted papers. As a result, information on retractions often stays within journals, he said.
Gagandeep Kang, adjunct professor of microbiology at the Christian Medical College, Vellore, said better communication alone is not enough. The more important question is whether the scientific system learns from each retraction. A reactive system, she says, removes a problematic paper but does little to understand why it passed scientific scrutiny in the first place.
She added that a stronger research integrity system would identify whether the problem was fraud, an honest error or failure of replication — and use those lessons to make peer review (the process where independent scientists check the work of others in their field), data-sharing practices, and institutional accountability better. There is no easy way to identify unreliable papers much earlier, although researchers are increasingly using artificial intelligence (AI) tools to flag problem papers, she added.
Mr. Devanath and Prof. Kumar also recommended dedicated research support offices, checks for retracted references during peer-review, and regular training in responsible research practices.
Correcting knowledge
Then again, even stronger research practices may not entirely solve the problem. Anchal Agrawal, founder of the independent research-integrity effort India Research Watch, said the challenge goes beyond scholarly publishing. As AI systems increasingly draw on published research, retracted studies may continue to influence AI-generated responses unless those models are also updated to flag retracted papers.
The noted philosopher of science Sundar Sarukkai said the matter of post-retraction citations encompasses the nature of scientific knowledge itself. Science, he continued, advances by constantly testing, revising, and challenging published work. In his view, being wrong is not the problem; deliberate falsification is.
Retractions sans repair
Thus, scholars agree that the way retractions happen at present also raise thorny challenges to preserving the integrity of the scientific record. Correcting the published literature alone is not enough.
In fact, Dr. Majumder said when a retraction can be said to have achieved its purpose is itself “very difficult” to determine. A retraction is meant to inform other scientists that the data and inferences reported in a study are incorrect. Because that information spreads slowly, he said he believes a retraction will have broadly achieved its purpose when about 90% of researchers in that field are aware of it, although complete awareness is impossible to ensure.
And even then, even widespread awareness of a retraction would not mean that the scientific record has been corrected.
“When errors in science are inadvertent, the errors get corrected when the experiment is repeated by the scientist who committed the error or by another scientist. Thereafter, the erroneous results of the experiment do not influence scientific practice,” Dr. Majumder said.
That process, however, may never occur in every case. According to Dr. Majumder, no scientist may be willing to study the same problem addressed in a retracted paper and generate correct data and inferences. “Therefore,” he added, “the scientific record may never be corrected.”
Rohan Singh is an independent science journalist in Lucknow.
Published – August 20, 2026 09:15 am IST
