AUTHORS: Michael F. Haney et al
The likelihood of detecting fraudulent research remains low, despite using tools for detection as discussed elsewhere. One purpose of this article is to convey that editors are actively evaluating options to increase deterrence against bad actors, who might otherwise continue to believe that the likelihood of their being caught is low. International conferences focused on scientific integrity (such as the Oxford FAIRS meeting on which this paper is based) bring together editors, sleuths who identify research fraud, publishers, university integrity officers, and others to address certain dilemmas. Although a comprehensive solution remains elusive, efforts are underway and future solutions might resemble one or more of the options discussed here. Prospective fraudsters should recognize that processes are likely to be tightened—both to increase detection and to facilitate information sharing across journals.
Anesthesia is notably well-represented in the retraction “leaderboard” (see retractionwatch.com), presenting fraudulent research reports identified after publication. Review of earlier publications from the same authors and their immediate collaborators has uncovered and led to the retraction of additional fraudulent articles. Yet, while postpublication identification of misconduct remains important, it is cumbersome, resource-intensive, and often takes years. Prevention before publication is preferable and needs to become more effective.
We write mainly from the perspective of editors-in-chief (M.F.H. and J.J.P.), where editors are in the privileged position of having first sight of submitted material. This point of “first contact” is often where suspicions of misconduct may initially arise, and there are several potential avenues for editors at that point (Figure). Concerns are most readily communicated to the authors, and, where the authors engage with the process of enquiry, there may be a satisfactory—albeit not necessarily speedy—resolution. For example, the authors’ co-operation and provision of additional data or material may show that the concerns are unfounded. Alternatively, flaws in data or analysis may be identified that warrant correction but fall short of justifying or requiring additional sanctions.
The main dilemma however arises when authors do not engage, and this, as is so often the case. A common experience is that authors simply cease contact with the journal at this point. In some cases, there is concern that some authors may not have even existed. Editors then face a dilemma. On the one hand, authors are likely to have submitted their manuscript on the understanding that the material will be kept confidential to the editors and reviewers. This is important as there may be commercial or professional interests for the authors in the contents of the paper. On the other hand, editors have a wider responsibility toward the public and patients to uphold rigorous and trustworthy scientific reporting. Fraudulent science can be harmful not just to public trust in science but also to patient care. The Figure depicts this journal and editorial ethical dilemma, which arises at this point of author disengagement.
The Committee on Publication Ethics (COPE), which is a nonprofit organization whose stated mission is to define and assist editors and publishers to achieve best practice in the ethics of scholarly publishing, has published discussions and recommendations for this type of situation. COPE advises that when authors disengage, the editor should contact the author’s employer (eg, university research integrity officer) to suggest or request further investigation. Thus, COPE recommendations move responsibility back to the author’s employer, recognizing that journals and editors have no formal investigatory powers or resources, and no authority to compel authors to respond to questioning or even to provide further data.
A key weakness in these COPE recommendations is that they depend on institutional (employer or research sponsor) willingness to engage, as well as on author’s co-operation with their own employer. Some employing institutions may have little interest in pursuing concerns raised by a journal editor: they too may lack capacity or resources to investigate (eg, there is no research integrity officer); may face conflicts of interest or fear repercussions. For example, the author(s) in question may be very senior or responsible for significant research funding. When an institution cannot or will not proceed, the editors gain no further insight and the process effectively reaches a dead end.
In theory, journal editors can “hold onto” a concerning manuscript in the journal’s review process until a university’s investigation concludes and the findings are made known. This can take a long time. This is to try to prevent authors from submitting the same work to another journal, given the usual journal submission requirement to declare that the work is not contemporaneously submitted elsewhere. Authors can request to withdraw their manuscript, but an editor is not compelled to do this if there is an ongoing investigation concerning the work. Of course, authors who are willing to falsify or fabricate data in the first place are likely to resubmit elsewhere, regardless, or revise the manuscript sufficiently to be able to claim it is “not the same.” At some point, however, the path of least resistance for an editor is to reject the article, given uncertainty concerning potential investigations, leaving the authors free to resubmit.
All this seems unsatisfactory. A journal may have diligently applied available methods for assessing data integrity (these methods are discussed in accompanying articles,). The journal may then have reported the resulting concern to the author (with no response) and then to the author’s employers. But at that point, the journal loses control of the nature and pace of the investigation. Even if there is a local sanction applied to the authors, there is no requirement for this to be communicated to the journal. It could be kept private between the authors and their employer, thereby with no way for the journal from currently being able to share concerns with other journals. The authors cannot be “marked” as miscreants as a warning to other journals or the public.
Effective prepublication prevention could be strengthened if there was sharing of information across the journals to which authors submit. However, the dilemma, as illustrated in the Figure, is the framework under which that sharing should occur. The remainder of this article discusses the options for prepublication sharing and the relative pros and cons of each option. First, we address the risks of sharing at all.
SOME RISKS OF PREPUBLICATION SHARING
In the absence of an acknowledgment by an author that suspicions about research integrity have some foundation, editors face the risk that they are mistaken. It is one thing to exercise editorial discretion and simply to reject a manuscript suspected to contain fraudulent material, and quite another thing to disseminate that suspicion. Mistaken allegations of misconduct may be defamatory and cause significant reputational injury to an author or researcher group, even if they are later found to be unfounded. In turn, raising concerns that are later proven baseless may cast doubt on an editor’s judgement.
Early suspicions may also cloud the review process through “groupthink,” compromising independent analysis. There is perceived to be “no smoke without fire”: a researcher whose manuscript is circulated among editors because an initial editor had concerns about research integrity will likely be viewed as “suspect” from the outset. Supplemental Digital Content, Supplemental file S1, https://links.lww.com/AA/F758, provides a real-life example of this concern.
The COPE guidance partially mitigates these risks by recommending that editors contact authors and provide an opportunity to respond to an editor’s concerns. One approach can be to state: “Failure to respond with X weeks will be deemed to be the author accepting that their submission can be shared with other journals to assist with investigation.” While such wording should prompt engagement, it does not resolve the issue if authors respond only to request more time, and insist that any sharing would be considered a breach of confidentiality.
These risks imply that journals should share suspicions only when there are strong grounds for doing so, which may, in turn, be dependent on the performance of the tools to detect fraud, as discussed in accompanying articles in this series. At one extreme, a journal may face legal action (eg, for defamation) from authors or their institutions; or an editor may themselves be referred to their professional regulatory body (eg, for malpractice in breach of confidentiality). It is therefore understandable that editors often choose the pragmatic option: reject the manuscript and shift the residual risk to other journals.
INFORMAL (VERBAL) PREPUBLICATION SHARING
Journal editors-in-chief could simply share relevant information with each other, verbally, with no paper trail and therefore unlikely legal repercussions. To some extent, informal sharing occurs already. Journal editors-in-chief can discuss manuscript concerns with each other verbally, leaving no paper trail and reducing legal exposure. For example, about a dozen editors-in-chief (in the specialty) meet biannually at the American and European society meetings and share issues of concern. This has led to joint statements being published.,
However, informal sharing can have drawbacks,., Sharing confidential information, even verbally, may itself be a probity issue. For this reason, editors generally anonymize discussions, focusing on the suspicious manuscript content, rather than naming authors. If a university has been alerted to concern about misconduct, and has initiated its own formal investigation, under the framework of the COPE guidance, editors can be more specific and discuss matters in the journals related to the enquiry.
Nevertheless, verbal communication can be as damaging as written communication. It may further introduce inconsistency and bias in subsequent investigations, and its impact is constrained by the limited network of participants. Despite there being over 150 anesthesia journals and over 80 intensive care journals, it is mainly only the dozen or so relatively high-impact journals that communicate regularly enough to facilitate informal sharing.
In summary, while informal sharing does occur in a limited, proportionate, and anonymized manner, it is hardly a robust solution to the dilemma.
SHARING DEIDENTIFIED MANUSCRIPTS
A more systematic extension to informal sharing, and in keeping with anonymity, would be the dissemination of deidentified manuscripts. This could enable receiving editors to determine whether they had already seen the same manuscript, or be vigilant of multiple submissions. The primary limitation here, however, is that few papers can be truly deidentified: trial registrations, details in the methods, sample population details, or reference lists may reveal the source. Moreover, a deidentified manuscript cannot include a “conflicts of interest” section, though this is potentially relevant to the analysis of misconduct.
EDITORS ACTING AS REVIEWERS
Another variant could be to exploit the peer review process itself. Editors are frequently asked to review papers for their own or other journals as content expert reviewers. There is no formal hindrance to one editor sending a suspicious paper to any number of other journal editors, with a cover letter explaining the sending editor’s concern and desire for more intensive peer review. This way, other editors-in-chief are alerted of the issues, while constructively having the opportunity to contribute to peer review. At the same time, they gain knowledge of the concern, and have the material to assess if there arise multiple submissions by the authors. This approach has not been tested legally or in practice, but it appears possible.
Nevertheless, there remain concerns about this approach. The “reviewer” (in reality, another editor) receives an explicit hint that the sending journal has a particular concern, generating conditions for “groupthink.” The proposal seems to lack transparency and can be perceived to be disingenuous.
PREPRINT SERVERS
Many authors use preprint servers and, by doing so, make their work publicly available before publication. When these servers were first used for clinical medical reports, some journals prohibited consideration of manuscripts posted as preprints. This has changed, and we are not aware of any current prohibitions.
We might turn this on its head and require all submissions to be posted on publicly accessible preprint servers. This is the norm in some other scientific fields, especially mathematics and the physical sciences. In this way, suspicious data could be scrutinized freely and independently.
However, there are limitations. Some cautious authors may legitimately prefer confidentiality until peer review, recognizing that review may improve a manuscript that otherwise may be less than optimally presented. Preserving intellectual property rights or dataset confidentiality may also be a concern for others who avoid preprint posting. Furthermore, the solution would have an impact only if all journals required it.
DEVELOPING A “PUBLIC INTEREST DEFENCE” FOR PREPUBLICATION SHARING
If the dilemma is that prepublication sharing breaches confidentiality, exposing the journal or editor to legal or regulatory consequences, then one solution could be to strengthen a “public interest defence” in law. Under such a defense, editors could argue that when they shared material or information that was provided to them in confidence, and that this was to protect patients or the public. The position would be that the actions were proportionate and based on careful consideration and advice, including legal advice.
Public interest defences exist in other domains of medical practice in many countries, where patients might reveal information during a medical consultation about their intentions to harm others, or exhibit behaviors that might do so. It is acceptable for the doctor to share this with the police, for example. An extreme example in publishing would be where an author submits a paper describing how to undertake some criminal act (eg, organ removal without consent), or confessing to such an act. In most jurisdictions, editors would not only be defended if they reported this to authorities, but in fact obliged to do so.
This public interest defence could be strengthened by explicit author submission conditions and instructions to which authors agree when submitting their manuscript, thus making it part of the submission contract. However, relatively few problematic manuscripts, if published, would cause direct harm to patients. At least, this would be difficult to prove in many cases. Even when published research evidence supports new interventions, general implementation can be slow, and it can take time for fraud to cause harm. The editor may thus struggle to convince that there would be direct patient harm from publication of a suspect paper, and to specify the mechanism of harm. The most immediate injury from fraudulent research is to scientific integrity and the process of knowledge generation, and of course, public trust in the scientific process. Whether this is a sufficient basis on which to mount a public interest defence remains untested.
REVISED SUBMISSION CONDITIONS
Amending Confidentiality Agreements
One way of looking at the Figure is to appreciate that the dilemma arises from the obligation to keep submitted material confidential. A fundamental solution is to revise conditions of submission explicitly to permit manuscript sharing under specified conditions, including when well-grounded concerns about research integrity exist.
Implementing this would require substantial changes to journal policies and practices. Submitting authors would need to agree that editors and journals may share concerns about suspected research fraud and, in doing so, waive certain rights (eg, to sue for defamation), provided that editors and journals act in good faith and follow clear processes. This would require explicit criteria for triggering concern. The COPE guidance currently falls short of defining these thresholds. Given that there are no generally accepted precedents, establishing widely acceptable thresholds would be needed before new guidelines could be developed.
Declaration of Prior Submissions
Authors are not currently required to declare whether, or to which journals, they have previously submitted a manuscript paper. They attest only that the manuscript is not under active consideration for publication elsewhere. A stronger approach would require that authors declare which journals have previously considered the manuscript, and authorize that or those journals to confirm this history, and even share previous peer review reports. This would create a more robust and traceable record that could improve the detection of research problems. For example, a prior journal review may have raised concerns about data integrity. However, disreputable authors may provide false or incomplete declarations, and at present, editors and journals do not have the means to verify the accuracy of these kinds of possible submission details. “We forgot” might become routine, or some authors could argue that revisions rendered in the manuscript make it “not the same” as a previously submitted manuscript to another journal—akin to the Ship of Theseus paradox; all its timbers are replaced over time, so it can be argued it is no longer the same ship.
A Central Database
The paper trail of all submissions implied by this proposal point toward the need for development and maintenance of a central, shared repository, a sort of nonpublic preprint server. Editors could consult this repository at the point of submission to review prior consideration and reports.
It is self-evident that such an infrastructure would be complex and incur significant costs. It is also unclear who would fund, govern, and curate this repository. On a technical level, the database would need to be organized and searchable by author primarily, rather than by manuscript content, for reasons discussed earlier: manuscript content may evolve, whereas author names are presumed to usually follow specific authors.
Summary of Revised Submission Conditions
The real limitations of revising submission conditions are that they would be effective only with universal journal adoption. Since these conditions are all more stringent, nonadopting journals might gain a competitive advantage in a crowded publishing market space. One might hope that because prepublication sharing and screening are intended to prevent post-publication retraction, journals that opt out would incur a higher retraction rate over time.
Two broad ownership models for journals further complicate the implementation of these approaches. Many journals are owned by professional organizations, where publishers for these journals are independent business entities that enter into a publishing agreement with these societies/journal owners. Journal editors also report to a professional society executive, which can also have thoughts about editorial policies. Publishers can advise and support the editorial policies chosen by the specialty editorial boards who represent the journal owners. Publishers provide the manuscript handling platforms used to process manuscript content, either with their own system or by subcontracting with a third party for a manuscript handling platform (eg, Anesthesia and Analgesia uses a system known as Editorial Manager). Thus, the relationships are complex. On the one hand, authors have an “intellectual contract” with the editorial board that governs how their manuscript content is scientifically managed. On the other hand, they have a “technical contract” with the publisher because their work is physically held and handled within a publishing platform. The duty of confidentiality at the heart of the dilemma therefore exists on two levels: one for the editorial board but also one for the publisher. In other words, any new approach to managing the dilemma in the Table will need the agreement of publishers.
Table.
Summary of Some Problems, Proposed Solutions, and Barriers to Adoption for Information-Sharing Prepublication
| Problem | Proposed solution | Barriers to adoption (and how they might be overcome) | How barriers may be overcome |
|---|---|---|---|
| Authors submit fraudulent paper to another journal, after its rejection (for suspect data) | A tracking system of manuscript system across all journals to identify where fraudulent papers are submitted to different journals in sequence | May not be embraced by publishers | May require legislation to enforce a tracking system |
| Possible data concerns identified by the first journal editor, which is deemed important to share with other editors for assistance, or to forewarn | Informal (verbal) information sharing, short of manuscript sharing, between journal editors-in-chief | Already occurs, but does not create a database that can be shared; verbal sharing is not a consistent process and is not transparent | Caution needed with groupthink and reputational damage: will need training for editors and terms of reference for when to share |
| Publisher does not support sharing of concerns across journals/publishers | Sharing of suspicious manuscripts across journal editors, under guise of seeking peer review | May be perceived as violation of author-journal confidentiality agreements, under a guise of seeking peer review | Barriers could be overcome by seeking prior consent from authors, or requiring that such other editors called as peer reviewers provide “open” (ie, signed) reviews |
| There is no accessible database of suspicious data to use for future learning | Requirement for all authors to have placed submitted manuscripts on a preprint server | Author reluctance to share a preliminary version of their work, or may wish to preserve their intellectual property | All journals would need to agree to mandatory preprint deposits for the system to become uniform |
| Editors are criticized—or face legal sanction—for sharing manuscripts with other editors | Editors exploit “public interest” clauses in legal or professional regulatory frameworks as protection | Public interest is determined on a case-by-case basis and may not be deemed to apply to any given case | Statutory or professional regulatory clarification that editors may use public interest defence rather than only author confidentiality |
| The history of submissions to previous journals is not accessible, in case that fraud was previously suspected | A common database is created that editors can access at point of submission to their journal, to review complete submission history | Funding, maintenance, and curation for such a database would need to be agreed and identified | Cross-journal and cross-publisher agreement to create such a database |
A second common model involves journals that are owned directly by a publisher, with no professional society involved in the editorial process (most common), or with a professional society becoming affiliated ex post facto in some form. The “intellectual” versus “technical” contracts apply, but the financial dynamics differ. In the society-owned model, the professional society historically provides the journal’s financial foundation (102 years in the case of Anesthesia and Analgesia) through subscriptions. In the publisher-owned model, professional society affiliation to the journal is effectively “purchased” through subscriptions. Where research integrity follow-up activities by journals are time-consuming, costly, and legally risky, even publishers that own a journal concerned (like editors sometimes) may rationally favor the simplest path—that of simply rejecting the suspect manuscript—rather than seeking to solve the dilemma outlined in the Figure. These dynamics pose an additional barrier.
CONCLUSIONS
Early detection and prevention of scientific misconduct and compromised data integrity—before publication—is highly desirable, yet difficult to achieve. Effective prevention before publication relies on many detection steps, and would be more effective if there was sharing of manuscript information prepublication across journals. Currently, this is inhibited by the need to maintain author and data confidentiality during the review process. Several approaches could mitigate this dilemma, or the balance between risks with confidentiality breach for authors and risk for harm based on publication of results of research misconduct. Each of these routines or risks has costs and benefits. Ultimately, the most effective and durable solutions will require co-operation among journals and publishers, and may also necessitate infrastructure change and funding.
In conclusion, scientific publishing depends on trust. But verification is also important. Authors can expect respectful and confidential handling of their manuscripts from scientific journals, and in return, journals (representing the community through peer review) expect honesty in conduct and co-operation from authors, and integrity of the data.
REFERENCES
-
1.
Brown NJL, Carlisle JB. Detecting inconsistencies and fraud in research data: time for authors to share the data underlying their summary statistics as a matter of course. Anesth Analg. 2026;143:1002–1005.
-
2.
Pandit JJ. On statistical methods to test if sampling in trials is genuinely random. Anaesthesia. 2012;67:456–462.
-
3.
Carlisle JB, Dexter F, Pandit JJ, Shafer SL, Yentis SM. Calculating the probability of random sampling for continuous variables in submitted or published randomised controlled trials. Anaesthesia. 2015;70:848–858.
-
4.
Benford F. The law of anomalous numbers. Proc Am Phil Soc. 1938;78:551–572.
-
5.
Brown NJL, Heathers JAJ. The GRIM test: a simple technique detects numerous anomalies in the reporting of results in psychology. Social Psycholog Personal Sci. 2017;8:363–369.
-
6.
Anaya J. The GRIMMER test: a method for testing the validity of reported measures of variability. PeerJ Preprints. 2016;4:e2400v1.
-
7.
Nuijten MB, Hartgerink CHJ, Van Assen MALM, Epskamp S, Wicherts JM. The prevalence of statistical reporting errors in psychology (1985–2013). Behav Res Methods. 2016;48:1205–1226.
-
8.
DeMets DL. Distinctions between fraud, bias, errors, misunderstanding, and incompetence. Control Clin Trials. 1997;18:637–650.
-
9.
Buyse M, George SL, Evans S, et al. The role of biostatistics in the prevention, detection and treatment of fraud in clinical trials. Stat Med. 1999;18:3435–3451.
-
10.
Kirkwood AA, Cox T, Hackshaw A. Application of methods for central statistical monitoring in clinical trials. Clin Trials. 2013;10:783–806.
-
11.
Pogue JM, Devereaux P, Thorlund K, Yusuf S. Central statistical monitoring: detecting fraud in clinical trials. Clin Trials. 2013;10:225–235.
-
12.
George SL, Buyse M. Data fraud in clinical trials. Clin Investig. 2015;5:161–173.
-
13.
Sajid M, Sanaullah M, Fuzail M, Malik TS, Shuhidan SM. Comparative analysis of text-based plagiarism detection techniques. PLoS One. 2025;20:e0319551.
-
14.
Bik EM. Seeing is believing? Scientific misconduct and the detection of problematic images. Anesth Analg. 2025;143:1019–1025.
-
15.
McHugh UM, Yentis SM. An analysis of retractions of papers authored by Scott Reuben, Joachim Boldt and Yoshitaka Fujii. Anaesthesia. 2019;74:17–21.
-
16.
Picano E. Who is the author: genuine, honorary, ghost, gold, and fake authors? Expl Cardiol. 2024;2:88–96.
-
17.
Hoch R, Mika K, Nancarrow C, Osman A, Puebla I; FORCE11 Research Data Publishing Ethics WG. Joint FORCE11 & COPE Research Data Publishing Ethics Working Group Flowcharts. Zenodo. 2023.
-
18.
Horbach SPJM, Breit E, Mamelund SE. Organisational responses to alleged scientific misconduct: sensemaking, sensegiving, and sensehiding. Sci Pub Policy. 2019;46:415–429.
-
19.
Horbach SPJM, Breit E, Halffman W, Mamelund SE. On the willingness to report and the consequences of reporting research misconduct: the role of power relations. Sci Eng Ethics. 2020;26:1595–1623.
-
20.
Higgins J, Reitz M. If whistleblowing is the answer, ask a better question. J R Soc Med. 2019;112:453–455.
-
21.
Goldenring JR. Perspective: innocence and due diligence: managing unfounded allegations of scientific misconduct. Acad Med. 2010;85:527–530.
-
22.
Bajraktari R, Booth F, Munafo M, Beazley-Long N. The most prevalent perceived barriers to sharing research data at the point of manuscript acceptance in health and life sciences: a survey at the University of Bristol. F1000Research. 2025;14:482.
-
23.
Gurumeta AA, Davidson A, Landau R, et al. Comments on alleged editor misconduct from editors-in-chief of anaesthesia journals. Br J Anaesth. 2025;134:1181–1182.
-
24.
Palermo TM, Bouhassira D, Davis KD, et al. Editorial commitment to trust and integrity in science: implications for pain and anesthesiology research. Europ J Pai. 2025;29:e70052.
-
25.
Chung KC, Shauver MJ, Malay S, Zhong L, Weinstein A, Rohrich RJ. Is double-blinded peer review necessary? The effect of blinding on review quality. Plast Reconstr Surg. 2015;136:1369–1377.
-
26.
Blatch-Jones AJ, Recio Saucedo A, Giddins B. The use and acceptability of preprints in health and social care settings: a scoping review. PLoS One. 2023;18:e0291627.
-
27.
Kranke P, Apfel CC, Roewer N, Fujii Y. Reported data on granisetron and postoperative nausea and vomiting by Fujii et al. are incredibly nice! Anesth Analg. 2000;90:1004–1007.
-
28.
Hauptman PJ, Armbrecht ES, Chibnall JT, Guild C, Timm JP, Rich MW. Errata in medical publications. Am J Med. 2014;127:779–785.e1.
-
29.
Vaught M, Jordan DC, Bastian H. Concern noted: a descriptive study of editorial expressions of concern in PubMed and PubMed Central. Res Integrity Peer Rev. 2017;2:10.
-
30.
The Lancet Editors. Correcting the scientific literature: retraction and republication. Lancet. 2015;385:394.
-
31.
Nallamothu BK. Better: reflections on retraction and republication in science. Circ Cardiovasc Qual Outcomes. 2023;16:e010579.
-
32.
The Lancet Editors. Retraction and republication: 6-month consequences of COVID-19 in patients discharged from hospital: a cohort study. Lancet. 2023;401:2025.
-
33.
Loadsman JA. Why does retraction take so much longer than publication? Anaesthesia. 2019;74:3–5.