Authors, beware of plagiarism in medical writing

Introduction

Scientific research is crucial for advancing medical knowledge, driving innovation and improving patient care. Research publications are not just professional academic achievements, but also opportunities for sharing unique scientific discoveries. The fundamental principle underlying all scientific endeavor is originality and intellectual integrity. However, these values are threatened by plagiarism – a practice that has progressed from seemingly benign symptoms of copying others’ ideas (respect via imitation) to a full-blown malignant disease, thus undermining the very foundation of scientific progress. In this editorial, we will delve into its numerous definitions, root causes, and consequences, examine the role of technology in aiding and detecting plagiarism, and also discuss the responsible conduct required at various levels to minimize this intellectual threat.

The word “plagiarism” originates from the Latin word “plagiarius,” which means “kidnapper,” essentially translating into intellectual theft.[1] Multiple authoritative organizations provide comprehensive definitions of plagiarism. The United States Office of Research Integrity (ORI) defines plagiarism as “the theft or misappropriation of intellectual property and the substantial unattributed textual copying of another’s work.”[2] World Association of Medical Editors (WAME) defines plagiarism as “the use of others published and unpublished ideas or words (or other intellectual property) without attribution or permission and presenting them as new and original rather than derived from an existing source,” while the Committee on Publication Ethics (COPE) defines plagiarism as “the unreferenced use of others’ published and unpublished ideas.”[3,4] Thus, plagiarism essentially refers to the act of using someone’s ideas or content and taking credit for it without acknowledging the original author/researcher. However, many gray areas remain in its interpretation – How many consecutive or overlapping words can constitute plagiarism? How much content can be quoted or cited before it becomes “too much”? What would constitute common knowledge not requiring a citation?[1] To minimize misinterpretation and consequently unintended/undetected plagiarism, there should be a universally accepted, non-controversial definition of plagiarism. This will ensure clarity and consistency across different fields and contexts.[5]

Plagiarism is one of the three “cardinal sins” of scientific misconduct, besides data fabrication and data falsification.[6] The ethical responsibility of researchers extends beyond initial methodology and conduct of research, to credible documentation, reporting, and dissemination of scientific findings.[7] Plagiarism undermines the validity and reliability of research by misrepresenting intellectual contributions and wasting valuable time and resources through repetitive, non-original work. It has not only ethical, but also legal implications, potentially involving copyright infringement.[8] The prevalence of plagiarism is deeply concerning; a meta-analysis and systematic review on biomedical research misconduct by Phogat et al.[9] revealed a plagiarism prevalence of 0.2%–49.4% in the studies reviewed by them. An increasing number of journal retractions due to plagiarism has also been noted in China, Iran, India, and USA.[10] So, what drives researchers to plagiarize? Young researchers may resort to unintentional plagiarism due to poor training in research writing skills, language barriers, educational background promoting rote learning, lack of mentorship, and misconceptions about the scope and severity of plagiarism and intellectual property.[11,12,13] However, intentional/deliberate plagiarism stems from intense professional pressures (for career advancement, promotions, and academic recognition) and an institutional tolerance toward plagiarism creating a permissive environment for unethical academic practices.[8,11,12,13] Technological advancements have paradoxically improved access to literature while simultaneously making plagiarism easier, through “copy–paste” mechanisms. Many authors remain unaware of sophisticated plagiarism detection tools that could help them maintain ethical standards.[11] Thus, the pathophysiology of plagiarism is complex and deep-rooted, and requires educational and attitudinal reforms, besides stringent policies.

Researchers engaging in unattributed copying or misuse of intellectual work risk severe professional repercussions, ranging from manuscript rejection and article retraction to more serious outcomes. Serial offenders may be banned from publication in a particular journal and reported to the institution/employers/professional bodies for disciplinary action, leading to professional disrepute, loss of research funding, and potential job termination.[14] Plagiarism noted post-publication can affect the reputation of not only the authors, but also the editors, reviewers, and publishers involved in the whole process.[14] As the consequences of plagiarism are far-reaching and may extend beyond individual researchers, all stakeholders should bear responsibility and collaborate efficiently to minimize its threat.

Types of Plagiarism

Different terminologies/classifications are used to describe the spectrum of plagiarism, and they are broadly similar in their descriptions of level. Its severity can be determined using multiple dimensions such as the extent of copied material (a few words to entire sections or complete works), context in which plagiarism has occurred (similar methodology/terminology to directly copying data/results), novelty of the material (drawing from common themes to copying others’ original ideas), acknowledgment and referencing (incomplete, incorrect referencing to entirely absent citations), and intent (an honest error to a more malignant intention).[5] COPE also categorizes plagiarism into three types – clear plagiarism (large portions of text without acknowledgment or copying others’ data as own work), minor copying, and redundancy (reusing published work or self-plagiarism).[5] The American Medical Association identifies four common types: direct plagiarism, mosaic plagiarism, paraphrasing, and insufficient acknowledgment.[15] Direct copying may involve ideas without acknowledgment (such as ideas copied during reviewing manuscripts or using unpublished student dissertations) or verbatim plagiarism (word-for-word reproduction without credit or quotation marks). In mosaic plagiarism, the content is assembled from multiple sources and patched into a seemingly original piece. Paraphrasing may be used strategically to disguise copied content by selective word substitution.[15] Besides insufficient acknowledgment, citation may be manipulated by introducing non-existent (invalid source plagiarism) or incorrect references to mask the plagiarized content.[16] Zimba and Gasparyan[16] classified common plagiarism as plagiarism of ideas, direct copying, paraphragiarism, text recycling (reusing own intellectual property without referencing the original), translational plagiarism, graphical plagiarism, citation manipulation, and compound plagiarism. Thus, any original work can be plagiarized, encompassing a broad range of intellectual property including text, ideas, methodology, concepts, data, images, tables, graphs, figures, and even creative works from varying sources such as abstracts, institutional review board applications, grant applications, and unpublished/published works.[3,5]

Self-plagiarism presents a complex ethical challenge as it involves an author’s unauthorized reuse of their own previously published work without proper/appropriate citation. Despite arguments about whether using one’s own work constitutes actual theft, the practice is generally considered unethical because published content is typically owned by the publisher, potentially constituting a copyright infringement.[17] Several degrees of self-plagiarism exist, including duplication (republishing identical content in different journals), “salami slicing” or segmentation (fragmenting research into multiple stand-alone publications), translational plagiarism (translating a published article into another language and presenting it as original work), and content augmentation (refurbishing previous publications by adding new data, patients, follow-up periods, or outcomes).[13,14] Masic[18] has innovatively summarized plagiarism types for the technology savvy generation by introducing terms which have been further explained in Table 1.

Table 1.

Types of plagiarism and terminologies[13,15,16,18]

Term	Type of plagiarism	Description
CLONE	Plagiarism of ideas	Submitting another’s ideas as own
CTRL-C	Verbatim plagiarism	Word-to-word copying from single source
FIND–REPLACE	Paraphragiarism	Paraphrasing words, but using substantial content from original
REMIX	Mosaic	Paraphrasing from different sources and rearranging as own work
RECYCLE	Self-plagiarism by duplication	Using one’s own work without proper attribution
HYBRID	Patchwork	Using cited and uncited sources to hide plagiarism
MASH UP	Mosaic	Patching different sources to make new content
ERROR 404	Invalid source	Non-existent/inaccurate referencing
AGGREGATOR	Almost all the content is copied from elsewhere, but the citation is correct
RE–TWEET	Using a large chunk of content/words from the original work, with proper referencing

“Duplicate publication” is a significant form of self-plagiarism, whereby substantial content from a previous publication is reproduced without proper referencing, raising serious concerns about international copyright laws, ethical research conduct, and scientific integrity.[6] This practice not only wastes academic resources and time, but also misrepresents scientific evidence and conclusions through double counting of research results.[6] The International Committee of Medical Journal Editors (ICMJE) provides guidance on acceptable repeated publications, such as complete reports that follow publication of preliminary writings like letters to editors, scientific meeting presentations and posters, health authority assessment reports, meeting reports, conference proceedings, preprint server publications, and clinical trial registry results.[6] Secondary publications can be acceptable under specific circumstances, particularly when they serve the broader public interest, such as disseminating important information through government guidelines or professional organization recommendations.[6] The critical factors in determining the legitimacy of such republication are transparency and mutual editorial approval, ensuring that the republication adds value and serves the academic and public knowledge without compromising research integrity.

Another complex issue is the use of artificial intelligence (AI) in scientific writing and its contribution to plagiarism. AI software is being increasingly used to produce scientific manuscripts at a click, thus threatening the fundamental research principles of originality, validity, and intellectual accountability. Though AI tools produce seemingly “original” content, they utilize information from existing online literature, often without appropriate acknowledgment or attribution.[19] AI models, trained on pre-existing data sources, can inadvertently reproduce inaccuracies and biases and generate non-existent references, referred to as “AI hallucinations.”[19,20] The term “AIgiarism” emphasizes the substantial risks associated with AI-generated content, including potential intellectual property infringements, misinformation, content fabrication, and inherent algorithmic biases.[19,20] Professional organizations like WAME and ICMJE strongly caution against recognizing AI as legitimate authors, emphasizing that AI tools cannot fulfill essential authorship criteria of accountability, responsible content creation, and acknowledgment of sources. Consequently, journals have clear policies requiring disclosure when AI is used in manuscript development, with detailed documentation such as specific prompts used, timestamps, and tools employed.[6] While AI can offer valuable assistance – particularly for non-native English speakers in generating outlines, summarizing, or paraphrasing – its use demands absolute transparency, complete source attribution, and meticulous verification of accuracy and originality to prevent academic misconduct.[20] The consensus is clear: AI may complement the scientific writing process, but it must never substitute human academic judgment, critical analysis, and ethical conduct.

Plagiarism Detection Software

Plagiarism detection has now evolved from chance discovery by editors, reviewers, authors, or readers to vigorous screening using advanced dedicated software. These plagiarism detection tools range from free platforms like Viper, Small SEO tools, PlagAware, Dupli Checker, and Grammarly to sophisticated paid services such as Turnitin, iThenticate, and Crossref Similarity Check, varying in their target audience, vocabulary, accuracy, and database sources.[21,22,23,24,25,26,27] The ideal plagiarism detection software should have a combination of extensive vocabulary, vast source databases, multi-language support, and precise algorithmic analysis to ensure comprehensive and accurate content verification.

“iThenticate” (https://www.ithenticate.com), owned by Turnitin, is a well-known, paid plagiarism detection platform primarily targeting researchers, professional writers, and academic institutions.[22] Unlike Turnitin, which provides only institutional access, iThenticate offers additional flexible individual subscriptions with pricing ranging from US $125 for 25,000 words to US $300 for manuscripts up to 75,000 words. Licensed by over 1500 publishers, the platform provides comprehensive content comparison against an extensive database by partnering with Crossref. iThenticate provides similarity scoring, color-coded source matching, and compatibility with multiple file types and languages.[22]

“Crossref Similarity Check” (https://www.crossref.org/) is another specialized plagiarism detection service exclusively for Crossref members, combining Crossref’s publishing network with iThenticate’s advanced plagiarism technology. Designed for academic publishers, the service provides content similarity checks through a tiered subscription that includes annual and per-document fees.[23]

Free individual options like Grammarly (https://www.grammarly.com/plagiarism/), PlagAware (https://www.plagaware.com/), Viper (https://www.scanmyessay.com/), Dupli Checker (https://www.duplichecker.com/), Small SEO Tools (https://smallseotools.com/), PaperRater, etc. offer limited plagiarism checking capabilities, while premium services provide more extensive features.[21,24,25,26,27] Grammarly Premium is priced between US $12 and 30 monthly, PlagAware ranges from €25 to 50 per month, Viper has pay for credits (US $7.62 for 5000 words/1 credit to US $19.05 for 15,000 words/3 credits), and Dupli Checker has a limited free version with premium subscription staring at US $10/month.[24,25,26,27]

Free tools provide individual researchers with the benefits of pre-submission primary plagiarism check, in addition to other writing tools such as grammar check, AI detector, paraphrasing tool, image checker, and citation generator.[21,24,27] An institutional access allows researchers to further refine their work before submission to reduce chances of rejection due to text similarity. What would be an incidental discovery has now become a routine screening process and set higher norms for publication acceptance.

Plagiarism detection tools identify textual matches and commonly produce color-coded reports with percentage-based text similarities and source identification; however, these reports cannot determine plagiarism. Academic judgment remains crucial in interpreting how much is actual plagiarism as many factors need to be considered beyond simple text matching, such as the percentage of text similarity, origin of copied text (same or multiple sources), and inclusion of bibliography/quoted text, methodology, results, or technical phrases in text matches.[28] Plagiarism software depends on digitized databases, and if the academic work is not digitized (e.g., student dissertation) or database sources are not comprehensive, similarity against plagiarized text will be missed.[29] Thus, free software may result in missed instances or inaccurate results due to restricted access to comprehensive databases.[21] Free platforms also have other disadvantages such as limited trial versions, restricted word count limits, fluctuating similarity levels due to random sampling of text, and security concerns where uploaded text might become part of a permanent repository, potentially exposing academic work.[28,29] Well-known software are expensive and often have only institutional access, limiting access to efficient plagiarism detection. Plagiarism detection software may also fail to detect translational plagiarism, advanced paraphrasing, AI-generated content modifications, images/figures, and spelling errors used to mask plagiarism.[29] Ultimately, while technological tools provide valuable initial screening, comprehensive plagiarism assessment requires critical human evaluation and academic judgment.

There is no universal standards/consensus on acceptable text similarity; most sources suggest 5% or less text as acceptable, while editors, journals, and institutions generally consider 10%–20% as the threshold.[17] It may also depend on the context, for example, methods sections may tolerate up to 10% similarity. However, it is important to emphasize that similarity percentages alone cannot definitively determine plagiarism – the context, the intent, and the nature of the copied text are crucial. Zero similarity does not guarantee originality, and high similarity does not automatically warrant rejection.[28] Though screening using similarity reports may aid the editor/s when plagued by numerous submissions, it should not serve as an easy ground for rejection.

How to prevent plagiarism?

Preventing plagiarism requires a multifaceted approach focused on researcher/author education, awareness, and ethical practices, as it often stems from ignorance rather than intentional misconduct. Studies have demonstrated a lack of awareness or clarity on what constitutes plagiarism in students and young researchers, making it important to introduce research skills early in the medical curriculum.[12,28,30] Education and training of young researchers (workshops) on copyright laws, legal regulations on copyright, appropriate usage of literature sources, and intellectual property formats are important.[12] Online ethical training on plagiarism is being organized by COPE, ICMJE, ORI, and National Institutes of Health at regular intervals.[31] Providing additional support in English writing skills to non-native speakers is also vital.

Authors should not copy the text matter/content “as it is” from any journal, book, or internet sources. They should read, understand, assimilate, and synthesize the matter in their own language.[14] Authors also need to learn how to effectively summarize and paraphrase, use quotations for citing verbatim content, and cite the sources correctly.[14]

Authors need to always give due credit to the original source via appropriate referencing. Meticulously acknowledging sources (ideas/text/illustration), following journal citation guidelines, and ensuring accuracy in referencing are important steps in preventing plagiarism.[13,14] Spelling errors or incorrect citations cannot be excused as minor errors, as they can constitute plagiarism.[14] Authors must insert provisional references while making the initial draft for source of ideas/text. They must obtain appropriate permissions for figures and illustrations (from the copyright holder/s) to avoid legal and copyright infringements, besides referencing them accurately. Maintaining transparency and a healthy communication with the editor/s in case of duplication or use of preprints is important – authors need to clearly state the overlap from prior research and ask the first and second publishers for their permission and editor/s for guidance (as may become necessary).[6,14]

Despite the most stringent measures, the risk of plagiarism remains. Authors should run a plagiarism software check on the manuscript before submitting to a medical journal. Since most advanced software are expensive, institutional support is necessary for access to such resources.[12] Using anti-plagiarism checks regularly will encourage vigilance and train researchers on early remedial measures, but it should not be the sole determinant of academic integrity. While plagiarism software can detect text similarity for human plagiarism, detecting AI plagiarism is difficult, especially due to complexity of their models. Most journals and ICJME recommendations do not recognize AI as authors. Journals should have clear policies regarding use of AI in submitted manuscripts, with clear declaration of use of AI in the writing process and its exact use.[6]

Preventing plagiarism is a collective responsibility that requires contribution from everyone involved from the concept to publication of the research. Senior researchers must mentor junior colleagues, guiding them in ethical writing practices.[32] Institutions should implement zero tolerance policies, with clear consequences and penalties for intentional academic misconduct.[12] Reviewers play an important role as facilitators in detecting plagiarism and should be trained in what constitutes plagiarism, how to verify the authenticity of references, and how to report any such instances, particularly in an era where AI-generated content presents new challenges to research originality.[14] While AI-based advanced technologies are available to detect AI-generated text, they have their own drawbacks.

Role of editors and how to deal with suspected plagiarism

Editors and publishers of journals bear responsibility to maintain high ethical standards in their journals, ensuring that the articles published are original, legitimate, and accurate.[32] The journal’s “instructions to authors” should have strict and clear policies on definition and consequences of plagiarism.[31] Just as copyright forms, a self-declaration form for plagiarism during submission can ensure that the authors check and bear the responsibility for plagiarism in their research work.[32] Advanced anti-plagiarism software should be available to editors for efficient screening; while some journals screen for similarity index of all manuscripts at submission, others check only accepted articles for plagiarism before production.[12] Similarity checks should not serve as an opportunity for easy initial rejection of articles without peer review, especially when editors are overburdened by excess submissions. They should be verified by human judgment and allowed the process of peer review, especially if the manuscript otherwise is academically suitable. Organizations like ICMJE, COPE, and WAME provide essential guidance and frameworks for maintaining research ethics, including preventing and addressing plagiarism. Thus, the editor’s responsibilities extend beyond simple identification, involving careful assessment of the plagiarism’s extent, intent, and context.[12] The editorial response varies based on the magnitude of plagiarism and stage of publication. For submitted articles with minor plagiarism, editors may allow authors to revise (rephrase or correct citations/references or remove overlapping material), recognizing that methodological descriptions or standard techniques may naturally show some similarity. In published articles, they may issue corrections/corrigendum or apologies. Editors must distinguish between intentional plagiarism and honest mistakes, especially for junior researchers who may lack proper training. While leniency might be shown to inexperienced authors, senior researchers are held to stricter standards. The process also involves confidential communication with authors, their institutions, and professional bodies.[5,12]

Major plagiarism (clear plagiarism or major overlap including duplication/self-plagiarism) warrants more serious actions, including rejection (pre-publication) or potential retraction (post-publication), institutional notification, and comprehensive investigation involving all the authors. The author’s intention, admission to guilt, and response to editor’s communication determine the severity of transgression. All investigations of redundancy/major plagiarism are communicated to relevant stakeholders, such as research supervisors and institutional authorities, through confidential channels and followed up three to six monthly for updates on the action taken, escalating to other governing authorities if unresolved.[33,34] Retraction notices must comprehensively detail the reasons for retraction, specifying whether the author or the publisher initiated the process. The publisher is responsible for ensuring retraction is noted in both electronic and print versions, with retraction notices appearing in all the electronic searches of the publication.[12]

While journals may publish retractions or expressions of concern following investigations, the ultimate responsibility for disciplining researchers and ensuring research integrity lies with the respective research institutions. Professional associations, educational institutions, and publishers can implement penalties, suspensions, and even expulsions of the authors.[13] The ultimate goal here is not punishment, but developing communication with institutions, so that they can recognize causes and institute remedial measures at the root level.

Concluding remarks

Plagiarism is a chronic disease aggravated by modern technology. Multiple definitions of plagiarism and the resulting vague interpretations need to be replaced by standard working definitions to clearly guide researchers against plagiarizing their work and to guide editors/reviewers in the detection of plagiarism and subsequent action that is needed. Anti-plagiarism software should be complemented by human context-based evaluation, rather than a rigid percentage-based decision. Education, firm mentorship and peer guidance, stringent policies, high ethical standards, and a research-conducive environment are necessary to minimize the (rising) threat of plagiarism. Comprehensive preventive measures at the root level include introduction of research and scientific writing early in the educational curriculum, ongoing workshops to train budding researchers, resources for anti-plagiarism check, clear institutional policies (for dealing with plagiarism once it has been detected), and a shift from prioritizing quality over quantity for academic publications. The responsibility of ethical research conduct lies with all the stakeholders – researchers, reviewers, editors, journals/publishers, institutions and readers – who should remain vigilant and bear zero tolerance toward plagiarism. Strict punitive measures should be instituted against serial plagiarists. Editors play a strong role in the entire process from screening for plagiarism to determining if it is minor/major with pertinent consequences. With increasing globalization and collaborative research, there should be strict adherence to ethical standards in publishing to ensure the integrity of academic and scientific pursuits.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.