CASRAI Dictionary

Tag: ICMJE

  • How to write a CRediT statement for medical research in 2026

    Medical-research contributorship sits at an awkward intersection. The International Committee of Medical Journal Editors (ICMJE) still defines who may sign as an author of a clinical paper through its four-part test: substantial contribution to conception/design or acquisition/analysis/interpretation; drafting or critical revision; final approval; and accountability. CRediT, the Contributor Roles Taxonomy that CASRAI helped steward into NISO Z39.104-2022, sits underneath and describes what each named contributor actually did. In 2026, after another wave of journal adoption and the long-anticipated alignment with ORCID’s contributor affiliation model, a CRediT statement is no longer a discretionary nicety. It is the contributorship record of the paper.

    This post walks through how to write a CRediT statement that satisfies a medical journal’s submission system in 2026, with attention to the editorial conventions of NEJM, The Lancet, JAMA, and The BMJ. It assumes you have already worked out who meets the ICMJE authorship threshold; see our medical-research authors guide for that step.

    Authorship versus contributorship: not the same question

    The first error we see in submissions is conflating ICMJE authorship with CRediT contributorship. ICMJE answers a binary: does this person qualify to be listed as an author and to be accountable for the work? CRediT answers a granular: of the people who are listed, who did what? A statistician who ran the analysis but did not draft or revise may not meet ICMJE criteria and is acknowledged separately; if they do meet ICMJE criteria, then their CRediT role assignment would include Formal analysis, possibly Methodology, possibly Software, and they would be named on the byline. Liz Allen and the team that originated CRediT at Wellcome Trust were explicit on this distinction; the taxonomy was designed to complement, not replace, journal authorship rules.

    For medical research the second confounder is the guarantor. The BMJ has long required a named guarantor in addition to authors, and other ICMJE-following journals encourage the convention for clinical trials. The guarantor sits outside CRediT; it is closest in spirit to the Supervision role plus an accountability commitment, but it is not encoded in the taxonomy. In your CRediT statement, name the guarantor in a separate sentence; do not invent a Guarantor role.

    The 14 roles in medical-research context

    CRediT’s 14 roles were drafted for general research and need a brief translation when applied to clinical work. The full role definitions are normative; what follows is interpretive guidance, not a redefinition.

    • Conceptualization. The research question. For a registered clinical trial this is often a Principal Investigator role; for a secondary analysis it may be a junior contributor with a novel hypothesis.
    • Methodology. Study design, choice of endpoints, statistical-analysis-plan structure. A trial statistician contributing to the SAP earns this role even if a different person ran the final analysis.
    • Software. Programming for data capture (REDCap configuration counts), randomisation code, custom statistical packages, any analytic script that materially shaped results.
    • Validation. Reproduction of analyses, sensitivity analyses, cross-checks against an independent dataset. Often a co-author who replicates the lead analyst’s work.
    • Formal analysis. The statistical analysis itself.
    • Investigation. Recruitment, screening, consenting, clinical assessments, sample collection. Often the largest list of contributors in multi-site trials.
    • Resources. Provision of patient samples, biobanks, animal models, instrument time. Distinct from Funding acquisition.
    • Data curation. Data cleaning, harmonisation, query resolution, lock-down.
    • Writing – original draft. First-draft authorship of the manuscript.
    • Writing – review & editing. Substantive editorial revision, not copy-editing.
    • Visualization. Figures, including Kaplan-Meier curves, forest plots, CONSORT flow diagrams.
    • Supervision. Mentorship and oversight, often the senior author. A PI typically combines Supervision with Conceptualization and Funding acquisition.
    • Project administration. Coordination across sites, ethics submissions, sponsor liaison.
    • Funding acquisition. Grant-writing for the funded work.

    The lead/equal/supporting qualifier

    Adopted formally into NISO Z39.104 and now widely supported, the degree-of-contribution qualifier resolves a recurring source of disputes. For each role, exactly one contributor may be marked Lead, or several may be marked Equal; everyone else for that role is Supporting. In a multi-site oncology trial it is realistic to have a Lead Investigator (the coordinating PI), several Equal Investigators (site PIs), and a longer list of Supporting Investigators (sub-investigators, research nurses who meet ICMJE thresholds). The qualifier exists precisely so that the byline order does not have to encode contribution magnitude.

    Writing the statement

    A 2026-compliant CRediT statement is rendered as prose in the manuscript and as structured data in the submission system. Most major medical journals now extract the structured form from their submission portal directly; the prose paragraph is for the published version. Here is a worked example for a four-author RCT report:

    CRediT author statement. Sarah Chen: Conceptualization (lead), Methodology (lead), Funding acquisition (lead), Supervision (lead), Writing – review & editing (equal). Marcus Okonkwo: Investigation (lead), Project administration (lead), Data curation (lead), Writing – original draft (lead). Priya Raman: Formal analysis (lead), Software (lead), Validation (lead), Visualization (lead), Writing – review & editing (equal). David Holcombe: Methodology (supporting), Investigation (supporting), Writing – review & editing (supporting), Supervision (supporting). Guarantor: Sarah Chen.

    Note the explicit guarantor statement, separate from CRediT. Note also that not every role appears; Resources was inapplicable here and should be omitted rather than padded.

    JATS XML output

    For machine-actionable contributorship, journals serialise CRediT into JATS XML using the <role> element with the vocab="credit" attribute and the canonical role URI. The 2022 NISO version pinned the URIs at https://credit.niso.org/contributor-roles/<role-slug>/ with the qualifier expressed via specific-use="lead|equal|supporting". As an author you do not write the JATS by hand; you fill in the submission portal and the publisher’s tooling renders the XML. Where things go wrong is the round-trip: if the published HTML drops the qualifier, the JATS may also drop it and downstream Crossref deposits will be incomplete. If you care about the persistent record, check the published JATS via the publisher’s content syndication endpoint after acceptance.

    Journal-specific notes

    NEJM

    The New England Journal of Medicine adopted CRediT in late 2023 and integrated it into its Editorial Manager workflow in 2024. NEJM’s idiosyncrasy is that it still asks separately for the prose contribution statement, then asks each author to confirm their CRediT roles, and finally requires a writing-assistance declaration that is not CRediT (it covers professional medical writers funded by sponsors). Do not list a paid medical writer who does not meet ICMJE criteria under CRediT Writing – original draft; declare them in the acknowledgements with the funding source per Good Publication Practice (GPP 2022).

    The Lancet

    The Lancet was an early CRediT adopter and was unusual in coupling the taxonomy to a long-standing requirement for each author to write a one-sentence prose contribution statement in their own words. Both are retained in 2026. The prose statement is what readers see in the printed acknowledgements; the structured CRediT data lives in the JATS and in Crossref. For a Lancet submission, write the structured assignment first and then have each author translate their own roles into a single readable sentence.

    JAMA

    JAMA and the JAMA Network journals adopted CRediT in 2022 and tied it tightly to ORCID; an author without a verified ORCID iD cannot complete the contributorship form. JAMA also asks for explicit role assignments for Statistical analysis, Obtained funding, and Administrative, technical, or material support; these are journal-specific role labels that overlap with CRediT but are tracked separately for editorial QA. If you have a Formal analysis role under CRediT you must also tick Statistical analysis on the JAMA form, otherwise the submission will not validate.

    The BMJ

    The BMJ adopted CRediT in 2023 and retained its long-standing guarantor requirement on top. BMJ’s submission system asks for the CRediT roles in structured form and then asks the corresponding author to identify the guarantor by name. The published article carries both: the CRediT statement as prose, and the guarantor sentence beneath it. BMJ also continues to require declarations of relationships and activities (the BMJ-specific competing interests format) which sit alongside but separately from CRediT.

    Common failure modes

    Three patterns recur in submissions to medical journals. First, role inflation: assigning Conceptualization to every author by reflex. CRediT is a record, not a recognition device; if a co-author did not contribute to conceptualisation, do not assign that role. Second, byline order substituting for qualifiers: a paper with five equal first-authors should mark all five as Equal on the roles they share, not just rely on a footnote saying “these authors contributed equally.” Third, missing the writing roles: every paper has someone who wrote the first draft. If your CRediT statement omits Writing – original draft, the editor will ask.

    Adoption status and trajectory

    As of early 2026 the CRediT adoption ledger records 70+ publishers with active CRediT support and structured submission workflows in most major medical and biomedical journals. The ICMJE has not made CRediT mandatory across its full membership, but its 2024 update to the Recommendations explicitly endorses CRediT as an acceptable mechanism for describing contributions, and several ICMJE journals require it. Outside ICMJE, the trajectory is the same: PLOS, Cell Press, Springer Nature, Wiley, Taylor & Francis, Elsevier, OUP, CUP, and a long tail of society publishers now require structured CRediT at submission.

    What to do next

    If you are preparing a submission, work through these in order: (1) settle the authorship list against ICMJE criteria; (2) draft the CRediT role assignment in a shared document with qualifiers; (3) have each author confirm their roles in writing before submission; (4) enter the structured data in the submission portal and copy the prose statement into the manuscript; (5) declare the guarantor and any medical writers separately. The CASRAI CRediT authors guide contains a downloadable role-assignment worksheet that has saved more co-author disputes than any other artefact we publish.

    Related dictionary entries

    References

    ICMJE, Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals (2024 update). NISO Z39.104-2022, CRediT, Contributor Roles Taxonomy. Allen et al., Nature (2014), Publishing: Credit where credit is due. Brand et al., Learned Publishing (2015), Beyond authorship: attribution, contribution, collaboration, and credit. Holcombe, Publications (2019), Contributorship, not authorship.

  • GenAI in scholarly authorship: the 2026 disclosure landscape

    The 2023 ICMJE position that generative AI cannot be a co-author has aged into a stable consensus across scholarly publishing, but the implementation surface around it has grown fast. In 2026 the question is no longer whether to disclose AI use in a manuscript; it is how, where, and with what evidence. This post maps the current disclosure landscape, the technical mitigations that publishers expect authors to apply, and the residual uncertainty around detection.

    The ICMJE 2023 position and its echoes

    In January 2023 the ICMJE updated its Recommendations to add that chatbots cannot be authors because they cannot meet the accountability criterion: an LLM cannot take responsibility for the integrity of the work, cannot approve the final version in any meaningful sense, and cannot be contacted by readers seeking clarification. The position was endorsed within weeks by the World Association of Medical Editors and by COPE. By mid-2023 every major publisher had aligned. The AI co-authorship rejection is now treated as a settled norm.

    What replaced the brief flurry of “ChatGPT as co-author” papers was a more nuanced question: how should authors disclose AI use when the system is a tool? This is where 2024 and 2025 brought significant fragmentation, and where 2026 has begun to consolidate.

    The publisher landscape in 2026

    Nature and the Springer Nature stable

    Nature requires authors to declare any use of LLMs in the Methods section (for research articles) or in the acknowledgements (for editorial and review content). The declaration must specify the model, the version, the date of use, and the purpose. Nature does not permit AI-generated images or figures except where the AI generation itself is the subject of the research. Springer Nature has cascaded a similar policy across its journals with light variation.

    Cell Press and Elsevier

    Cell Press and Elsevier journals require disclosure of AI-assisted writing in a dedicated declaration that sits alongside competing interests and funding. The declaration is structured: type of tool, purpose (e.g., language polishing, literature search, code generation, image analysis), and a confirmation that the authors take full responsibility. Elsevier additionally requires that AI-generated text be reviewed and edited by the authors and explicitly forbids using AI for peer review.

    Wiley

    Wiley’s policy distinguishes between using AI as a tool (allowed with disclosure) and using AI to generate substantive intellectual content (not allowed). The distinction is fuzzy at the boundary, and Wiley’s submission system asks authors to self-classify. Wiley also publishes its Best practice guidelines on research integrity and publishing ethics which were updated in 2024 to cover GenAI in detail.

    PLOS, eLife, F1000Research

    The open-publishing platforms have generally taken the position that AI use must be disclosed and that authors are responsible for verification, but they have been more permissive about disclosed and reviewed AI use than the closed-access incumbents. eLife in particular has experimented with AI-assisted peer review summaries, with disclosure to authors and readers.

    What “disclosure” actually requires

    The fragmentation across publisher policies has converged on a common five-element disclosure, which CASRAI’s AI disclosure helper assembles into a publisher-specific declaration:

    1. Tool and version. Not “ChatGPT” but “GPT-4o (OpenAI), version of 2025-12-04.”
    2. Purpose. One of: language polishing, translation, literature search, code generation, data extraction, image analysis, hypothesis generation, draft writing. If the use spanned multiple purposes, list each.
    3. Scope. Which sections or artefacts were involved. “Abstract and discussion polished” is meaningfully different from “first draft written.”
    4. Human verification. A statement that named authors have reviewed and verified the output and take responsibility for it.
    5. Prompt and output retention. Increasingly, journals are asking authors to retain prompts and outputs for audit. Cell Press now formally asks; Nature recommends. Treat this as a 5-year retention obligation.

    See our AI disclosure for authors guide for the publisher-by-publisher decision tree.

    The hallucination problem

    The single largest editorial concern in 2026 remains hallucination: an LLM fabricating a citation, a method, or a result and the authors failing to catch it. Retraction Watch tracked over 200 retractions in 2024 and 2025 attributable in whole or part to undisclosed AI-generated fabrications, primarily fictitious references but increasingly fabricated quantitative results in tables.

    The mitigations are well-known and surprisingly under-applied:

    • Citation verification. Every citation in an LLM-generated draft must be checked against the actual source. Tools like Scite, Semantic Scholar’s citation graph, and Crossref’s metadata API help. The bare minimum: every DOI must resolve and the paper at that DOI must say what the LLM claims.
    • Numerical verification. If an LLM produces a number, the human author must reproduce the number from the underlying source. “The LLM said it” is not provenance.
    • Retrieval-augmented generation (RAG). Grounding an LLM in a fixed corpus of verified sources, with citation chaining, reduces but does not eliminate hallucination. RAG-based research-writing tools (Elicit, Consensus, scite Assistant) have an accuracy edge over raw LLMs precisely because they constrain the model to a verifiable corpus.

    Munafò and colleagues at the UK Reproducibility Network have argued, correctly in our view, that AI-assisted writing should sit inside the same reproducibility envelope as the rest of the work: prompts and outputs are part of the methods, not part of the prose.

    Detection and watermarking

    The detection problem has not been solved. Tools that claim to identify AI-generated text by perplexity or burstiness have unacceptable false-positive rates against careful human writers and are easily defeated by simple paraphrasing. AI-assisted writing is, on the open web, essentially undetectable in 2026.

    Three more promising directions exist. First, watermarking: the major LLM providers have prototyped statistical watermarks (Google’s SynthID-Text, OpenAI’s research-stage text watermark) that embed a detectable signal in token-selection statistics without affecting fluency. Adoption has been slow because authors can defeat watermarks by re-rolling with a different model, and because no publisher has committed to refusing un-watermarked submissions. Second, provenance metadata: the C2PA standard (originally for images) is being extended to text, with cryptographically signed assertions of generation source. Third, process auditing: rather than detecting AI in the output, audit the authors’ process artefacts (version history, prompt logs, draft trail). This is the direction in which institutional integrity offices are moving.

    For authors, the practical takeaway is that you should not rely on undetectability. The conservative path is disclosure plus verification.

    What about peer review?

    The 2026 consensus is that peer reviewers may not paste unpublished manuscripts into a third-party LLM. The reason is confidentiality, not anti-AI sentiment: a paper under review is privileged information and most LLM providers retain inputs in some form. NIH and several large funders have made this an explicit policy for proposal review; publishers are catching up. eLife and a handful of others are experimenting with publisher-hosted LLM tooling that does not exfiltrate the manuscript, which threads the needle.

    Where this is going

    Three trajectories are visible. First, the disclosure form will converge: expect a NISO or COPE-led standardisation of GenAI disclosure within 18-24 months, modelled on the structured CRediT statement. Second, prompt-and-output retention will become mandatory for high-stakes journals (clinical, regulatory-relevant), and audited at random. Third, the line between “AI as tool” and “AI as substantive contributor” will be tested by hybrid systems where the human author’s contribution is curation, framing, and verification rather than generation. We expect the integrity community to draw a harder line on quantitative and methodological substance than on prose: an AI may polish your discussion section with disclosure, but an AI may not propose your analytic method without that proposal being independently validated and disclosed.

    For now, the operating rule is straightforward. If you used AI, disclose it specifically. If the AI produced text or numbers in your paper, verify them yourself. If a publisher asks for prompts and outputs, retain them. If you are reviewing a paper, do not paste it into a chatbot. The GenAI disclosure domain at CASRAI tracks the publisher-by-publisher policy text for authors who need to comply across multiple submission targets.

    Related dictionary entries

    References

    ICMJE, Recommendations (January 2023 update, defining authorship to exclude AI). WAME, Chatbots, ChatGPT, and Scholarly Manuscripts (2023). COPE, Authorship and AI tools position statement (2023, reaffirmed 2025). Nature editorial, Tools such as ChatGPT threaten transparent science; here are our ground rules for their use (2023). Munafò et al., The reproducibility debate is an opportunity, not a crisis (PLOS Biology, 2022).