Reporting guidelines (EQUATOR Network)

Last updated: 2026-10-20

Synthesis

Reporting guidelines are checklists that specify, study design by study design, what information a paper must report to be evaluable. They are not style advice; they are minimum-information standards, analogous to MIAME for microarrays or DICOM for imaging. The EQUATOR Network (equator-network.org), an international consortium founded in 2008, maintains the canonical registry. Its online library listed roughly 700 records across study designs and reporting domains when checked on 2026-10-20 [10]. (That figure is a count of the library listing, not a number EQUATOR publishes.)

For a scriptorium-style writing system, reporting guidelines are the validation contract. Where narrative frameworks (narrative-frameworks) tell you whether a paper is readable, reporting guidelines tell you whether it is evaluable. A trial paper that omits randomization details is not recoverable by good prose. Several scriptorium skills planned for v0.3 — in particular statistics-consistency, sample-size-validation, and a future reporting-guideline-compliance skill — are grounded directly in these checklists.

The major reporting guidelines have been published with explanation-and- elaboration companion documents that work as worked examples and are useful as the substrate for LLM-driven compliance checks. Adoption by journals is mixed: high-impact medical journals largely require CONSORT, PRISMA, and STROBE in their author guidelines; specialty and methods journals are patchier; AI-specific extensions are very new and not yet uniformly required.

Two 2025–2026 developments matter most here. First, the two core trial guidelines were revised: CONSORT 2025 [11] replaces CONSORT 2010 and renumbers the checklist, and SPIRIT 2025 [12] does the same for trial protocols. Any tool that cites CONSORT item numbers must therefore name the version. Second, there is now direct validation evidence for LLM-assisted adherence assessment. Zero-shot or lightly prompted LLMs reach about 90% agreement with expert raters when assessing trial reporting against CONSORT- and SPIRIT-based items [20, 21, 22]. AutoReporter, the only one of these studies to report chance-corrected agreement, gives Cohen’s κ of 0.70–0.77 [22]. On the 93-item TRIPOD checklist the best models reach F1 ≈ 0.73–0.74 [23]. The authors recommend these tools for semi-automated use with human checking, not full automation.

Evidence and frameworks

CONSORT 2010 and CONSORT 2025 — randomized controlled trials

Schulz KF, Altman DG, Moher D (for the CONSORT Group). CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332. DOI: 10.1136/bmj.c332. [1]

CONSORT is a 25-item checklist plus a participant flow diagram. The check covers trial design, randomization details, blinding, sample size justification, primary and secondary outcomes, statistical methods, harms, and limitations. It is required (or “strongly encouraged”) by most major clinical journals and by ICMJE-member journals.

The extensions are now numerous: cluster trials, non-inferiority/equivalence trials, pragmatic trials, herbal interventions, and — most relevant for modern manuscripts — CONSORT-AI (see below).

Hopewell S, Chan A-W, Collins GS, Hróbjartsson A, et al. CONSORT 2025 statement: updated guideline for reporting randomised trials. BMJ. 2025;389:e081123. DOI: 10.1136/bmj-2024-081123. [11]

CONSORT 2025 (published 14 April 2025, simultaneously in several journals) is a 30-item checklist plus a flow diagram. The developers “added seven new checklist items, revised three items, deleted one item, and integrated several items from key CONSORT extensions”, and restructured the checklist around a new Open science section [11]. That section covers trial registration (item 2), access to the protocol and statistical analysis plan (item 3), data sharing (item 4), and funding and conflicts of interest (items 5a–b). Patient and public involvement is a new item (8). Harms items from the CONSORT Harms extension are integrated (items 7, 15, 21a, 23a, 27).

Item numbers shifted. Sample-size determination was item 7a in CONSORT 2010 and is 16a in CONSORT 2025. Statistical methods were item 12 and are now 21a–d, with missing-data handling (21c) as its own sub-item. Sequence generation is 17a–b. A tool that cites “CONSORT 7a” or “CONSORT 17b” without naming the version will point at the wrong item. A companion explanation-and-elaboration paper was published alongside the statement (BMJ 2025;389:e081124).

SPIRIT 2025 — trial protocols

Chan A-W, Boutron I, Hopewell S, Moher D, et al. SPIRIT 2025 statement: updated guideline for protocols of randomised trials. BMJ. 2025;389:e081477. DOI: 10.1136/bmj-2024-081477. [12]

SPIRIT 2025 is a 34-item minimum checklist for trial protocols plus a schedule-of-enrolment diagram. The update added two new items, revised five, and deleted or merged five. It also introduced a new open-science section, put more emphasis on harms and on describing interventions and comparators, and added a new patient-and-public-involvement item [12]. It is the protocol-stage counterpart to CONSORT 2025. A manuscript that is a trial protocol should be audited against SPIRIT 2025, not CONSORT.

STROBE — observational studies

von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP (for the STROBE Initiative). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. DOI: 10.1371/journal.pmed.0040296. [2]

A 22-item checklist for cohort, case-control, and cross-sectional studies (18 items common; 4 design-specific). STROBE is the de facto standard for epidemiological reporting and the foundation for several extensions (STROBE-ME for molecular epidemiology, RECORD for routinely collected health data, STROBE-Vet for veterinary studies).

PRISMA 2020 — systematic reviews

Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. DOI: 10.1136/bmj.n71. [3]

PRISMA 2020 replaced the 2009 statement (Moher et al., DOI: 10.1136/bmj.b2535). The 27-item checklist (with an expanded abstract checklist and revised flow diagrams) addresses search strategy, study selection, risk-of-bias assessment, certainty assessment, and synthesis methods. PRISMA extensions exist for network meta-analyses, scoping reviews, diagnostic test accuracy reviews, and individual participant data.

As of October 2026, PRISMA 2020 is still the current version. The official PRISMA site lists two AI-related items as under development: a PRISMA extension for systematic reviews of AI interventions, and a “partial update of PRISMA 2020 to incorporate guidance on use of AI tools in the systematic review process” [19]. It also lists updates of the network-meta-analysis, scoping-review and protocol extensions as under development. No full PRISMA successor has been published.

ARRIVE 2.0 — animal research

Percie du Sert N, Hurst V, Ahluwalia A, et al. The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. PLoS Biol. 2020; 18(7):e3000410. DOI: 10.1371/journal.pbio.3000410. [4]

ARRIVE 2.0 reorganized the 2010 original into two priority tiers: the “Essential 10” (the absolute minimum, including study design, sample size, inclusion/exclusion criteria, randomization, blinding, outcome measures, statistical methods, experimental animals, experimental procedures, and results) and a Recommended Set covering ethics, housing, contextual detail.

STARD 2015 — diagnostic accuracy

Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. DOI: 10.1136/bmj.h5527. [5]

A 30-item checklist for studies estimating the accuracy of a diagnostic test against a reference standard. Co-published in BMJ, Radiology, and Clinical Chemistry in October 2015. A STARD-AI extension was published in 2025 (Sounderajah V, et al. Nat Med. 2025;31(10):3283–3289; DOI: 10.1038/s41591-025-03953-8) [13] to address artificial-intelligence-based diagnostic tools.

TRIPOD and TRIPOD+AI — prediction models

Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): the TRIPOD Statement. BMJ. 2015;350:g7594. DOI: 10.1136/bmj.g7594. [6]

TRIPOD covers the development, validation, or update of multivariable prediction models. The 22-item checklist addresses outcomes, predictors, sample size, missing data, model specification, performance, and external validation.

Collins GS, Moons KGM, Dhiman P, et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. 2024;385:e078378. DOI: 10.1136/bmj-2023-078378. [7] (The earlier citation here, BMJ 2024;385:q824, is the accompanying editorial by Cohen and Bossuyt, not the statement itself.)

TRIPOD+AI expanded the checklist to 27 items and unified reporting for regression-based and ML-based models. It places new emphasis on trustworthiness, fairness, and reproducibility — including reporting of training/validation/test splits, hyperparameter selection, and fairness across demographic subgroups.

Generative-AI and LLM reporting guidelines (2025–2026)

LLM-specific guidance has arrived quickly since 2025:

• TRIPOD-LLM [14]: an extension of TRIPOD+AI for studies that develop, tune, prompt-engineer, or evaluate LLMs in biomedicine. It has 19 main items and 50 subitems. The format is modular, with 14 main items applicable across all study categories, and the authors describe it as a “living document”.
• REFINE [15]: a 44-item, six-section checklist for foundation-model and LLM studies in medical research, including imaging AI. It was published ahead of print in February 2026.
• CHART [16]: the Chatbot Assessment Reporting Tool, for studies that evaluate generative-AI chatbots giving health advice. It was published in 2025, jointly across several journals, with an explanation-and-elaboration paper in BMJ.
• A reporting checklist for LLMs in behavioural science [17] (Nat Hum Behav, 2026). This is the first such checklist outside biomedicine in this set.
• GAMER [18] is different from the others. It is not a study-design guideline but a nine-item checklist for disclosing how generative-AI tools were used in producing medical research. Its items cover a general declaration, tool specifications, prompting techniques, the tool’s role, any new models developed, AI-assisted manuscript sections, content verification, data privacy, and impact on conclusions.

GAMER is directly relevant to scriptorium users. An author who used scriptorium skills while drafting is the kind of author GAMER asks to disclose.

Evidence on LLM-assisted adherence assessment (2025–2026)

Four validation studies now bear directly on any skill that walks a checklist with an LLM:

• Wrightson et al. 2025 [20]: 113 sports-medicine trial reports, nine reporting questions. GPT-4 Turbo reached accuracy 90% (F1 0.89). Llama 2 70B started at 64% and reached 83% after fine-tuning. GPT-4 Vision identified every participant-flow diagram but was only 57% accurate at detecting missing details in them.
• Srinivasan et al. 2025 [21]: a zero-shot GPT-4o-mini pipeline matched experts on 91.7% of 2,210 item decisions (macro-F1 0.86 on the CONSORT-TM benchmark). Applied to 21,041 open-access RCTs, it found mean CONSORT compliance rising from 27.3% (1966–1990) to 57.0% (2010–2024). Critical items stayed rare: allocation concealment was reported in 16.1% of trials and external validity was discussed in 1.6%.
• Chen et al. 2026, AutoReporter [22]: a zero-shot, no-retrieval o3-mini prompt reached 90.1% accuracy on CONSORT (κ = 0.70) and 92.1% on SPIRIT (κ = 0.77). It matched or exceeded fine-tuned domain models without annotated training data.
• Spiero et al. 2026 [23]: five LLMs scored the 93 TRIPOD items for 70 manually scored prediction-model papers. Gemini-2.5-pro (F1 0.74) and GPT-5 (F1 0.73) performed best. Gemini-2.5-pro’s supporting quotes were correct in 80–97% of cases. The authors conclude these models suit “semiautomated” rather than fully automated use.

Interpretation (researcher inference). These studies support the present / partial / missing / not-applicable design with quoted anchors. Agreement around 90%, with κ of 0.70–0.77 where it was reported [22], is useful, but it means roughly one item decision in ten will be wrong. Requiring a verbatim quote per present row is the check that lets an author catch those errors. All four studies evaluate published papers against established checklists. None evaluates pre-submission drafts or the 2025 checklists.

CONSORT-AI and SPIRIT-AI — AI-enabled trials

Liu X, Cruz Rivera S, Moher D, Calvert MJ, Denniston AK; SPIRIT-AI and CONSORT-AI Working Group. Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. Nat Med. 2020;26:1364–1374. DOI: 10.1038/s41591-020-1034-x. [8]

Cruz Rivera S, Liu X, Chan A-W, et al. Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nat Med. 2020;26:1351–1363. DOI: 10.1038/s41591-020-1037-7. [9]

These were the first international standards for trial reporting and protocols of AI-based interventions. SPIRIT-AI covers protocols (before the trial); CONSORT-AI covers final trial reports. Both add items for AI intervention description, input data handling, error analysis, and human/AI interaction.

CONSORT-AI and SPIRIT-AI were built on the 2010 and 2013 base checklists. CONSORT 2025 and SPIRIT 2025 do not fold in AI-specific items [11, 12]. SPIRIT 2025 addresses the gap directly. Its developers say they will work with extension leaders to align the extensions with SPIRIT 2025, and “In the meantime, we recommend that the existing version of the relevant SPIRIT extensions be used” [12]. SPIRIT-AI is among the extensions it cites. An AI-intervention trial therefore needs both the 2025 base checklist and the 2020 AI extension, with the version mismatch stated. (Researcher inference: applying the same approach to CONSORT-AI is by analogy, and we found no published CONSORT-AI or SPIRIT-AI update as of 2026-10-20.)

Beyond these: the EQUATOR Network

The EQUATOR Network maintains the registry at equator-network.org. Other relevant guidelines include:

• CARE — case reports (DOI: 10.7453/gahmj.2013.008)
• SRQR / COREQ — qualitative research synthesis and reporting
• AGREE II — practice guidelines
• REMARK — tumor marker studies
• CHEERS 2022 — health economic evaluations (DOI: 10.1136/bmj-2021-067975)

For any given manuscript, the EQUATOR search tool returns the applicable guidelines by study design. A scriptorium skill that auto-detects design type from manuscript content could route to the right checklist automatically; this is a candidate v0.3 capability.

How this informs scriptorium

Reporting guidelines are the cleanest validation substrate available because they are enumerable. Each guideline ships as a numbered checklist, and each checklist item has a canonical answer pattern. This is ideal ground for structured-output skills.

Direct connections:

• statistics-consistency (planned, v0.3). Should ground its checks in the statistics items of the relevant guideline: CONSORT 2010 items 7a and 12, which are CONSORT 2025 items 16a and 21a–d; STROBE item 12; TRIPOD items 8 and 9. Inconsistencies between methods and results in those specific items are higher-severity than free-text arithmetic mismatches. See statistical-inconsistency for the deterministic-recomputation discipline.
• sample-size-validation (planned). Should detect whether sample-size justification is present and whether it names the assumed effect size, power, and alpha. The relevant items are CONSORT 2010 7a / CONSORT 2025 16a, ARRIVE Essential 10 item 2, and STROBE item 10. The skill should not invent these values; it should only flag their absence.
• reporting-guideline-compliance (v0.3). A direct mapping from manuscript sections to checklist items. Output: per-item status (present / partial / missing / not-applicable) with a span pointer for each “present” claim. This is a critique skill, not transformative. Since April 2025 the default trial checklist should be CONSORT 2025, with CONSORT 2010 offered only when the author or venue requires it. Trial protocols should be audited against SPIRIT 2025. Item numbers must always carry the checklist version. The 2025–2026 validation studies [20–23] support checklist walking by LLM, but with an error rate of roughly one item in ten. Quoted anchors are therefore essential, not optional. Checks on figure-borne items, such as completeness of the flow diagram, are weaker still and should default to partial with a request to verify [20].
• reporting-guideline-fit (v0.2). The routing step should know the 2025–2026 additions. Route trial protocols to SPIRIT 2025, trial reports to CONSORT 2025 (plus CONSORT-AI for AI interventions), LLM-evaluation studies to TRIPOD-LLM or REFINE, and chatbot health-advice studies to CHART. GAMER is an additional disclosure checklist that applies whenever generative AI was used in the research. It does not replace the design checklist.
• journal-style-conversion (planned). Reporting guidelines are part of journal style. A conversion skill should know which guideline each target journal requires and check that the manuscript complies before reformatting citations or section titles.
• reviewer-simulation (current). The simulator should consult the applicable guideline as a baseline of reviewer concerns — many real reviewer comments are paraphrases of checklist items the manuscript didn’t satisfy. Srinivasan et al. [21] suggest where to start: allocation concealment and external-validity discussion are the most chronically under-reported CONSORT items.

A practical design note: the manuscript-state schema should be able to record which reporting guideline(s) apply, so skills don’t have to re-detect. Adding an optional reporting_guidelines: [CONSORT-AI, TRIPOD+AI] field to MANUSCRIPT_STATE.yaml would let any skill consume the answer. Entries should carry a version, for example CONSORT 2025 rather than CONSORT.

Implementation priority for scriptorium

Verdict: Yes. The current skills are reporting-guideline-fit (v0.2) and reporting-guideline-compliance (v0.3), and reviewer-simulation consumes the guideline as a baseline.

If Yes:

• reporting-guideline-compliance: update the default trial checklist from CONSORT 2010 to CONSORT 2025, and add SPIRIT 2025 for protocols. Version-qualify every item number. Keep quoted anchors mandatory, which the LLM-adherence evidence [20–23] now directly supports.
• reporting-guideline-fit: add routing for SPIRIT 2025, TRIPOD-LLM, REFINE and CHART. Surface GAMER as an add-on disclosure checklist when generative-AI use is declared.
• Required data: the checklist text for each named version. The CONSORT 2025 and SPIRIT 2025 checklists are published open access [11, 12].
• Not yet: automated flow-diagram auditing. The one vision-model result (57% accuracy at detecting missing flow-diagram details [20]) is too weak to rely on.

Open questions / weak evidence

• Adoption rates vary widely, and reporting remains incomplete even where CONSORT is endorsed. A Cochrane review found that endorsing journals reported trials more completely on 25 of 27 outcomes, five of them statistically significant. Allocation concealment, for example, was adequately reported in 45% of RCTs in endorsing journals and 22% in non-endorsing ones [24]. A large LLM-assisted audit put mean CONSORT compliance at 57% for 2010–2024 [21]. Compliance for newer guidelines is lower. (This replaces an earlier citation to “Turner et al. Trials 2012”, which we could not locate. The Turner et al. 2012 review appeared in the Cochrane Database of Systematic Reviews and in summary form in Systematic Reviews.)
• Whether guideline compliance improves outcomes (reduced research waste, more accurate meta-analyses) is plausible but not directly established by RCT-grade evidence — the natural experiment is hard to run.
• The AI extensions are very new. CONSORT-AI / SPIRIT-AI / TRIPOD+AI have been adopted by Nature Medicine, BMJ, and Lancet Digital Health but field-wide uptake is still patchy. The 2025–2026 wave of LLM guidelines (TRIPOD-LLM, REFINE, CHART, GAMER, and the behavioural-science checklist) overlaps in scope, and it is not yet clear which one journals will require.
• How CONSORT-AI and SPIRIT-AI fit with the 2025 base checklists is unresolved (see above).
• The LLM-adherence studies [20–23] all evaluate published papers against established checklists. Their performance on pre-submission drafts, on the renumbered 2025 checklists, and on non-English or non-biomedical manuscripts is untested.
• EQUATOR catalogs hundreds of guidelines; many overlap, many are dormant. Auto-detecting which guideline applies to a given manuscript is non-trivial. A naïve routing rule (e.g., design-type keyword match) will misroute prospectively-collected cohort studies that look like trials.

References

1. Schulz KF, Altman DG, Moher D; CONSORT Group. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332. DOI: 10.1136/bmj.c332.
2. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. DOI: 10.1371/journal.pmed.0040296.
3. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. DOI: 10.1136/bmj.n71.
4. Percie du Sert N, Hurst V, Ahluwalia A, et al. The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. PLoS Biol. 2020;18(7):e3000410. DOI: 10.1371/journal.pbio.3000410.
5. Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. DOI: 10.1136/bmj.h5527.
6. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): the TRIPOD Statement. BMJ. 2015;350:g7594. DOI: 10.1136/bmj.g7594.
7. Collins GS, Moons KGM, Dhiman P, et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. 2024;385:e078378. DOI: 10.1136/bmj-2023-078378. PMID: 38626948. (Correction: BMJ 2024;385:q902. Accompanying editorial: Cohen JF, Bossuyt PM. BMJ 2024;385:q824, DOI: 10.1136/bmj.q824.)
8. Liu X, Cruz Rivera S, Moher D, Calvert MJ, Denniston AK; SPIRIT-AI and CONSORT-AI Working Group. Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. Nat Med. 2020;26:1364–1374. DOI: 10.1038/s41591-020-1034-x.
9. Cruz Rivera S, Liu X, Chan A-W, et al. Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nat Med. 2020;26:1351–1363. DOI: 10.1038/s41591-020-1037-7.
10. EQUATOR Network. Enhancing the QUAlity and Transparency Of health Research. https://www.equator-network.org/ (accessed 2026-10-20; library listing at https://www.equator-network.org/reporting-guidelines/).
11. Hopewell S, Chan A-W, Collins GS, Hróbjartsson A, et al. CONSORT 2025 statement: updated guideline for reporting randomised trials. BMJ. 2025;389:e081123. DOI: 10.1136/bmj-2024-081123. PMID: 40228833. Explanation and elaboration: BMJ 2025;389:e081124, DOI: 10.1136/bmj-2024-081124.
12. Chan A-W, Boutron I, Hopewell S, Moher D, et al. SPIRIT 2025 statement: updated guideline for protocols of randomised trials. BMJ. 2025;389:e081477. DOI: 10.1136/bmj-2024-081477. PMID: 40294953.
13. Sounderajah V, Guni A, Liu X, Collins GS, et al.; STARD-AI Steering Committee. The STARD-AI reporting guideline for diagnostic accuracy studies using artificial intelligence. Nat Med. 2025;31(10):3283–3289. DOI: 10.1038/s41591-025-03953-8. PMID: 40954311.
14. Gallifant J, Afshar M, Ameen S, Aphinyanaphongs Y, et al. The TRIPOD-LLM reporting guideline for studies using large language models. Nat Med. 2025;31(1):60–69. DOI: 10.1038/s41591-024-03425-5. PMID: 39779929.
15. Mese I, Akinci D’Antonoli T, Bluethgen C, Bressem K, et al. Reporting checklist for foundation and large language models in medical research (REFINE): an international consensus guideline. Diagn Interv Radiol. 2026 (ahead of print). DOI: 10.4274/dir.2026.263812. PMID: 41742713.
16. CHART Collaborative. Reporting guidelines for chatbot health advice studies: explanation and elaboration for the Chatbot Assessment Reporting Tool (CHART). BMJ. 2025;390:e083305. DOI: 10.1136/bmj-2024-083305. PMID: 40750271. (The CHART statement itself was co-published in several journals, including BMJ Medicine and BMC Medicine.)
17. Feuerriegel S, Barrie C, Crockett MJ, Globig LK, et al. A reporting checklist for large language models in behavioural science. Nat Hum Behav. 2026;10(7):1182–1186. DOI: 10.1038/s41562-026-02492-7. PMID: 42265331.
18. Luo X, Tham YC, et al.; GAMER Working Group. Reporting guideline for the use of Generative Artificial intelligence tools in MEdical Research: the GAMER Statement. BMJ Evid Based Med. 2025;30(6):390–400. DOI: 10.1136/bmjebm-2025-113825. PMID: 40360239.
19. PRISMA Executive. PRISMA extensions (including extensions under development). https://www.prisma-statement.org/extensions (accessed 2026-10-20).
20. Wrightson JG, Blazey P, Moher D, et al. GPT for RCTs? Using AI to determine adherence to clinical trial reporting guidelines. BMJ Open. 2025;15(3):e088735. DOI: 10.1136/bmjopen-2024-088735. PMID: 40107689.
21. Srinivasan A, Berkowitz J, Friedrich NA, et al. Large Language Model Analysis of Reporting Quality of Randomized Clinical Trial Articles: A Systematic Review. JAMA Netw Open. 2025;8(8):e2529418. DOI: 10.1001/jamanetworkopen.2025.29418. PMID: 40875232.
22. Chen D, Li P, Khoshkish E, et al. AutoReporter: development of an artificial intelligence tool for automated assessment of research reporting guideline adherence. J Am Med Inform Assoc. 2026;33(3):724–731. DOI: 10.1093/jamia/ocaf223. PMID: 41435508.
23. Spiero I, Leeuwenberg AM, Kamperman J, et al. Using large language models for automated assessment of reporting quality and completeness of prediction model studies. J Clin Epidemiol. 2026;196:112320. DOI: 10.1016/j.jclinepi.2026.112320. PMID: 42134730.
24. Turner L, Shamseer L, Altman DG, et al. Consolidated standards of reporting trials (CONSORT) and the completeness of reporting of randomised controlled trials (RCTs) published in medical journals. Cochrane Database Syst Rev. 2012;11:MR000030. DOI: 10.1002/14651858.MR000030.pub2. PMID: 23152285. Summary version: Syst Rev. 2012;1:60. DOI: 10.1186/2046-4053-1-60.