COREQ+LLM

COREQ+LLM extends the established COREQ checklist (Consolidated Criteria for Reporting Qualitative Research) with reporting items for qualitative studies that use large language models (LLMs) as research tools. It was developed through a systematic scoping review, a two-round Delphi consensus process, and a virtual consensus conference with an international, interdisciplinary panel of experts in qualitative research, AI, medicine, ethics and journal editing.

Why it is needed

LLMs are increasingly used across the qualitative research process: to design interview guides, transcribe and translate data, support coding and thematic analysis, and even interact directly with qualitative datasets. Because LLMs generate text probabilistically rather than deterministically, can reproduce or amplify biases in their training data, and can confidently produce incorrect content ("hallucination"), their use raises new questions for transparency, reproducibility and interpretive trust. COREQ covers software use in a single item and does not capture these LLM-specific considerations.

How it was developed

A scoping review of peer-reviewed literature (January 2020–May 2025), conducted according to PRISMA-ScR, identified 149 studies using LLMs in qualitative health research. From this evidence, 29 candidate reporting items were drafted across seven categories. An international panel of experts then rated these items in two online Delphi rounds (141 and 115 valid responses), with remaining disagreements resolved in a virtual consensus conference. The resulting COREQ+LLM checklist comprises 31 items across seven categories. The full methods and Delphi results are reported in the COREQ+LLM manuscript; the scoping review methodology is described in the published study protocol (JMIR Research Protocols, 2025).

Scope

The guideline was developed with a focus on generative LLMs, such as the GPT family of models. The expert panel agreed that its principles are broadly applicable to other natural-language-processing techniques and architectures as well, including encoder-only models such as BERT. Not every item will be relevant to every study; authors can record an item as "not applicable" together with a brief justification.

Use it when

Your qualitative study uses an LLM as a research tool at any stage, for example to:

• formulate research questions or draft interview guides
• transcribe or translate data
• code data or identify themes
• interact with qualitative datasets in dialogue

Who it is for

• Authors planning or writing up LLM-supported qualitative studies
• Reviewers and journal editors appraising such submissions

How it fits with COREQ

COREQ+LLM is used together with COREQ, not instead of it. COREQ covers the general reporting of interviews and focus groups; COREQ+LLM covers the LLM-specific parts.

Outside its scope

Studies in which an LLM is only the object of evaluation (for example, benchmarking model performance) or which examine only perceptions of LLMs are not the target.

Report alongside the original COREQ

COREQ+LLM is used together with the original COREQ checklist, not instead of it. Report your study against the 32 original COREQ items for the general conduct of interviews or focus groups, and against the COREQ+LLM items for the LLM-specific parts of your workflow.

Working through the checklist

• Go through each item and report the requested information in your manuscript or in a supplementary reporting table
• If an item does not apply to your study, mark it as "not applicable" and briefly say why
• If you used more than one model, report each model and its role in the workflow

For a detailed explanation of every item, see the item-by-item guide.

Download and cite

Download the checklist using the button above. See “How to cite the guideline?” below for the recommended citation.

While the COREQ+LLM manuscript is under review, please cite the preprint:

Fehring, L., Frings, J., Rust, P., Kempny, C., Thuermann, P., Tamerler, E., & Meister, S. (2026). Reporting of Qualitative Research Using Large Language Models (COREQ+LLM): A Delphi-based Extension of the COREQ Reporting Guideline. medRxiv., 2026.2010.2001.26364479. doi: 10.64898/2026.10.01.26364479

COREQ+LLM was developed by a steering committee at Witten/Herdecke University, Germany.

Contact

Leonard Fehring
Health Care Informatics, Faculty of Health, School of Medicine
Witten/Herdecke University
Pferdebachstrasse 11, 58455 Witten, Germany
leonard.fehring@uni-wh.de