The proposed guideline aims to expand COREQ’s established framework for reporting qualitative research by addressing the emerging use of large language models (LLMs) across multiple stages of qualitative research. Although COREQ, endorsed by the EQUATOR network and cited more than 30,000 times, provides essential criteria for transparent, standardized qualitative research, it does not yet reflect recent advances in artificial intelligence (AI), especially LLMs.The integration of LLMs is rapidly transforming qualitative research. LLMs generate coherent, including why a new reporting token prediction. This enables their use at multiple research stages: from formulating research questions and designing interview protocols to supporting data processing (e.g., transcription and translation), facilitating data analysis (e.g., coding and theme identification), drafting manuscript sections, and enabling interactive exploration of qualitative datasets.The increasing adoption of LLMs stems from their potential to enhance analytical efficiency by rapidly processing large datasets, allowing broader and faster research investigations. Additionally, LLMs offer novel analytical perspectives, revealing latent semantic patterns, proposing thematic frameworks, and distilling complex narratives. These capabilities may complement human interpretive efforts.
Popular qualitative analysis software platforms, such as MAXQDA, now integrate LLM functionalities to leverage these advantages.However, the use of LLMs introduces notable methodological risks that require rigorous reporting standards. Unlike deterministic software, LLMs operate probabilistically, making their semantic comprehension uncertain, particularly concerning nuanced contexts. LLMs may also replicate biases present in their training data, produce opaque internal reasoning, omit information, and generate plausible yet incorrect content (“hallucinations”). Specific model characteristics, including model type, version, training data, parameter settings, context window, and prompting strategies, substantially impact research outcomes. Together, these factors pose challenges to reproducibility, validity, and trustworthiness of qualitative research, revealing a critical gap in the existing COREQ guidelines. Currently, COREQ’s single relevant item (Item 27 “What software, if applicable, was used to manage the data?”) inadequately captures essential details needed to evaluate LLM integration comprehensively.To address this gap, we are developing COREQ+LLM, an extension to the original COREQ checklist designed explicitly for qualitative research employing LLMs. COREQ+LLM will introduce detailed reporting requirements for model selection, training parameters, prompting strategies, task definitions, evaluation metrics, and human oversight mechanisms. By providing clear and practical guidance, this extension aims to uphold rigorous, transparent, and ethically sound qualitative health research in an era increasingly shaped by AI.
Protocol available here
Registered: 18 June 2025
Last updated:
Contact: Priv. Doz. Dr. Leonard Fehring Email: [email protected]