COREQ Checklist: Reporting Qualitative Research

The COREQ checklist is a 32-item reporting guideline for interview and focus group research, published by Tong, Sainsbury and Craig in 2007. Its name stands for Consolidated Criteria for Reporting Qualitative Research, and it was built by consolidating the reporting recommendations already scattered across the qualitative literature into one list. Its three domains cover the research team and reflexivity, the study design, and the analysis and findings.

What makes COREQ distinctive among reporting guidelines is that it treats the researcher as part of the method. A trial report can leave the personalities of the investigators out entirely, because randomisation is supposed to make them irrelevant. In an interview study the interviewer is the instrument, so who they were, what they believed, and how participants saw them are data a reader needs. That is the part authors most often skip, and the part reviewers most reliably ask about.

Which guideline your qualitative study needs

COREQ is narrower than it is often assumed to be. It was written for studies collecting data through individual interviews and focus groups, and its items assume that shape. If your study did something else, a different guideline fits better:

• Interviews or focus groups. COREQ, 32 items. The best fit and the one most journals in health research expect.
• Broader qualitative designs, including ethnography, observation, document analysis, or mixed qualitative approaches. SRQR, 21 items, published by O'Brien and colleagues in 2014. Less prescriptive about data collection, and a more honest fit when your design is not interview-based.
• Qualitative evidence synthesis, where you are synthesising other people's qualitative studies rather than collecting data. A synthesis-specific guideline applies, and the reporting job is different again.
• Mixed methods studies. Report the qualitative component against COREQ or SRQR and the quantitative component against its own guideline rather than trying to force both into one checklist.

Choosing between COREQ and SRQR is a genuine decision rather than a formality, and stating which you used and why takes one sentence in the methods. The EQUATOR Network library lists both alongside the extensions.

Domain one: the research team and reflexivity

Eight items, and the domain that separates a thorough qualitative report from a thin one. Described in our own words, it asks who conducted the interviews or focus groups, what their credentials and occupation were, their gender, their experience and training in qualitative methods, whether they had a relationship with participants before the study began, what participants knew about the researcher, and what characteristics of the researcher might have influenced the research, including assumptions and reasons for interest in the topic.

Authors often read this as an invitation to declare bias and then apologise for it. It is not. The purpose is interpretive: a reader who knows the interviewer was a clinician in the ward where participants were patients will read the transcripts differently from one who assumes an independent researcher. Withholding that context does not make a study more objective, it makes it harder to use. The official item wording sits with the guideline and on EQUATOR, and should be taken from there.

Domain two: study design

Fifteen items, covering the methodological spine of the study. It asks for the theoretical framework and methodological orientation underpinning the analysis, whether that is grounded theory, phenomenology, content analysis or something else; how participants were selected and by what sampling strategy; how they were approached; the sample size and the number who declined or dropped out with reasons; the setting of data collection and who else was present; the characteristics of the sample; whether an interview guide or topic list was used and whether it was pilot tested; whether repeat interviews were carried out; whether the data were audio or video recorded; whether field notes were made; the duration of interviews or groups; whether data saturation was discussed; and whether transcripts were returned to participants for comment.

Two of these carry most of the weight in review. The theoretical framework matters because it determines what counts as a finding, and a study that names no orientation leaves a reader unable to judge whether the analysis was applied consistently. Saturation matters because it is the commonest justification offered for a sample size and the least often described. If you claim saturation, say at what point new codes stopped appearing and how you verified it. If your sample size was determined by pragmatics or by an information power argument, that is defensible too, and saying so is better than an unsupported saturation claim.

Domain three: analysis and findings

Nine items, covering how you got from transcripts to themes and how the reader can check it. The domain asks how many coders coded the data, whether a coding tree or framework was provided, whether themes were derived from the data in advance or identified during analysis, what software was used to manage the data, whether participant checking of the findings took place, whether quotations are presented to illustrate themes and whether each is identified to a participant, whether the data presented and the findings are consistent, whether major themes are clearly presented, and whether there is a description of diverse cases or minor themes.

The items on coders and coding trees are where qualitative and quantitative reporting expectations meet, and where a number is genuinely useful. If two people coded independently, a reader will want to know how agreement was handled. That can be a reconciliation process described in prose, or a statistic: Cohen's kappa for two coders on categorical codes, Fleiss' kappa for more than two, or Krippendorff's alpha where you have multiple coders, missing data, or ordinal categories. Our Krippendorff's alpha guide explains which coefficient fits which coding setup, and the inter-rater agreement calculator computes all of them.

A note on reporting agreement honestly: an agreement statistic describes how consistently two people applied a codebook. It says nothing about whether the codebook captured anything meaningful. Presenting a high kappa as evidence that the themes are valid is a category error, and an experienced qualitative reviewer will say so.

What COREQ does not do

COREQ is a reporting checklist. It asks whether your study is adequately described, not whether it is trustworthy. Those are separate assessments, and conflating them causes real problems in two directions.

Authors sometimes present a completed COREQ checklist as evidence of quality. It is not; a poorly conceived study can be described in full compliance with all 32 items. Reviewers sometimes score COREQ items and report a total as a quality score. That is worse, because the items are not weighted and were never designed to sum. Trustworthiness in qualitative work is argued through credibility, transferability, dependability and confirmability, or through a comparable framework, and it draws on the study's design and conduct rather than on the completeness of its write-up.

When you are appraising qualitative studies for inclusion in a synthesis, use an appraisal instrument built for the job, such as the JBI critical appraisal tools, and keep the reporting judgement separate from the conduct judgement. The same separation applies across designs: STROBE for observational studies and CONSORT for trials are likewise reporting guidelines rather than quality scores.

Completing COREQ for submission

Work through the 32 items against your final manuscript and record the page and section where each is addressed, rather than ticking them. Where an item does not apply, for example repeat interviews in a single-interview design, write not applicable with a brief reason. Submit the completed checklist as a supplementary file and name the guideline in your methods.

Used this way the checklist earns its keep before anyone else sees it. The items that cannot be pointed at a page are, in our experience, almost always the same three: the interviewer's relationship with participants, the basis for the sample size, and the number of coders and how disagreements were resolved. Those are also the first three questions a qualitative reviewer asks.