Reporting guidelines and the EQUATOR Network

Why reporting guidelines exist

A study that is not reported clearly cannot be assessed, repeated or used. Reviews of published research have repeatedly found that essential information is missing: how participants were allocated, how outcomes were measured, how many were lost to follow-up, which analyses were planned. Without it, readers cannot tell whether a result is trustworthy, systematic reviewers cannot extract data or judge bias, and the effort invested in the research is partly wasted. Reporting guidelines respond by listing the minimum information that should be included in a report of a given type of study, as agreed through a consensus process involving methodologists, researchers and editors.

The guidelines are not a measure of quality and do not tell researchers how to conduct a study. They tell them what to say about what they did. A well-reported poor study is easier to criticize, which is as it should be, and a well-conducted study that is badly reported may be dismissed. Evidence from comparisons of journals that endorse the guidelines and those that do not suggests that reporting is better where they are applied, though it remains far from perfect. The authoritative collection is maintained by the EQUATOR Network, whose name stands for Enhancing the QUAlity and Transparency Of health Research.

A map of the main guidelines

This is a selection, not a complete list, and guidelines are revised. Authors should always check the EQUATOR Network library and the journal's instructions for the current version and for guidelines for designs not shown here.

How to find the right guideline

1. Name the design

   Decide exactly what kind of study it is. A randomized trial, a cohort study and a cross-sectional survey have different guidelines, and the label used in the paper should match.

2. Search the EQUATOR library

   The EQUATOR Network library can be searched by study type, subject area and other filters, and it links to the official checklist and the explanation and elaboration paper.

3. Check the journal

   The instructions for authors name the guidelines the journal requires and may supply its own checklist form.

4. Use the current version

   Guidelines are updated, and an outdated checklist can be rejected or incomplete. Look for the most recent official version.

5. Read the explanation, not only the checklist

   The companion paper explains the reasoning for each item and gives examples of good reporting.

Using a guideline well

The most useful time to open the guideline is before the study is written up, and for some designs before it begins. Trial protocols follow SPIRIT, and knowing what CONSORT will require of the report helps researchers record the necessary information while the trial runs, such as the flow of participants and the reasons for withdrawal. When writing the manuscript, the checklist works as an outline. At the end, it is a self-audit: for each item, the author finds the place in the manuscript where it is addressed and records the page or section. If it is not there, it is added. An item that does not apply, such as blinding in a study where it was impossible, is marked as not applicable with the reason, or addressed by an explanation in the text.

A checklist completed after the fact, with page numbers that do not lead to the information, defeats the purpose, and editors notice. Equally, following the checklist does not require every item to be satisfied in the study itself. It requires that the report says what was done, including when something was not done. Honesty about what was not done is part of good reporting.

Example: what CONSORT asks of a trial report

CONSORT, the guideline for randomized trials, shows how the checklists work. Its items cover the title, which should identify the study as randomized, and a structured abstract. The introduction explains the background and the objectives or hypotheses. The methods describe the trial design, any changes after the trial began, the participants and setting, the interventions in enough detail to replicate, the outcomes defined with their time points, the sample size and how it was determined, the method of random allocation including the sequence generation, allocation concealment and implementation, the blinding, and the statistical methods. The results give the flow of participants through each stage, ideally in a diagram, the dates of recruitment and follow-up, the baseline data, the numbers analyzed, the results for each outcome with the estimated effect size and its precision, and harms. The discussion covers limitations, generalizability and interpretation. Finally, the report gives the trial registration number, the protocol and the funding.

Each item is something a reader needs in order to judge the trial and use it in a review. Studies of reporting have found that, for many of these items, a large share of published trials fail to give the information, and that trials published in journals that endorse CONSORT report somewhat better. The guideline for the other designs works in the same way.

Example: the participant flow diagram

One of the most useful things a reporting guideline asks for is easy to describe: a flow diagram. For a trial, CONSORT asks for a diagram showing how many people were assessed for eligibility, how many were excluded and why, how many were randomized to each group, how many received the allocated intervention, how many were lost to follow-up or discontinued and why, and how many were included in each analysis. The numbers should add up from one box to the next. A reader can see at a glance whether the groups were comparable at the end, whether attrition was balanced, and how many people the result rests on, which is much of what is needed to judge bias. The same device appears in other guidelines: PRISMA has the flow of records through a review, and STARD has the flow of participants through a diagnostic study. A missing or inconsistent diagram is a warning sign that reviewers learn to notice.

Beyond the checklist: registration, data and protocols

Reporting is one part of research transparency. Several other practices are tied to the guidelines and to what journals ask. Prospective registration of trials and reviews makes the plan public, and the guidelines ask authors to report the registration number. Publication of the protocol, or access to it, lets readers compare the plan with the report. Sharing data and code, increasingly expected, allows others to check and reuse the results. Declarations of funding, the role of the funder and competing interests complete the picture. A study that meets the reporting guideline and has done these other things is much easier to trust. Researchers who plan for all of them at the start find the burden at the end smaller.

When to use which guideline in a project's life

Planning

Use the protocol guideline for the design, such as SPIRIT for a trial or PRISMA-P for a review, and note the reporting requirements of the final paper so that the necessary information is recorded.

Conduct

Keep records that the report will need: participant flow, reasons for exclusion, protocol deviations, amendments and the dates of key decisions.

Writing

Use the checklist as the outline of the manuscript and the explanation paper for guidance on each item.

Submission

Complete the checklist with the location of each item, attach it if the journal asks, and state in the methods which guideline was followed.

Revision

Check that changes made in response to reviewers do not remove required items, and update the checklist.

Extensions and variants

Many guidelines have extensions for particular forms of a design or for parts of a report. CONSORT has extensions for cluster-randomized trials, for non-inferiority trials, for the reporting of harms, for abstracts, for patient-reported outcomes and for interventions that are not drugs, among others. STROBE has extensions for specific fields such as genetic association, molecular epidemiology and infectious diseases. PRISMA has the extensions described in the guide on PRISMA extensions. TRIPOD has an update for studies of models that use machine learning methods. When an extension applies, it is used together with the base guideline, and the manuscript says so. The EQUATOR library lists them under the main guideline.

Journals and endorsement

Many journals endorse reporting guidelines, and some require a completed checklist at submission, as an upload or in the submission form. Requirements vary: some journals specify the guidelines they expect for each article type, some ask authors to state which they followed, and some check compliance during editorial review. A paper prepared without regard to the guideline for its design is likely to be returned with a request for the checklist, which delays publication. Authors should find out the requirement of the journal at the start and plan the manuscript accordingly. Choosing a journal is discussed in the guide on choosing a journal.

What reporting guidelines cannot do

A guideline improves the completeness of a report, not the quality of the study. It does not tell researchers what the correct design or analysis is, and a fully compliant report can describe a flawed study. It also does not protect against deliberate misrepresentation. Some guidelines have been criticized for length and for treating every item as equally important, and checklists can encourage box-ticking. The fair view is that they set a floor, not a ceiling: they are a minimum standard of transparency. Methodological quality is assessed separately, using tools such as those described in risk-of-bias tools, and the two assessments complement each other. Some guidelines are also used by reviewers as markers of reporting quality when they appraise primary studies.

Common mistakes

• Choosing a guideline that does not match the design, for example using STROBE for a trial.
• Using an outdated version.
• Completing the checklist after writing, without changing the manuscript.
• Marking items as met when the page reference does not lead to the information.
• Treating the guideline as a quality score or a methodological standard.
• Ignoring extensions that apply to the study.
• Not saying in the manuscript which guideline was followed.

Support

Frequently asked questions