To publish a clinical audit, report it as improvement work rather than as research: complete at least one full audit cycle so you have before-and-after data, structure the write-up to the SQUIRE 2.0 checklist rather than STROBE, and state that the project was registered with your clinical governance department rather than claiming ethics committee approval.
An audit that stops at the first data collection has measured a gap and nothing more. A write-up built on research headings misses what improvement work has to show: the standard, the change, and the reasoning between them.
Audit, quality improvement and research answer three different questions
A clinical audit is a measurement of actual care against an existing, agreed standard, followed by a change in practice and a re-measurement. Quality improvement is the iterative testing of changes to a care process, in short cycles, to shift a measured outcome. Research is designed to produce new, generalisable knowledge.
| Feature | Clinical audit | Quality improvement | Research |
|---|---|---|---|
| Question asked | Does our care meet an agreed standard? | Can we make this process better? | What was not previously known? |
| Comparator | An agreed standard | Your own baseline | A control group or hypothesis |
| Reporting checklist | None dedicated; SQUIRE 2.0 once you act | SQUIRE 2.0 | CONSORT, STROBE, PRISMA |
| Approval route | Governance registration, local rules vary | Governance registration, local rules vary | Ethics committee or IRB review first |
The audit cycle, not the first data collection, is what gets published
The audit cycle is a loop, most often drawn in four stages: agree a standard, measure practice against it, change something, then measure again against the same standard. A first round tells your department it has a gap; a completed cycle tells other departments what happened when somebody closed one.
So keep the denominator identical between rounds, allow the change time to take effect, and add one balancing measure to show whether improving your target made something else worse.
Before you publish a clinical audit, check it against these six tests
- The standard is external and named. Cite the guideline, its issuing body, version and date.
- At least one cycle is complete. Post-change data exist, collected as the baseline was.
- The change is described well enough to copy. Who did what, to whom, how often, from when.
- The denominator is defined and stable. Same criteria, same source, same window in both rounds.
- Something transfers. Another unit can read your context and judge whether your result would hold.
- The project is registered. You hold a governance registration number or a written exemption.
Fail test 2 and you have an incomplete audit cycle, reportable as a baseline but not publishable as an audit. Fail 1, 3 or 4 and the gap has to be closed with new data: an unnamed standard cannot be checked, an undescribed change cannot be copied, and a shifting denominator makes the two rounds incomparable.
Why SQUIRE 2.0 rather than STROBE applies to improvement work
SQUIRE 2.0 is an 18-item reporting guideline for work whose purpose is to improve the quality, safety or value of healthcare. Its name expands to Standards for QUality Improvement Reporting Excellence, and the full item list is published as the SQUIRE 2.0 statement; the EQUATOR Network lists it alongside the study-design checklists.
STROBE reports observational studies — cohort, case-control and cross-sectional — so its items are eligibility criteria, bias, confounding and study size. None of those items fits a comparison against an agreed standard. STROBE asks for the setting and dates only as background, and has no item at all for the intervention or for the reasoning that connects a change to the outcome it is meant to shift — the two things that make improvement work interpretable. SQUIRE 2.0 asks for both, and asks you to keep assessing context while the work runs.
If your project allocated patients to different care, you are running a study; our guide to reporting guidelines including CONSORT, PRISMA and STROBE covers that choice.
A structure map for an audit write-up, mapped to SQUIRE 2.0
Write to this map before you write prose: one row per SQUIRE 2.0 item.
| SQUIRE 2.0 item | What belongs here | Omission to check for |
|---|---|---|
| 1. Title | The change, the setting, the outcome | No improvement-work signal |
| 2. Abstract | Problem, context, intervention, measures, results | Only one round of numbers |
| 3. Problem description | The gap as it appeared locally | National statistic, no local evidence |
| 4. Available knowledge | What is known about the problem and about fixing it | The disease reviewed, not the fix |
| 5. Rationale | Why you expected this change to work | Absent, so the change looks unexplained |
| 6. Specific aims | The target, as a number, by a date | Compliance improved, with no threshold |
| 7. Context | Setting, staffing, case mix, systems that shaped the result | One line, transferability unjudgeable |
| 8. Intervention | The change in replicable detail, and who delivered it | Sessions held, no content or dose |
| 9. Study of the intervention | How you assessed whether the change caused the effect | What else changed goes unsaid |
| 10. Measures | Outcome, process and balancing measures | No balancing measure, so harms hide |
| 11. Analysis | Methods, including how variation over time was handled | A p-value on tiny numbers |
| 12. Ethical considerations | Registration or exemption, who granted it, data handling, and any conflicts of interest | Approval claimed but never sought |
| 13. Results | Round-by-round numbers with denominators | Percentages without the counts |
| 14. Summary | Key findings and their strengths | Results restated, not synthesised |
| 15. Interpretation | Comparison with other reports, and why | Association presented as proof |
| 16. Limitations | Threats to internal validity and generalisability | Small sample size offered alone |
| 17. Conclusions | What is usable elsewhere, and what next | A recommendation wider than the data |
| 18. Funding | Funding for the work, and its role | Dropped when there was none |
How to write the rationale item
Item 5 separates a report of improvement work from a table of before-and-after numbers. State, before the results, why your change ought to have shifted the measure:
Baseline case-note review suggested documentation was omitted at handover rather than at assessment, so the intervention targeted handover with a structured proforma, on the reasoning that a prompt at the point of recording is more reliable than teaching delivered weeks earlier.
If you cannot write that sentence, the audit may have changed the wrong thing.
Audit usually needs governance registration, not ethics committee approval
Clinical audit is generally treated as service review rather than research: it uses routinely generated data, judges care against a standard already agreed for it, and allocates nobody to anything. The step usually required is registration with the clinical audit or governance department. Requirements vary by institution, so ask yours.
Four features take a project beyond routine audit, and any one of them usually means a research ethics committee, or your information governance office, should see it first:
- Patients are allocated, randomised, or given something other than usual care.
- Data are collected from patients solely for the project.
- Identifiable data leave the institution or the care team.
- The design aims at generalisable knowledge, not local practice.
State the position in one sentence, naming the body that gave it:
The project was registered as a clinical audit with the Clinical Governance Department of [institution], reference [number]. Ethics committee review was not required, as the work used routinely collected data to assess care against an existing standard.
If a committee ruled the work service evaluation, say so instead, with its name and date. Never write that approval was obtained when only registration was. Note that our Author Guidelines ask you to state institutional review board or ethics committee approval with the committee name and reference number, and say nothing about audit registration. Where the project never went to a committee, give the governance registration in that field and add one line saying why committee review was not sought. Our post on ethics approval and informed consent sets out what to report when the work is research.
Name the article type in your cover letter and frame the work as an audit
Directive Publications lists its types on the Article Types page. That page lists nine types, and none of them is clinical audit or quality improvement. Name the closest type in your cover letter, and say there that the work is a completed audit cycle reported to SQUIRE 2.0. Our Author Guidelines ask you to match the manuscript to the relevant EQUATOR checklist, and name eight: CONSORT, PRISMA, STROBE, ARRIVE, CARE, STARD, SPIRIT and CHEERS. SQUIRE 2.0 is not on that list, because the list covers study designs and improvement work is not one. Report to SQUIRE 2.0 anyway, and say in your cover letter that you have done so.
Before submitting, read our guidance on the types of research articles, writing a cover letter to the editor and writing a limitations section, where an audit has to address the threats a before-and-after design cannot rule out. Our Publication Ethics and Editorial Policies pages cover approvals and disclosure, and Submit a Manuscript explains how to reach the editorial office.