Classification comes first
Clinical audit, quality improvement, service evaluation and research can overlap in methods but have different governance routes. The website should never classify a project solely from a few questions. Obtain the formal local determination before accessing or using data.
Approval and governance
Usually needs formal review / authorisation
- Authorised governance/department approval for access to clinical operational data.
- Ethics/research review if the activity is classified as research or if publication/generalisation changes the regulatory route.
- Privacy review when identifiable patient/staff data are extracted or linked.
May follow a lighter or different route
- A local audit against an established standard or a service-improvement project may use an audit/QI governance pathway rather than full research ethics review, depending on policy.
- Publication of QI does not automatically make it research, but intended dissemination should be discussed with the authorised office.
Do not do this
- Do not label a research project “QI” simply to avoid ethics review.
- Do not introduce unapproved interventions that may create patient safety risk.
- Do not extract more patient identifiers than the improvement activity needs.
Mediclinic Middle East publicly states that research projects carried out at MCME are to receive approval from its internal Research and Ethics Committee and applicable local regulatory authorities before initiation. The exact route varies by project, facility and emirate. Dubai projects may involve DSREC depending on applicability. This hub must therefore route users to the Research Office and current local forms rather than declaring a project “ethics exempt.” Institution-specific forms and contacts will be inserted after verification.
Audit cycle: step by step
Choose a clinically meaningful topic
Prioritise patient safety, quality, variation, complaints, guideline adherence or a known service problem.
Choose an explicit standard
Use a guideline, policy, regulatory standard or agreed benchmark. Write the exact numerator/denominator or criterion before measuring performance.
Define population, period and data source
Specify which patients/episodes are audited, the sampling method and where each variable comes from.
Obtain governance/data authorisation
Confirm classification and permission before extracting records. Use the minimum identifiable data necessary.
Collect baseline data
Use a standard extraction form. Validate a sample of records and report missing/uncertain information.
Compare performance with the standard
Calculate compliance and identify where and why gaps occur. Avoid attributing causes without evidence.
Design the improvement intervention
Choose changes targeted to the identified barriers: workflow redesign, reminders, training, standardised order sets, feedback or another evidence-based approach. Define who owns each action.
Implement with process measures
Track whether the intervention actually occurred and monitor unintended consequences/balancing measures.
Re-measure using the same definitions
Repeat measurement after an appropriate interval so baseline and post-change results are comparable.
Decide the next cycle
If improvement is inadequate, refine the intervention and repeat. If successful, plan sustainability and periodic re-audit.
Report transparently
If publishing a QI study, use SQUIRE 2.0. Describe context, intervention, measures, study of the intervention, ethics/governance, results, interpretation and limitations.
Final checklist
- Formal project classification obtained.
- Standard/target defined before data collection.
- Population and denominator explicit.
- Data access authorised.
- Baseline measurement complete.
- Intervention linked to identified barrier.
- Balancing/unintended effects considered.
- Re-measurement uses same definitions.
- SQUIRE used if publishing QI.
Measurement plan
- Define one or more outcome measures that reflect patient/service results, process measures showing whether the intended change occurred, and balancing measures for possible unintended harm.
- Create a data dictionary so baseline and re-audit use identical definitions.
- Where data are collected repeatedly over time, consider run charts or statistical process control rather than comparing only one before and one after percentage.
- Record concurrent service changes that could explain improvement or deterioration independently of the QI intervention.
- If staff behaviour is measured, consider the Hawthorne effect and whether observation itself changes performance.
Publication caution
A successful local improvement does not automatically prove that the intervention will work elsewhere. If publishing, describe the local context, implementation details and other system changes so readers can judge transferability. SQUIRE reporting is designed to make these contextual features visible.