12 Audit and quality improvement

Choose a problem worth understanding

Improvement work is most useful when it addresses a real problem in care or service delivery. A project selected only because an application needs an audit can become an exercise in producing a report rather than understanding and changing practice.

Start with a concern that matters to the service and can be examined within a feasible scope. Ask who experiences the problem, what is already known and whether someone else is addressing it. A modest project with clear ownership and useful learning can be more convincing than an ambitious project that never becomes operational.

The doctor's contribution should be explicit from the beginning. You may define the question, collect or analyse information, coordinate colleagues, test a change or help sustain it. You do not need to claim sole leadership of every stage to demonstrate meaningful work.

T08 provides a record of the aim, method, measures, findings, limitations and next action. The method should fit the question, not be chosen simply because a familiar template is available.

Audit improvement evaluation and research

Clinical audit examines practice against an identified standard. Service evaluation asks about the service as it is delivered. Research seeks to generate knowledge through an appropriate research design and governance process. Quality improvement involves purposeful work to improve a process or outcome, often through iterative testing. These activities can overlap in practice, but the terms are not interchangeable labels. [S20, S21, S23]

If classification is uncertain, use the relevant decision tools and local audit, improvement or research advice before proceeding. Do not choose a label to avoid oversight. Data access, consent, information handling and approvals depend on the actual activity and setting.

The Institute for Healthcare Improvement's Model for Improvement connects an aim, measures and changes, using iterative tests to learn and adapt. That is one useful methodological resource; it does not guarantee that a proposed local change will work. [S22]

For career evidence, the important point is to show that you understood the question, selected an appropriate method and interpreted the findings honestly. A project called QI but lacking a clear measure is not strengthened by its label.

Define the question and measure

A broad concern such as poor discharge communication needs a more precise question. Which part of communication is unclear? Who uses it? What would a satisfactory record contain? How will the team decide whether the problem has improved?

Define the measure before collecting data. If different people interpret success differently, comparisons may be misleading. A measure should be understandable enough that another person can apply it consistently.

Also consider what the measure does not show. Documentation of a follow-up action is not the same as completion of that action. A shorter process is not necessarily safer or better for patients. A measure can still be useful if its limits are explicit.

Choose a feasible scope. A small project may be a sensible starting point, but the sample and time period should be described. Avoid presenting convenience as representativeness without examination.

QI01 a hypothetical discharge communication project

The following teaching example uses invented events and figures to explain improvement evidence.

A Specialty Doctor and ward team notice that some discharge summaries do not clearly identify both the follow-up action and the responsible recipient. They agree with the local audit lead to review a defined sample against a locally approved documentation standard. The doctor helps define the measure and discusses the project with nursing, therapy, pharmacy and administrative colleagues.

At baseline, 12 of 30 summaries meet the agreed definition. The team tests a revised prompt in the template. In the next sample, 22 of 30 meet the definition. A later sample shows 24 of 30. Staff feedback identifies duplication and extra work, so the team removes a redundant field and continues monitoring.

The project demonstrates a change in a documentation process measure within these samples. It does not establish fewer readmissions, improved survival or a general reduction in harm. There is no control group, the samples are small and staffing or other circumstances may have changed.

What the doctor actually contributed

The evidence account should identify the doctor's role rather than simply attach the team report. In QI01, the doctor helped define the measure, discussed the proposed change with colleagues and contributed to the response to feedback. Another person may have built the template, collected some data or managed the reporting system.

That shared work does not diminish the doctor's contribution. It makes the account accurate. A statement that distinguishes personal action from team achievement is more credible than a claim to have transformed the whole service alone.

Suitable corroboration might include the agreed project summary, aggregate results, version history and a relevant colleague's account of contribution. Identifiable discharge documents do not need to be copied into a personal portfolio to demonstrate the work.

Involve the people who use the process

A change can be technically sensible and operationally awkward. The people who use the process often recognise problems that the project lead misses. In QI01, feedback about duplication changes the design. That response is part of the evidence of learning.

Involvement should occur early enough to influence the work. Asking colleagues to endorse a completed proposal is different from asking them to help define the problem and test the change. Where patients or carers can appropriately contribute, clarify how their experience will inform the project.

Do not claim involvement merely because people were informed. Record what contribution changed the proposal or interpretation. If a group could not be involved, state that limitation and consider whether it matters to the next stage.

The practical skill is to connect different perspectives to a workable decision. This is also leadership evidence, but it should be described through the actual actions rather than a general claim of excellent teamwork.

Interpret change cautiously

An improvement between samples may be encouraging, but it needs explanation. Were the same definitions used? Were the samples comparable? Did the people collecting data change? Was another service change occurring? These questions do not invalidate the work; they make the conclusion more reliable.

In QI01, the increase from 40 to 80 per cent is a 40 percentage-point difference in the measured proportion. It should not be described casually as a 40 per cent improvement without clarifying the calculation. More importantly, the numbers do not establish the mechanism or durability of the change on their own.

Use plain language that fits the evidence: the later sample showed a higher proportion meeting the documentation definition. Explain the limitations and the need for continued review. You do not need to add an unsupported causal claim to make the work valuable.

If statistical analysis is appropriate, seek relevant support rather than add a test because it makes the report look scientific. The method should answer the project's question and fit the data.

Consider unintended consequences

A change can improve one measure while creating another problem. The duplicated field in QI01 is a simple example. Additional documentation may consume time or lead users to copy information without thinking. A faster process may shift work to another team.

Ask what burden or risk the change could introduce and how the team will notice it. Feedback can be a useful part of that review, alongside relevant measures. Do not treat resistance as evidence that colleagues dislike improvement; it may identify a real design problem.

Document the response. Removing the redundant field shows that the team used feedback rather than defended the original idea. That adaptation may be more educationally valuable than presenting the first version as perfect.

When the result is disappointing

A project can be worthwhile even when the intended measure does not improve. The account should then explain what was tested, what happened and what was learned. Do not change the aim retrospectively to make the result look successful.

Imagine a separate hypothetical test in which a reminder is rarely used and the measure remains unchanged. The team discovers that the reminder appears at the wrong point in the workflow. A reasonable next action may be to redesign it or stop the intervention. The evidence can show careful inquiry and responsible adaptation.

An incomplete project also needs an honest account. State which stages were completed, why the work stopped and what remains unresolved. Do not describe an intention to re-audit as though the re-audit has occurred.

The value for progression lies in method, contribution, interpretation and learning. A series of exaggerated success stories is weaker evidence than a fair account of useful work with limits.

Sustainability and ownership

Before concluding a project, ask who will maintain the change and review it. A process that depends entirely on one doctor's enthusiasm may disappear when that doctor rotates, changes duties or takes leave.

Identify the service owner, where the revised process is recorded and what would trigger another review. If continuing measurement is not feasible, explain how the team will know whether the change remains in use. Avoid claiming sustainability merely because the intervention was implemented once.

The handover itself can be evidence of service contribution. It shows that the doctor considered what happens after the portfolio entry is written. It should remain proportionate to the scale of the project.

Turn the project question into a collection rule

Before collecting data, write a short definition that another reviewer can apply. For QI01, a summary meets the hypothetical measure only when both the follow-up action and the responsible recipient are identifiable. A document with an action but no recipient does not meet that definition. The numerator is the number meeting both elements; the denominator is the number reviewed in that sample.

The project team must also define which summaries are eligible, how the sample is selected and what happens when information is missing or ambiguous. Those details are not supplied by QI01's illustrative results. In a real project, agree them before the first collection and retain the definitions so the later comparison uses the same rules.

Trial the collection rule on a small permitted set with a colleague. If you disagree about how to classify a record, refine the definition before collecting the main sample. Record any change that affects comparability. This is a suggested practical quality check, not an additional mandatory audit stage.

Keep the portfolio record focused on the method and aggregate findings. The underlying patient information should remain within the authorised project arrangements. You can demonstrate a careful collection method without exporting the reviewed documents into a personal folder.

Plan a small test before wider implementation

A proposed change should include a prediction: what do you think will happen, and why? For the hypothetical discharge-summary prompt, the prediction is that a clearer place to record the action and recipient will make those elements easier to complete. The test should examine that prediction and whether the change creates extra work or confusion.

Describe what was actually tested, who used it and what feedback was obtained. Review the findings with the people affected, then decide whether to retain, adapt or abandon the change. If it is adapted, record why. The PDSA method described by IHI provides a useful structure for this cycle of learning. [S22]

In QI01, staff report duplication and the team removes a redundant field. That is a design decision worth explaining. The portfolio should not simply list the three percentages and omit the problem that changed the intervention. The response to difficulty helps demonstrate that the doctor understood improvement as a learning process.

Choose measures that answer different questions

A process measure asks whether the intended step occurred. An outcome measure asks what happened to the result the service cares about. A balancing measure examines an unwanted consequence elsewhere. These categories help a team avoid assuming that one convenient figure answers every question.

QI01's documented action and recipient are a process measure. Whether follow-up actually happened would require a separate definition and data source. Time spent completing the summary could be considered as a balancing measure, but the hypothetical project has only feedback about duplication; it does not contain a measured time result. Preserve that distinction when writing the report.

For a small real project, select measures that are feasible and meaningful. Explain the trade-off if a desired outcome cannot be measured. A precise process conclusion with an honest limitation is more useful than an unsupported claim about patient benefit.

Write the conclusion before choosing the most impressive headline

A defensible QI01 conclusion is: “The proportion of reviewed summaries containing both defined elements was higher in the later samples. The team revised the prompt after feedback about duplication. The small samples and absence of a comparison group limit interpretation, and sustained use requires further review.”

The portfolio contribution statement then adds what the doctor did: helped define the measure, involved colleagues and contributed to adapting the change. The method, result and contribution are related, but they answer different questions. Keep all three visible.

If the project stops before another cycle, state the stopping point and the reason. Retain the learning and any agreed handover. A future plan to repeat measurement belongs under next steps until it has actually occurred.

What the service needs to arrange

Provide appropriate advice on classification, governance, data and method. Make time and access feasible, involve relevant colleagues and identify ownership. A doctor should not be asked to produce improvement evidence without access to the information or authority needed to undertake the work responsibly.

Do not require every doctor to lead a large project or impose an arbitrary number of audits without a basis in the relevant requirements. Assess the contribution and learning in context. A focused project can be strong evidence when its method and limitations are clear.