Best Practice in Data Collection Across Audits is not simply about gathering information efficiently. As Nancy Dixon, Director of Strategic Services at HQQ, explained during the Clinical Audit – Elevating Patient Care Standards Conference 2026 On Demand session, high-quality audit data depends on careful audit design, clear measures, reliable collection methods and meaningful review of what the data actually shows.
Drawing on decades of experience teaching and conducting clinical audits, Dixon’s session focused mainly on local clinical audits and the practical lessons learned from getting data collection right—sometimes, as she put it, “the hard way”. Her central message was clear: clinical teams are far more likely to trust and act on audit findings when the data are accurate, valid and directly connected to improving patient care.
Why Data Quality Matters in Clinical Audit
Dixon began by setting out the characteristics of good audit data. For data to be useful, it must accurately represent what happened to patients, identify the right cases, be complete, relevant, reliable, timely, and valid. In other words, audit data should provide a true picture of the quality of care being delivered.
This matters because clinical auditing is not just a counting exercise. If data are incomplete, unclear, or open to inconsistent interpretation, clinicians may not accept the findings. When that happens, the audit is less likely to lead to the practice changes needed to improve outcomes.
Best Practice in Data Collection Across Audits Starts with the Objective
Dixon distinguished between three common approaches: counting, comparing and changing. Counting how many patients received a particular aspect of care may be useful for service evaluation, but it is not necessarily a clinical audit unless there is a defined standard for comparison.
Comparing actual practice with best practice is closer to clinical audit. However, Dixon argued that the strongest audits go further by being designed with change in mind. If a team already suspects that care is not where it should be, the audit should help identify where improvement is needed and support evidence-based change.
Moving Beyond Counting Towards Change
Dixon distinguished between three common approaches: counting, comparing and changing. Counting how many patients received a particular aspect of care may be useful for service evaluation, but it is not necessarily a clinical audit unless there is a defined standard for comparison.
Comparing actual practice with best practice is closer to clinical audit. However, Dixon argued that the strongest audits go further by being designed with change in mind. If a team already suspects that care is not where it should be, the audit should help identify where improvement is needed and support evidence-based change.
Getting Case Selection Right
Clear case selection is another essential part of reliable data collection. Teams need to be explicit about which patients are included and which are excluded before data collection starts.
Dixon warned that when inclusion and exclusion criteria are vague, teams may later discover that some patients did not belong in the audit at all. These patients should not be treated as exceptions within the audit results; they should be excluded from the audit process entirely. This distinction helps protect the validity of the findings.
What Makes a Good Clinical Audit Measure?
Clinical audit measures may also be called criteria, indicators or key performance indicators. Dixon’s view was that the terminology matters less than the structure. Each measure should define what good practice looks like and how many patients should receive that care.
The Four Parts of a Strong Measure
According to Dixon, every clinical audit measure should answer four practical questions for the data collector:
- What evidence should be found to show that the patient received the right care?
- How many patients should that evidence apply to?
- Are there clinically justifiable exceptions?
- Where should the data collector look, and what definitions or terms do they need to apply?
Without these definitions, data collectors may rely on personal judgement. That creates inconsistency and can reduce clinicians’ confidence in the results. Clear definitions and instructions make it easier for different people to reach the same decision when reviewing the same case.
Using Protocols to Improve Consistency
Dixon recommended using a data collection protocol for each audit. This does not need to be a long document, but it should explain how data collection decisions will be made. The protocol should guide data collectors on where to find evidence, how to record compliance, how to identify exceptions, and how to handle missing information.
A case should be considered compliant when the evidence of quality is found or when a valid exception applies. A case should be considered non-compliant when the evidence is absent, no exception is found, or there is no available information to confirm what happened.
Testing Reliability Before Trusting the Results
For high-profile audits especially, Dixon highlighted the value of reliability testing. If there is more than one data collector, each person can independently review the same cases and compare decisions. If there is only one data collector, that person will need to review the same cases at different times and compare his/her own decisions.
The aim is to understand the extent of agreement in data collection. Dixon noted that the evidence base refers to at least 85% agreement for activities such as clinical audit, while her organisation prefers 95% or better. The reason is practical: higher reliability gives clinicians more confidence that the data accurately reflects care quality.
The Two-Stage Approach to Clinical Audit Data
One of the most important takeaways from the session was Dixon’s explanation of a two-stage approach. In many audits, the data collected against the measures are presented as the result, with no further review of patients whose care appeared not to meet the measure.
Dixon cautioned that this could underestimate the true quality of care. Care may have differed from the measure for valid reasons: a patient may have declined treatment, a rare clinical exception may apply, comorbidities may have required a different approach, or the evidence base may be unclear.
In the two-stage model, the first stage uses audit measures to screen records efficiently. The second stage involves structured clinician review of cases that did not appear to comply, looking for additional valid exceptions before moving on to analyse shortcomings and plan improvement. This balances standardised measurement with clinical judgement.
Why Trustworthy Data Supports Better Public Services
Although Dixon’s presentation focused on clinical audit, the broader lesson applies across public services: reliable data are essential for accountability, trust and improvement. Whether in healthcare, education or policing, organisations need evidence that stakeholders can understand and act upon. Similar themes around trust and improvement can be seen in discussions on maintaining public trust and preventing police misconduct, as well as work on supporting wellbeing across education settings.
Conclusion: Best Practice in Data Collection Across Audits Builds Confidence and Drives Change
Best Practice in Data Collection Across Audits is about much more than collecting complete datasets. As Nancy Dixon’s session showed, good audit data begins with a clear objective, explicit case selection, well-defined measures, a practical protocol, reliability testing and structured review of apparent non-compliance.
For clinical teams, the benefit is straightforward: reliable and valid data are more likely to be trusted, discussed and acted upon. For patients, that means audits are more likely to lead to meaningful improvements in care. To explore Dixon’s session in full, the Clinical Audit – Elevating Patient Care Standards Conference 2026 is now available to watch On Demand.