From Delphi Consensus to Measurable Audit Standards
An agreed recommendation identifies what a panel considers important. An audit standard also needs a clear population, observable action, measurement period, and data source. Keep those decisions connected as the wording becomes more precise.
A recent protocol makes operational review an explicit stage
West and colleagues’ October 20, 2026 National Clinical Audit of Perioperative Care protocol uses patient and healthcare-professional panels. Its planned progression rule can be met by either panel in round 1, while final inclusion in round 2 requires both. A later meeting will compare revised audit wording with the original statements and vote on whether the intended meaning is preserved.
This is a protocol: it describes planned methods, with study results still pending. It offers a useful example of separating priority setting from the work of making a standard measurable.
The workflow and examples below are Calibrum’s original planning suggestions. Adapt them to the project’s approved methods and intended use.
Record agreement, priority, and measurability as separate decisions
Start by asking what the panel actually rated. Importance, appropriateness, feasibility, and priority are different judgments. An item can meet its importance threshold while still needing a definition that can be applied consistently to routine records.
Keep three linked outputs: the original item and its consensus result; any later selection or ranking decision; and the operational definition proposed for measurement. Identify the people and rules responsible for each output.
When criteria differ between rounds, label each rule by its purpose. A permissive rule for retaining discussion topics does not establish final acceptance. Show the result for every required stakeholder group before combining it into a final classification. See stakeholder-group analysis and separate rating dimensions.
Write down exactly what will be counted
Consider this original example: “Offer patients a preoperative discussion about their recovery goals.” The research team wants to examine delivery using clinical records. The table describes a proposed specification, not a finding from the cited protocol.
| Field | Proposed definition | Question requiring review |
|---|---|---|
| Eligible population | Adults undergoing elective major surgery in participating services. | Does this narrower population fit the panel’s intended scope? |
| Denominator | Eligible surgical episodes during the audit month. | How are cancellations, repeat procedures, and transfers handled? |
| Numerator | Eligible episodes with a documented offer of a recovery-goal discussion before surgery. | Can the record distinguish an offer, a completed discussion, and a declined offer? |
| Exceptions | Explicitly defined exclusions, each with a recorded reason. | Could exclusions systematically remove harder-to-reach patients? |
| Data source | A specified field or reviewed section in the clinical record. | Is the information captured consistently across services? |
| Version and approval | A07 version 1; linked to R07 and the recorded review decision. | Who can approve changes, and which changes require another rating? |
Distinguish the eligible population from exclusions and missing information. A missing record should have a defined status. Treating every blank as failure, success, or an exclusion can change the result substantially.
A documentation measure is evidence about the record. Establish separately how well it reflects the offer or conversation that occurred. Also decide whether a performance target is needed: agreeing that an action matters does not automatically establish a numerical compliance target.
Make changes visible before asking for approval
Present the original recommendation beside the proposed measure. Explain each addition: population, timing, setting, action, exception, and evidence of completion. A short change log helps reviewers focus on the effect of the revision.
Original review prompts
“Does this definition preserve the action and population intended by R07?”
“Would the proposed numerator count an episode in which the intended action had not happened?”
“Can each participating service identify the denominator and exceptions in the same way?”
Use the protocol to decide when clarification is sufficient and when a substantive change needs renewed assessment. A new requirement about timing or eligible patients can change the recommendation’s scope even if the sentence reads more smoothly.
Record who reviewed the definition, the materials they saw, the voting denominator, the decision rule, and the outcome. Attribute a steering-group approval to that group. Describe any additional panel vote with its own item version and result.
Test the measure and preserve the complete chain of decisions
Before a large audit, try the specification on a suitable sample of records. Check whether independent reviewers identify the same eligible episodes, exclusions, and numerator events. Investigate disagreements and missing data, then document any resulting revision.
Define what a pilot must establish, including data availability, consistent interpretation, extraction burden, and usefulness for the intended comparison. A consensus vote on wording does not itself provide those results. Where case-based interpretation needs examination, see testing agreed recommendations with vignettes.
For reporting, retain a chain from recommendation ID and round, through stakeholder results and selection decisions, to measure ID, specification version, approval, and pilot findings. Report dropped or unresolved measures and explain the reason for each disposition.
Plan these records alongside Surveylet’s item ratings, comments, and exports. Agree any custom mapping, meeting record, or audit specification as part of the project’s reporting scope. The implementation-planning guide covers the later work of assigning actions, owners, and evaluation measures.
Sources and scope
West MA, Parkington T, Moore J, et al. Development of consensus-based standards for a National Clinical Audit of Perioperative Care: protocol for a modified Delphi consensus study. BMJ Open. 2026;16:e123840. Published October 20, 2026.
For transparent reporting of the consensus process, consult the ACCORD reporting guideline. The recommendation, measure, review prompts, and workflow examples on this page are original illustrations; they require local methodological and operational review.
Plan the output beyond the final vote
Define the ratings, review stages, and reporting records your team will need to turn an agreed recommendation into a usable measurement proposal.