Choosing an investigation and interpreting its result are different clinical tasks. iRefer and the ACR Appropriateness Criteria address the first: whether an imaging investigation is appropriate for a specified clinical situation. Their usefulness depends on matching the situation correctly, understanding the recommendation and checking the pathway available to the referring clinician.
As checked on 26 September 2026, the Royal College of Radiologists describes iRefer as imaging referral guidance, while the American College of Radiology's Appropriateness Criteria organise recommendations around defined clinical scenarios. Neither should be confused with a radiology image-recognition course or a universal instruction to order the highest-rated test.
This comparison is published by iatroX and includes iatroX among the complementary reference resources considered. Ask-iatroX can support a source-linked clinical question, but it should not be presented as identical to either specialist imaging-appropriateness resource.
Start with the decision, not the modality
A referral beginning with "MRI please" can conceal several unresolved questions. What problem is being investigated? Which findings make it plausible? What previous investigations have been performed? What would change if the proposed scan were normal, abnormal or inconclusive?
A better starting point is a short clinical question. For a fictional patient with persistent knee symptoms, the question might distinguish an initial assessment from a problem that persists after previous examination and imaging. These situations should not be assumed to share the same recommendation merely because both contain the word "knee".
This is a referral-preparation example, not a recommendation for a particular investigation. Its purpose is to show why an indication is more informative than a body part and a requested modality.
Before searching, identify the care setting, relevant time course, important findings, previous tests and the uncertainty the investigation is meant to resolve. That information helps select the correct scenario and makes any subsequent discussion with radiology more productive.
What iRefer and ACR each contribute
The RCR's iRefer service information, checked on 26 September 2026, describes a UK-facing resource intended to support appropriate imaging referrals. It can therefore be a useful starting point for clinicians working within UK referral arrangements, alongside the applicable local pathway.
The ACR's public description, reviewed on the same date, presents evidence-based guidance developed through expert review for defined clinical conditions and variants. The variant matters: an investigation can have a different place at an initial assessment than after a previous test or in a different patient subgroup. ACR's overview explains the resource and its methodology.
These are different editorial and implementation contexts, not proof that one resource is invariably more accurate. A UK clinician may find ACR's scenario structure educationally useful while still needing to follow local referral arrangements. An international learner may use both to understand why the same symptom does not always imply the same investigation.
How to interpret ACR's ratings
The ACR's rating-process document, revised in April 2021 and checked on 26 September 2026, groups its ordinal ratings into "Usually not appropriate" for 1 to 3, "May be appropriate" for 4 to 6, and "Usually appropriate" for 7 to 9. These are appropriateness categories, not percentages or predicted diagnostic yields.
The middle category can reflect uncertainty, disagreement or circumstances requiring more careful interpretation. It should not be translated automatically into "order it only after everything else". Similarly, a usually appropriate investigation is not necessarily the only reasonable option, and a rating does not replace examination of the relevant variant.
When reading a table, inspect the clinical scenario first, then the proposed investigation, supporting discussion and relevant limitations. Do not collect the highest numbers across unrelated variants and create a personalised imaging menu. That would remove the context on which the recommendations depend.
The useful question is: does the category apply to this defined situation, and what still needs to be decided locally?
A referral-preparation checklist
The following checklist is an original working aid, not an official RCR or ACR referral form. It is designed to make the clinical question inspectable before a request is submitted.
| Referral element | Information to establish | Why it matters to the discussion |
|---|---|---|
| Clinical question | The uncertainty the investigation should resolve | Avoids a request based only on a suspected label |
| Relevant history | Time course and features that change the scenario | Helps distinguish otherwise similar indications |
| Examination | Findings relevant to the question | Shows the basis for the proposed investigation |
| Previous tests | What was done, when and with what result | Prevents treating a follow-up problem as an initial presentation |
| Patient considerations | Relevant circumstances affecting suitability or preparation | Allows appropriate discussion of the individual investigation |
| Intended consequence | What the result would change | Connects the test to a clinical decision |
| Local route | Referral eligibility, urgency and available pathway | Separates published appropriateness from practical access |
A useful request does not need to reproduce an entire consultation. It needs enough information to justify the selected scenario and make the intended decision clear. Irrelevant detail can obscure the important question just as easily as an incomplete history can leave it unexplained.
Where a significant patient consideration is uncertain, do not assume it is normal or absent merely to complete the referral. Clarify it through the appropriate clinical process.
A fictional referral before and after preparation
Consider a request reading: "Persistent pain. Please scan." The receiving clinician cannot readily identify the anatomical concern, the duration, what has already been assessed or what an investigation is intended to change.
A more useful structure would be: "Persistent symptoms in the specified region despite the documented assessment; relevant examination findings are recorded below; previous imaging and its date are attached; the question is whether there is evidence of the suspected structural explanation that would change the next management discussion."
The second version is intentionally not a completed clinical referral or a recommendation for a scan. It demonstrates how the request can expose its reasoning. The referrer would still need to select the actual clinical variant, check the appropriate resource and follow local arrangements.
If the resource suggests that another investigation or a non-imaging pathway may be more appropriate, the response should be to revisit the decision rather than rewrite the request to obtain a predetermined test. Appropriateness guidance is useful precisely because it can challenge the original assumption.
Access is a separate comparison
As checked on 26 September 2026, the RCR describes several iRefer access routes, including member access and arrangements for specified national health services and organisations. The public information names NHS Scotland, NHS Wales, Health and Social Care Northern Ireland and Ireland's HSE among relevant access arrangements. It also distinguishes limited access from the full resource. That does not establish that every NHS employee has identical entitlement. Check the current iRefer access instructions through your own organisation.
ACR provides public routes to its criteria, while particular search tools, integrations and organisational implementations can have separate access conditions. As checked on 26 September 2026, its clinical-resource page distinguishes the criteria from associated delivery tools.
For a department, public readability is not the same as permission to integrate content into a referral system. For an individual, an attractive demonstration is not proof that the same function is included in their account. Confirm the actual route rather than relying on a colleague's different entitlement.
Three reader scenarios
For a UK GP preparing a referral, begin with the applicable local pathway and use the relevant iRefer guidance where available. ACR can provide an additional way to examine the scenario, but should not be used to assume access to an investigation outside the local referral arrangement.
For an emergency clinician, a decision-support resource can help clarify a defined imaging question, but its use must fit the clinical situation and the need for timely senior or radiology discussion. A rating should not become a reason to ignore information that makes the selected scenario inapplicable.
For an international learner, comparing the resources can be an exercise in clinical reasoning. Identify what remains constant, what depends on the guideline context and which additional patient detail changes the recommendation. The aim is not to memorise a worldwide hierarchy of scans.
Where iatroX belongs
Per iatroX product information, September 2026, Ask-iatroX provides free source-linked reference grounded in NICE, CKS, SIGN and SmPC information from emc, without a trial expiry or verification gate. It can support the broader question around an investigation, such as clarifying the clinical context or finding relevant UK guidance.
Its published methodology is a design for retrieving and checking referenced answers, not proof that it reproduces iRefer's recommendations or ACR's rating system. For a specialist imaging-appropriateness decision, retain the relevant specialist source and local clinical discussion.
The scenario-based verdict is therefore straightforward: use a resource that matches the jurisdiction and task, and preserve the reasoning that connects the patient to the recommendation. A better scan request begins with a better question, not a more confident modality preference.
Frequently asked questions
Is iRefer mainly a resource for interpreting images?
No: the RCR's description, checked on 26 September 2026, positions it around appropriate imaging referral. Learning to interpret a completed scan is a separate educational task.
Does an ACR rating of 9 mean a scan is 90% accurate?
No: the ACR rating scale expresses appropriateness for a specified clinical scenario. It is not a percentage measure of diagnostic accuracy or a prediction that the investigation will be positive.
Can a UK clinician use ACR Appropriateness Criteria?
The criteria can inform learning and discussion, but their application still requires the correct clinical scenario and attention to local guidance and referral arrangements. Published appropriateness does not establish local access or authorisation to request a test.
Explore the clinical question behind an investigation with Ask-iatroX →