How the ACR Appropriateness Criteria Turn Evidence Into an Imaging Recommendation

The ACR Appropriateness Criteria convert medical evidence into an imaging recommendation through a structured two-part engine. First a systematic literature review assembles and grades the available studies for a defined clinical question; then a multispecialty expert panel votes each clinical scenario onto a scale of 1 to 9 using the RAND/UCLA Appropriateness Method, where 7 to 9 reads as "Usually Appropriate," 4 to 6 as "May Be Appropriate," and 1 to 3 as "Usually Not Appropriate." The single number you see on an ACR document is a graded, transparent summary of expert judgment tied to that evidence, not a mandate that a specific test must or must not be performed for an individual patient.

The American College of Radiology maintains this library at scale. As of its current release, the criteria span 277 diagnostic imaging and interventional radiology topics covering roughly 4,100 clinical scenarios and more than 1,350 clinical variants, reviewed annually by panels drawing on more than 700 volunteer physicians. Knowing how a single rating is assembled lets you read it as graded evidence plus disciplined consensus, rather than as a verdict handed down from on high.

The definition that anchors everything

Before any voting, the method fixes what "appropriate" means. Borrowing directly from the RAND/UCLA Appropriateness Method User's Manual, the ACR defines an appropriate imaging procedure as one for which the expected health benefit exceeds the expected negative consequences by a margin wide enough to justify doing it, considered separately from cost. That definition tells you what a high score is claiming and what it is not. A "Usually Appropriate" rating is a statement about the benefit-to-harm balance for a typical patient who fits the scenario. It is not a cost-effectiveness verdict, an insurance coverage decision, or a promise about any one person.

Step one: building a graded evidence base

Each topic begins with a systematic literature search, and the results are distilled into an evidence table that summarizes every included study and rates its quality by study type. The ACR aligns this appraisal with the GRADE framework (Grading of Recommendations Assessment, Development and Evaluation) and the Institute of Medicine's standards for trustworthy guidelines. GRADE separates two things people often blur: how certain we are about the evidence, and how strong the resulting recommendation is. High-certainty evidence can still support a cautious recommendation, and a strong recommendation can rest, honestly labeled, on lower-certainty evidence when the balance of benefit and harm is clear.

The appropriateness number sits on top of an evidence table you can actually open. When the underlying studies are thin or conflicting, the panel documents it, and that uncertainty is meant to travel with the rating rather than vanish behind it.

Step two: the RAND/UCLA vote

With the evidence in hand, the panel rates appropriateness through a modified Delphi process. Members score each procedure for each scenario privately, on the 1-to-9 scale, without a dominant voice steering the room. The anonymous ratings are tabulated and redistributed, and the panel votes again across successive rounds. The scenario's final category is set by the panel's median rating, so the published number reflects where the group's collective judgment settled after seeing the evidence and one another's scores.

Two design choices matter here. The private, iterative voting is built to blunt the influence of the most senior or loudest person in the room. And because the panels are multispecialty, including both the clinicians who order the test and the radiologists who perform it, a rating carries more than one professional vantage point.

Reading the middle of the scale honestly

The most misread part of the scale is the middle. "May Be Appropriate" does not always mean "roughly a coin flip on the merits." Under the RAND/UCLA logic the ACR uses, a scenario lands in that band for several distinct reasons: the benefits and harms genuinely are equivocal, the evidence is contradictory or sparse, special subpopulations complicate the balance, or the panel itself could not agree. When ratings scatter too far from the panel median, the ACR flags the scenario as "May Be Appropriate (Disagreement)" and assigns it a 5. That label is a feature. It signals that the number reflects unresolved expert conflict, not settled equipoise, and those are very different messages to act on.

So a 5 is not one thing. Reading it well means opening the narrative to see which of those reasons produced it.

Where radiation dose fits

Alongside each appropriateness rating, the criteria carry a Relative Radiation Level, a symbolic scale in which more symbols signal a higher range of expected effective dose. Dose is presented next to appropriateness rather than folded into it, so a reader can weigh a test's expected yield against its radiation burden as separate facts. The dose ranges are assigned by an ACR radiation-exposure subcommittee and have been validated against real-world exposure data in the ACR Dose Index Registry, which grounds the symbols in measured practice rather than estimate alone.

Appraising a score as a reader

Put together, an ACR appropriateness rating is best read as three layers stacked in the open: a graded body of evidence, a structured expert vote summarized as a median, and a companion radiation estimate. A high number tells you a benefit-to-harm case is strong for the typical patient in that scenario; a middle number is an invitation to read why; the radiation symbols keep dose in view. None of it replaces the clinical judgment that adapts a category to a real person with a real history.

This is educational information about how a guideline is constructed, not medical advice for any individual situation. The strength of the ACR framework is that it shows its work. When a recommendation is this legible, from literature search to graded evidence to a documented vote, you are equipped to appraise it as evidence rather than accept or dismiss it as an edict.