An AJR expert panel has reviewed breast cancer risk stratification — the models in current clinical use, what the professional societies say about using them, and what stops the calculated number turning into a screening decision. Their framing of the problem is the useful part: societal recommendations vary, so there is no settled answer on which tool to use, when in a woman's life to run it, or what threshold should change her screening.
The panel proposes a practical stepwise approach to embedding risk assessment in routine work, and reviews the evidence for deep learning models that estimate risk from the mammogram itself rather than from a questionnaire. That last strand is genuinely different in kind: a model reading the image needs no family history to be collected, which is the step that fails most often in a busy service.
For a radiology department the decision is not which model is best but whether risk assessment happens at all and who owns it. A risk figure calculated and never acted on is worse than none, because it creates a record of a risk nobody addressed. Decide who runs the model, at what visit, what threshold triggers supplemental MRI or earlier screening, and who has the conversation — before adopting a tool. In Indian practice the constraint is different again: screening is largely opportunistic, family history data are thin, and the models in widest use were derived in Western populations, so a calculated risk should be treated as one input to a discussion rather than as a number to act on mechanically.
- Decide who runs the risk model and at which visit before adopting one.
- Set the local threshold that triggers supplemental imaging, and write it down.
- Do not record a risk score with no pathway attached to it.
- Check the derivation population of any model before applying it to Indian women.
- Image-based risk models avoid the family history step that most often fails in practice.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for radiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free