The overnight radiologist burden no CMR program budgets for
How a CMR triage program's overnight radiologist burden balloons when nobody budgets for after-hours pages: every borderline QA alert becomes an on-call paging event, and the read queue grows before morning. The pre-filter breaks the cycle.
The alert math
Daytime reads are easy to budget. A 7T cardiac MR program books a fixed number of slots and the on-site radiologist works through them. After-hours work behaves differently. Every borderline QA find, every flagged arrhythmia strip, every cine flagged as a rescue read turns into a page to the on-call list. A typical 60-study evening produces 8 to 12 pages before sunrise — the workload the radiologist did not sign up for and the program did not budget for.
What the QA pre-filter catches
The intake QA rubric grades each cine against motion blur, signal homogeneity, and gating failure modes. Roughly half of borderline scores resolve into "this can wait until morning" and get parked in a structured review queue. The other half — missed arrhythmia the AI saw, missed vasculature, gated failures that hide real findings — escalate immediately. The filter is the difference between 12 pages an evening and 1.
What still gets paged
Even with a tuned QA filter, a small number of cases always crosses the threshold: genuine missed pathology, arrhythmia strips the rubric under-scored, freezes the technologist marked as rescue read during acquisition. These are the irreducible paging load. The job is to make this list narrow, predictable, and the radiologist's first action when they open the page — not a triage step after they wake up.
After-hours read latency
Median time-to-read on a paged case runs 35 to 45 minutes from alert to signed report, because the on-call radiologist has to spin up the full study context from scratch. The same case reads in under 8 minutes during the daytime queue, where the QA verdict and the suspected root cause are pre-loaded. Closing that 30-minute gap is what the program actually buys when it eliminates overnight paging noise.
What this changes for your program
A right-sized on-call budget looks like one to two paged cases an evening, not twelve. The metric to publish in the dashboard is post-intake read latency and page rate per study — those are the numbers that tell you whether the QA rubric is doing its job. They also move the vendor conversation: a 7T program that pages its radiologist twelve times an evening is buying a reading service, not an imaging pipeline.
Related reading: The 7T CMR workflow, end to end and Why CMR protocol consistency is the hardest multi-site problem.
Ready to retriage your on-call paging? Tell us about the program.