TrialsgrambetaRCTs are beautiful

NOTIFY-1

Incidental Coronary Artery Calcium: Opportunistic Screening of Previous Nongated Chest Computed Tomography Scans to Improve Statin Rates (NOTIFY-1 Project)

Patient / Population Intervention / Exposure Comparison Outcome

In 173 patients with incidental coronary artery calcium on a previous chest CT and no statin, deep-learning detection with notification of clinician and patient increased statin prescription within 6 months compared with usual care.

N
173patients
Design
Randomised quality-improvement project, Stanford Health Care, USA
Endpoint
Statin prescription within 6 months
Relevance
2Important — one of several pillars.
ResultStatin prescription 51.2% (44/86) vs 6.9% (6/87) (P<0.001); coronary artery disease testing 15.1% (13/86) vs 2.3% (2/87) (P=0.008). 194 were randomised, of whom a radiologist confirmed CAC in 173 (89.2%).
Sandhu AT, et al. Circulation. 2023;147(9):703-714. 10.1161/CIRCULATIONAHA.122.062746
Discussion & critique

The template for opportunistic AI screening of imaging already in the record: a deep-learning read of old chest CTs, a patient-specific image and a guideline note produced a large rise in statin prescribing. Small and from a single health system, with a prescription rather than an event endpoint; of 2,113 eligible patients the algorithm flagged 424, and the radiologist did not confirm the finding in about one in ten of those randomised (89.2% confirmed), so human confirmation remained part of the pathway.