AHA PREVENT Risk

The Pooled Cohort Equations estimate 10-year ASCVD risk only, from age 40, using race as an input. PREVENT drops race, starts at age 30, adds BMI, eGFR and current statin use, reports 30-year as well as 10-year risk, and predicts heart failure as a separate outcome. It was derived on 3.28 million US adults across 25 datasets, and is better calibrated than the PCE — in practice it usually estimates lower risk.

Because the PREVENT paper defines no such categories. It reports absolute risk and stops there. The familiar 5 / 7.5 / 20% cut points belong to the 2018 ACC/AHA cholesterol guideline and were set against the Pooled Cohort Equations, not against PREVENT — carrying them over to a differently calibrated model would shift who gets treated without any source saying so. Apply whichever guideline threshold you are working from.

The coefficients are read straight out of the supplemental workbook published with the paper, by an offline script (scripts/prevent/build_prevent.py) rather than typed by hand — the base model alone is 460 coefficients. The test suite then reproduces the authors’ own worked examples for all five outcomes, both sexes and both horizons. Nothing here is transcribed from another calculator.

For a 35-year-old, almost any 10-year risk looks reassuring, which is exactly when a long-horizon estimate changes the conversation. Note the 30-year equations are a looser approximation of the underlying competing-risks model than the 10-year ones, so treat them as the coarser figure.

Published, and on the same supplement sheets, but not built here yet. They need a UACR and a zip-code-level deprivation index you rarely have in the room. The generator already extracts only the base model; adding one is a matter of pointing it at another sheet.