PREVENT vs PCE: Refining ASCVD Risk Prediction in Statin-Treated Populations

Estimating atherosclerotic cardiovascular disease (ASCVD) risk has never been a static exercise. A recent Kaiser Permanente cohort analysis of nearly 194,000 adults without prior ASCVD or diabetes tested how the Predicting Risk of Cardiovascular Disease Events (PREVENT) and Pooled Cohort Equations (PCE) models perform at estimating ASCVD risk and stratifying results by statin exposure during follow-up.

The study included adults aged 40 to 75 years with LDL cholesterol levels between 70 and 189 mg/dL, followed for 10 years. Researchers calculated each participant’s 10-year ASCVD risk using both models, then compared predicted versus observed outcomes across three exposure groups: no statin, any statin use, and high adherence (>80% of follow-up). Model discrimination was measured by C statistics, and calibration assessed whether predicted risk matched observed events.

Where the Models Aligned and Diverged
Discrimination was nearly identical (C statistic 0.725 for PCE, 0.723 for PREVENT), showing both models could rank risk reasonably well. The main differences emerged in calibration:

• In the overall cohort, PCE tended to overestimate risk, while PREVENT aligned more closely with observed outcomes.
• Among patients not on statins, PREVENT underestimated observed risk, whereas PCE predictions were closer to actual event rates.
• In patients on statins, especially adherent users, observed ASCVD events fell below predictions from both models—consistent with the protective effect of therapy rather than a model flaw.

These results suggest that PREVENT’s lower estimates may reflect its design as a model tuned to modern, often treated populations, whereas PCE retains the calibration of an untreated baseline. PREVENT intentionally removes race as a predictor, replacing it with social deprivation index and incorporating kidney function and body mass index. Those adjustments make its estimates more representative of contemporary practice but may also shift risk downward when applied to untreated individuals. Compared with PCE, PREVENT reclassified many patients into lower estimated-risk categories, particularly those near guideline thresholds.

There are a few limitations to consider here. This was an observational analysis within a single integrated health system, and statin initiation was not randomized. Differences in adherence, follow-up duration, and residual confounding could influence calibration. Cardiovascular events were identified from health records rather than centralized adjudication, and findings may not extend to uninsured or higher-risk populations.

Clinical Takeaway
PCE remains better aligned with untreated risk, while PREVENT reflects the evolving epidemiology of a statin-treated population. Understanding where each model fits can refine how clinicians and health systems communicate risk and help ensure that the conversation about ASCVD prevention reflects both baseline vulnerability and the expected effect of contemporary therapy.

Reference
Lee M-S, Onwuzurike J, Wu Y-L, Palmer-Toy DE, An J, Chen W. PREVENT and PCE Models for Estimating ASCVD Risk Stratified by Statin Exposure. JAMA Network Open. 2025;8(9):e2532164. doi:10.1001/jamanetworkopen.2025.32164