In a large cohort study of nearly 194,000 adults, researchers found that the American Heart Association’s new Predicting Risk of Cardiovascular Disease Events (PREVENT) equations underestimated atherosclerotic cardiovascular disease (ASCVD) risk in patients who did not receive statins, while the long-standing Pooled Cohort Equations (PCE) more closely reflected untreated risk.
Ming-Sum Lee, MD, PhD
Courtesy of Kaiser Permanente Southern California
The findings, published in JAMA Network Open, suggest that PREVENT may be better suited for contemporary populations receiving therapy, whereas PCE remains more reliable for estimating baseline risk without treatment exposure.
Short History of ASCVD Risk Prediction Tools
Risk prediction tools are central to shared decision making discussions about starting statin therapy for primary prevention of ASCVD, first author Ming-Sum Lee, MD, PhD, department of cardiology, Kaiser Permanente Los Angeles Medicine Center, and colleagues wrote.
The PCEs, introduced in 2013, have been the guideline-endorsed standard but has been criticized for overestimating risk in modern, diverse populations. They are based on cohort studies that included predominantly White and Black populations.4-6 PREVENT, introduced by the American Heart Association in 2023, incorporates kidney function, body mass index, and the Social Deprivation Index in place of race, aiming for greater accuracy and ethnic and racial equity. However, the equations were developed without fully accounting for the effect of statins, which can significantly reduce event rates during follow-up. Because risk calculators are intended to reflect untreated risk, the question remained how PREVENT performs when statin use is considered.
Study Details
For the retrospective cohort analysis, Lee et al tapped data from the Kaiser Permanente Southern California health plan for adults aged 40–75 years with LDL cholesterol levels between 70 and 189 mg/dL in 2013. Inclusion required having no prior ASCVD or diabetes; no recent statin use; and complete baseline data. Participants were followed for 10 years through December 2023. The final cohort included 193,885 individuals (median age 55 years; 58.5% women), representing a racially and ethnically diverse population. About half (51.2%) did not receive statins during follow-up, while the remainder filled at least 1 prescription.
The primary outcome was incident ASCVD, defined as acute myocardial infarction, fatal or nonfatal stroke, or death from coronary heart disease. Researchers generated 10-year risk estimates using both the PCE and PREVENT base models and compared with observed event rates. Discrimination was assessed with C statistics, and calibration was examined across risk categories.
Findings
Over 10 years, 6,528 ASCVD events occurred. Lee and colleagues observed higher event-free survival among individuals with statin exposure, with the highest event free-survival observed among individuals exposed to statin therapy for more than 80% of the follow-up period
Both models demonstrated reasonable discrimination (C statistic: 0.725 for PCE, 0.723 for PREVENT). However, their calibration differed depending on statin exposure.
In the overall population, regardless of statin exposure, PCE consistently overestimated risk. For example, among participants with predicted ASCVD risk of 7.5% to less than 10%, PCE estimated 8.7% while the observed risk was 4.5%. PREVENT estimates aligned more closely: predicted 8.6% vs observed 8.1%.
On the other hand, in participants not treated with statins, PREVENT underestimated risk. Those in the 7.5% to less than 10% risk group had an observed rate of 13.5% compared with 8.6% predicted. By contrast, PCE estimates were closer, with a predicted rate of 8.7% and observed 8.3%.
Among statin-treated patients, both models overestimated risk, but PREVENT predictions more closely tracked observed outcomes. Patients with estimated 10-year risk of 10% or greater had observed risk of 9.3% with statins, compared to 12.0% predicted by PREVENT and 16.7% predicted by PCE.
The patterns, authors said, indicate that PREVENT reflects risk levels closer to those seen in patients receiving modern preventive therapies, while PCE better captures baseline, untreated risk.
The Yin and Yang
Among the study’s limitations, the authors acknowledged the inevitable risk for confounding in observational research, in this case particularly because statin initiation was not randomized. Patients who began therapy may have differed from those who did not in ways not captured by risk models. The results may also not be generalizable beyond the study population, they wrote. Finally, while PREVENT provides both 10-year and 30-year risk estimates, only the 10-year predictions were evaluated.
The study highlights an important distinction between the two calculators. PREVENT appears to better estimate outcomes for patients who receive statins, but it underestimates untreated risk. PCE, while known to overestimate risk overall, remains closer to true baseline risk in untreated populations.
“From a practical standpoint, clinicians using the PCE should recognize that the actual event rates may be lower than estimated if patients adhere to risk-reducing interventions. There is an evolving body of literature emphasizing the importance of accounting for treatment in clinical risk prediction models,” Lee et all stressed.
References
-
Lee M, Onwuzurike J, Wu Y, Palmer-Toy DE, An J, Chen W. PREVENT and PCE models for estimating ASCVD risk stratified by statin exposure. JAMA Netw Open. 2025;8(9):e2532164. doi:10.1001/jamanetworkopen.2025.32164
-
DeFilippis AP, Young R, Carrubba CJ, et al. An analysis of calibration and discrimination among multiple cardiovascular risk scores in a modern multiethnic cohort. Ann Intern Med. 2015;162(4):266-275. doi:10.7326/ M14-1281 5.
-
Cook NR, Ridker PM. Calibration of the pooled cohort equations for atherosclerotic cardiovascular disease: an update. Ann Intern Med. 2016;165(11):786-794. doi:10.7326/M16-1739