Evaluating the American Heart Association’s PREVENT Equations: Enhancing CVD Risk Prediction

The American Heart Association's (AHA’s) Predicting Risk of CVD EVENTs (PREVENT) equations showed excellent discrimination in identifying cardiovascular disease (CVD) risk and only modest discrepancies in calibration, a study finds.¹

The researchers believe the findings support the use of the AHA’s PREVENT equations for application in the intended target population. | Image credit: Pana Studio - stock.adobe.com

The prognostic cohort study is published in JAMA Network Open.

“In the overall cohort and with the removal of extreme clinical variable values, the C statistics were similar to, or greater than, those observed by the PREVENT working group,” wrote the researchers of the study. “These findings support the incorporation of the PREVENT equations in clinical settings to inform patient-clinician discussions on CVD risk and appropriateness of intervention.”

The AHA’s PREVENT equations were developed to improve the accuracy of CVD risk prediction in adults aged 30 to 79 years without known CVD.² These multivariable models address limitations in existing risk assessment tools by incorporating traditional risk factors, such as smoking, blood pressure, cholesterol, diabetes, and kidney function alongside competing risks like non-CVD death. The equations were developed using data from over 3.2 million participants across 25 studies and were externally validated in more than 3.3 million additional individuals. PREVENT models are sex-specific and race-free, enhancing their applicability across diverse populations.

The new study utilized data from the National Health and Nutrition Examination Survey (NHANES) spanning 1999 to 2010, including a 10-year follow-up period for cardiovascular CVD risk assessment.¹ Data were analyzed in accordance with the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) guidelines.

Key variables, such as age, blood pressure, cholesterol levels, and estimated glomerular filtration rate (eGFR), were extracted and integrated into the PREVENT equation models. Sensitivity analyses were also conducted to ensure robustness of the results across various risk factor definitions and participant cohorts.

From the original cohort of 62,160 NHANES participants, a final dataset of 24,582 participants representing 172.9 million US adults was analyzed after excluding individuals with missing data or prevalent atherosclerotic CVD. 

The analysis showed significant differences in CVD mortality across PREVENT risk categories, with each 1% increase in risk linked to higher CVD mortality (HR, 1.090; 95% CI, 1.087-1.094). Additionally, the PREVENT model demonstrated excellent discrimination (C statistic, 0.890), and calibration was slightly underfitted, particularly in women.

Comparatively, the Pooled Cohort Equations (PCE) model performed worse, with overfitting (slope, 0.77) and lower discrimination (C statistic, 0.880). Furthermore, The PREVENT model significantly improved risk reclassification, particularly for men, highlighting its utility over PCE in predicting CVD mortality.

However, the researchers acknowledged some limitations to the study. First, NHANES does not track nonfatal CVD events, limiting the analysis to fatal outcomes and potentially underrepresenting the full burden of CVD. Additionally, some of the more recent statistical approaches applied could not fully accommodate the NHANES complex survey design. Lastly, while sensitivity analyses were performed to mitigate these limitations, the researchers noted that these findings could benefit from validation in other large, longitudinally monitored cohorts that capture both fatal and nonfatal CVD events.

Despite these limitations, the researchers believe the findings support the use of the AHA’s PREVENT equations for application in the intended target population.

“The PREVENT risk scores demonstrated excellent discrimination of CVD mortality risk in the overall cohort,” wrote the researchers. “Despite moderate discrepancies observed in the calibration assessments, the C statistics consistently supported good-to-excellent discrimination in the present results.”

References

1. Scheuermann B, Brown A, Colburn T, et al. External validation of the American Heart Association PREVENT cardiovascular disease risk equations. JAMA Netw Open. 2024;7(10):e2438311. doi:10.1001/jamanetworkopen.2024.38311

2. Khan SS, Matsushita K, Sang Y, et al. Development and validation of the American Heart Association’s Prevent Equations. Circulation. 2024;149(6):430-449. doi:10.1161/circulationaha.123.067626