Study Validates Accuracy of PREVENT for CVD Mortality Risk Prediction

The PREVENT equations from the American Heart Association (AHA) accurately predict 10-year risk of mortality related to cardiovascular disease in the US general population, according to an external validation study.

In fact, the risk-prediction tool performed better than the prior pooled cohort equations (PCEs), which have been the gold standard in assessing risk of atherosclerotic cardiovascular disease (ASCVD) over the past decade, researchers led by Britton Scheuermann, MS (Kansas State University, Manhattan), report in a paper published online recently in JAMA Network Open.

Video Topical Discussion

Quick Takes: Britton Scheuermann on the Accuracy of the PREVENT Equations

November 12, 2024

The PREVENT equations, first presented by Sadiya S. Khan, MD (Northwestern University Feinberg School of Medicine, Chicago, IL), about a year ago at the AHA 2023 Scientific Sessions, were developed to update the PCEs. Their creators aimed to provide a broader assessment of risk by incorporating heart failure and a wider array of cardiovascular, kidney, and metabolic risk factors, and sought to ultimately improve patient-physician conversations and treatment decisions.

This new tool expands assessments of risk beyond typical factors like cholesterol and blood pressure, which “could be really exciting for opening the door for better understanding how all of the organ systems contribute to cardiovascular risk,” Scheuermann told TCTMD.

Moreover, added senior author Carl Ade, PhD (Kansas State University), “with this validation study, we hope to have provided a little additional confidence that the PREVENT calculator is an important tool and an accurate tool for evaluating cardiovascular disease risk, really building on the work of Khan et al.”

PREVENT vs PCEs

To provide external validation of the PREVENT equations, which were initially developed using data on more than 6.6 million adults, Scheuermann, Ade, and colleagues turned to the National Health and Nutrition Examination Survey (NHANES), which provides data that are both representative of the US population and publicly available. They examined data spanning 1999 to 2010 on 24,582 adults followed for at least 10 years. This group was representative of 172.9 million Americans (mean age 45.0 years; 52.1% women).

The investigators showed that each 1% increase in the PREVENT risk estimate was associated with a greater risk of CVD mortality at 10 years (HR 1.090; 95% CI 1.087-1.094). The tool had excellent discrimination, with a C-statistic of 0.89 overall; this was slightly higher in women versus men (0.90 vs 0.87).

The area under the receiver-operator characteristic curve was 0.81 overall, 0.83 in men, and 0.80 in women. In terms of model calibration, there was “modest underfitting” (calibration slope 1.13; 95% CI 1.06-1.21).

We hope to have provided a little additional confidence that the PREVENT calculator is an important tool and an accurate tool for evaluating cardiovascular disease risk. Carl Ade

The PREVENT models performed significantly better than the PCEs for discriminating risk of CVD mortality at 10 years (P < 0.001). The net reclassification index was 0.093, indicating a significant improvement when using PREVENT versus the PCEs. This was mostly related to reclassifying patients to a lower risk level than what was estimated by the PCEs.

The better performance compared with the PCEs suggests that the PREVENT equations “really may be a next step forward,” Scheuermann said, noting that the use of more-recent data to develop and validate PREVENT captures the changing demographics of the country and temporal trends in CV risk.

Overall, the validation highlighted the reproducibility of the PREVENT equations and indicated “that this is going to be really a game changer in patient communication,” Ade said. A large healthcare system in Kansas that his team works closely with, Stormont Vail Health, has already started integrating the PREVENT equations into its electronic health records system, he added.

How Will This Change Discussions With Patients?

Scheuermann noted that recent studies have shown that use of PREVENT versus the PCEs can be expected to reclassify many patients as lower risk and perhaps lessen use of medications like statins and antihypertensives. There could potentially be shifts in use of certain tests, like assessments of lipoprotein(a) concentrations and calcium scoring, he added.

Moreover, in terms of patient communication, the ability to provide estimates of 10- and 30-year risks allows for different types of discussions and decision-making, Ade said.

In an accompanying editorial, Sridhar Mangalesh, MBBS (Jacobi Medical Center, Albert Einstein College of Medicine, Bronx, NY), and Michael Nanna, MD (Yale School of Medicine, New Haven, CT), say the study “is a commendable effort toward the adoption of the base PREVENT equations in current practice,” but note that there are some limitations.

They point to the look at only one outcome (CVD mortality), a consequence of what was available within the database used, as well as other issues related to the complexity of the NHANES population.

“Nonetheless, the present investigation represents an excellent use of NHANES data for evaluating the PREVENT equations against an aging criterion standard,” Mangalesh and Nanna write. They also note that “there is a substantial lack of US cardiovascular data that may lend sufficient power to similar investigations and remain nationally representative while not relying on weighted methods, so as to allow for conventional robust model comparison methods used in the present study.”

Ade and Scheuermann agreed with the limitations highlighted by the editorialists. “There’s still work to be done, but we’re just very excited to be part of the process,” Scheuermann said.

Ade added: “Anybody who was maybe on the fence [about using PREVENT], this maybe helps them have more confidence moving forward with this equation.”