Clinical Challenges: Have Pooled Cohort Equations in ASCVD Stood the Test of Time?

The Pooled Cohort Equations (PCEs) have held up in atherosclerotic cardiovascular disease (ASCVD) risk prediction despite initial concerns, though further refinements may be expected.

Since the American College of Cardiology (ACC) and American Heart Association (AHA) introduced them in 2013 guidelines on assessment of cardiovascular risk, the PCEs have offered sex- and race-specific estimates of 10-year absolute rates of ASCVD events (i.e., coronary heart disease and stroke) for primary prevention populations.

The PCEs are the starting point for risk stratification in the latest cholesterol guidelines. Where there is uncertainty, clinicians may use them in conjunction with other potential risk modifiers such as coronary calcium scoring, advanced lipid testing, and other factors.

“It’s been a very nice, logical, effective way to improve decision-making about statins and other preventive therapies,” according to AHA President-Elect Donald Lloyd-Jones, MD, of Northwestern University Feinberg School of Medicine in Chicago. The risk calculator has been “very widely disseminated” and is now incorporated into most electronic health records, he noted.

Early on, however, critics had questioned the external validity of the PCEs, given that these were based on older, population-based cohort studies of Black and white people in the U.S. and may not be applicable to certain subgroups.

All risk scores have the same problem: they are derived from populations but applied to individuals, Lloyd-Jones said in an interview. “Certainly, in some groups it overestimates risk; and in some it underestimates risk.”

Indeed, the official ACC ASCVD risk calculator explicitly warns users that estimates may underestimate 10-year risk for American Indians, South Asians, and Puerto Ricans but overestimate risk for East Asians and Mexican Americans.

Yet the PCEs are meant to start a conversation about ASCVD risk between clinician and patient, Lloyd-Jones emphasized.

“The equations are just part of the process. They don’t make any decisions,” he said. “We would do far more good work to implement these approaches than continue to tweak the risk scores ad nauseam.”

Accordingly, the U.S. Preventive Services Task Force backs use of the PCEs for risk assessment despite their shortcomings.

“In general, the PCEs may overestimate risk at the high end of the risk spectrum but are fairly well calibrated at the usual decision threshold of 7.5% for statin initiation,” according to Ian Neeland, MD, of University Hospitals Cleveland Medical Center and Case Western Reserve University School of Medicine.

Concerns that the PCEs overestimate risk were assuaged by a 2018 analysis from the Women’s Health Study showing that when investigators did a better job counting events (by adding Medicare claims data to self-reported events in the study), the PCE estimates did come fairly close to observed event rates.

“People have taken more time to evaluate these equations, and they’ve stood up quite well,” Lloyd-Jones said. “There were some early concerns, and some of those data have been walked back.”

Researchers have also investigated how the PCEs fare in several subgroups, including by African American race, obesity status, South Asian ethnicity, and HIV-positive status. The PCEs showed adequate validity in each case.

“To my knowledge, although there is always variation around the extremes, the PCEs perform similarly across broad populations, even among those who were not necessarily included in the derivation cohorts,” Neeland told MedPage Today in an email.

Neeland noted that his group recently tested the performance of the PCEs by BMI, as the tool does not include metrics for obesity.

“Although BMI and waist circumference were associated with ASCVD risk, addition [of these] to the PCEs did not meaningfully improve the performance of the PCEs calculator. This is likely why BMI was never incorporated into the final PCEs algorithm,” he reasoned.

“One thing that is not clear, given the obesity epidemic and the knowledge that visceral and ectopic fat are the primary drivers of the cardiometabolic complications of obesity, would incorporation of a clinical assessment of visceral and/or ectopic (e.g. liver) fat into the PCEs improve the discrimination and risk stratification performance of the test? This has yet to be tested,” noted Neeland, “but stay tuned!”

Variables currently included in the PCEs are age, sex, race (i.e., white, Black, or other), smoking status, systolic blood pressure, hypertension treatment status, diabetes status, and total and HDL cholesterol levels.

Lloyd-Jones acknowledged that the tool may change or be used differently in the future as risk prediction continues to be refined.

In 2016, the PCEs were supplemented with the Million Hearts Model Longitudinal ASCVD Risk Assessment tool, a project of the ACC and AHA working with CMS. The calculator extension added how a person’s estimated ASCVD risk may change over time as preventive treatments are initiated.

In the future, “my expectation is that we will have more data for other races and ethnic groups, which may be helpful — certainly more data for groups such as Latinx and South Asians. I hope that will help us refine the equations further,” according to Lloyd-Jones.

The next round of cardiovascular prevention guidelines are expected in 2023. Writing committee members will be free to keep or to toss the quantitative approach to ASCVD risk estimation then, Lloyd-Jones said.

Ultimately, however, the last 8 years have shown the PCEs to be a “very well calibrated” risk prediction tool, he said. “The original concerns have largely not panned out.”