Risk equations yield similar observed, predicted 5-year ASCVD risks

WASHINGTON — The new American College of Cardiology/American Heart Association Pooled Cohort risk equations were well calibrated in a cohort of US adults and yielded similar observed and predicted 5-year atherosclerotic CVD risks, researchers reported at the American College of Cardiology Scientific Sessions.

Paul Muntner, PhD, from the University of Alabama at Birmingham, and colleagues evaluated the Pooled Cohort risk equations in US adults aged 45 to 79 years enrolled in the REGARDS study. The risk equations were developed to estimate atherosclerotic CVD (ASCVD) and guide statin initiation.

Considered for statin therapy

Paul Muntner

Muntner said the researchers analyzed the risk equations in members of the REGARDS population likely to be considered for initiation of statin therapy based on their results from these equations: those without ASCVD or diabetes and LDL 70 mg/dL to 189 mg/dL who are not taking statins (n=10,997). They also performed an analysis of Medicare beneficiaries in the cohort (n=3,333), using data on ASCVD in Medicare claims and adjudicated outcome data from REGARDS, he said.

“As any new risk equation needs to be externally validated, our objective was to evaluate the validity of the Pooled Cohort risk equation in a contemporary US population for whom the equations are intended to inform discussions about initiating statins,” Muntner said.

The researchers compared predicted risk with the following observed adjudicated ASCVD in the REGARDS population: nonfatal MI, CHD death and fatal/nonfatal stroke. Although the risk equations attempt to predict risk for CVD for the next 10 years, the analysis covered 5 years because REGARDS participants have not yet been followed for 10 years, Muntner said.

There were 338 adjudicated events (192 CHD events, 146 strokes) during the 5-year period, according to the results.

The C-index for the main analysis was 0.72 (95% CI, 0.7-0.75); more than 0.7 is considered good discrimination, Muntner said.

Most accurate for those at low risk

The predictions of the risk equations were most accurate in lower-risk participants, Muntner said. The observed and predicted 5-year ASCVD incidence per 1,000 person-years were as follows:

For those with a 10-year predicted risk <5%: 1.9 observed incidence (95% CI, 1.3-2.7), 1.9 predicted incidence.
For those with a 10-year predicted risk 5% to <7.5%: 4.8 observed incidence (95% CI, 3.4-6.7), 4.8 predicted incidence.
For those with a 10-year predicted risk 7.5% to <10%: 6.1 observed incidence (95% CI, 4.4-8.6), 6.9 predicted incidence.
For those with a 10-year predicted risk ≥10%: 12 observed incidence (95% CI, 10.6-13.6), 15.1 predicted incidence (Hosmer-Lemeshow goodness-of-fit-test=19.9; P=.01).

The results “suggest some undercalibration” in those with high predicted risk, Muntner said.

In the Medicare population analyzed, there were 234 ASCVD events (120 CHD events, 114 strokes) and better calibration than in the larger population (Hosmer-Lemeshow goodness-of-fit-test=5.4; P=.71). The C-index was 0.67 (95% CI, 0.64-0.71).

In the Medicare population, “if anything, the Pooled Cohort risk equations are underestimating [CVD risk],” he said.

“Based on our results, we think that the previous results of overestimation of CVD risk are likely due to incomplete capture of CVD events and inclusion of participants taking statins,” Muntner said. “We believe the current study supports the validity of the Pooled Cohort risk equations to inform clinical management decisions.” – by Erik Swain

For more information:

Muntner P. Featured Clinical Research I. Presented at: American College of Cardiology Scientific Sessions ; March 29-31, 2014; Washington, D.C.

Muntner P. JAMA. 2014;doi:10.1001/jama.2014.2630.

Disclosure: Muntner reports financial ties with Amgen.