The Accuracy of Cardiovascular Pooled Cohort Risk Estimates in U.S. Older Adults

Background: The ACC/AHA guidelines for primary prevention rely on the Pooled Cohort Risk Equations (PCE) risk estimates of atherosclerotic cardiovascular disease (ASCVD) to guide treatment decisions. In light of the PCE being derived in younger populations, their accuracy in older adults is uncertain. Objective: To evaluate the predictive accuracy and calibration of the PCE in older individuals. Design and Setting: We estimated CVD predicted and observed risk among individuals from four large prospective cohort studies: Cardiovascular Health Study, Multiethnic Study of Atherosclerosis, Framingham Original, and Framingham Offspring. Participants: 12,527 overall individuals without ASCVD, including 9864 individuals aged 40–74 years and 2663 aged ≥75 years. Measurements: We examined the operating characteristics of the PCE to estimate 5-year risk of stroke, MI, and CHD death overall and by age and sex strata. The associations between individual components of the PCE and cardiovascular events by age group (≥75 vs 40–74 years) were also evaluated. Results: The PCE had low discrimination for 5-year ASCVD risk in older (≥75 years) (c-statistic = 0.62, 95% CI 0.60–0.65) vs. younger (40–74 years) adults (c-statistic = 0.75, 95% CI 0.73–0.76). Calibration of the PCE was suboptimal in both older and younger adults, overestimating risk in the highest risk groups. Performance of the PCE in older adults was similarly poor when stratified by sex and age ≥ 80 years. Limitations: Since the PCE were derived from similar cohorts, though using different age groups and exams, this analysis likely overestimates the performance of the PCE. Conclusion: The performance of the PCE for ASCVD risk estimation in older adults is suboptimal; new models to effectively risk-stratify older adults are needed.