Abstract P077: Validation of the Pooled Cohort Equations for CVD Risk Assessment in African Americans: The Jackson Heart Study
Background: Cardiovascular disease (CVD) risk assessment tools such as the Framingham Risk Score are useful to identify population high risk subgroups for targeted intervention. However, no CVD risk score specific for African Americans (AAs) were available until the Pooled Cohort Equations (PCE) was introduced in 2013 for calculating sex- and race-specific10-year predicted risk of atherosclerotic cardiovascular disease (ASCVD). This study evaluated the performance of PCE in the Jackson Heart Study (JHS), a prospective cohort study of CVD in AAs.
Methods: The analytic sample included 2,191 JHS participants who were 40-79 years old without a history of CVD or CVD procedures at baseline (2000-2004) and who were not a shared participant in the Atherosclerosis Risk in Communities (ARIC) Study. ASCVD events (CHD and stroke) were ascertained by active surveillance with medical records abstraction. Because all participants were followed at least 8 years through 2012, validation of the PCE was based on 8-year observed and predicted risks of ASCVD. The PCE was evaluated for discrimination and calibration properties using c-index and Hosmer-Lemeshow (HL) x2 statistic, respectively. Overall and subgroup analysis among participants (no diabetes, LDL between 70 and 189 mg/dL and not taking statins) for whom CVD risk assessment may be applied to guide treatment for high blood cholesterol were performed. Stratified analyses evaluating the performance of PCE by baseline characteristics, including sex, age (<50/≥50 years), income (affluent*/not affluent), education (<high school/≥high school), BMI (<30/≥30), diabetes status (yes/no), self-reported use of hypertension medications (yes/no), self-reported use of statins (yes/no) and current smoking status (yes/no) were also performed.
Results: There were a total of 63 incident ASCVD events (29 CHD; 34 stroke). The PCE predicted total number of event was 130, with a c-index=0.78 (95% CI 0.54-0.96) and a HL x2=38.2 (p<0.001). The PCE showed a similar discrimination but better calibration property in the subset of participants for whom CVD risk assessment may be applied to guide treatment for high blood cholesterol (n=1,576, c-index=0.78, 95% CI 0.43-1; HL x2=21.9, p=0.005). In stratified analyses, PCE had better calibration in participants who were younger (n=831, HL x2=10.5, p=0.23), not affluent (n=1,120, HL x2=12.6, p=0.12), less educated (n=186, HL x2=6.6, p=0.58), those with lower BMI (n=833, HL x2=8.5, p=0.39), those with diabetes (n=287, HL x2=11.1, p=0.20), and statin users (n=159, HL x2=10.8, p=0.21).
Conclusions: Overall, the PCE showed good discrimination but did not calibrate well and overestimated the risk of ASCVD. In the subset of participants for whom CVD risk assessment may be applied to guide treatment for high blood cholesterol, the PCE showed improved calibration but still overestimated risk.
*Defined as 3 times above poverty level.
Author Disclosures: Y. Min: None. S.K. Musani: None. M. Sims: None. P. Anugu: None. G. Adu-Boateng: None. E.C. O’Brien: None. A. Correa: None.
- © 2017 by American Heart Association, Inc.