Abstract WE553: Discrimination and Calibration of the Pooled Cohort Equation Among Hispanics/Latinos Stratified by Background group and Genetic Ancestry: Insights from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL)

Abstract

Background: The Pooled Cohort Equation (PCE), a central tool in preventive cardiology, was developed without representation of Hispanic/Latino adults, a diverse and heterogenous group. We leveraged HCHS/SOL, the largest population-based study of Hispanic/Latinos, to evaluate performance of the PCE, stratified by genetic ancestry, self-reported background group and race. Methods: We included 5416 participants aged 44–74 years without prior ASCVD, statin use, diabetes, or LDL-C ≥190 mg/dL. Ten-year atherosclerotic cardiovascular disease (ASCVD) risk (2008–2019) was calculated using race-specific (non-Hispanic Black [NHB] and non-Hispanic White [NHW]) PCEs, with ASCVD events (myocardial infarction and stroke) adjudicated using hospital records. Genetic ancestry (European, African, and Amerindian) was estimated using ADMIXTURE. Kaplan-Meier estimates determined observed ASCVD event probabilities. Calibration was assessed using predicted-to-observed (P/O) ratios and C-statistics provided model discrimination. All analyses were weighted and accounted for HCHS/SOL complex survey design. Results: Mean age (SE) was 53.6 (0.2) years, 53.2% were female. Over a 10-year mean follow-up, 1.5% developed ASCVD. Both PCEs overestimated risk (predicted: 6.9% [NHB], 6.0% [NHW]), with P/O >1 across all Hispanic background groups (Table) . Miscalibration was most pronounced in Hispanics of Dominican background (P/O: 6.43 [NHB], 5.11 [NHW]) and participants with greater-than-median Amerindian Ancestry (P/O: 6.89 [NHB], 5.89 [NHW]). Calibration was more accurate among people who reported Black/African American race (P/O: 1.52 [NHB], 1.24 [NHW]), and Puerto Rican background (P/O: 2.96 [NHB], 2.49 [NHW]). Discrimination was moderate overall (C=0.70 [NHB], 0.71 [NHW]), varied by Hispanic background (Table 1), and was highest among individuals with greater-than-median African Ancestry (0.77 [NHB], 0.77 [NHW]) (Table) . Conclusions: While both the NHB and NHW PCEs overestimated ASCVD risk across all Hispanic background groups, calibration was more accurate in Hispanic/Latino individuals self-identified as Black, and discrimination was strongest with the NHW PCE. These findings highlight the need for more inclusive and tailored risk prediction tools in diverse populations.