Performance of the American Heart Association's PREVENT Equations Among Disaggregated Racial and Ethnic Subgroups.

Abstract

Importance In the original validation, the Predicting Risk of Cardiovascular Disease (CVD) Events (PREVENT) equations demonstrated good discrimination and calibration among racial and ethnic groups, but the model performance among Asian and Hispanic disaggregated subgroups has not been previously described. Objective To assess the performance of the PREVENT equations by race and ethnicity, including disaggregated Asian and Hispanic subgroups. Design, Setting, and Participants This was an electronic health record-based retrospective cohort study of primary care patients aged 30 to 79 years across Sutter Health, a large integrated health system in Northern California, from January 2010 to September 2023. Patients who had at least 2 primary care visits during the study period were eligible for the study (1 484 582). Those outside of the study age range, with prior CVD events in the washout period, missing key predictors, or having at least 1 predictor out of the allowed normal range for the American Heart Association's PREVENT equations, were excluded, leaving a study population of 361 778. Exposure Eligible patients had complete baseline data required for the PREVENT equations, including non-high-density lipoprotein cholesterol, high-density lipoprotein cholesterol, systolic blood pressure, body mass index, estimated glomerular filtration rate (or creatinine), diabetes, and current smoking status, and were free from CVD at baseline. Main Outcomes and Measures The primary outcomes were CVD events, identified using International Classification of Diseases, Ninth and Tenth Revisions, codes described in the PREVENT derivation. Results Among 361 778 patients who met the inclusion criteria, mean (SD) age was 54.6 (12.2) years; 191 151 (53%) were female; and 81 424 (22%) were non-Hispanic Asian and 40 897 (11%) were Hispanic. Over a mean (SD) follow-up of 8.1 (3.2) years, there were 22 648 (6.3%) CVD events. The C statistic for total CVD was 0.83 (95% CI, 0.82-0.84) for the Asian population and 0.80 (95% CI, 0.79-0.81) for the Hispanic population. The calibration slopes were 0.84 (95% CI, 0.78-0.90) and 1.02 (95% CI, 0.94-1.10) for Asian and Hispanic patients, respectively. Within the Asian population, C statistics for total CVD among disaggregated Asian subgroups ranged from 0.79 (95% CI, 0.77-0.81) in Filipino patients to 0.85 (95% CI, 0.83-0.87) in Asian Indian patients. The calibration slope for total CVD was less than 1 for all Asian subgroups except Asian Indian. Among disaggregated Hispanic subgroups, the C statistics were similar and between 0.80 and 0.82 for total CVD, and the calibration slope for total CVD included 1 for all subgroups. There were small differences in the performance of atherosclerotic CVD and heart failure PREVENT equations among racial and ethnic groups and subgroups. Conclusions and Relevance The PREVENT equations appropriately predicted risk in contemporary diverse Asian and Hispanic subgroups with modest variation in performance across disaggregated subgroups.