AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Multidomain Cardiovascular Risk Assessment Using PREVENT Score, ECG, and Neighborhood Environment
By King R, Alkhaleefah M, Ardakani J et al.
Background: Cardiovascular risk is influenced by clinical risk factors, subclinical atherosclerosis, and environmental factors. The American Heart Association PREVENT equations estimate ASCVD risk using traditional clinical variables, but the independent contributions of electrocardiogram (ECG)-derived coronary artery calcium (CAC) and neighborhoid factors remain incompletely characterized. Hypothesis: PREVENT risk, deep learning based ECG-derived CAC, and community vulnerability index (a multidomain neighborhood environment index) each provide independent and additive prognostic information for incident ASCVD. Methods: We conducted a retrospective cohort study within the Houston Methodist CVD Registry. Patients without prevalent ASCVD were included. For each patient, we calculated the 10-year PREVENT ASCVD risk scores, ECG-derived CAC classifications, and census tract-level Community Vulnerability Index measure. PREVENT risk was categorized as <3%, 3-5%, 5-10%, and >10%; CAC as 0, 1-99, 100-399, and ≥400; and CVI into quartiles. Cox proportional hazards models evaluated independent associations with incident ASCVD after mutual adjustment. Model discrimination was assessed using Harrell’s C-index. Results: Among 177,728 patients (mean age 56.4 years; 57.4% women), 11,871 incident ASCVD events occurred during a median follow-up of 2.65 years. In combined models, higher PREVENT risk, ECG-derived probability of CAC >=1000, and community vulnerability index were each independently associated with incident ASCVD. Per standard deviation increase, the hazard of ASCVD increased by 56% for PREVENT risk (HR 1.56, 95% CI 1.54-1.59), 60% for CAC probability (HR 1.60, 95% CI 1.56-1.64), and 23% for CVI (HR 1.23, 95% CI 1.21-1.25). Using categorical reference groups, patients with PREVENT >10% had nearly six-fold greater risk than those with PREVENT <3% (HR 5.99, 95% CI 5.53-6.50), while CAC ≥400 (HR 1.72, 95% CI 1.59-1.86) and highest-versus-lowest CVI quartile (HR 1.79, 95% CI 1.70-1.89) remained independently associated with ASCVD after adjustment for the other modalities. Seven-year ASCVD event rates increased progressively across both PREVENT and CAC strata, ranging from 2.8% among patients with PREVENT <3% and CAC=0 to 32.2% among those with PREVENT >10% and CAC ≥400. Model discrimination improved from 0.716 for PREVENT alone to 0.734 when PREVENT, CAC, and CVI were combined. Conclusions: Clinical risk estimated by PREVENT, ECG-derived markers of subclinical atherosclerosis, and neighborhood factors each contribute independently to ASCVD risk. Integration of these complementary domains identifies substantial heterogeneity in ASCVD risk and may enable more comprehensive cardiovascular risk assessment than clinical risk scores alone.