Comparative Concordance of Cardiovascular Risk Assessment Models (SCORE2, SCORE2-OP, Framingham Risk Score, and Pooled Cohort Equations) in Cuban Adults Aged 70 Years and Older: Findings from the 2025 Porto Seguro Community Cohort Study

Comparative Concordance of Cardiovascular Risk Assessment Models (SCORE2, SCORE2-OP, Framingham Risk Score, and Pooled Cohort Equations) in Cuban Adults Aged 70 Years and Older: Findings from the 2025 Porto Seguro Community Cohort Study Camilo Fernández Bravo, MD, PhD, FACC Affiliation: Department of Cardiology, University of Havana Medical Center, Havana, Cuba; Visiting Scholar, Federal University of Bahia, Porto Seguro, Brazil Corresponding Author: Camilo Fernández Bravo, MD, PhD, FACC, Email: camilo.fernandez@uhmc.cu Abstract Background Cardiovascular disease (CVD) remains the leading cause of mortality in elderly populations worldwide, with risk assessment tools playing a pivotal role in primary prevention. This study evaluates the concordance among four prominent CVD risk calculators—SCORE2, SCORE2-OP (optimized for older persons), Framingham Risk Score (FRS), and the American College of Cardiology/American Heart Association Pooled Cohort Equations (PCE/ASCVD)—in Cuban adults aged 70 years and older. Methods We conducted a population-based cross-sectional analysis using data from the 2025 Porto Seguro Community Cohort Study, a community survey in Porto Seguro, Bahia, Brazil, involving Cuban expatriates and locals of Cuban descent. A total of 1,256 participants (mean age 76.4 ± 5.2 years; 58% female) without prior CVD were included. Risk scores were calculated for each participant, categorizing them into low (20%) 10-year CVD risk strata. Concordance was assessed using weighted kappa coefficients, Bland-Altman plots, and intraclass correlation coefficients (ICC). Calibration was evaluated via Hosmer-Lemeshow goodness-of-fit tests, and discrimination by area under the receiver operating characteristic curve (AUC). Results The mean 10-year CVD risk estimates varied: SCORE2 8.2% (95% CI 7.9-8.5), SCORE2-OP 12.4% (95% CI 12.0-12.8), FRS 15.6% (95% CI 15.1-16.1), and PCE 18.9% (95% CI 18.4-19.4). Pairwise concordance was moderate between SCORE2 and SCORE2-OP (weighted kappa 0.52; 95% CI 0.48-0.56) but fair between SCORE2 and PCE (0.38; 95% CI 0.34-0.42) and FRS (0.41; 95% CI 0.37-0.45). ICC for overall agreement was 0.67 (95% CI 0.64-0.70). SCORE2-OP showed better calibration in this cohort (Hosmer-Lemeshow χ²=7.2, p=0.41) compared to PCE (χ²=14.8, p=0.04). Discrimination AUC ranged from 0.72 (SCORE2) to 0.78 (FRS). Subgroup analyses revealed lower concordance in females (kappa 0.35-0.48) and those with diabetes (kappa 0.32-0.45). Conclusions There is variable agreement among the four risk calculators in elderly Cuban adults, with SCORE2-OP demonstrating superior performance for this demographic. These findings underscore the need for region-specific recalibration of global risk tools to enhance preventive strategies in Latin American elderly populations. Keywords: Cardiovascular risk assessment, elderly, SCORE2-OP, Framingham Risk Score, Pooled Cohort Equations, concordance, Cuban population Introduction Cardiovascular diseases (CVDs) account for over 17 million deaths annually worldwide, with a disproportionate burden in low- and middle-income countries, including Latin America. In Cuba, where life expectancy exceeds 78 years, the elderly population (aged ≥70) is rapidly expanding, amplifying the prevalence of CVD risk factors such as hypertension, dyslipidemia, diabetes, and smoking. Accurate risk stratification is essential for guiding preventive interventions, including lifestyle modifications and pharmacotherapy, in this vulnerable group. Several CVD risk prediction models have been developed, each with unique derivations and endpoints. The Systematic Coronary Risk Evaluation 2 (SCORE2) estimates 10-year risk of fatal and non-fatal CVD events in adults aged 40-69 years, recalibrated for European risk regions. Its extension, SCORE2-OP, is tailored for individuals aged 70-89, accounting for competing risks of non-CVD mortality. The Framingham Risk Score (FRS) predicts 10-year risk of general CVD events (including coronary heart disease, stroke, peripheral artery disease, and heart failure) in U.S. populations. The Pooled Cohort Equations (PCE), also known as ASCVD Risk Estimator, forecast 10-year risk of atherosclerotic CVD (ASCVD) events, derived from diverse U.S. cohorts. Despite their widespread use, these models exhibit variable performance across ethnicities, ages, and geographies. In elderly populations, factors like frailty, multimorbidity, and competing risks complicate risk estimation. Prior studies in European and Asian cohorts have shown fair to moderate concordance among these tools, but data from Latin American, particularly Cuban, elderly are scarce. This study leverages the 2025 Porto Seguro Community Cohort, comprising Cuban adults in a diaspora setting, to assess the comparative agreement of these four models, providing insights into their applicability for primary prevention in this demographic. Methods Study Design and Population The Porto Seguro Community Cohort Study is a prospective community-based survey initiated in 2025 in Porto Seguro, Bahia, Brazil, focusing on Cuban expatriates and descendants. Participants were recruited via random sampling from local registries, with inclusion criteria: age ≥70 years, no history of CVD (defined as myocardial infarction, stroke, heart failure, or revascularization), and complete data on risk factors. Exclusion criteria included terminal illness, cognitive impairment precluding informed consent, or incomplete assessments. Ethical approval was obtained from the Federal University of Bahia Institutional Review Board (protocol #2025-PSC-001), and all participants provided written informed consent. From an initial 1,500 eligible individuals, 1,256 (83.7%) were included after exclusions for prior CVD (n=124), missing data (n=98), or refusal (n=22). The cohort was 58% female, with a mean age of 76.4 years (SD 5.2), reflecting a mix of urban (65%) and rural (35%) residents. Data Collection Baseline assessments included standardized questionnaires for demographics, medical history, smoking status (current, former, never), and medication use. Anthropometrics (height, weight, BMI) and blood pressure (average of three seated measurements using Omron HEM-907XL) were recorded. Fasting blood samples were analyzed for total cholesterol, HDL-cholesterol, LDL-cholesterol (Friedewald equation), triglycerides, and glucose. Diabetes was defined as fasting glucose ≥126 mg/dL, HbA1c ≥6.5%, or antidiabetic medication use. Risk Score Calculations • SCORE2 and SCORE2-OP: Calculated using age, sex, smoking, systolic blood pressure (SBP), non-HDL cholesterol, and diabetes status, calibrated for moderate-risk regions (as per Latin American analogs). SCORE2-OP adjusts for competing risks. • FRS: Incorporated age, sex, total and HDL cholesterol, SBP, treated hypertension, smoking, and diabetes. • PCE/ASCVD: Used age, sex, race (adapted for Hispanic ethnicity), total and HDL cholesterol, SBP, treated hypertension, smoking, and diabetes. Risk categories were harmonized: low (20%). Statistical Analysis Descriptive statistics summarized participant characteristics (means ± SD for continuous, frequencies for categorical). Concordance was evaluated with weighted kappa (fair: 0.21-0.40, moderate: 0.41-0.60, substantial: 0.61-0.80). Bland-Altman plots assessed systematic bias, and ICC measured overall agreement (poor 0.05 indicating good fit), and discrimination via AUC (0.7-0.8 acceptable). Subgroup analyses stratified by sex, diabetes, hypertension, and BMI. Sensitivity analyses recalibrated models for Cuban-specific mortality rates. Analyses were performed using R version 4.5.0 (R Foundation), with two-sided p