Key result
PREVENT classified 19.8% of first-AMI patients as high risk compared to 20.2% with FRS, 15.0% with WHO, 12.3% with ASCVD 2013, and 11.7% with JBS-3, with poor intermodel agreement.
Why the study?
Most cardiovascular risk prediction models were derived in Western populations and may not generalize to the South Asian population, necessitating comparison in first AMI patients.
Does the novel PREVENT score improve pre-event high-risk classification and agreement compared with legacy cardiovascular risk prediction models (FRS, ASCVD 2013, WHO, JBS-3) in Indian patients presenting with a first acute myocardial infarction?
Cohort (n=4,975)
No
Does the novel PREVENT score improve pre-event high-risk classification and agreement compared with legacy cardiovascular risk prediction models (FRS, ASCVD 2013, WHO, JBS-3) in Indian patients presenting with a first acute myocardial infarction?
Major Western-derived cardiovascular risk prediction models (including PREVENT, FRS, ASCVD 2013, WHO, and JBS-3) show poor categorical agreement and fail to identify at least 80% of Indian patients as high risk prior to their first acute myocardial infarction, highlighting the urgent need for South Asia-specific risk algorithms.
Answers are grounded in this paper and cite their sources.
Over 80% of first-AMI patients not flagged high-risk by any model; leaves open South Asia-specific risk equations.
BACKGROUND: Cardiovascular (CV) risk prediction models guide primary prevention, yet most were derived in Western populations and may not generalize to the South Asian population. OBJECTIVES: To compare risk classification, agreement, and discriminative patterns of major atherosclerotic cardiovascular disease (ASCVD) risk prediction models in first acute myocardial infarction (AMI). METHODS: In this retrospective analysis, 4,975 consecutive patients aged 40 to 79 years presenting with first AMI were evaluated. Pre-event demographics, clinical characteristics, and laboratory data were extracted from outpatient records. Ten-year pre-event CV risk was calculated using 5 prediction models: Framingham Risk Score (FRS), American College of Cardiology/American Heart Association ASCVD 2013, World Health Organization (WHO) CVD charts, Joint British Societies' cardiovascular risk score version 3 (JBS-3), and PREVENT equations. Patients were categorized as low (<7.5%), intermediate (7.5% to <20%), or high (≥20%) risk. Statistical analyses included Pearson correlation, intraclass correlation, weighted Cohen's kappa, and repeated-measures generalized linear modeling with Bonferroni-adjusted pairwise comparisons. RESULTS: PREVENT classified 19.8% of patients as high risk, compared with 20.2% using FRS, 15.0% using WHO charts, 11.7% using JBS-3, and 12.3% using ASCVD 2013. PREVENT demonstrated the widest risk distribution (0.2%-91%). Agreement between PREVENT and other models was poor (κ = 0.228; 95% CI: 0.204-0.252 with ASCVD, κ = 0.311; 95% CI: 0.288-0.33 with FRS, and κ = 0.063; 95% CI: 0.038-0.088 with WHO), despite moderate positive correlations. CONCLUSIONS: Major CV risk algorithms classify Indian AMI patients very differently, with substantial potential for misclassification. Although PREVENT and FRS showed nearly similar classification of high risk, 80% of patients were still not identified as high risk before their event, underscoring the need for South Asia-specific risk score development and validation and alternative preventive strategies.
No takes yet. Share an insight, caveat, or question.
Gupta et al. (2026) conducted a cohort in First episode of acute myocardial infarction (AMI) (n=4,975). PREVENT 10-year cardiovascular risk prediction score vs. Framingham Risk Score (FRS), ACC/AHA ASCVD 2013, WHO CVD risk charts, and JBS-3 score was evaluated on Proportion of patients classified as high risk (≥20% 10-year risk) by each model before their AMI event (p=<0.001). PREVENT classified 19.8% of first-AMI patients as high risk compared to 20.2% with FRS, 15.0% with WHO, 12.3% with ASCVD 2013, and 11.7% with JBS-3, with poor intermodel agreement.
Synapse has enriched 10 related papers. Consider them for comparative context: