Agreement Between Cardiovascular Risk Assessment Models (SCORE2, Framingham, ASCVD-2013, and SCORE2-OP) in Adults Aged 70 Years or Older: A Population-Based Study

Background: Despite several prediction models to estimate cardiovascular risk (CVR) being available, most have been developed and validated in populations under 70 years of age. To improve this estimation in older adults aged 70 years or older, the SCORE2-OP model was developed; however, its application outside Europe may under- or overestimate true CVR. Objective: To assess the concordance between the CVR measured by 4 widely used CVR tools (Framingham, ASCVD-2013, SCORE2) and the SCORE2-OP in older adults aged ⩾ 70 years. Methods: Secondary analysis of adults aged ⩾ 70 years from the SABE Colombia study, which was conducted between April and September 2015. The concordance between the 4 different CVR prediction models and SCORE2-OP was assessed via Cohen's quadratically weighted kappa coefficient and Lin's concordance correlation coefficient. Results: Among the 23 694 participants in the SABE Colombia study, 23 108 were excluded due to insufficient data to estimate the CVR for any of the tools evaluated. 586 individuals met the inclusion criteria and were analyzed. The concordance between the ASCVD-2013 (weighted kappa 0.24; 95% CI 0.22-0.26) and Framingham algorithms (0.22; 95% CI 0.20-0.24) calibrated to the Colombian population and SCORE2-OP were both classified as fair. In contrast, agreement between SCORE2 calibrated for Colombia and SCORE2-OP reached a moderate level (0.43; 95% CI 0.40-0.46). Higher values were observed compared with the SCORE2-OP model developed for intermediate-risk regions. Conclusions: Our results suggest that, among adults aged ⩾ 70 years, there is variable agreement between widely used CVR prediction models and SCORE2-OP, with the latter overestimating CVR compared to nationally validated CVR scores. This highlights the need for region-specific validation to ensure accurate estimation of CVR in this population.