Key result
Adapted Framingham model using self-reported hypertension accurately reproduces original risk scores with ~0.87 concordance.
Why the study?
Do adapted Framingham Risk Score models using self-reported data on hypertension, dyslipidemia, and diabetes accurately reproduce 10-year cardiovascular risk classification compared to the original Framingham Risk Score in Brazilian adults aged 30–74 years?
Cross-Sectional (n=6,555)
Yes
Do adapted Framingham Risk Score models using self-reported data on hypertension, dyslipidemia, and diabetes accurately reproduce 10-year cardiovascular risk classification compared to the original Framingham Risk Score in Brazilian adults aged 30–74 years?
Adapted Framingham Risk Score models utilizing self-reported data on hypertension, dyslipidemia, and diabetes demonstrate high discrimination and agreement with the original laboratory-based score, supporting their utility in large-scale epidemiological surveys where clinical and biochemical measurements are unavailable.
Answers are grounded in this paper and cite their sources.
May enable CVD risk estimation in surveys without labs; leaves open prospective validation before clinical use.
Objective To adapt the Framingham Risk Score (FRS) for the general population and for adults with higher education (HE), and to evaluate its performance and agreement with the original model using self-reported data from the Brazilian National Health Survey (PNS). Methods We analyzed data from 6,555 PNS participants aged 30–74 years. Three adapted models replacing clinical and laboratory variables with self-reported data were evaluated using the original FRS as reference: FRS-SAH, FRS-DLP-DM, and FRS-DLP-SAH-DM. Model performance across cardiovascular risk cutoffs (≥ 5%, ≥ 10%, and ≥ 20%) was assessed using ROC/AUC analysis, Brier score evaluation, intraclass correlation coefficients (ICC), Bland–Altman agreement, and bootstrap analysis. Results: In the total population, FRS-SAH showed the highest capacity to reproduce the original score across all cutoffs (AUCs: 0.963 at ≥ 5%, 0.954 at ≥ 10%, and 0.917 at ≥ 20%) with the lowest Brier scores (0.4110, 0.2443, and 0.1375). FRS-DLP-DM exhibited strong performance (AUCs: 0.967, 0.952, and 0.914), whereas FRS-DLP-SAH-DM presented slightly lower discrimination at higher cutoffs (AUCs: 0.928, 0.902, and 0.834). High continuous agreement was observed across models, with FRS-SAH demonstrating the highest ICC. Mean bias was + 0.42% for FRS-SAH, + 2.25% for FRS-DLP-DM, and + 2.65% for FRS-DLP-SAH-DM, with Bland-Altman plots indicating proportional bias at risk levels above 15–20%. All models demonstrated high internal stability in bootstrap analysis and maintained good discrimination. Conclusion The adapted FRS models using self-reported data showed high agreement compared with the original score for both the general population and the highly educated group. These findings support the use of simplified versions of the FRS in epidemiological studies and surveillance contexts in which clinical and biochemical measures are not available. Clinical trial number Not applicable.
No takes yet. Share an insight, caveat, or question.
Rocha et al. (2026) conducted a cross-sectional in Cardiovascular disease risk estimation (n=6,555). Adapted Framingham Risk Score models using self-reported data (FRS-SAH, FRS-DLP-DM, and FRS-DLP-SAH-DM) vs. Original Framingham Risk Score based on measured blood pressure, total cholesterol, HDL cholesterol, and HbA1c was evaluated on Agreement and discriminatory performance reproducing original 10-year Framingham Risk Score across cutoffs (≥5%, ≥10%, ≥20%) (MD -0.42 points, p=0.272). The adapted FRS-SAH model using self-reported hypertension data reproduced the original Framingham score with high discrimination (AUCs 0.917–0.963) and an ICC of 0.867.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: