**Background:** Cardiovascular diseases (CVD) remain the leading cause of global disease burden, with prevalence nearly doubling from 1990 to 2019. Primordial prevention—preventing risk factors before they develop—is increasingly recognized as key. The American Heart Association's Life's Simple 7 (LS7) cardiovascular health (CVH) score, comprising four behavioral metrics (non-smoking, BMI, physical activity, diet) and three biological metrics (fasting glucose, blood pressure, total cholesterol), is a validated tool for monitoring population CVH. However, widespread use is limited by the need for clinical examinations and blood tests, which are particularly challenging in low- and middle-income countries (LMICs). Person-centered approaches, where individuals self-report their health status, could overcome this barrier, but their validity compared to measured CVH and their association with incident CVD had not been directly compared.
**Methods:** The study included 18,714 CVD-free adults from the NutriNet-Santé e-cohort (launched 2009 in France), who underwent clinical examinations and blood draws between 2011 and 2014. Participants completed validated online questionnaires on smoking, physical activity (IPAQ), diet (three 24-h dietary records), and medical history. Measured biological metrics were defined by clinical thresholds: fasting plasma glucose <100 mg/dL, blood pressure <120/80 mmHg, and total cholesterol <200 mg/dL, all untreated. Person-centered biological metrics were defined as absence of self-reported diagnosis, medication, or treatment for type 2 diabetes, hypertension, or hypercholesterolemia. Behavioral metrics were identical for both approaches. Concordance was assessed using weighted Gwet AC2 and Brennan-Prediger coefficients, sensitivity, specificity, PPV, NPV, and diagnostic accuracy. Cox proportional hazard models (age as timescale, stratified by birth year, adjusted for sex, cohabitation, education, occupation, alcohol use, CVD family history) estimated hazard ratios for incident CVD per additional ideal metric and for CVH categories (low: 0–2, intermediate: 3–4, high: 5–7 ideal metrics). Multiple imputation handled missing data.
**Key Results:** The study population had a mean age of 51 years (73% women). According to measured CVH, 16.52% had high CVH (5–7 ideal metrics), compared to 38.75% by person-centered CVH. Weighted concordance was high: Gwet AC2 = 0.92 [0.91; 0.93]; Brennan and Prediger = 0.87 [0.86; 0.88]. Sensitivity of person-centered CVH for identifying ideal measured CVH was 96.94% [96.28; 97.49], specificity 72.83% [72.12; 73.52], PPV 41.52% [40.39; 42.66], NPV 99.17% [98.99; 99.32], diagnostic accuracy 76.83% [76.22; 77.43]. For individual biological metrics, sensitivity ranged from 93.44% (hypertension) to 99.87% (T2D); specificity ranged from 15.37% (T2D) to 39.89% (hypertension); diagnostic accuracy was 88.29% for T2D, 72.51% for hypertension, and 46.86% for hypercholesterolemia. Over median 8.05 years follow-up, 749 incident CVD events occurred (incidence rate 5.31 per 1000 person-years). Per additional ideal metric, risk reduction was 7% for measured CVH (HR 0.93 [0.88; 0.99]) and 13% for person-centered CVH (HR 0.87 [0.83; 0.92]). High vs. low CVH showed 29% risk reduction for measured (HR 0.71 [0.53; 0.95]) and 40% for person-centered (HR 0.60 [0.49; 0.74]). No sex interaction was found. Sensitivity analyses (excluding first-year CVD, complete case, competing risk) yielded consistent results. Change analysis in 126,871 participants over 5 years showed that each additional ideal person-centered metric was associated with 13% CVD risk reduction independent of baseline CVH.
**Clinical Implications:** Person-centered CVH demonstrates high concordance with measured CVH and shows a consistent, albeit slightly stronger, inverse association with incident CVD. The overestimation of risk reduction (13% vs. 7%) is likely due to misclassification of biological metrics, particularly hypercholesterolemia (diagnostic accuracy only 46.86%). Despite this, person-centered CVH offers a practical, economical, and scalable tool for primordial prevention, especially in LMICs and underserved areas where clinical resources are limited. The approach empowers individuals to self-monitor their cardiovascular health and may facilitate population-level screening and intervention targeting. The estimated population attributable preventive fraction suggests 15.54% of CVD events could be prevented if all individuals achieved high person-centered CVH. Limitations include overrepresentation of women and higher socioeconomic status participants, single-time measurement of biological metrics, and conduct in a high-resource setting, which may limit generalizability to LMICs.