Frailty Index Dynamic Changes and Incident Cardiovascular Disease in Middle-Aged and Older Adults: Dose-Response, Effect Modification, and Clinical Translation for Primary Care Screening — A Nationwide Prospective Cohort Study

<title>Abstract</title> <p>Background The frailty index (FI) is associated with cardiovascular disease (CVD) risk, yet previous studies predominantly relied on single measurements, lacking quantitative assessments of dynamic frailty changes and clear clinical translation pathways for primary care screening. This study aimed to evaluate the associations of baseline FI and dynamic FI changes with CVD, and to develop a simplified screening tool for primary care. Methods Based on CHARLS data from 2011–2020, 7,999 middle-aged and older adults (≥ 45 years) without baseline CVD were included, with a median follow-up of 7.0 years. The FI comprised 41 health deficits. Cox models, restricted cubic splines, Fine-Gray competing risk models, Shapley value decomposition, LASSO cross-validation, Bootstrap optimism correction, and decision curve analysis were used to evaluate baseline FI-CVD associations. Logistic regression, NRI, and IDI assessed the incremental predictive value of FI changes. Results During follow-up, 1,516 incident CVD events (18.95%) occurred. Each 0.1 increase in baseline FI was associated with a 42% increased CVD risk (HR = 1.42, 95% CI: 1.374–1.468). Restricted cubic splines revealed a significant non-linear relationship (P &lt; 0.0001), with an optimal threshold of 0.24. Shapley decomposition showed non-cardiovascular deficits contributed 15.8% of independent predictive value. Adding FI change classification increased AUC from 0.634 to 0.761 (ΔAUC = + 0.127), with IDI of 10.89% and NRI of 20.18% (P &lt; 0.001). Ten-fold LASSO identified seven core variables (hypertension, dyslipidemia, asthma, high waist circumference, poor self-rated health, low back pain, arthritis) with selection frequencies ≥ 80%. The simplified FI showed a Bootstrap-corrected AUC of 0.646. Decision curve analysis demonstrated clinical net benefit within the 11%–30% threshold range. Conclusions Baseline FI exhibits a significant non-linear dose-response relationship with CVD risk. Dynamic FI changes provide substantial incremental predictive value (ΔAUC = + 0.127, IDI = 10.89%), supporting incorporation of dynamic frailty monitoring into cardiovascular risk stratification. The seven-item simplified FI (assessment time &lt; 2 minutes, AUC = 0.615–0.646) is a practical, cost-effective primary care screening tool for population health applications. Trial registration Not applicable (observational cohort study).</p>