Frailty and Dementia Among Middle‑Aged and Older Adults: Evidence from the Global Burden of Disease 2021 and the China Health and Retirement Longitudinal Study

Objective This study examined the association between frailty, measured by the Frailty Index (FI), and incident dementia through triangulation of evidence from two independent data sources: the Global Burden of Disease (GBD) Study 2021 and the China Health and Retirement Longitudinal Study (CHARLS). Methods Temporal trends in the global dementia burden were analyzed using GBD 2021 data. The association between FI and incident dementia was evaluated among Chinese adults aged ≥ 45 years using CHARLS (2011–2021). The FI was treated as a continuous variable and further categorized into quartiles representing robustness levels (robust, pre-frail, frail). Multivariable logistic regression, restricted cubic spline (RCS), and receiver operating characteristic (ROC) analyses were performed to assess risk associations and predictive performance. Comprehensive sensitivity analyses were conducted to address potential self-report bias, and subgroup analyses tested effect modification across demographic and clinical strata. Results GBD 2021 analyses revealed a marked global increase in dementia prevalence and disability-adjusted life years (DALYs), with disproportionate burdens among older adults across all socio-demographic regions. Age-standardized rates remained stable in high-SDI countries but increased notably in several low- and middle-SDI regions. In the CHARLS cohort, higher FI levels were significantly associated with increased dementia risk. Multivariable regression showed clear dose–response relationships: pre-frailty increased dementia odds by 22% (OR = 1.22, 95% CI: 0.94–1.58), and frailty by 78% (OR = 1.78, 95% CI: 1.02–3.10; p = 0.04). Each one-unit increase in FI corresponded to an 8.44-fold higher dementia risk (OR = 8.44, 95% CI: 1.45–49.12; p = 0.02), whereas a per-IQR increase was associated with 15% higher odds (OR = 1.15, 95% CI: 1.03–1.30). Quartile analysis indicated threshold-like patterns, with non-significant associations in Q2 (OR = 1.15, 95% CI: 0.83–1.60) and Q3 (OR = 1.14, 95% CI: 0.82–1.58), but a significantly elevated risk in Q4 (OR = 1.67, 95% CI: 1.18–2.36; p = 0.004). RCS analysis confirmed an approximately linear dose–response pattern (P_overall = 0.0365; P_nonlinear = 0.4895), showing progressive risk increases across frailty levels without clear threshold effects. ROC analyses demonstrated incremental improvements in model performance—from FI alone (AUC = 0.547) to demographically adjusted (AUC = 0.732), lifestyle-adjusted (AUC = 0.782), and fully adjusted models (AUC = 0.785)—representing a 43% relative gain in discrimination. Subgroup analyses provided consistent validation across diverse populations, showing significant effect modification by educational attainment and cardiovascular disease status (p_interaction < 0.05). The associations were directionally consistent across demographic and clinical strata, with stronger effects among individuals with cardiovascular disease and intermediate education levels. These findings support frailty as a robust predictor of cognitive decline across heterogeneous populations. Conclusion The FI is a robust, dose-dependent predictor of incident dementia among middle-aged and older Chinese adults. Integrating frailty with demographic and cardiovascular risk factors substantially improves predictive accuracy and supports frailty-based screening to guide targeted prevention and resource allocation in aging populations facing a growing global dementia burden.