Association Between Non-Cardiovascular Frailty Trajectories and Incident Stroke in Middle-Aged and Older Chinese Women: A National Longitudinal Cohort Study

Background Frailty is associated with adverse cardiovascular outcomes, yet whether longitudinal trajectories of non-cardiovascular deficit accumulation predict incident stroke independently of baseline risk factors remains unclear. We examined the association between frailty trajectories derived from repeated non-cardiovascular frailty index (FI) measurements and incident stroke in a nationally representative cohort of Chinese women. Methods We used the China Health and Retirement Longitudinal Study (CHARLS, 2011–2018). A 14-item non-cardiovascular FI (six activities of daily living, five non-cardiovascular chronic conditions, three cognitive tests) was constructed at Waves 1–3. Individual linear trajectories were estimated and grouped by K-means clustering (k = 4). The primary outcome was incident self-reported physician-diagnosed stroke between Waves 3 and 4 (2015–2018), assessed in women free of cardiovascular disease at Waves 1–3. Cox proportional hazards models adjusted for age, education, marital status, residence, smoking, alcohol use, and body mass index. Sensitivity analyses included composite cardiovascular disease, heart disease, an 18-item full FI, alternative cluster numbers, competing mortality, and E-values for unmeasured confounding. Mediation by baseline hypertension and diabetes was explored. Results Among 4,954 women (mean age 58.9 years), four trajectories emerged: Low progressing (n = 2,349, 47.4%), Moderate stable (n = 2,201, 44.4%), High stable (n = 567, 11.4%), and Rapidly progressing (n = 937, 18.9%). Three-year stroke incidence ranged from 2.7% (Low progressing) to 7.2% (High stable). In fully adjusted models (reference: Low progressing), High stable was associated with higher stroke risk (HR 1.57, 95% CI: 1.05–2.37), and Rapidly progressing showed the strongest association (HR 1.83, 95% CI: 1.31–2.55). The gradient persisted for composite cardiovascular disease and heart disease, and was robust to the 18-item full FI (rapid HR 1.65), k = 3 (HR 1.60), and k = 5 (HR 1.47). Baseline hypertension and diabetes mediated only 2.4% and 1.2% of the rapid-progression association, respectively. An unmeasured confounder would need a risk ratio exceeding 3.0 to explain the primary association. Conclusions In Chinese women, trajectories of non-cardiovascular deficit accumulation were associated with incident stroke in a graded pattern, largely independent of baseline hypertension and diabetes. Serial frailty assessment may help identify women at elevated stroke risk, though the modest effect sizes and observational design preclude causal inference.