Frailty, Insulin Resistance, and Stroke Risk: A Prospective Cohort Study and Mendelian Randomization Analysis

Stroke is currently the second leading cause of death worldwide, especially among aging populations. Frailty and insulin resistance (IR) have both been associated with stroke risk, but their longitudinal relationship and potential mechanistic links remain unclear.

This prospective study assessed 8,582 Chinese adults aged ≥ 50 years in the China Health and Retirement Longitudinal Study (CHARLS) from 2011 to 2018. Frailty was assessed using a validated 29-item frailty index (FI), and IR was estimated by the triglyceride-glucose (TyG) index. Incident stroke was identified by self-reported physician diagnosis. Cox proportional hazards models estimated associations of baseline frailty, frailty transitions, and TyG with incident stroke. Mediation analysis explored whether TyG statistically accounted for part of the frailty-stroke association. Exploratory two-sample Mendelian randomization (MR) was further conducted using European-ancestry GWAS datasets.

During follow-up, 584 (6.8%) participants developed stroke. Baseline frailty independently predicted stroke (HR 1.94, 95% CI 1.58-2.39). Compared with participants who remained robust, stroke risk was higher among those with worsened frailty (HR 2.56, 95% CI 1.78-3.67), stable frailty (HR 3.42, 95% CI 2.70-4.34), or improved frailty (HR 1.55, 95% CI 1.21-1.98). Higher TyG levels were associated with stroke risk (Q4 vs. Q1: HR 1.97, 95% CI 1.52-2.55). TyG statistically mediated 4.5% of the baseline frailty-stroke association and 3.4% of the association between frailty worsening and stroke. Exploratory MR analyses based on European- ancestry GWAS datasets yielded directionally consistent genetic evidence (pooled IVW OR 1.41, 95% CI 1.13-1.77).

Frailty was associated with incident stroke both at baseline and through longitudinal transitions, with the highest risks observed among participants with stable or worsening frailty. Participants with improved frailty remained at higher risk than those who remained robust, although their lower risk estimate compared with stable frail participants may suggest partial risk attenuation and requires cautious interpretation. The small mediation proportion by TyG suggests that IR explains only a minor fraction of the frailty-stroke association, with most of the association likely operating through non-TyG pathways. These findings support careful consideration of frailty assessment and metabolic profiling within multidomain stroke prevention strategies for older adults.

Frailty, particularly persistent or worsening frailty, was associated with increased stroke risk among middle-aged and older adults. TyG explained only a small fraction of this association, while exploratory MR findings require cautious interpretation. Frailty and metabolic assessment may help identify individuals at elevated cerebrovascular risk.