Frailty modifies the association between long-term fasting glucose variability and incident type 2 diabetes in older adults

Background: Type 2 diabetes (T2D) is a major contributor to morbidity in older adults, yet accurate identification of high-risk individuals remains difficult. Although long-term fasting plasma glucose (FPG) variability predicts incident T2D, whether frailty modifies this association is unclear. We examined this in a large elderly cohort. Methods: We conducted a population-based retrospective cohort study including 71,849 Chinese adults aged ≥65 years without diabetes at baseline. Frailty was assessed using an electronic Frailty Index (eFI) derived from routinely collected electronic health records. Long-term FPG variability was quantified using the standard deviation (SD), coefficient of variation (CV), and variability independent of the mean (VIM) from three serial health examinations. Incident T2D was identified through electronic health records during follow-up. Cox proportional hazards models were used to examine associations and test interactions between frailty and glucose variability. Results: During a median follow-up of 3.15 years, 4,635 participants developed T2D. Higher long-term FPG variability was independently associated with increased T2D risk. Frailty significantly modified this association. Each standard deviation increase in FPG-SD was associated with a 24% higher risk of incident T2D among non-frail participants (HR 1.24, 95% CI 1.21–1.27), compared with a 13% higher risk among frail participants (HR 1.13, 95% CI 1.02–1.25). Similar interaction patterns were observed for FPG-CV and FPG-VIM. Conclusions: Frailty substantially influences the predictive value of long-term glucose variability for incident T2D in older adults. These findings suggest that frailty should be incorporated into glycaemic risk stratification and support a more individualized approach to diabetes prevention in ageing populations.