Abstract / Summary
Abstract Background Glycated hemoglobin (HbA1c) reflects long-term average glycemia, yet stable inter-individual differences in hemoglobin glycation exist. The hemoglobin glycation index (HGI) quantifies the gap between measured HbA1c and the HbA1c predicted from blood glucose. Whether HGI predicts incident stroke in the general population, particularly independently of HbA1c, remains unclear. Methods Within the China Health and Retirement Longitudinal Study, 10,643 stroke-free participants aged 45 years or older with baseline (2011) fasting plasma glucose and HbA1c were followed to 2020. The outcome was incident self-reported physician-diagnosed stroke. HGI equaled measured HbA1c minus the value predicted from a baseline-population regression of HbA1c on fasting plasma glucose; the primary exposure was high HGI (at or above the 90th percentile). We fitted sequentially adjusted Cox proportional hazards models, including a prespecified HbA1c-adjusted key test; restricted cubic splines assessed the dose–response relationship; a Fine–Gray model addressed the competing risk of death; and discrimination was compared against the stress hyperglycemia ratio, triglyceride–glucose index, HbA1c, and fasting plasma glucose. Eight prespecified sensitivity analyses were performed. Results Over a median follow-up of about 9 years, 841 incident strokes occurred (7.9%). In the complete-case main analysis (n = 8,684; 699 events), high HGI was associated with increased stroke risk (adjusted hazard ratio 1.392, 95% confidence interval 1.122–1.728, P = 0.003) and persisted after HbA1c adjustment (hazard ratio 1.333, 1.018–1.746, P = 0.037), meeting the prespecified criterion. Nonlinearity was suggested overall (P = 0.060) and significant among participants without diabetes (P = 0.017). Competing-risk results were consistent (subdistribution hazard ratio 1.386, 1.119–1.716). The association was stronger among participants without diabetes (hazard ratio 1.478, 1.118–1.954). All sensitivity analyses were directionally consistent, and no elevated risk at low HGI was observed after excluding anemia. Discrimination did not differ significantly among the five indices (C-index 0.658–0.663). Conclusions High HGI was independently associated with incident stroke in middle-aged and older Chinese adults, beyond HbA1c and more pronounced in those without diabetes. As a trait-like marker of glycation propensity computable from routine measurements, HGI may help identify individuals at elevated stroke risk whom HbA1c does not flag, although its stand-alone predictive value is limited.