Background: Effective risk stratification tools are scarce for intensive care unit patients with atrial fibrillation (AF). Although the hemoglobin glycation index (HGI) predicts poor outcomes in various conditions, its clinical utility for critically ill AF populations is undetermined.
Methods: Data from 10,159 ICU patients with AF were extracted from the MIMIC-IV database. The HGI was calculated as the residual difference between observed and predicted glycated hemoglobin based on fasting glucose levels. Associations with 30-, 90-, and 365-day mortality were assessed using Cox proportional hazards models and restricted cubic splines. Causal mediation analysis was performed to explore underlying pathways.
Results: The lowest HGI quartile (Q1) had significantly higher mortality risks versus Q2 (reference) across 30-day (HR = 1.46), 90-day (HR = 1.42), and 365-day (HR = 1.26) follow-ups. A non-linear, inverse association was confirmed, with risk concentrated at low HGI levels. This association was stronger in patients with elevated BMI (p-interaction < 0.01). Causal mediation analysis revealed that red blood cell distribution width (RDW) significantly mediated the risk of low HGI, explaining 3%-8% of the total effect.
Conclusions: The study demonstrates a non-linear inverse association between HGI and mortality in critically ill AF patients, a relationship partially mediated by RDW and particularly strong in patients with elevated BMI. The findings highlight that the increased risk is primarily concentrated among patients with the lowest HGI values.
Keywords: atrial fibrillation; critical care; hemoglobin glycation index; mortality; risk stratification.
© 2026 The Author(s). Clinical Cardiology published by Wiley Periodicals LLC.