Joint association of the triglyceride–glucose index and stress hyperglycemia ratio with incidence and mortality risks of new-onset atrial fibrillation during sepsis: a retrospective cohort study

The triglyceride-glucose (TyG) index and stress hyperglycemia ratio (SHR) have been linked to the cardiovascular risks in critical ill patients. However, little is known about the predictive power of the TyG index, SHR and their combination on the incidence and mortality risks of new-onset atrial fibrillation (NOAF) in patients with sepsis. This retrospective study included patients from the Medical Information Mart for Intensive Care (MIMIC)-IV database. Primary outcomes were defined as the incidence and 360-day mortality of in-hospital NOAF among patients with sepsis. Logistic model, Cox proportional hazard model, Kaplan-Meier analysis and receiver-operating characteristic (ROC) were performed to explore the association between the indices and clinical outcomes. Machine learning approach also was constructed to evaluate and compare the indices in predicting mortality risks. 4276 patients meeting the inclusion criteria were enrolled and 764 individuals developed NOAF during hospitalization. The multivariable adjusted odds ratios (95%, CI) of incidence of NOAF in patients with sepsis in the highest group versus the lowest group were 1.36 (1.10-1.69), 1.35 (1.09-1.67) and 1.58 (1.23-2.02), respectively, for the TyG index, SHR and the TyG index-SHR combination. However, the predictive powers of these indices were relatively low. Among septic patients who developed in-hospital NOAF, those in the highest TyG index group and the highest SHR group exhibited an increased risk of 360-day mortality compared with those with the lowest TyG index and the lowest SHR (the TyG index: hazard ratio [HR] 1.59, 95% CI 1.00-2.62; SHR: HR 1.67, 95% CI 1.03-2.70). Patients with both the highest the TyG index and the highest SHR demonstrated the highest risk of 360-day mortality (HR 1.72, 95% CI 1.08-2.72). The ROC also confirmed the TyG index-SHR combination had more robust predictive power for 360-day mortality among septic patients with NOAF than the TyG index and SHR itself (p < 0.05). The random forest model validated that the predictive capability was significantly enhanced with the integration of the TyG index and SHR. The TyG index and SHR were associated with the incidence of in-hospital NOAF during sepsis, although their predictive powers were limited. In septic patients with in-hospital NOAF, high levels of the TyG index and SHR were significantly associated with increased 360-day mortality risks, with their combination demonstrating superior predictive power. Joint assessments of the TyG index and SHR could help identify individuals at high risks of mortality post-discharge, enabling clinicians to prioritize follow-up care and improve patient management.