Abstract
BACKGROUND: Stress hyperglycemia ratio (SHR) has been associated with poor outcomes in patients with acute myocardial infarction (AMI). Modified SHR (mSHR) is defined as the highest SHR observed within the first 24 h of admission. However, the association between mSHR and early cardiovascular complications (ECC) following AMI is unclear. METHODS: This multicenter observational study incorporated retrospective and prospective analyses across two independent cohorts. The discovery and validation cohorts each included consecutive AMI patients admitted to intensive care units. The blood glucose and hemoglobin A1c levels were used to calculate mSHR. The primary outcome was the occurrence of ECC during the hospital stay. RESULTS: In the discovery cohort, ECC occurred in 322 (23.9%) of 1,349 patients, and mSHR was independently associated with ECC (adjusted odds ratio: 1.164; 95% confidence interval (CI): 1.124-1.206; P < 0.001). Machine learning approaches identified mSHR as the most important feature. In the validation cohort, 303 patients were divided into three groups according to mSHR tertiles. Modified Poisson regression analysis showed that patients with mSHR ≥ 1.202 (tertile 3) had a significantly higher risk of ECC compared to those in tertiles 1-2 (adjusted risk ratio: 2.337; 95% CI: 1.479-3.693; P < 0.001). The results of multivariate analysis were consistent before and after applying inverse probability of treatment weighting. CONCLUSION: In AMI patients, mSHR is an accurate risk-stratification tool for identifying ECC. It may provide exploratory evidence to optimize current glycemic control strategies and early discharge pathways.