The U shape relationship between glucose and potassium ratio and mortality in patients with subarachnoid hemorrhage in the US population
摘要
The prognostic value of the glucose-potassium ratio (GPR) in non-traumatic subarachnoid hemorrhage (SAH) remains undetermined. We investigated the association between the GPR at admission and all-cause mortality (ACM) in critically ill patients with SAH. We identified critically ill patients with SAH from the Medical Information Mart for Intensive Care database and stratified them into quartiles based on the GPR levels at admission. To evaluate mortality risk associations, we employed Cox proportional hazards models along with restricted cubic splines (RCS) to assess non-linear relationships. Survival curves were generated using the Kaplan–Meier (K–M) method. The robustness of the results was assessed through prespecified subgroup analyses and interaction tests, with effect modifications evaluated using likelihood ratio testing. The study cohort comprised 855 patients (median age: 61 years), with cumulative ACM rates of 18.5% at 30 days, 22.7% at 90 days, and 26.4% at 1 year. Cox regression analysis results revealed that higher GPR was significantly related to ACM at 30 days (hazard ratio (HR): 1.42; 95% confidence interval (CI): 1.13–1.80), 90 days (HR: 1.31; 95% CI 1.05–1.64), and 1 year (HR: 1.25; 95% CI 1.00–1.54). RCS analysis revealed a non-linear U-shaped association with an inflection point at GPR = 2.3. Below this threshold, GPR revealed no mortality association (HR: 0.96, 95% CI 0.57–1.63), while values above exhibited progressive risk elevation (HR: 1.67, 95% CI 1.25–2.22). Subgroup analyses confirmed consistent associations across patient characteristics (all interactions, p > 0.05). Moreover, the combination of GPR and GCS performed better than GPR and GCS alone in predicting ACM. A U-shaped relationship was found between GPR and mortality in critically ill patients with SAH. This easily available biomarker holds potential for risk stratification and clinical decision-making, though optimal thresholds require prospective validation.