Abstract
OBJECTIVE: To explore the feasibility of serum albumin (Alb) and left ventricular ejection fraction (LVEF) in predicting all-cause death (ACD) in patients with stable coronary artery disease (SCAD). METHODS: Patients with SCAD were divided into 4 groups according to their Alb and LVEF levels: Group A: Alb ≤4 g/dL and LVEF > 50%; Group B: Alb ≤4 g/dL and LVEF ≤50%; Group C: Alb >4 g/dL and LVEF ≤50%; Group D: Alb >4 g/dL and LVEF >50%. The K-M curve and log-rank test were used to compare ACD among the four groups over three years. Receiver operating characteristic (ROC) curves were used to compare the efficacy of predicting ACD among the combination of Alb and LVEF and either Alb or LVEF alone. Cox regression analysis identified the influencing factors of ACD in patients with SCAD and detected the correlation between Alb and LVEF. RESULTS: ACD occurred in 18 (8.9%) of 203 patients with SCAD, with an average follow-up of 26.53 ± 14.34 months. In the Kaplan‒Meier analysis, the risk of ACD in the four groups ranged from high to low: Group B (17.6%) > Group A (26.7%) > Group D (0.9%) > Group C (0%, P < 0.001). The ROC curve showed that the combination of Alb and LVEF (AUC = 0.888) had better predictive value for ACD than either Alb (AUC = 0.879) or LVEF alone (AUC = 0.651), P < 0.001. Multivariate Cox regression analysis showed that Alb ≤4 g/dL predicted ACD events after adjusting for baseline (HR: 12.16, 95% CI: 1.57 to 94.41; P=0.017) and treatment (HR: 19.36, 95% CI: 2.53-147.78, P=0.004). Alb was positively correlated with LVEF (r = 0.22, P=0.002). CONCLUSIONS: Alb combined with LVEF is more effective than a single index in predicting ACD in SCAD and could be used as a new model to judge the prognosis of SCAD.