Association between atherogenic index of plasma and ICU 28-day mortality in geriatric diabetes: an EICU-based retrospective analysis

BACKGROUND: The Atherogenic Index of Plasma (AIP), defined as log10 (triglycerides/high-density lipoprotein cholesterol), is a well-established indicator of adverse metabolic profiles and cardiovascular risk. While its role in chronic cardiovascular disease is established, its prognostic value in acute critical illness among elderly diabetic patients remains underexplored. METHODS: This multicenter retrospective cohort study analyzed data from the eICU Collaborative Research Database (eICU-CRD) across U.S. hospitals (2014–2015). Elderly diabetic patients (≥ 60 years) admitted to the ICU were included (n = 7,873), excluding those with non-first ICU admissions, ICU stays < 48 h, missing AIP components, or incomplete mortality records. Admission AIP values were analyzed. Multivariable-adjusted regression, generalized additive models (GAM), and threshold analyses assessed the AIP–28-day mortality association, with covariate adjustment for demographics, comorbidities, and physiological markers. RESULTS: Among 7,873 elderly diabetic ICU patients, 28-day mortality was 4.8% overall, with mortality increasing across AIP tertiles (T1:4.0%, T2:4.5%, T3:6.4%; P < 0.001). Multivariable analysis demonstrated a nonlinear threshold effect at AIP = 0.1: above this threshold, each unit increase conferred 1.9-fold higher mortality risk (95% CI:1.3–2.7, P < 0.001), persisting after full adjustment for physiological/clinical confounders. CONCLUSION: ICU admission AIP independently associates with mortality risk in elderly diabetic patients, with a threshold value of 0.1 identifying individuals at heightened risk. These findings implicate atherogenic dyslipidemia in critical illness outcomes. Prospective validation of this threshold and investigation of underlying lipid dysregulation-mortality mechanisms are warranted.