Abstract
BACKGROUND: Previous observations have reported inconsistent results on the association between low-density lipoprotein cholesterol (LDL-C) at presentation and long-term outcomes after acute myocardial infarction (AMI). We aimed to clarify the potential impact of baseline characteristics on the inverse association between LDL-C and all-cause mortality, known as the lipid paradox. METHODS: A total of 1,305 critically ill patients with AMI from the Medical Information Mart for Intensive Care IV database were included in the analysis. Patients were stratified according to LDL-C quartiles. The primary outcome was 180-day and 360-day all-cause mortality. Baseline characteristics were included in stepwise Cox regression models. Restricted cubic spline analyses across multiple models and subgroups were performed to assess the influence of baseline characteristics on the association between LDL-C and long-term outcomes. RESULTS: A total of 244 (18.7%) and 291 (22.3%) mortality events occurred at 180 and 360 days of follow-up, respectively. Patients in the lowest LDL-C quartile had the highest all-cause mortality at both 180 and 360 days (28.7% and 35.2%, respectively). After stepwise adjustment for baseline covariates, the J-shaped relationship observed in the unadjusted model was gradually attenuated and disappeared. The inverse association between LDL-C and mortality was more pronounced in subgroups characterized by elevated mortality risk, including patients with low albumin levels, elevated neutrophil-to-lymphocyte ratio, and higher SOFA scores. Nevertheless, in-hospital statin use was consistently associated with reduced all-cause mortality across nearly all subgroups. CONCLUSIONS: The lipid paradox observed in critically ill patients with AMI is attributed to differences in baseline characteristics across LDL-C strata. After adjusting for potential confounders, baseline LDL-C was not an independent predictor of long-term mortality in AMI. Lipid-lowering therapy was associated with favorable long-term outcomes irrespective of baseline LDL-C levels. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12944-026-02871-z.