Platelet–albumin ratio predicts 28-day mortality risk in ICU heart failure patients: A retrospective cohort study

Heart failure (HF) in intensive care unit (ICU) patients carries high mortality, necessitating reliable prognostic biomarkers. This study evaluates the platelet-albumin ratio (PAR) as a predictor of 28-day mortality. A retrospective analysis of 6490 ICU HF patients was conducted. Baseline characteristics were compared between survivors and non-survivors. Cox regression identified independent mortality predictors. Machine learning models (LASSO and Boruta) assessed PAR's importance. Restricted cubic spline analysis determined the optimal cutoff. Predictive performance was evaluated via receiver operating characteristic and decision curve analysis. External validation was performed using the MIMIC-III database. PAR independently predicted 28-day mortality (hazard ratio = 1.113, 95% confidence interval: 1.101-1.124, P < .001). The optimal cutoff was 5.97, with higher PAR associated with lower survival probability. Receiver operating characteristic analysis yielded an area under the curve of 0.733 (95% confidence interval: 0.718-0.748). Decision curve analysis confirmed its clinical utility. Sensitivity analysis demonstrated robust predictive performance across subgroups. External validation reaffirmed these findings, showing similar survival trends and predictive accuracy. PAR is a clinically useful predictor of 28-day mortality in ICU HF patients. Its predictive power, confirmed through statistical and machine learning approaches, highlights its potential for risk stratification and clinical decision-making.