Baseline characteristics.

Background
The purpose of this article is to assess the relationship between serum albumin level and long length of stay (LOS) of inpatients with acute heart failure (AHF) in the intensive care unit (ICU).
Methods
We retrospectively analyzed data of 2280 patients with AHF from the medical information mart for intensive care IV (the MIMIC-IV) database. Multivariate logistic regression was performed to evaluate the association between serum albumin and long LOS, and the development of the predictive model was based on independent predictors of long LOS.
Results
According to the statistical results, A negative linear relationship was presented between albumin and long LOS of AHF patients in the ICU (P for trend <0.001), and serum albumin could predict long LOS (AUC 0.649, 95%CI 0.616–0.683, P <0.001). Based on independent predictors, including respiratory failure (OR 1.672, 95%CI 1.289–2.169, P<0.001), WBC (OR 1.046, 95%CI 1.031–1.061, P<0.001), creatinine (OR 1.221, 95%CI 1.098–1.257, P<0.001), glucose (OR 1.010, 95%CI 1.007–1.012, P<0.001), lactic acid (OR 1.269, 95%CI 1.167–1.381, P<0.001), and albumin (OR 0.559, 95%CI 0.450–0.695, P<0.001), identified by multivariable logistic regression analysis, we developed the nomogram to predict the probability of long LOS of AHF patients in the ICU. The nomogram accurately predicted the probability of long LOS (AUC 0.740, 95%CI 0.712–0.768, P<0.001). The calibration suggested the predictive probability was highly consistent with the actual probability of long LOS. Decision curve analysis (DCA) also suggested that the nomogram was applicable in the clinic.
Conclusion
Serum albumin level was negatively associated with LOS among AHF patients. The predictive model based on serum albumin has predictive value for evaluating the length of stay in AHF patients.