# Prognostic utility of the lactate-to-albumin ratio for predicting 28-day all-cause mortality in critically ill cases with acute sepsis: A retrospective study on the basis of MIMIC-IV critical care database

## TL;DR

This study shows that the lactate-to-albumin ratio (LAR) is a better predictor of 28-day mortality in sepsis patients than existing models like SOFA.

## Contribution

The study introduces LAR as a novel, objective biomarker for early mortality risk assessment in sepsis patients.

## Key findings

- LAR was significantly higher in non-survivors compared to survivors (median 0.9 vs. 0.6).
- LAR outperformed lactate, albumin, and SOFA score in predicting mortality (AUC of 64.71%).
- LAR was confirmed as an independent predictor of 28-day mortality (OR = 1.32).

## Abstract

Sepsis constitutes a systemic dysregulated host response to infection and remains a predominant cause of ICU mortality globally. Given the limitations of conventional prognostic models (e.g., SOFA and APACHE II), incorporating variably subjective parameters, there is a pressing need to identify robust, objective biomarkers for early mortality risk stratification. This investigation delineated the prognostic significance of the lactate-to-albumin ratio (LAR) in predicting 28-day all-cause mortality (28-DACM) among critically ill septic cases.

We performed a retrospective analysis utilizing the MIMIC-IV database (2008-2019), comprising 5,398 adult cases who met Sepsis-3 diagnostic criteria. Clinical and laboratory data within the initial 24-h post-ICU admission were extracted. The LASSO regression algorithm was implemented as a regularization technique to mitigate multicollinearity, enhance model generalizability, and facilitate high-dimensional feature selection. It was made to evaluate the prognostic utility of LAR through Kaplan-Meier (KM) survival estimation, receiver operating characteristic (ROC) curve analysis, and multivariate logistic regression modeling.

LAR values were remarkably escalated in non-survivors relative to survivors (median, 0.9 vs. 0.6; P < 0.001). ROC curve analysis unveiled that LAR outperformed lactate (AUC: 63.52% ), albumin (AUC: 43.34% ), and the SOFA score (AUC: 59.87% ), achieving the highest discriminatory capacity (AUC: 64.71% ; 95% CI: 62.85-66.58%). An optimal LAR threshold of 1.032 was identified, attaining sensitivity and specificity of 45.1% and 76.6% , respectively. KM analysis uncovered remarkably attenuated 28-day survival in cases with LAR >1.032 (P < 0.001). Multivariate logistic regression confirmed LAR as an independent predictor of 28-DACM (OR = 1.32; P < 0.001), following adjusting for confounding variables.

The LAR serves as a clinically accessible, objective biomarker with superior prognostic performance relative to established indicators in association with sepsis. Its integration into early risk assessment algorithms may enhance prognostication and inform timely therapeutic decision-making. Prospective, multicenter investigations are warranted to validate its external generalizability and clinical utility.