.png){kind=icon}
.png){kind=icon}
Return
Lactate dehydrogenase-to-albumin ratio predicts mortality in mechanically ventilated patients with CAP-associated sepsis: a MIMIC-IV study with external validation
.png){kind=icon}
OA
AI
K
T
X
R
H
L
J
Q
DOI：10.1186/s40001-026-05037-w.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Community-acquired pneumonia (CAP) complicated by sepsis is associated with high mortality, particularly among critically ill patients requiring mechanical ventilation. Reliable and easily obtainable biomarkers for early risk stratification remain limited. This study aimed to evaluate the prognostic value of the lactate dehydrogenase-to-albumin ratio (LDAR) for predicting mortality in mechanically ventilated patients with CAP-associated sepsis. This retrospective cohort study included 866 mechanically ventilated patients with CAP-associated sepsis from the MIMIC-IV database, with external validation using an independent cohort of 232 patients from Hebei Provincial People's Hospital. Patients were stratified according to LDAR quartiles. Cox proportional hazards models and logistic regression analyses were performed to assess the association between LDAR and mortality outcomes. Restricted cubic spline analysis was used to explore the dose–response relationship between LDAR and mortality. Kaplan–Meier survival analysis, subgroup analyses, and receiver operating characteristic (ROC) curve analysis were conducted to evaluate prognostic performance. Machine learning models were further developed, and model interpretability was assessed using SHAP analysis. Higher LDAR levels were significantly associated with increased risks of both 28-day ICU mortality and in-hospital mortality. In the multivariable Cox model, patients in the third and fourth LDAR quartiles had significantly higher risks of 28-day mortality compared with those in the lowest quartile (HR 1.85, 95% CI 1.22–2.81; HR 1.68, 95% CI 1.09–2.58, respectively). Logistic regression analysis showed similar associations with hospital mortality. Restricted cubic spline analysis demonstrated a positive relationship between LDAR and mortality risk. Kaplan–Meier analysis revealed significantly lower survival probabilities in patients with higher LDAR levels. ROC analysis indicated that incorporating LDAR into the baseline clinical model improved predictive performance (AUC 0.688). External validation confirmed the robustness of these findings (AUC 0.724). Machine learning analysis identified LDAR as one of the most important predictors of mortality. LDAR is an independent and readily available biomarker associated with mortality risk in mechanically ventilated patients with CAP-associated sepsis. Incorporating LDAR into conventional clinical models may improve early risk stratification and prognostic assessment in critically ill patients. Clinical trial number This article is a retrospective study on the MIMIC database. Data from Hebei Provincial People's Hospital were incorporated into this study as an external validation data source for this dataset, in accordance with 2025-LW-0170, to get patient consent.
Keywords:
Community-acquired pneumonia
Sepsis
Mortality
Lactate dehydrogenase-to-albumin ratio
MIMIC-IV database
Journal
IF:
3.4
Papers:
2.0K
Citations:
5.9K
Organization
Cited Papers