What question did this study set out to answer?

May 24, 2026Open Access

Enhanced mortality prediction in pneumonia-associated acute respiratory distress syndrome: a model integrating lymphocyte subsets with clinical parameters in non-immunosuppressed adults

Key Points

This study aims to develop a predictive model for in-hospital mortality in non-immunosuppressed adults with pneumonia-associated ARDS.
Single-center retrospective study enrolling 126 adults with pneumonia-associated ARDS.
Data collected within 48 hours of admission analyzed using BIC-based selection and LASSO regression.
Constructed a combined model using selected predictors alongside the SOFA score, with bootstrap validation.
In-hospital mortality rate was 35.7% (45/126).
The final model had an AUC of 0.805 (95% CI: 0.730–0.879), outperforming the SOFA-only model (AUC: 0.707; p=0.020).
Bootstrap validation showed robustness with an optimism-corrected C-statistic of 0.788.

Abstract

Background Acute respiratory distress syndrome (ARDS) is a common clinical syndrome among critically ill patients and is associated with high mortality. Although scoring tools based on organ dysfunction currently exist, their ability to assess the immune status of patients remains limited. This study aimed to develop a predictive model for in-hospital mortality in non-immunosuppressed adults with pneumonia-associated ARDS by integrating clinical and host immune parameters. Methods This single-center retrospective study included 126 adults with p-ARDS. Data collected within 48 h of admission were analyzed. Candidate predictors ( p 0.10 in univariate analysis, excluding post-admission treatments) underwent simultaneous selection by BIC-based best subset selection and LASSO regression; variables identified by both methods were retained. A multivariable model combining selected predictors with the SOFA score (Model 2) was constructed, and a SOFA-only model served as the baseline (Model 1). Performance was evaluated for discrimination, calibration, and clinical utility. Bootstrap internal validation (500 iterations) was performed. Results The overall in-hospital mortality rate was 35.7% (45/126). The final combined model incorporated the following variables: heart failure (aOR = 5.207), CD3 + CD4 + T-cell count (aOR = 0.719 per 100 cells/μL), D-dimer (aOR = 1.113 per 1 μg/mL) and SOFA score (aOR = 1.126 per point). The model demonstrated excellent predictive performance, with an AUC of 0.805 (95% CI: 0.730–0.879), which was significantly higher than that of the baseline model (AUC: 0.707; 95% CI: 0.618–0.797; p = 0.020). Calibration curves and decision curve analysis further confirmed its good calibration and superior clinical net benefit. Bootstrap internal validation demonstrated robustness (optimism-corrected C-statistic: 0.788). Conclusion We developed a multidimensional model for predicting in-hospital mortality in non-immunosuppressed adult patients with p-ARDS. The model’s performance was superior to the SOFA score. External validation is needed for the widespread application of this model.

AI에게 질문

Bookmark

View Full Paper