Lactate dehydrogenase-to-albumin ratio as a potential prognostic indicator in glucocorticoid-treated severe pneumonia: a multicenter retrospective study with external validation.

Background: Glucocorticoids are widely used in severe pneumonia but can mask clinical symptoms and traditional severity scores. The lactate dehydrogenase-to-albumin ratio (LAR) reflects the balance between tissue injury and metabolic reserve, yet its prognostic value in steroid-treated pneumonia remains undefined. Methods: This multicenter study utilized a primary cohort (n=500) and an independent external validation cohort (n=354) of pneumonia patients receiving glucocorticoid therapy. The primary endpoints were 30-day and 90-day all-cause mortality. We employed Cox regression, restricted cubic splines (RCS), and ROC analysis to evaluate LAR performance. Results: High admission LAR (≥10.48) was identified as a potential prognostic indicator of mortality, associated with a more than twofold risk increase at 30 days (Adjusted HR 2.54; 95% CI: 1.56-4.12) and 90 days (Adjusted HR 2.44; 95% CI: 1.56-3.82). RCS analysis confirmed a non-linear risk escalation with a biological threshold of 10.22 (P non-linearity < 0.001). LAR demonstrated improved predictive discrimination (AUC 0.742) compared to PSI (AUC 0.700) and CURB-65 (AUC 0.654). Although the high LAR group received lower median cumulative glucocorticoid doses (3.0g vs. 5.8g; P < 0.001), this finding was associated with more rapid clinical progression and truncated treatment windows in high-risk patients, reflecting a survival-time-dependent exposure rather than a lack of therapeutic intensity. Conclusions: Admission LAR is a potential, easily accessible biochemical marker that provides prognostic value independent of typical inflammatory signs masked by steroids. In patients exceeding the 10.22 threshold, our findings suggest a potential efficacy bottleneck for conventional steroid dosing, highlighting a high-risk subpopulation that may require optimized early intervention.