A combined model based on lactate clearance rate and SOFA score in prognostic assessment of patients with severe pneumonia.

Abstract

OBJECTIVE: To investigate the role of lactate clearance rate (LCR) and Sequential Organ Failure Assessment (SOFA) score in prognostic assessment of patients with severe pneumonia (SP). METHODS: A retrospective study of 185 SP patients was conducted. Baseline characteristics were compared between survivors (n = 128) and non-survivors (n = 57). Following univariate and correlation analyses, significant variables were used to build a combined prediction model via binary logistic regression. Model calibration and predictive performance for 28-day mortality were assessed using Hosmer-Lemeshow and ROC curve analyses, respectively. Dynamic changes in LCR and SOFA scores at admission (T0), 6 h (T6), and 24 h (T24) were compared. RESULTS: The 28-day mortality rate was 30.81% (57/185). Univariate analysis revealed statistically significant differences in age, chronic lung disease, procalcitonin (PCT), LCR, and SOFA score (P < 0.05). Spearman correlation analysis showed a moderate negative correlation between LCR and SOFA score (r = -0.401, P < 0.001). Multivariate analysis identified LCR [0.911 (0.861–0.965)] and SOFA score [1.508 (1.278–1.779)] as independent predictors (P < 0.05). The combined model demonstrated superior predictive accuracy (AUC: 0.882, 95% CI: 0.833–0.932) compared to LCR (AUC: 0.793) or SOFA score (AUC: 0.832) alone (Delong test, P < 0.05). Survivors exhibited significantly higher LCR and lower SOFA scores at T6 and T24, with improving trends over time, unlike non-survivors. CONCLUSION: The combined model integrating LCR and SOFA score outperforms either parameter alone in predicting 28-day mortality risk in SP, offering a potential tool for early risk stratification.