The incremental value of aspartate aminotransferase/alanine aminotransferase ratio combined with CURB-65 in predicting treatment outcomes in hospitalized adult community-acquired pneumonia patients with type 2 diabetes mellitus

Abstract

Background

The features of community-acquired pneumonia (CAP) patients with type 2 diabetes mellitus (T2DM) differ from those without. This study aims to spot a routinely tested parameter with discriminative, predictive and prognostic value to enhance CURB-65’s prognostic accuracy in CAP patients with T2DM.

Methods

We retrospectively studied consecutive CAP patients from 2020 to 2021, comparing laboratory parameters between patients with and without T2DM. Receiver operating characteristic (ROC) curve analysis, univariate and multivariate logistic regression were used to identify key parameters. The area under the ROC curve (AUC), Fagan’s nomogram, net reclassification improvement (NRI), and integrated discrimination improvement (IDI) evaluated the added predictive accuracy.

Results

A total of 720 patients were included, comprising 180 diabetic CAP patients and 540 non-diabetic controls after matching for age, gender, and comorbidities through propensity score matching. In diabetic CAP patients, the aspartate aminotransferase/alanine aminotransferase (AST/ALT) ratio showed the highest AUC (0.676, 95% CI, 0.575–0.776) among laboratory parameters with different distributions between the groups. AST/ALT was also identified as an independent predictor of poor treatment outcome (OR = 3.672, 95% CI, 1.455–9.268, p = 0.006). Adding AST/ALT to CURB-65 slightly increased the AUC, but remarkably enhanced NRI and IDI (AUC, 0.756 vs. 0.782, p = 0.017; continuous NRI, 0.635, 95% CI, 0.304–0.966, p < 0.001; categorical NRI, 0.175, 95% CI, 0.044–0.307, p = 0.009; IDI, 0.043, 95% CI, 0.006–0.080, p = 0.021). An AST/ALT ratio of ≥ 1.625 conferred a 74% post-test probability of poor treatment outcome, while < 1.625 predicted 21%. AST/ALT also predicted outcomes for all the CAP patients enrolled (OR = 1.771, 95% CI, 1.231–2.549, p = 0.002). Predictive accuracy improved after incorporating AST/ALP into CURB-65 in these population (AUC, 0.615 vs. 0.645, p = 0.038; continuous NRI, 0.357, 95% CI, 0.196–0.517, p < 0.001; categorical NRI, 0.264, 95% CI, 0.151–0.376, p < 0.001; IDI, 0.019, 95% CI, 0.008–0.029, p < 0.001).

Conclusions

AST/ALT was identified as a discriminative, predictive and prognostic factor for CAP patients with T2DM. The integration of AST/ALT into CURB-65 enhanced outcome prediction for both diabetic and non-diabetic CAP patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-025-03488-1.