Abstract
OBJECTIVES: We developed a new predictive model to more accurately assess the risk of patients developing severe pneumonia (SP) after hospital admission. METHODS: We retrospectively analyzed patients with pneumonia admitted between June 2022 and May 2024. According to the 2019 American Thoracic Society/Infectious Diseases Society of America guideline, patients were classified into SP and non-severe pneumonia (NSP) groups. Basic clinical information at admission and laboratory results, including complete blood count, coagulation function, biochemical parameters, and bacterial co-infection, were collected. Quantitative real-time polymerase chain reaction (qRT-PCR) was used to determine the cycle threshold (Ct) values of respiratory pathogen nucleic acids to estimate the pathogen load. Host immune biomarkers were measured in fasting serum collected on the morning following the first positive pathogen detection. RESULTS: Among 241 patients (NSP group=139, SP group=102), patients with SP showed significantly lower pathogen Ct values (influenza A virus [IVA]: 24.32 ± 4.56 vs. 28.45 ± 3.21, P<0.001), higher C-peptide levels (2.72 ± 0.84 vs. 2.25 ± 0.68 ng/mL, P<0.001), and higher ferritin levels (590.67 ± 102.78 vs. 498.32 ± 110.45 μg/L, P<0.001). The area under the curve (AUC) was 0.906 in the training set and 0.926 in the test set, indicating high predictive accuracy of the model for SP risk. CONCLUSIONS: This study demonstrats that a predictive model combining quantitative pathogen load with host immune-metabolic biomarkers can effectively predict the risk of severe pneumonia.