Microbial spectrum and resistance profiles in elderly patients with pulmonary infections: A tNGS-based retrospective study.

Pulmonary infections in the elderly are difficult to diagnose due to immunosenescence, atypical symptoms, and frequent polymicrobial or opportunistic infections. Traditional methods often miss pathogens. Targeted next-generation sequencing (tNGS) enables rapid, sensitive, and comprehensive detection of pathogens and resistance genes, offering clear advantages. This retrospective study analyzed bronchoalveolar lavage fluid samples from 306 elderly inpatients with clinically suspected pulmonary infections using tNGS. Detected microbial species, antimicrobial resistance genes, and relevant clinical data were extracted from electronic medical records. Kaplan-Meier survival analysis and an eXtreme gradient boosting (XGBoost) machine learning model were employed to identify predictors of prolonged hospitalization. Viruses were the most frequently identified pathogens (29.6%), followed by bacteria (24.4%) and fungi (9.0%). Fungal co-infections were significantly associated with prolonged hospital stays and disrupted blood gas homeostasis, characterized by elevated partial pressure of carbon dioxide (PaCO₂) and bicarbonate (HCO₃⁻) levels. Antimicrobial resistance genes were detected in 7.5% of patients, with 23S rRNA mutations and mecA being the most prevalent. Inflammatory markers (C-reactive protein, procalcitonin) and gas exchange indices (PaCO₂, partial pressure of oxygen [PaO₂]) showed significantly correlations with overall pathogen burden. The eXtreme gradient boosting model identified PaCO₂, procalcitonin, HCO₃⁻, and fungal infection status as key predictors of prolonged hospitalization. tNGS facilitates comprehensive detection of pathogens and antimicrobial resistance genes in elderly patients with pulmonary infections. Our findings underscore the often overlooked clinical impact of fungal co-infection and respiratory dysfunction on patient outcomes. These results highlight the value of incorporating tNGS into routine diagnostic workflows for geriatric infection management, enhancing both diagnostic accuracy and prognostic assessment.