| Summary: | Zhuman Du,1 Dan Yang,2 Linhai Pan,1 Qianglin Zeng,1 Xiaoju Chen1 1Department of Respiratory and Critical Care Medicine, Clinical Medicine College & Affiliated Hospital of Chengdu University, Chengdu, People’s Republic of China; 2Center of Gerontology and Geriatrics, West China Hospital, Sichuan University, Chengdu, People’s Republic of ChinaCorrespondence: Qianglin Zeng; Xiaoju Chen, Department of Respiratory and Critical Care Medicine, Clinical Medicine College & Affiliated Hospital of Chengdu University, No. 82 North Section 2, Second Ring Road, Chengdu, Sichuan Province, People’s Republic of China, Email qlzeng@hotmail.com; cxj9592@163.comObjective: The management of multidrug-resistant Pseudomonas aeruginosa (MDR-PA) pneumonia remains challenging due to increasing antibiotic resistance and high mortality rates. Current prediction models often neglect three critical factors: comorbidities, medical interventions, and prior antibiotic exposure. This study created a practical risk assessment tool incorporating these clinical elements.Methods: This retrospective study collected 3132 bronchoalveolar lavage fluid specimens from a large tertiary comprehensive hospital in southwestern China over a two-year period (20232024). A total of 209 patients with Pseudomonas aeruginosa pneumonia were ultimately enrolled, including 94 cases of multidrug-resistant and 115 drug-sensitive cases. Data included demographics, comorbidities, invasive procedures, antibiotic histories, and laboratory findings. Key predictors were selected using LASSO regression, followed by logistic regression to build a predictive model. Performance was evaluated through ROC analysis, calibration tests, and bootstrap validation.Results: Thirteen predictors were identified: Age, prolonged hospitalization; Cor pulmonale, cardiac insufficiency, old cerebral infarction, chronic renal failure; ventilation, tracheostomy, nasogastric intubation; Cefoperazone-sulbactam, aminoglycosides, antifungals; Hemoglobin levels. The model showed strong predictive accuracy with AUC values of 0.853 (95% CI 0.793– 0.914) in training and 0.972 (0.931– 1.000) in validation cohorts. Calibration demonstrated excellent consistency (Hosmer-Lemeshow P=0.989).Conclusion: In this study, we developed and validated a predictive model for MDR-PA pneumonia risk by integrating host comorbidities, iatrogenic interventions, and antimicrobial exposure. The model demonstrates potential utility in early identification of high-risk patients and may inform antimicrobial stewardship strategies in regions with predominant cephalosporin/aminoglycoside resistance patterns.Keywords: pseudomonas aeruginosa, multidrug resistance, nomogram, risk factor
|