Prediction of risk for secondary lower respiratory tract fungal infection during the acute exacerbation phase of COPD

Abstract

Introduction: We aimed to investigate the risk factors for secondary lower respiratory tract fungal infection during acute exacerbation of chronic obstructive pulmonary disease (AECOPD).

Methodology: A total of 466 AECOPD patients diagnosed from March 2019 to November 2020 were divided into infection (n = 48) and non-infection (n = 418) groups. The risk factors for lower respiratory tract fungal infection were screened by logistic regression analysis, and a nomogram prediction model was established. The discriminability was validated by area under the receiver operating characteristic curve (AUC) and C-index, calibration was validated by GiViTI calibration belt and Hosmer-Lemeshow test, and clinical validity was assessed by decision curve analysis (DCA) curve.

Results: Thirty fungi strains were detected, including 18 strains of Candida albicans. Pulmonary heart disease, hypoalbuminemia, use of antibiotics within 3 months before admission, use time of antibiotics ≥ 14 d, invasive operation, blood glucose ≥ 11.10 mmol/L at admission, and procalcitonin (PCT) ≥ 0.5 ng/mL when diagnosed as fungal infection independent risk factors (p < 0.05). AUC was 0.891, indicating high discriminability of the model. The threshold probability in the DCA curve was set to 31.3%, suggesting that the model had clinical validity.

Conclusions: We identified the independent risk factors for lower respiratory tract fungal infection in AECOPD patients. The established model has high discriminability and calibration. Immediate intervention is beneficial when the predicted risk exceeds 31.3%.