Pathogen distribution, drug resistance risk factor analysis, and risk prediction model construction of drug-resistant bacteria infection of inpatients to respiratory department of a tertiary hospital around the time of the COVID-19 pandemic

Background: This study aimed to investigate the distribution and drug resistance of pathogens in hospitalized patients in the respiratory and critical care medicine department during the Coronavirus Disease 2019 (COVID-19) pandemic, analyze the risk factors of drug- resistance, and construct the risk prediction model.Methods: Patients who showed positive results in the bacterial culture in the Department of Respiratory and Critical Care Medicine of a large tertiary teaching hospital were enrolled using WHONET 5.6 software from December 2019 to June 2021. The patients were divided into training and validation sets based on a random number table method (8:2). A prediction model was then developed using the training set and verified using the validation set. Single factor analysis was used to compare the general situation and clinical characteristics of patients in the two groups. The risk prediction model of drug-resistant bacteria was constructed using the multi-factor logistic regression. A line diagram was then established based on the regression coefficient of the model. The model was internally and externally verified using receiver operating characteristic (ROC), area under the curve (AUC), and calibration curve.Results: Klebsiella pneumoniae (196/791, 24.78%), Pseudomonas aeruginosa (136/791, 17.19%), Acinetobacter baumannii (82/791, 10.37%), Escherichia coli (82/791, 10.37%), and Enterococcus faecalis (35/791, 4.42%) were the top five bacterial isolates. The isolated drug-resistant bacteria mainly included ESBL-producing E. coli (53/174, 30.46%) and K. pneumonia (28/174, 16.09%), carbapenem-resistant Acinetobacter baumannii (CR-Ab) (34/179, 19.54%), carbapenem-resistant Pseudomonas aeruginosa (CR-Pa) (17/174, 9.77%) and Klebsiella pneumoniae (CR-Kp) (7/174, 4.02%), and methicillin-resistant Staphylococcus aureus (MRSA) (11/174, 6.32%). gram-negative bacteria had a high resistance to ampicillin, ceftriaxone, cefotaxime, cefazolin, cefuroxime, aztreonam, cefepime, and ceftazidime. However, resistance rates of gram-negative bacteria to tigecycline, ertapenem, and cefoxitin were low. The nosocomial infection prediction model of drug-resistant bacteria was developed based on the combined use of antibiotics (antifungal drugs or respiratory quinolones), pharmacological immunosuppression, PCT > 0.5 ng/mL, CKD stage 4-5, indwelling catheter, and age > 60 years via multivariate logistic regression. The AUC under the ROC curve of the training and validation sets were 0.768 (95 % CI:0.624 - 0.817) and 0.753 (95 % CI: 0.657-0.785), respectively, indicating that the model had good discrimination. The predictive ability of the model was evaluated using calibration curve. The Hosmer-Lemeshow test showed that the model fitting had no significant difference (P > 0.05).Conclusions: E. coli, A. baumannii, K. pneumoniae, and P. aeruginosa are the main drug-resistant bacteria in nosocomial infection. COVID-19 does not increase the drug resistance pressure of the main strains. The combined use of antifungal, respiratory quinolone antibiotics, indwelling catheter, chronic renal failure, and age > 60 years are the independent risk factors of drug-resistant bacteria infection. The risk prediction model of drug-resistant bacteria infection can help in the prevention and control of hospital antibacterial-resistant bacteria infection.