Carriage of multidrug-resistant Gram-negative bacilli: duration and risk factors

Identification of risk factors influencing the duration of carriage of multidrug-resistant Gram-negative bacilli (MDR-GNB) may be useful for infection control. The aim of this study is to estimate the impact of several factors collected for routine hospital surveillance on the duration of carriage of selected MDR-GNB. From January 2015 to July 2021, patients with at least two clinical/surveillance samples positive for MDR-GNB different from ESBL-producing E. coli or AmpC - exclusively producing Enterobacterales were assessed. Microorganisms, age, number of admissions, clinical or rectal sample, sex, and admission service were evaluated as risk factors. Multivariate analysis was performed by a Cox proportional hazard model. A total of 1981 episodes of colonization were included. Involved microorganisms were ESBL-Klebsiella pneumoniae (KP) in 1057 cases (53.4%), other ESBL-non-E. coli Enterobacterales in 91 (4.6%), OXA-48-KP in 263 (13.3%), KPC-KP in 90 (4.5%), VIM-KP in 29 (1.5%), carbapenemase-producing non-KP Enterobacterales (CP-non-KP) in 124 (6.3%), and MDR Pseudomonas aeruginosa (MDR-PAER) in 327 (16.5%). No differences in duration of colonization were observed among ESBL-KP (median colonization time 320 days), ESBL-non-E. coli Enterobacterales (226 days), OXA48-KP (305 days), and MDR-PAER (321 days). For each group, duration of colonization was significantly longer than that of KPC-KP (median colonization time 60 days), VIM-KP (138 days), and CP-non-KP (71 days). Male sex (HR = 0.88; 95% CI 0.78-0.99), detection in Hepatology-Gastroenterology (HR = 0.71; 95% CI 0.54-0.93), clinical sample (HR = 0.61; 95% CI 0.53-0.69), and > 2 admissions after first detection (HR = 0.47; 95% CI 0.42-0.52) were independent predictors of longer carriage, whereas VIM-KP (HR = 1.61; 95% CI 1.04-2.48), KPC-KP (HR = 1.85; 95% CI 1.49-2.3), and CP-non-KP (HR = 1.92; 95% CI 1.49-2.47) were associated with shorter colonization time. Duration of colonization was significantly longer for ESBL-KP, other ESBL-non-E. coli Enterobacterales, OXA-48-KP, and MDR-PAER. For these microorganisms, prolonging surveillance up to 2.5-3 years should be considered. Male sex, clinical sample, multiple readmissions, admission service, and type of microorganism are independent predictors of the duration of carriage.