BackgroundInformation is limited about the effect of restricted carbapenem use on clearance of multi-drug resistant Acinetobacter baumannii (MDRAB). We sought to determine the time effect of antibiotic exposure on multi-drug resistant Acinetobacter baumannii (MDRAB) acquisition and clearance.MethodsWe conducted a retrospective observational study at the intensive care units of a tertiary medical center. Forty-two of a cohort of previously healthy young adults who were concurrently burned by a dust explosion was included. Cases consisted of those from whom MDRAB was isolated during hospitalization. Controls consisted of patients from whom MDRAB was not isolated in the same period. Use of antimicrobial agents was compared based on days of therapy per 1,000 patient-days (DOT/1,000PD). A 2-state Markov multi-state model was used to estimate the risk of acquisition and clearance of MDRAB.ResultsMDRAB was discovered in 9/42 (21.4%) individuals. The cases had significantly higher use of carbapenem (652 DOT/1,000PD vs. 385 DOT/1,000PD, P < 0.001) before MDRAB isolation. For the cases, clearance of MDRAB was associated with lower use of carbapenem (469 DOT/1,000PD vs. 708 DOT/1,000PD, P = 0.003) and higher use of non-carbapenem beta-lactam (612 DOT/1,000PD vs. 246 DOT/1,000PD, P <0.001). In multi-state model, each additional DOT of carbapenem increased the hazard of acquiring MDRAB (hazard ratio (HR), 1.08; 95% confidence interval (CI) 1.01–1.16) and each additional DOT of non-carbapenem beta-lactam increased the protection of clearing MDRAB (HR, 1.25; 95% CI 1.07–1.46).ConclusionsBoth acquisition and clearance of MDRAB were related to antibiotic exposure in a homogeneous population. Our findings suggest that early discontinuation of carbapenem could be an effective measure in antibiotic stewardship for the control of MDRAB spreading.