Acinetobacter baumannii is an important pathogen, causing healthcare-associated infections such as pneumonia, urinary tract infection, soft tissue infection, and bloodstream infection [1, 2]. The extraordinary metabolic versatility of A. baumannii contributes to this organism’s survival in the environment, including persistence for extended periods of time on dry surfaces [3]. Given this organism’s propensity to horizontally acquire resistance to multiple classes of antimicrobial agents, A. baumannii ranks among the most important nosocomial pathogens [4]. The increasing spread of international clones of multidrug-resistant Acinetobacter baumannii (MDRAB) with decreased susceptibility to carbapenem poses a great threat to the health care system; alternative therapeutic options are limited and outcome is poor in patients infected by beta-lactam-resistant A. baumannii [5, 6]. An integrated, multidisciplinary approach is advocated to control the growing threat of MDRAB-related infection or colonization. Proposed responses have included changes in hand hygiene, surveillance, cohort policy, environmental disinfection and cleaning, contact isolation, decolonization, the use of chlorhexidine baths, and antibiotic stewardship programs (ASPs), all with the intent of eliminating this bacterium’s reservoir, transmission, and source [7–10].

Studies have shown that antimicrobial use is associated with the emergence of drug-resistant pathogens; the implementation of an ASP may be effective in reducing resistance rates [11–17]. While there is increasing evidence that MDRAB acquisition may be related to prior carbapenem exposure [18–20], there is typically a time lag of one to several months between antimicrobial prescription and the emergence of bacterial resistance [16, 21]. The most affected patients are those of advanced age, immunosuppressed status, and/or comorbidities that create a need for broader antimicrobial coverage in clinical situations. However, the direct effect of restricted use of carbapenem at the individual patient-level remains poorly understood.

On June 27, 2015, a dust explosion occurred in the late evening at the Formosa Fun Coast water park in northern Taiwan, creating over 400 burn victims and representing a major challenge to the health care system [22]. Among the victims were forty-two previously healthy young adults who presented with severe burns and were admitted as a cohort to two intensive care units (ICUs) at Chang Gung Memorial Hospital. The dust explosion was caused by a flammable starch-based colored powder. The powder created an extremely dense dust cloud, which immediately caused partygoers to be engulfed by flames when the powder ignited accidentally [23]. This event therefore provided with an opportunity to study the effect of exposure to various antibiotics on the acquisition and clearance of MDRAB in a homogeneous group of patients.

We conducted this study to examine the association between carbapenem and MDRAB in an extremely homogeneous group of previously healthy young burn patients. Our results show that antibiotic exposure was related to both the acquisition and clearance of MDRAB. In particularly, MDRAB clearance was associated with restricted use of carbapenem. Each additional DOT of carbapenem increased the risk of MDRAB acquisition, and each additional DOT of non-carbapenem beta-lactam increased the chance of MDRAB clearance. Our findings highlight the impact of antimicrobial use on the occurrence of drug-resistant pathogens, suggesting that the early discontinuation of carbapenem use may help to control the emergence of MDRAB.

Munoz-Price et al. [27] reported that every additional defined daily dose (DDD) of carbapenem increased the risk of carbapenem-resistant A. baumannii colonization (among patients initially found to be non-colonized) by 5.1%. In the group of previously healthy young burn patients, we demonstrated that carbapenem was a time-dependent variable for MDRAB acquisition, and every additional DOT increased the risk of MDRAB acquisition by 8%. Both the studies explored the effect of carbapenem exposure as a continuous variable to further understand the association between time-dependent antibiotic exposure and acquisition of MDRAB. However, our cohort is uniquely positioned to evaluate the impact of antibiotic exposure on the occurrence of MDRAB infection/colonization, given that our patients were free of underlying conditions and were admitted at the same time due to the dust explosion; the results of these experiments provided additional data to facilitate the optimization of antibiotic prescription by clinicians.

The importance of detection of carbapenem-resistant A. baumannii colonization lies in the high predictive value with regard to the subsequent development of carbapenem-resistant A. baumannii infection [28]. Notably, infection by carbapenem-resistant A. baumannii is associated with a mortality rate of approximately 52%, compared with a 19% mortality rate observed following infection with carbapenem-susceptible A. baumannii [29]. Our findings support the short-term impact of an ASP on the clearance of drug-resistant pathogens; the early discontinued use of carbapenem may help to control an MDRAB outbreak. In the long term, carbapenem use may increase following the implementation of ASP due to the ballooning effect [15]. Carbapenems, which exhibit broad-spectrum activity against bacteria, may impose collateral damage and perturb the human indigenous microbiota [30]. Halting carbapenem use may be helpful for reframing and restoring the human microbiome, which may combat MDRAB, and for precluding the selection of resistant strains from a population of susceptible bacteria. Our results also suggest that the competitive advantage of A. baumannii in the microbial ecology may solely reflect this organism’s resistance to antimicrobial agents under selective pressure.

There has been a report that beta-lactam was a risk factor for acquisition of MDRAB [31]. Furthermore, differing from our findings, the usage of carbapenem tended to correlate to clearance of MDRAB in the multivariable analysis with borderline significance [31]. One possible reason is that our patient group was comprised of a cohort of relatively young burn victims without comorbid illnesses. There were few potential confounding variables. Most of the case patients (7/9) were cleared from MDRAB without antimicrobial therapy directed against MDRAB (Table 3). Still, the patients received other antibiotics due to critically ill conditions and burn injury. There were two of the nine cases who received specific treatment for MDRAB (Case No. 7 and case No. 9 in Table 3). Despite limited number of cases, our findings may imply the possibility of spontaneously clearance of MDRAB in immune competent hosts. It might be partly caused by less selective pressure and the clearance could occur without pharmacological intervention.

Both DDD and DOT are standardized methods for measurement of antibiotic use [32]. We chose DOT divided by time-at-risk (i.e., accumulated person days) to compare the rate of antimicrobial use between the case and the control group. As the measure of antimicrobial use that best predicts the prevalence of antimicrobial resistance has not yet been defined [33], one advantage of DOT is that it is not affected by changes in dosing regimen [21]. Moreover, DOT could be more helpful in comparing the use of different classes of antimicrobials within a given institution and indicate timely antimicrobial usage in a particular patient [21]. Because the clinicians might adjust the antimicrobial regimens promptly when treating critically ill patients, we think that measurement of the DOTs would be a better reflection of clinical scenario and usage of antimicrobial agents.

Our study does have some limitations. Due to this study’s retrospective nature and single-center design, our results may be not applicable in other settings or hospitals. There were more than 400 injuries due to the dust explosion and only 42 of the them who were admitted to the ICUs of the hospital were included. This has implications for possible selection bias. However, our study population consisted of young adults who were previously healthy and admitted as a concurrent group following a single event (the dust explosion). The patients were admitted to the hospital at the same time between June 27, 2015 and July 1, 2015. We acknowledged that medical procedures and environmental factors played important roles in the process of MDRAB colonization. The integrated and intensive care provided by the hospital at that time maximally eliminated the confounding factors that were measurable [23]. Hence, the results are expected to be relevant and may contribute to clinical decision-making. Another potential drawback is that we did not investigate the indications of antimicrobial use in our analysis. However, a well-established integrated ASP has been implemented in the hospital for a decade; each prescription was carefully reviewed daily by both pharmacists and infection diseases specialists [10, 12]. As a result, decisions regarding antimicrobial use were attributable to responsible clinicians; appropriate use of antimicrobial agents could be expected in the study. The definition of MDRAB clearance among the patients with prior MDRAB isolation may be concerned. We defined MDRAB clearance according to the hospital’s policy, which requires surveillance cultures to be obtained within 7 days of the previous culture. By definition, eight of nine MDRAB-positive patients were clear from MDRAB. Biases were minimized, given that the ASP were applied regularly and consistently across the hospital. A further limitation of our study was that we did not perform contemporaneous environmental surveillances in either of the two ICUs. This precluded the analysis of interactions between the patients and their hospital environment. Finally, we did not discern between infection and colonization status in the study. Yet the conclusion may be valid in either infection or colonization status.

In conclusion, our results suggest that antibiotic exposure is associated with both acquisition and clearance of MDRAB in this homogeneous population of previously healthy, critically ill young burn patients. Our findings reinforce the utility of an ASP in reducing the occurrence of multidrug-resistant pathogens, and highlight the appropriateness of early discontinuation of carbapenem when the use of carbapenem is inevitable.