We describe the persistence of MBL-producing organisms in a nosocomial setting, detected from both clinical and environmental sources, which has continued for six years. A small but significant proportion of patients admitted to BU were colonized and/or infected, 1.5%, and BU-related isolates were greatly overrepresented, accounting for 55% of overall hospital MBL isolates detected during this time. Case patients were significantly older, with greater extent of burns and hospital lengths of stay. Those with clinical infections had generally favorable outcomes, with no attributable deaths. Most cases reflected asymptomatic colonization rather than clinical infection, in contrast to a previously described outbreak in the Intensive Care Unit at another Australian hospital, where 75% of patients carrying MBL-organisms had attributable infections .
Between 2000 and 2006, the predominant carbapenem-resistant Gram-negative organism isolated from BU environmental screening was multidrug-resistant Acinetobacter baumannii, conferring resistance via a Class D OXA-23 enzyme. Its presence was curtailed with improved environmental sampling methods  together with an effective surface cleaning intervention, which included hiring of dedicated cleaners for the BU. Consequently MDR A. baumannii ceased to be detected from the environment after this time.
Environmentally-isolated CRE were first detected in our BU in January 2007, however molecular IMP-4 MBL confirmation in environmental isolates had not been instituted at that time, and these unconfirmed carbapenem-resistant isolates were excluded in this study. Since molecular confirmation began in early 2008, MBL organisms have been recurrently isolated, predominately within the shower facilities that are shared amongst the patients. Our findings of multi-resistant Gram-negative organisms persisting in mostly wet areas despite hospital disinfection, echoes the experiences described both locally , and overseas [8, 9, 14, 15], of outbreaks related to plumbing and water-borne sources . These studies, as well as genomic data from selected clinical and environmental isolates from our BU , suggest a close association between environmental sources, colonization and secondary infection. However, unlike outbreaks that have largely centered around handwashing sinks, the vast majority of positive environmental isolates originated from drains and sluices within the BU shower facilities, though not the shower heads themselves. We postulate that severe burns patients’ large surface areas of denuded and exposed skin, and requirement for regular bathing, puts them at greater risk of colonization by environmental Gram-negative organisms that may have a water-related reservoir , particularly if associated with biofilms .
While some studies have found implementation of infection control bundled interventions, such as universal standard precautions with antibiotic stewardship, have eradicated local outbreaks of multi-resistant Gram-negative organisms [6, 18], others have experienced outbreaks despite this . Existence of such measures in our BU well before outbreak onset did not prevent its establishment or dissemination, although it may have limited its potential magnitude. Although terminal cleaning within our BU terminated an MDR-Acinetobacter baumannii outbreak six years previously, repeated attempts at cleaning interventions targeting structural plumbing components have not resulted in elimination of MBLs from the environment. This may be related to the persistence of biofilms despite terminal cleaning . CRE outbreaks in other settings have involved physical removal of plumbing, together with other infection control measures, with subsequent success at terminating the outbreak [9, 15].
We recognize that the association of MBLs detected in clinical and environmental isolates cannot prove causality. In fact this has, until recently, hampered discussion of the role of the environment in transmission of Enterobacteriaceae between patients and the importance of hospital cleaning in control. Even though the MBL isolates in both clinical and environmental samples were shown to be genetically identical , this does not confirm the reservoir as the definitive source of transmission to newly admitted patients. Nonetheless, apart from the BU staff, the only environment shared between patients, were the shower and bathroom facilities. There was continuing environmental detection and despite concerted terminal cleaning, persistence of MBLs in these, independent of the concomitant presence in the BU of MBL positive patients. We believe the consistency of our findings in a unique patient niche, in association with similar findings emerging in the literature to echo our experience [9, 15] would suggest that environmental reservoirs for multi-resistant Enterobacteriaceae are a potential source for ongoing cross-contamination.
Novel approaches to environmental disinfection, such as hydrogen peroxide vapor decontamination [21, 22] are being trialed, however this approach may be more efficacious against surface environmental organisms such as MRSA and VRE, and efficacy against Enterobacteriaceae, especially those with an established water source, have not been demonstrated to date.
One limitation to this retrospective study was the irregular nature of surveillance from both clinical and environmental sources, which potentially missed a proportion of asymptomatic colonization, and potential environmental reservoirs, due to the selective nature of environmental sampling. This study highlights the importance for active surveillance during an outbreak and in high-risk settings such as the Burns Unit, as recommended by the Centre of Disease Control regarding the control of CREs . A recent carbapenem-resistant Klebsiella pneumoniae outbreak at a National Institutes of Health hospital involved whole-genome sequencing to track the origin of the outbreak and delineate its epidemiology . Real-time sequencing may prove a useful tool in the future as cost-effectiveness and technological improvements may lead to mainstream clinical use.
Laboratory detection methods carried out at the beginning of the outbreak may have under-detected the burden of MBLs in both clinical and environmental isolates, as automated systems may miss carbapenemase-producing organisms [24, 25], and CLSI breakpoints were lowered in 2010 to facilitate greater detection of CREs . Despite this, we observed that improving the sensitivity of laboratory CRE detection, particularly for environmental CRE organisms, required the addition of a lower MIC (0.5 mg/L meropenem) agar dilution plate. Developing rapid diagnostic tests to detect CREs is difficult, especially for environmental screening purposes, due to their multiple and complex resistance mechanisms , and also in screening a multitude of organisms, some of which may only exhibit low-level resistance. Such a test would allow prompt identification of carrier patients, incur appropriate infection control measures, and delineate the extent outbreak before its spread. We have recently evaluated the CarbaNP test as a simple, cost-effective test for CRE testing  and hope to implement this in routine environmental screening for bla-IMP-4 organisms.
Further studies are required to explore the long-term outcome of patients colonized with MBL organisms, and whether chronic carriage persists after hospital discharge. The persistence of CRE in the nosocomial setting despite traditional measures signal a need to pursue newer approaches, including novel cleaning methods, assessing the need for removing physical structures, such as plumbing and drains, vulnerable to biofilm formation, and the role active surveillance plays in monitoring and curtailing outbreaks.