“Down the drain”: carbapenem-resistant bacteria in intensive care unit patients and handwashing sinks

One of the most concerning emerging resistance traits among gram-negative bacteria is the ability of these organisms to produce carbapenem-hydrolysing β-lactamases, which confer resistance to almost all β-lactams.1 Carbapenem-resistant Enterobacteriaceae (CRE) are increasing in prevalence worldwide, causing growing concern, as they are often combined with non-β-lactam resistance to produce isolates that are multidrug resistant, have few treatment options available and are associated with high mortality rates.2

Although multiple resistance mechanisms have been identified, carbapenem resistance is often plasmid-encoded, allowing gene dissemination and a propensity to cause nosocomial outbreaks.3-6

We describe a CRE outbreak due to the presence of the metallo-β-lactamase gene blaIMP-4 in an intensive care unit (ICU) associated with contaminated sinks. This report highlights the key role of bacterial environmental contamination and sink design and usage in the propagation of CRE outbreaks.

Dandenong Hospital is a 440-bed tertiary referral hospital in Melbourne, Australia. The ICU has a 14-bed capacity, admitting both medical and surgical patients and averaging 1400 admissions yearly.

Ten ICU patients were found to have clinical specimens with Enterobacteriaceae harbouring the blaIMP-4 gene between November 2009 and July 2012. The ICU routinely screens for vancomycin-resistant enterococci (VRE) carriage using rectal swabs on admission, weekly and on discharge. During the 4-week period from 6 September to 4 October 2012, CRE screening was performed in addition to the routine screen for VRE.

Environmental screening for CRE targeting all 28 wet area locations (including 11 sinks and taps, as well as water fountains and ice machines) was performed. Samples were recollected from CRE positive locations after successive decontamination attempts. A total of 97 samples were collected from taps, water, grates and drains (10 cm down and above the water trap interface where biofilm may be expected to persist).

All environmental and patient screening swabs were cultured onto chromogenic agar (Brilliance CRE Agar, Oxoid). Suspect colonies were further characterised using the modified Hodge test and VITEK 2 (Biomérieux). All isolates with a meropenem minimum inhibitory concentration (MIC) ≥ 0.25 mg/L and a positive modified Hodge test result were forwarded for confirmation by polymerase chain reaction and molecular sequencing. Molecular typing using pulsed field gel electrophoresis (PFGE) was conducted on clinical and environmental isolates and interpreted as per the reference guidelines.7

Guidelines on ICU handwashing sink styles were reviewed to establish whether sinks met clinical design standards.8

Ten clinical isolates (Klebsiella pneumoniae [n = 5], Serratia marcescens [n = 4] and Enterobacter cloacae [n = 1]) and one screening isolate (Escherichia coli) containing the blaIMP-4 gene were detected over the 30-month period. There were four distinct CRE clusters, commencing with three cases of K. pneumoniae in November 2009, followed by two cases of S. marcescens 6 months later, three cases (two S. marcescens and one E. cloacae) after another 11 months, and three cases of K. pneumoniae in July 2012.

Clinical characteristics, patient demographics and ICU admission details of all 11 patients are summarised in Box 1. Patients acquired CRE after a median length of stay in ICU of 10 days (range, 3–134 days). No patient died due to clinical CRE infection. Patient 1 had a prolonged CRE bacteraemia that responded to removal of a central venous catheter, the presumed source of infection.

A total of 111 rectal swabs were collected from 71 patients. Only one patient (Patient 11) was CRE-positive with an E. coli isolate detected.

Antibiotic resistance profiles of clinical isolates indicated resistance to β-lactams and meropenem with MICs ≥ 1 mg/L. All S. marcescens clinical isolates were sensitive to piperacillin/tazobactam, ciprofloxacin and amikacin, while K. pneumoniae and E. coli were only sensitive to amikacin. S. marcescens environmental isolates showed a higher meropenem MIC of ≥ 16 mg/L.

S. marcescens was the only species recovered from environmental samples and was isolated persistently, even after six attempts to decontaminate the grates and drains of eight of the 11 central sinks in the ICU. Tap spout and water cultures were negative for CRE.

Three of the four S. marcescens clinical isolates from 2010 and 2011 were indistinguishable or closely related by PFGE to four isolates from sinks. S. marcescens was isolated from an intercostal catheter swab of Patient 4, matching two sink isolates; Patient 5 had an endotracheal aspirate identical to two different sink locations; and the urine isolate from Patient 6 was closely related to a different sink. The isolate from Patient 7 was unable to be typed by PFGE.

Sink inspection revealed aged and deteriorating porcelain, even though sinks were only installed in 2005. Sink design did not comply with Australasian clinical design standards,8 with a small, shallow sink and a tap that directed water over the drain (Box 2). The design of the ICU sinks led to poor use for hand hygiene (although measured hand hygiene compliance using alcoholic hand rub was around 70% within the unit) and the potential for organisms residing down the drain to be splashed back onto staff hands or contaminate patient areas. It was also revealed that the handwashing sinks had been used incorrectly, with staff disposing clinical waste and residual antibiotics directly into drains. Further enquiry also revealed that a single brush had been used to clean down the drains of all sinks in the ICU without disinfection between sinks.

Cleaning was attempted in an effort to rid the organisms from the sinks. First, cleaning of grates and drains using single-use, soft brushes was attempted, but repeat screening revealed continued CRE growth. Next, in addition to the brushes, hypochlorite deep cleaning was used after the scrub; however, heavy CRE growth was again evident 1 week later. Finally, an attempt using pressurised steam decontamination (Jetsteam Maxi with plunger tool attachment, Duplex) for 1 minute at 170°C on grates and drains appeared to eradicate almost all CRE at Day 1 (one sink remained colonised); however, repeat testing 3 days after steam treatment showed re-emergence of CRE in all previously affected sinks.

We report an outbreak of CRE in an ICU with identical organisms isolated from patients and an environmental source (sinks).

Dissemination of carbapenemase-resistant bacteria has been reported previously in an ICU in Australia, with 62 patients infected or colonised with gram-negative bacteria from multiple genera containing the blaIMP-4 gene.9 The gene is carried on a highly mobile plasmid making it efficient for nosocomial transmission. The gene can reside in multiple genera of hardy environmental organisms that can establish a reservoir within biofilm.

Identifying the outbreak using molecular methods allowed us to establish clonality between clinical and environmental isolates and to propose a mechanism whereby patient contamination may have occurred. Although we cannot prove that the sinks were the source of patient infection, the persistence of the organism within the environment was a concern. Attempts at sink sterilisation were futile, and complete eradication will require future sink removal and replacement with appropriately designed sinks.

Others have reported outbreaks of multiresistant bacteria such as extended-spectrum β-lactamase-producing K. pneumoniae linked to imperfect sink design requiring sink cleaning, replacement and/or improved sink practices to successfully control the outbreaks.10

Our data should act as a reminder on the importance of appropriate hospital design, adequate environmental cleaning and antimicrobial stewardship in the hospital setting.

2 Existing design of intensive care unit sink compared with a design that complies with Australasian standards8

A: Existing sink, showing water spray directly over drain. B: Compliant sink design, showing larger basin and less forceful water flow directed away from drain, to prevent splash back and contamination with drain contents.