In January 2012, a patient residing in temperate Western Australia who had a superficial soft tissue infection had a preliminary isolation of B. pseudomallei. We interviewed the patient to determine a detailed local, national, and international travel history; potential means of B. pseudomallei exposure; and melioidosis-associated concurrent conditions.

The patient’s home property was visited by staff from PathWest Laboratory Medicine (Nedlands, Western Australia) for inspection and environmental sampling of garden beds (1 sample), potable water (2 samples), storm water drainage (1 sample), and a nearby nature reserve (1 sample). Soil samples were processed by suspending 10 g of soil in 20 mL of sterile water and incubated overnight with agitation. Samples were kept stationary for 2 h to allow the soil to settle, and 50 μL supernatant was spread across Ashdown’s agar and B. pseudomallei selective agar (15,16). Solid media was incubated for 48 hours at 37°C followed by 5 days at room temperature. One milliliter of supernatant was also inoculated into 10 mL of Ashdown’s broth (17) and incubated at 37°C overnight before being spread onto selective solid media as described. Plates were checked every 24 h and suspect colonies picked to nonselective blood agar with a 10-μg gentamicin disk placed on the second sector.

In December 2013, after 2 additional cases of culture-confirmed cutaneous melioidosis had been detected in the same postal district associated with a local healthcare facility, we conducted environmental sampling in and around the facility. Additional environmental samples were collected from public accessed land and building excavations in the neighborhood to identify other potential sources of B. pseudomallei external to the health facility. These samples were processed as described previously. Samples included all wound care products in current use, whether sealed or already open. We also sampled fixed surfaces patients were likely to come into contact with during wound care and soil at locations surrounding the health facility that could provide either a primary source or reservoir for later distribution. Laboratory-based surveillance for melioidosis cases from the region was performed for 12 months after the conclusion of the field investigation. This process included referral of all suspected B. pseudomallei by all pathology service providers in Western Australia, multilocus sequence typing (MLST) of all confirmed B. pseudomallei, and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry identification of all wound swab bacterial isolates processed at the state public health laboratory (4).

Solid samples were inoculated onto 5% horse blood agar, Ashdown’s selective agar, and Burkholderia selective agar and then incubated for 48 h at 37°C. Bulk liquids were sampled in a class II biological safety cabinet not previously used for B. pseudomallei work, and 1.0 mL dispensed into thioglycollate broth in accordance with standard microbial contamination assessment methods. A ≈100-mL aliquot of each bulk liquid was filtered through a 0.22-µm membrane filter and used to inoculate the same series of selective and nonselective agars. Positive fluid samples were used for detailed bacterial count studies by using a serial 10-fold dilution to 1:10⁷, a spiral plating device, and triplicates of each dilution.

One environmental sample yielded B. pseudomallei: a 1,000-mL bottle of wound irrigation fluid. Turbidity of the residual wound irrigation fluid in the bottle was not visible because the container walls were semi-opaque. To determine the extent of bacterial colonization and identify any specific higher-density bacterial localization within it, the bottle was dissected from its screw top down to its bottom, sampling at 2.5-cm intervals. Plastic surfaces contaminated with B. pseudomallei were cut into ≈0.5-cm square portions and processed for scanning electron microscopy. Samples were fixed overnight (2.5% glutaraldehyde [vol/vol] in 0.05 mol/L cacodylate buffer pH 7.4), then washed in the same buffer before postfixation for 30 min (1% aqueous osmium tetroxide) and sequential dehydration in ethanol series for 5 min each. Reagents were supplied by the PathWest Electron Microscopy Unit (Nedlands, Western Australia). Critical point drying was achieved by using liquid carbon dioxide; the plastic squares were then attached to aluminum stubs by using double-sided carbon tape with edges painted with carbon solution and coated with 15 nm of carbon. Samples were viewed through a Zeiss SUPRA 55 Variable Pressure SEM operating at 3–5 kV with either in-lens or SE2 detectors, depending on magnification required.

Preliminary identification of B. pseudomallei was made by using a MALDI-TOF mass spectrometer with a 70% formic acid partial extraction protocol and a locally generated Burkholderia mass spectrum database (4). Definitive confirmation of B. pseudomallei was made by using a panel of real-time PCR assays targeting independent genes (18,19). The mutually exclusive B. thailandensis–like flagellar (BTFC)/Yersinia-like fimbrial (YLF) genetic markers were used for preliminary molecular characterization, which used primers and probes developed in-house based on the previously published sequences (20). MLST was performed by using the current PCR and sequencing primers as previously described (http://bpseudomallei.mlst.net/misc/info2.asp) (21). Sequencing was performed on an ABI 3130xl sequencer by using forward and reverse primers with ABI BigDye version 3.1 sequencing chemistry (Applied Biosystems, Foster City, CA, USA). MEGA5 was used to construct a neighbor-joining tree of MLST sequence from all outbreak-associated isolates, including the wound irrigation fluid isolate, to show their genetic relationship to reference B. pseudomallei isolates in the Western Australian Burkholderia Collection (22,23). B. thailandensis E264 was used as an outgroup and root for the neighbor-joining tree. Bootstrap values >50 (>1,000 replicates) were included next to the tree’s branches (24), and evolutionary distances were computed by using a maximum composite likelihood method with units of the number of base substitutions per site (25). The rate variation among sites was modeled with a gamma distribution (shape parameter = 4).

Recent events in the continental United States highlight the ability of B. pseudomallei to breach ecologic or biologic boundaries (26–30). A review of U.S. state and territory cases identified 3 persons with culture-positive melioidosis in the absence of relevant travel to melioidosis-endemic regions and concluded with a recommendation that physicians and healthcare workers should be more aware of the disease (31). A lack of familiarity with this bacterial species in an unusual clinical setting, such as occurred in the cluster we describe here, can cause difficulty in identifying the primary source. The previous Western Australia cluster occurred in a very different setting, a remote community in the tropical north of the state (7). Only 1 other melioidosis cluster has been reported in temperate Western Australia; that cluster was attributed to livestock transported from a melioidosis-endemic region (6). The melioidosis cluster we describe here was notable for its occurrence in urban Western Australia, which is not considered to be endemic for melioidosis, and for its association with a contaminated wound care product.

We recovered B. pseudomallei isolates with the same multilocus ST from 6 melioidosis patients in 1 postal district and excluded another 2 cases of melioidosis from our investigation on the basis of clinical features, travel history, biogeographic bacterial clade, and MLST genotype. Isolation of B. pseudomallei of the same ST from contaminated wound irrigation fluid explains the 5 cases in 2013 because the bottle of wound irrigation fluid was used without replacement throughout this period. Although the manufacturer’s instructions advise that the bottle should be discarded within 24 hours of opening and label the fluid as single-use, it is common practice to use such large volumes of fluid as a stock during wound care procedures, providing decanting is conducted as a no-touch procedure. This melioidosis cluster highlights the public health risks of such a practice. Because the contaminated wound irrigation fluid bottle had not yet been received when the 2012 patient had wound care at the same health facility, the wound irrigation fluid probably was not contaminated before opening. We note that the connection between the January 2012 case and the September–December 2013 cases remains unexplained. Given that other patients received wound care during this period without evidence of cutaneous melioidosis, the existence of an earlier contaminated bottle with subsequent transfer to the contaminated bottle is improbable. Possible explanations for this interval include an undetected past case of chronic, unresolved cutaneous melioidosis with multiple introductions of B. pseudomallei into medical products or an external environmental reservoir common to the 2012 patient and the first or first few 2013 patients. The previous report of nosocomial melioidosis in Australia identified environmental B. pseudomallei that was biochemically similar to the clinical isolates, but that report lacked the strength of molecular epidemiology evidence (10). All previous reports of nosocomial melioidosis come from locations in the tropics (8–12).

The melioidosis cluster we report represents an unusual healthcare-associated outbreak in a temperate suburban setting. B. pseudomallei was not isolated from any other solution or environmental sample from the facility. It is not clear whether the count of B. pseudomallei found in the contaminated irrigation fluid was the result of an initial seeding event with subsequent bacterial growth or by gross contamination without further growth. Previous in vitro studies indicate that B. pseudomallei will tolerate a wide range of nutrient-free aqueous environments (32), and survival of 10⁷ CFU/mL B. pseudomallei in sterile distilled water over 16 years has been reported (33). Electron microscopy showed complex colonization patterns and extensive bacterial adhesion, consistent with long-term bacterial colonization. The initial seeding event probably included sufficient nutrients to support bacterial proliferation. The senescent bacteria we observed could reflect a larger original bacterial population introduced to the bottle. Although not healthcare-associated, 2 cases of cutaneous melioidosis attributed to a contaminated hand wash solution noted in another report further highlight the potential for contamination and subsequent transmission of B. pseudomallei in an occupational setting (34).

We investigated cause and effect in this laboratory outbreak investigation by using a set of stringent rules for emerging infectious disease causality (35), establishing the laboratory evidence to link a series of cutaneous infections in a geographically restricted cluster, identifying a probable source, and introducing early environmental controls. We used a combination of molecular epidemiology, microscopy, and culture-based bacteriologic methods to identify and study a point source for healthcare-associated infection in this investigation of a cluster of cutaneous melioidosis cases in temperate southwest Australia. We obtained circumstantial evidence that bacterial contamination combined with incorrect use of wound irrigation fluid to form a chain of events necessary for subsequent infection. The contaminated wound irrigation fluid is a plausible vehicle for infection, but uncertainty exists about its essential role in the case of patient A. Although we halted the series of infections by removing the source of B. pseudomallei infection from use, we have not yet been able to identify the initial environmental reservoir. Therefore, the long duration of the previous temperate Western Australia cluster leads us to expect additional sporadic cases in the area over an extended period and represents a continuing public health risk (4). However, we have not identified any other cases of cutaneous melioidosis in the same area or ST-1112 infections elsewhere during the subsequent 12 months of laboratory-based surveillance.

Although the inoculum of B. pseudomallei causing human cutaneous infection was measured, the circumstances of wound contamination in this cluster did not allow us to determine the probability of subsequent dissemination from an already infected wound. Additional questions raised are 1) whether pouring contaminated wound irrigation fluid limited infection to an already damaged epidermis without generating sufficient aerosol for pulmonary infection, and 2) whether others in this suburban community were exposed to B. pseudomallei ST-1112 without clinical consequences. Specific aspects of virulence phenotype, genetics, and circumstance that resulted in the notable absence of pneumonia or septicemic infection are the focus of further study.