We collected a 5-L water specimen from the previously culture-positive yard tap and another 5-L specimen from the sample access point before the gas chlorinator unit, after flame-sterilizing the outlet and letting water run for a prolonged period. A clump of wild grass above the bore line entering the compound, near the place where an excavator punctured the pipeline, was dug up with its roots intact. Water specimens were filtered, and the filter membrane was cultured in Ashdown broth and subcultured onto Ashdown agar. A 5-g sample of root soil was suspended in sterile saline and processed as for water specimens. Suspect colonies were identified by standard laboratory methods and confirmed by polymerase chain reaction (PCR) amplification of B. pseudomallei DNA sequences (6). B. pseudomallei was isolated from all three tested specimens, although the species was recovered only from the tap water after enrichment with trypticase soy broth and prolonged incubation (2 weeks). Sediment from the bottom of the water tank and a 5-L specimen of tank contents were cultured. Both were positive for B. pseudomallei.

Sampling of the water supply system progressively further upstream drew our attention to an aerator in the treatment plant that during the initial outbreak investigation had been temporarily removed for renovation. The aerator is installed to raise water pH (acid due to dissolved carbon dioxide) by spraying groundwater through a series of open 1 m x 1 m wire mesh trays located in a frame on an elevated platform. A series of water samples were collected in the community, immediately after chlorination, before chlorination, from the ground tank, the aerator, bore lines, and all four capped bores. Aerator spray was collected from the top and lower down the tower. Biofilm was scraped from the sides of the aerator trays. Water and scum from the aerator sump were also sampled. The pH was measured in the spray generated at each level of the aerator. Roots and soil were collected from a shrub (Acacia coleii) growing around the stop valve at the entry to the treatment plant complex, in case roots had penetrated the pipeline and soil had entered the water supply. All water specimens were filtered and spread onto Ashdown and trypticase soy broth media. Isolates were identified as described above. B. pseudomallei was isolated from the spray at the top and bottom of the aerator in abundant growth on direct culture plates, without any prior enrichment (Table), and from soil around the acacia roots. Despite the use of selective media, the growth of various environmental bacterial and fungal species was so abundant in the lower aerator trays and sump that the presence of B. pseudomallei could not be excluded, although all cultures were negative. All samples from the bore lines and bore field upstream of the aerator were negative for B. pseudomallei. Chlorine levels were maintained above 1 ppm.

Figure

Figure. Results of pulsed-field gel electrophoresis (PFGE) of clinical and environmental isolates of Burkholderia pseudomallei from western Australia. Molecular typing was performed by electrophoresis of twice-digested 18h XbaI digests of chromosomal DNA...

Environmental and clinical isolates of B. pseudomallei were typed by PFGE of an XbaI restriction digest of bacterial chromosomal DNA (Figure). After completion of these studies, environmental isolates from parts of the potable water supply, nearby soil, and two distant locations (250 and 1,000 km away) were run simultaneously with human and animal clinical isolates. As two gels were required, clinical isolates were run on one gel with a duplicate of the initial tap water isolate, and environmental isolates were run with a duplicated digest from the fifth case in the cluster. Soil isolates were distinct from the outbreak strain, but the aerator spray and ground tank isolates were identical to the outbreak strain, WKo97 (Figure; Table).

Confirmation that a specific device in the water treatment plant had become heavily contaminated with B. pseudomallei prompted biocontainment questions. Arrangements were made to isolate the aerator from the water inflow, dismantle it, and remove its parts for incineration. Before being dismantled, the aerator was drained and soaked in concentrated hypochlorite, then wrapped in a polythene sheet. Workers wore submicron particle filter masks, protective overalls, and heavy-duty gloves. The ground-level tank was drained, treated with concentrated hypochlorite solution, and cleaned before refilling. Samples collected from the ground-level tank 2 months after these measures had been completed contained no detectable B. pseudomallei.

In our investigation of the persistence of B. pseudomallei in the potable water supply, we traced the source of contamination upstream to the water treatment plant and identified the aerator as a possible source. If it had been connected during the initial outbreak investigation and had been sampled then, the aerator might have been identified as a potential source 1 year earlier. Moreover, failure to fully digest the prechlorination isolate in this and a second reference laboratory led to the erroneous belief that the prechlorination point result was unrelated to the case-cluster. We cannot prove beyond doubt that the aerator was the primary source of contamination without contemporaneous bacteriologic evidence. It is unlikely the aerator became contaminated later as a result of retrograde flow, especially from a point downstream to the chlorinator unit. There are thus several possible sources of contamination: inflowing water from a leak in the bore pipeline, rupture of the pipeline during renovation work, dust or vegetation blown through the mesh walls of the aerator by wind, or direct contact with contaminated soil during refurbishment of the dismantled aerator. The 1997 outbreak may have occurred because a shower of bacteria was released from the aerator into the potable water supply during renovation work on the aerator at a time when chlorine levels were at or near zero. The aerator unit had only a modest effect on pH but clearly provided optimal conditions for multiplication of aquatic bacteria with an aerobic metabolism (7). Whether or not the aerator became contaminated more recently than the outbreak, its ability to act as a persistent source or amplifier of the outbreak strain of B. pseudomallei justified its removal.

Aerators are used to correct acidic, ferric, or unusually warm potable groundwater supplies in many remote communities in northern Australia and represent a potential melioidosis risk. Deliberate or natural aeration of a contaminated water supply could amplify B. pseudomallei growth. In northeastern Thailand, where melioidosis is endemic and the groundwater is unusually acidic (8), natural or artificial aeration of the potable water supply should be considered as an additional contributory factor.

Measures taken after the initial western Australian outbreak investigation appear to have prevented further cases, despite the presence of multiple B. pseudomallei types in soil. Initial water engineering measures have cleared detectable B. pseudomallei from the water treatment plant.