We conducted an environmental investigation focused on water supplies and other sites conducive to biofilm formation. We collected 85 samples from plumbing fixtures and environmental surfaces in patient care areas as well as from water intake and storage areas, and evaluated for the presence of VIM-CRPA (Appendix).

Figure 1

Figure 1. Environmental sampling scheme at hospital A from a study of CRPA in acute-care hospital specialty care unit, Texas, USA. A) Collection location in hospital and number of each sample type...

We identified VIM-CRPA in 3 sites in room A related to a sink: the drain, bulk water (1 L of tap water), and the interior surface of a faucet serving as the water source for the reverse osmosis unit of portable dialysis machines (Figure 1, panel A). Of note, this faucet did not have an aerator, which has been implicated as a source of contamination in prior outbreaks. We did not recover VIM-CRPA from the interior of a second faucet intended for hand hygiene, the bulk water or point-of-use filter from that faucet, surface samples of the sink basin, or nearby areas.

The dialysis faucet in room A was installed in October 2017. After we identified VIM-CRPA from associated samples, it was disassembled, revealing a rubber gasket trapped in the gooseneck. We collected swab samples from the interior of the disassembled faucet, valve, and core; we then instilled water into the fixture, agitated it using a sonic device, and filtered it onto culture medium with 4 rinses from the faucet, connection supply line, core, and gasket. We cultured VIM-CRPA from the gasket, faucet, and water supply line (Figure 1, panel B). Three patients with VIM-CRPA infections received care in room A over a 6-month period; 1 had an infection identified on hospital day 46 and was discharged 1 month before the dialysis faucet was installed. None of the infected patients received dialysis.

We considered that portable dialysis machines attachable to the contaminated faucet could spread VIM-CRPA to other dialysis hook-ups and sink drains where effluent reverse osmosis water was discharged. We cultured bulk water and swab samples from the sink in the biomedical room where dialysis machines were cleaned, the dialysis machine connector, and the tubing from the reverse osmosis unit of the dialysis machine connected to the contaminated faucet to test for CPOs, but identified no VIM-CRPA. After removing the dialysis faucet and adopting measures intended to mitigate spread of organisms from premise plumbing (11), we identified no additional VIM-CRPA clinical cultures in patients admitted to the specialty unit.

Figure 2

Figure 2. Whole-genome sequencing dendrogram of VIM-CRPA clinical (N = 20) and environmental (N = 13) isolates from hospital A, Texas, USA. Location of culture collection, isolate source, and patient hospital day...

We performed whole-genome sequencing on 33 VIM-CRPA clinical isolates from 20 patients at multiple locations in hospital A and 13 environmental isolates from room A (Figure 2; Appendix); sequences are available at the National Center for Biotechnology Information (BioProject ID PRJNA288601). All 33 isolates were sequence type 308 and harbored VIM-2; clinical isolates varied by 3–255 (median 54, mean 60.3) high-quality single-nucleotide variants (hqSNVs) (12–14). Environmental isolates from the faucet and clinical isolates from 3 patients admitted to room A varied from 0–24 (median 10, mean 11.8) hqSNVs. The hqSNVs were derived from a conserved core of 6.5 Mb, which covered on average 91% of the assembled genome for the 33 isolates in the analysis.

The mechanism by which the faucet became contaminated is unknown. We considered it might have been through the water supply, considering a recent report of carbapenemase-producing organisms, although not VIM-producing, in US municipal water systems (15); however, none of the 24 water samples collected from other hospital locations grew any CPOs in cultures. Given the stay in room A of a patient with VIM-CRPA before the faucet was installed, we hypothesize that the sink drain became contaminated first, followed by retrograde contamination from the sink drain to the faucet, either during installation or through patient care activities. Although the misplaced rubber gasket provided a nidus for contamination, whether that was necessary for persistent faucet contamination is unclear.

We could not ascertain relative contributions of the faucet interior and upper part of the sink drain toward patients acquiring VIM-CRPA. Although none of the patients underwent dialysis, the faucet was not labeled for dialysis use and may have been used for hand hygiene and other purposes. Infection prevention efforts at hospital A focused on improving sink hygiene, including removing patient care supplies from sink splash zones and regularly cleaning splash zones to prevent future transmission to patients from wastewater plumbing. During the 18-month period after the sink hygiene interventions began in October 2018, another 2 cases were identified, but in units unrelated to the specialty care unit.

We identified VIM-CRPA from a dialysis faucet, in water from that faucet, and from the associated sink drain in an ICU room where VIM-CRPA infections of the same strain developed in 3 patients. Although healthcare facility wastewater plumbing and faucet aerators are well-documented reservoirs of CPOs, our findings highlight the importance of considering other plumbing sources as well.