Extensive contamination of CRKP in patient surroundings.

A total of 3,699 environmental samples were collected, including 2,964 from the surrounding of 82 patients. CRKP was detected from 76 samples (positive rate, 2.1%), including 67 from the immediate surroundings of 21 patients, six from two sinks in patient rooms, and three from the surface of instruments (two noninvasive ventilators and one nurse cart) in the corridor outside patient rooms (Table 2 and Fig. 1). These CRKP isolates were detected at one or more sites from the surroundings of 21/82 (25.6%) patients occupying 17 of the 20 (85.0%) beds. Except for the three instruments in the corridor, CRKP was not detected in any samples collected from areas outside patient rooms.

TABLE 2

Environment sampling results

Area	Surface typea	Sampling site	Sampling no.	No. positive	% Positive
Patient care area
Bed surroundings	Noninstrument	Bedrails	257	12	4.7
Bed surroundings	Noninstrument	Hanging towers	257	5	2.0
Bed surroundings	Noninstrument	Light switches	257	2	0.8
Bed surroundings	Noninstrument	Bed air pumps	255	4	1.6
Bed surroundings	Noninstrument	Lockers	255	5	2.0
Bed surroundings	Noninstrument	Stethoscopes	253	6	2.4
Bed surroundings	Instrument	Cardiovascular monitors	257	4	1.6
Bed surroundings	Instrument	Air-purifying disinfectors	249	3	1.2
Bed surroundings	Instrument	Infusion pumps	248	9	3.6
Bed surroundings	Instrument	Ventilators	239	6	2.5
Bed surroundings	Instrument	Nurse carts	234	6	2.6
Bed surroundings	Instrument	Nebulizers	203	5	2.5
Patient rooms		Computer mouse/keyboards	150	0	0.0
Patient rooms		Drinking water dispenser buttons	130	0	0.0
Patient rooms	Sink	Sink top surface	82	1	1.2
Patient rooms	Sink	Faucets	82	1	1.2
Patient rooms	Sink	Internal surfaces	82	3	3.7
Patient rooms	Sink	Drains	82	1	1.2
Corridor	Instrument	Vibration sputum ejection machines	10	0	0.0
Corridor	Instrument	Emergency rescue carts	6	0	0.0
Corridor	Instrument	Noninvasive ventilators	6	2	33.3
Corridor	Instrument	Steppers	4	0	0.0
Corridor	Instrument	Nurse carts	3	1	33.3
Corridor	Instrument	Walk helpers	3	0	0.0
Outside patient area
Nurse station	Sink	Sink top surface	3	0	0.0
Nurse station	Sink	Faucets	3	0	0.0
Nurse station	Sink	Internal surfaces	3	0	0.0
Nurse station	Sink	Drains	3	0	0.0
Nurse station		Computer mouse/keyboards	9	0	0.0
Nurse station		Telephones	6	0	0.0
Nurse station		Barcode scanners/printers	5	0	0.0
Nurse station		Medical record holder	3	0	0.0
Nurse station		Wireless interphones	3	0	0.0
Treatment room		Apparatus storage cabinets	25	0	0.0
Treatment room		Liquid storage cabinets	5	0	0.0
Treatment room		Refrigerator doorknobs	4	0	0.0
Treatment room		Light switches	3	0	0.0
Bronchoscope room		Bronchoscopes	5	0	0.0
Bronchoscope room		Drying tabletop	3	0	0.0
Bronchoscope room		Bronchoscope cabinet doorknob	3	0	0.0
Bronchoscope room		Bronchoscope cabinet internal wall	3	0	0.0
Bronchoscope room		Washing machine buttons	3	0	0.0
Bronchoscope room		Internal of washing tanks	3	0	0.0
Cleaner room		Cleaner vehicle body	3	0	0.0
Cleaner room		Cleaner vehicle cap	3	0	0.0
Cleaner room		Cleaner vehicle hand shank	3	0	0.0
Cleaner room		Mop handles	3	0	0.0
Meeting room		Computer mouse/keyboards	7	0	0.0
Meeting room		Drinking water dispenser buttons	3	0	0.0
Staff toilets		Doorknobs	6	0	0.0
Total			76	3,699	2.1

Open in a separate window

^aInstrument and noninstrument surfaces are assigned to surface types in the patient care area.

Among the 21 patients with CRKP detected in their surroundings, six had CRKP from their rectal/oral swabs and/or clinical samples. The interval between the first positive rectal/oral swab and the first positive environmental sample was 0 (the same day; patients P4, P6, P7, and P17-1), 5 days (patient P2-2), or 13days (patient P2-1) (Table 1). For five of the six patients, CRKP was always detected from their environment in each weekly sampling during their entire ICU stay, suggesting continuous contamination. In particular, 33 out of the 76 CRKP-positive environment samples were recovered from multiple sites in the surroundings of a single patient (patient P6) across a period of 7 weeks. Among the different sampling sites around patient P6, the panel of the ventilator, sink, bedrail, and hanging tower remained CRKP positive for the whole period of 7 weeks during the ICU stay. In contrast, for the 15 patients without CRKP colonization or infection, CRKP was only detected once in the weekly environment sampling. This suggests that the environmental contamination for patients without CRKP colonization or infection is usually transient and restricted.

Most CRKP isolates belonged to ST11, either the KL47 or KL64 capsular type.

A total of 49 CRKP isolates from 21 patients, including 9 from screening swabs (five from rectal swabs and four from oral), five from clinical samples, and 35 from the environment, were sequenced. The vast majority of the 49 CRKP isolates belonged to ST11 (n=45, 91.8%), either to the KL47 (n=22) or KL64 (n=23) capsular type, while the remaining four belonged to ST16 KL51 (n=3) and ST37 KL55 (n=1). All of the ST11 isolates had the carbapenemase-encoding gene bla_KPC-2 and an extended-spectrum β-lactamase (ESBL)-encoding gene bla_CTX-M-65. The three ST16 CRKP isolates had the carbapenemase-encoding gene bla_NDM-5 and an ESBL-encoding gene, bla_CTX-M-15, while the ST37 isolate exhibited low-level resistance to meropenem (MIC, 4mg/liter). No known carbapenemase-encoding genes were identified in this ST37 isolate, but its ompK36 was interrupted by an insertion sequence of the IS5 family. This isolate also had an ESBL-encoding gene, bla_CTX-M-27, and an AmpC-type cephalosporinase-encoding gene, bla_DHA-1. The combination of OmpK36 alteration and production of ESBL or AmpC has previously been reported to lead to low-level carbapenem resistance (10,–12).

Interpatient transmission and environmental contamination by ST11 KL47 KPC-2-producing CRKP clones.

The 22 ST11 KL47 KPC-2-producing CRKP isolates could be assigned to two clones based on single-nucleotide polymorphism (SNP) analysis (Fig. 3). Clone A consisted of 15 isolates associated with 10 patients, with 0 to 13 SNPs between them. Five isolates were recovered from a single patient (P2-1) from the rectal and oral swabs, a clinical sample (sputum), an instrument, and a non-instrument surface. The first isolate from clone A originated from a rectal swab of patient P2-1 that was collected on admission day, providing evidence that the clone was introduced into the ICU by this patient. A sputum sample from this patient with this clone was collected 16days after admission, suggesting that the patient developed a health care-associated infection after gut colonization. The first appearance of this clone in the environment of the ICU was 13days after admission of this patient, and subsequently clone A was also detected in the surroundings of 9 other patients scattered across the ICU in different beds (P5, P11, P12, P14, P15, P17-1, P17-2, and P19) across a period of 9weeks (Fig. 1), suggesting extensive and prolonged contamination of the ICU by this single clone.

Open in a separate window

FIG 3

Phylogenomic trees of ST11 CRKP isolates. Isolate name, patient number, bed number, and sample type (see Table 1 for details) are shown. All ST11 CRKP isolates had bla_KPC-2.

Clone B comprised seven isolates associated with four patients with 0 to 21 SNPs between them but which differed from the isolates in clone A by 27 to 44 SNPs (see Data Set S1 in the supplemental material). The first four isolates were from the same patient (P17-1) and were collected on the second day of ICU admission. The CRKP-positive instrument surface sample of this patient was collected on the same day, suggesting rapid contamination of CRKP from the patient to the environment. Clone B then was also found on an instrument surface of another patient (P19) and in a sink in the room of P8 (Fig. 1). In addition, an isolate of clone B was identified from an oral swab of patient P4 9weeks after the first isolation of the clone, suggesting prolonged transmission of the clone in the ICU.

The hospital acquisition and environment contamination of an ST11 KL64 KPC-2-producing CRKP clone.

Twenty-two of 23 ST11 KL64 KPC-2-producing CRKP isolates were assigned to clone C, as their core genomes differed by 0 to 9 SNPs (Data Set S2), while the remaining ST11 KL64 KPC-2-producing CRKP isolate had 30 to 36 SNPs compared to clone C (Data Set S2). Isolates of clone C were recovered from samples or the environment of seven patients, the environment of three other unoccupied beds, and three instruments in the corridor across a period of 7 weeks. This suggests extensive environment contamination of this clone in the ICU. The first isolate of clone C was recovered from a rectal swab of a patient (P7) on the first day of the study, suggesting that the clone had already been in the ICU before the start of the study. Isolates of clone C were also found in sputum, rectal, and oral swabs from another patient (P6) 26days after study initiation. This shows a hospital acquisition from either patient P7 or the contaminated environment (Fig. 1).

Environmental contamination of a ST16 KL51 NDM-5-producing CRKP clone.

There were no SNPs among the three ST16 isolates. The three isolates were recovered from a clinical sample (sputum) and an instrument surface (an infusion pump) of one patient (P4) and from a noninstrument surface (a light switch) of another patient (P3). The two patients were in two single rooms next to each other (Fig. 1), suggesting environment contamination and transmission of the CRKP clone. The isolate from the instrument surface was recovered 5 days previous to that from sputum of the same patient. Therefore, it is more likely that the patient had acquired the clone from the environment rather than vice versa. The three isolates were recovered in August, 2 months after the study start, suggesting that the ST16 NDM-5-producing strain was introduced into the ICU through an unknown source.

Environmental sampling.

In patient rooms, for each bed unit comprising a patient bed and its surroundings, we performed weekly sampling for the entire surface of all bedrails (right, left, head, and foot), the light switch, the stethoscope, the locker handle and buttons of the bed air pump, the button panel of ventilators, cardiovascular monitors, infusion pumps, air-purifying disinfectors, nebulizers plus their hand shanks, the 1,200-cm² (30cmby40cm) surface of the hanging tower, and nurse carts plus the entire surface of handles of nurse cart drawers. For each patient room, we also sampled the entire surface of computer mouse/keyboards as well as buttons of the drinking water dispenser weekly. Each of the 12 ward rooms has a handwashing sink. All sinks, including faucets, all surfaces, including the entire internal surface of the bowl including overflows, and drains were sampled once every 2 weeks.

Instruments such as ventilators, nurse carts, emergency rescue carts, vibration sputum ejection machines, Zimmer frames, and steppers (step trainers) are often left in the corridor outside patient rooms for temporary storage or when awaiting repair. These were also sampled if present when weekly sampling was performed. Outside patient rooms, we performed monthly sampling of the nurse station, the bronchoscope room, the treatment room, the cleaner room, the meeting room, and staff toilets. For the nurse station, we sampled the entire surface of computer mouse/keyboards, telephones, wireless interphones, barcode scanners/printers, and the medical record holder. The handwashing sink close to the nurse station in the corridor was also sampled at four sites as described above. For the bronchoscope room, we sampled the entire surfaces of the drying tabletop, the handle, the internal wall of the bronchoscope storage cabinet, the button panel of the washing machine, and inside washing tanks. We also sampled the lumen of the three bronchoscopes in addition to their operating buttons once during the 3-month period. For the treatment room, we sampled the entire surfaces of the light switch, the doorknob of the refrigerator, and a 30-cmby40-cm surface on the storage cabinets. For staff toilets, we sampled the entire surface of doorknobs. For the meeting room, we sampled the entire surface of computer mouse/keyboards and buttons of drinking water dispensers. For the cleaner room, we sampled the entire surface of mop handles and the hand shank of the cleaner vehicle and a 10-cmby10-cm area of the vehicle’s lid and body, respectively.

Surfaces and sinks were sampled using sterile rayon swabs (Copan; Brescia, Italy) moistened with tryptic soy broth (TSB; Hopebio). The swabs from surfaces were streaked onto Simmon’s citrate agar plates (supplemented with 1% inositol) containing 2μg/ml meropenem to screen CRKP. Swabs from sinks were immediately placed into 15-ml sterile tubes containing 6ml TSB after sampling and were processed as described below. The lumen of bronchoscopes was washed using 50ml phosphate neutralizing buffer containing 3 g/liter glycine, 10 g/liter peptone, 8.5 g/liter sodium chloride, and 1 g/liter Tween 80 (16), which was then passed through 0.22-μm filters (Millipore, Burlington, MA, USA). The filters were then incubated into 15-ml sterile tubes containing 6ml TSB. The tubes were incubated at 37°C overnight and centrifuged. Supernatants were discarded and cell pellets were resuspended in 1ml TSB. A 50-μl suspension aliquot was streaked onto Simmon’s citrate agar plates (supplemented with 1% inositol) containing 2μg/ml meropenem.

WGS and analysis.

For CRKP isolates from patients, the first isolate from each rectal swab, oral swab, and clinical sample was subjected to whole-genome sequencing (WGS). For CRKP isolated from environment sampling, up to three isolates were sequenced for each patient, including the first isolate from any instrument surface, the first isolate from any non-instrument (bedrails, bed air pumps, light switches, stethoscopes, locker handles, and hanging towers) surface, and the first isolate from any sink samples. Genomic DNA was prepared using the QIAamp DNA minikit (Qiagen, Hilden, Germany), and DNA sequencing libraries were prepared using the NEBNext Ultra II DNA library prep kit for Illumina (NEB, Ipswich, MA, USA). Reads were assembled into contigs using SPAdes v3.14.1 (17) with careful mode. Quality check was performed on the assembled genomes using CheckM v1.0.18 (18) to determine the existence of contamination. Sequence type (ST) was determined using the genome sequence to query the multilocus sequence typing database (http://bigsdb.pasteur.fr/klebsiella/klebsiella.html). Capsule (KL) typing and virulence factor prediction were performed using Kleborate v1.0.0 (19). Acquired antimicrobial resistance genes were identified using ResFinder (http://genomicepidemiology.org/). Gene sequences of outer membrane porins (OMPs) OmpK35- and OmpK36-encoding ompK35 and ompK36 of an ST37 CRKP isolate without known carbapenemase-encoding genes were compared with those of four carbapenem-susceptible ST37 isolates in our isolate collections to identity potential mechanism for carbapenem resistance by Clustal Omega (20).

SNP calling.

SNPs among CRKP isolates of the same ST and KL type were determined by mapping individuals against the complete genome of strains 090329 and 090357 and the draft genome of strain 090482 for strains of ST11-KL47, ST11-KL64, and ST16-KL51, respectively, using Snippy v4.6.0 (https://github.com/tseemann/snippy). Pseudo-multialignments of chromosomes were generated using scripts provided by Snippy, followed by recombination prediction and filtering using Gubbins v2.4.1 (21) under the GTRGAMMA model with a maximum of 50 iterations. Pairwise SNP distances were calculated from recombination-free chromosome alignments using snp-dists v0.7.0 (https://github.com/tseemann/snp-dists). Although the exact genome mutation rate of CRKP is unknown, several previous studies have estimated that it ranges from 3.9 to 18.4 SNPs per genome per year (22, 23). Therefore, we used a cutoff of≤21 high-quality core genome SNPs, as described previously (24), to define clones within the same ST and KL type.

Antimicrobial susceptibility testing.

MICs of meropenem were determined using the broth microdilution method of the Clinical and Laboratory Standards Institute (CLSI) (25).

Survival assay.

The survival of CRKP clones on polyvinyl chloride (PVC) surfaces at room temperature and humidity in the real ICU situation (temperature, 24 to 28°C; relative humidity, 56% to 60%) up to 43days was performed as we described previously (26). Briefly, each strain was incubated to mid-exponential growth phase, centrifuged, washed, and resuspended in 1ml phosphate-buffered saline (PBS) and finally adjusted to 0.5 MacFarland turbidity-equivalent standard (10⁸ CFU/ml). Aliquots (10μl) of each strain were added in wells (three wells per strain) of a sterile PVC 96-well plate (Corning, New York). After drying in ambient air overnight in a clean biosafety cabinet, the plates were put into a paper box and were then placed in a corner of the corridor in the ICU away from patient care areas. ICU staff were notified of the place to avoid any potential touches. Bacterial cells were recovered by adding 200μl PBS into each well, with shaking with 85±5rpm for 60min by a miniature orbital shaker (BETS-M1; Qilinbeier, Haimen, Jiangsu, China), streaking a 100-μl suspension on LB agar, and incubating at 37°C overnight. For clones with three or more isolates (clones A, B, C, and E), three isolates of each clone were selected for the assay, while for clones with only one isolate (clones D and F), the sole isolate was included. Escherichia coli strain ATCC 29922 and Klebsiella quasipneumoniae strain ATCC 700603 (carbapenem-susceptible) were used as controls.

We are grateful to the nurses in the ICU for performing active screening swabs. We thank Zongan Liang and He Yu for their helpful comments.

Z.Z., A.M., and W.V.S. designed the study. L. Wei, H.W., and S.Z. performed sampling. L. Wu and L.T. performed swab collection. L. Wei and H.W. performed microbiological procedures. Y.L. performed survival assays. S.Z. performed case investigations. Y.F., E.D., A.M., and Z.Z. performed analysis. Z.Z., E.D., A.M., and W.V.S. wrote the paper.

The work is part of the DETECTIVE project, supported by the National Natural Science Foundation of China (81861138055) and the Medical Research Council (MR/S013660/1).

We have no financial involvement with any organization or entity with a financial interest in or conflict with this study.