A Cluster of Sphingomonas koreensis Infections
Epidemiologic Investigation data-collection time). PacBio reads were assem-
Patient records were reviewed for possible com- bled and polished with the use of Canu.25 Further
mon sources of exposure, including hospital details on sequencing and bacterial culturing
rooms, wards, invasive procedures, dialysis, respi- methods are described in the Supplementary Ap-
ratory treatments, and therapeutic baths. Because pendix.
sphingomonas species reside in aqueous reser-
voirs, we cultured potable water (100 samples), Comparative Genomic Analyses
sink faucets (56 samples), ice machines (7 sam- To assess the genetic relatedness of the sphin-
ples), and other plumbing components (52 sam- gomonas isolates, the average nucleotide identity
ples) from October 2016 through December 2017. of each pairwise genome comparison was calcu-
Samples were obtained from water and from lated with the use of the Mash algorithm, ver-
faucets in the rooms in which inpatients were sion 2.0, on the basis of 1000 21-bp substrings
staying when they acquired S. koreensis infections. (k-mers) selected from each genome.26 Reads
Large-volume water samples were also collected from each S. koreensis isolate were mapped to the
from the main municipal intake pipe and its reference S. koreensis PacBio genome to assess
branches, from the heat exchanger, from down- genome content. For fine mapping, SNVs were
stream of the heat exchanger, from the recircu- identified in genomic regions shared among all
lating hot-water loop, and from pipes supplying isolates (the core genome). In brief, the MiSeq
rooms with known culture-positive sinks. With reads for each isolate were aligned to the PacBio
the aid of plumbers, three culture-positive sinks reference genome, and SNVs were identified
were disassembled into isolated components, and with the use of Snippy, version 3.2 (with the
any visible biofilm and attached pipe segments BWA-MEM algorithm, version 0.7.17, and Free-
were cultured. Bayes software, version 1.1.0; https://github.com/
tseemann/snippy), which requires a read depth
From November 2016 through December 2016, of at least 10×; we specified that at least 90% of
water samples were collected in triplicate from reads should support the variant nucleotide call.
25 sinks (samples of hot and cold water were A recombination-corrected core genome align-
collected from 13 manual sinks, and 12 mixed- ment was used to construct a phylogenetic tree
temperature samples were collected from auto- of the S. koreensis strains isolated from the NIH
matic sinks) (Table S2 in the Supplementary Clinical Center.
Appendix). The water samples were tested for
free and total chlorine concentrations with the To detect the underlying genetic heterogene-
use of a digital colorimetric assay (with the Hach ity of the primary environmental cultures, DNA
DR900 colorimeter) (Table S2 in the Supplemen- was prepared from an entire plate (or a quadrant
tary Appendix). Beginning in February 2017, the if densely populated), and ensuing sequence
chlorine concentrations were measured serially reads were aligned individually to the reference
from 13 faucets distributed throughout the hos- S. koreensis PacBio genome. Nucleotide frequency
pital, and hot-water temperature was measured at each unique SNV was assessed with the use of
at the heat exchanger two to five times weekly. Bam-readcount (https://github. c om/g enome/b am
-readcount). All data visualization was performed
Characterization of the Bacterial Isolates with the use of R Software, version 3.4.3 (R Foun-
Isolates were assessed by whole-genome sequenc- dation for Statistical Computing).27 An in-depth
ing, which was performed on an Illumina MiSeq description of the bioinformatic methods is pro-
instrument with 300-bp paired-end reads, as de- vided in the Supplementary Appendix. The se-
scribed previously.21,22 DNA sequence reads were quencing data for this study are linked to Na-
assembled with the use of SPAdes and polished tional Center for Biotechnology Information
with the use of Pilon.23,24 On average, each MiSeq BioProject number PRJNA445389.
genome assembly contained 93 contigs with a
mean N50 of 223,917 bp. To obtain a complete Results
genome, one S. koreensis isolate from the NIH
Clinical Center was sequenced with the use of Clinical Characteristics of the Patients
the PacBio RS II SMRT platform (with the P6 poly From 2006, a year after the opening of a new
merase and C4 sequencing kit, with a 240-minute inpatient hospital building, through 2016, S. kore-
n engl j med 379;26 nejm.org December 27, 2018 2531
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
The new england journal of medicine
ensis clinical isolates were identified in 12 pa- zole and ciprofloxacin, and therefore these agents
tients at the NIH Clinical Center (Table S1 in the were used for treatment. Other sphingomonas
Supplementary Appendix). Seven isolates from clinical isolates were resistant to fewer classes
patients were initially identified only at the ge- of antibiotics. Aside from a partially conserved
nus level, and the species was assigned retro- chloramphenicol acetyltransferase (catB) gene (96%
spectively. All the patients were hospitalized at coverage, 77% identity), no canonical resistance
the time of their positive cultures; 5 patients were genes could be identified within the S. koreensis
in the intensive care unit and 7 were in three genome, a finding that is consistent with intrin-
other wards. The median length of stay was 44 sic resistance or undiscovered mechanisms of
days (range, 0 to 374) before a positive culture antibiotic resistance.
was obtained. Four patients had a single positive
culture, and the other patients had either persis- Epidemiologic Investigation to Identify
tent or recurrent positive cultures over a range the Source of S. koreensis
of 2 to 43 days. One isolate per patient was se- S. koreensis grew from 22 of 56 faucets (39%) and
quenced. Among the 12 patients, 9 were recipients from 9 of 17 water samples (53%) collected from
of stem-cell transplants. Eight of the 12 patients faucets in patient rooms. Three positive water
had S. koreensis bacteremia, including 2 who had samples were collected from faucets with culture-
concurrent S. koreensis pneumonia and 3 who had negative swabs.
catheter-related bloodstream infections. In addi-
tion, 1 patient had pneumonia alone, 1 had cho- Environmental S. koreensis isolates showed re-
lecystitis that led to peritonitis, and 1 had a deep sistance to numerous antibiotics, similar to the
surgical-site infection. One of the 12 patients had isolates obtained from patients (Fig. S1 in the
S. koreensis cultured from urine with a low con- Supplementary Appendix). Isolates derived from
centration in the absence of pyuria; this patient sink components and water samples were geneti-
did not have a urinary tract infection, and the cally related to the S. koreensis isolates obtained
organism was believed to represent contamina- from patients in 2016 (>99.7% average nucleo-
tion or colonization. Of the remaining 11 patients, tide identity) (Table 1, and Fig. S1 in the Supple-
8 patients with S. koreensis infections recovered, mentary Appendix), which implicated sinks or
and 3 patients died (all 3 patients had S. koreensis water as the most likely source of nosocomial
sepsis as well as severe, unrelated infections) S. koreensis infections.
(Table S1 in the Supplementary Appendix).
The free chlorine concentration and water
Investigation of Clinical Isolates temperature of the samples were evaluated. Cold-
of Sphingomonas from 2016 water samples contained adequate free chlorine
Of the six clinical sphingomonas cases in 2016, concentrations (≥0.5 mg per liter, as recommend-
genomic sequencing classified four as S. koreensis, ed by the CDC28), but chlorine concentrations of
one as S. yanoikuyae, and one as S. trueperi. The hot-water samples were well below the recom-
four S. koreensis clinical isolates possessed an mended threshold. Starting in December 2016,
exceptionally high degree of genetic similarity hot water returning from the recirculating loops
(>99.92% average nucleotide identity), which sug- was flushed continuously at a rate that was cali-
gested that they belonged to the same clonal brated to maintain circulating free chlorine levels
strain. These S. koreensis isolates exhibited resis- above 0.5 mg per liter, effectively replacing water
tance to multiple classes of antibiotic agents, that had declining concentrations of free chlorine
including aminoglycosides (amikacin and genta- with more freshly chlorinated municipal water.
micin), beta-lactams (aztreonam, piperacillin– Concentrations of free chlorine in hot water rose
tazobactam, cefepime, ceftazidime, ceftriaxone, continuously from December 2016 through July
and meropenem), and fluoroquinolones (levo- 2017, reaching concentrations of approximately
floxacin) (Fig. 2) — a finding that reflects the 1.0 mg per liter. Water temperature was deter-
difficulty in treating these infections. The isolates mined at the heat exchanger and was adjusted
were susceptible to trimethoprim–sulfamethoxa- from a temperature of 46 to 49°C to a tempera-
ture of 60°C or higher in accordance with the
hospital standard of 51°C or higher.29
2532 n engl j med 379;26 nejm.org December 27, 2018
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
A Cluster of Sphingomonas koreensis Infections Resistant
Intermediate resistance
A Antibiotic Susceptibility of Sphingomonas Isolates Obtained from Patients Susceptible
No data
Amikacin
Gentamicin 100
Tobramycin 95
Aztreonam 90
Piperacillin–Tazobactam 85
Ticarcillin–Clavulanic Acid 80
75
Cefepime
Ceftazidime
Ceftriaxone
Imipenem
Meropenem
Ciprofloxacin
Levofloxacin
Colistin
Trimethoprim–Sulfamethoxazole
B Average Nucleotide Identity of Spingomonas Isolates Obtained from Patients
S−NIH.Pt3_0716
S−NIH.Pt1_0416
S−NIH.Pt20_1215
S−NIH.Pt19_1214
S−NIH.Pt18_1212
S−NIH.Pt17_1112
S−NIH.Pt15_0812
S−NIH.Pt13_0512
SK−NIH.Pt6_1016
SK−NIH.Pt5_1016
SK−NIH.Pt4_0816
SK−NIH.Pt2_0716
SK−NIH.Pt16_1112
SK−NIH.Pt14_0612
SK−NIH.Pt12_0112
SK−NIH.Pt11_0411
SK−NIH.Pt10_0210
SK−NIH.Pt9_0208
SK−NIH.Pt8_0208
SK−NIH.Pt7_0906
SSSSSSSSSSSSSSKKSSSSKKKKKKKKSS−KK−−−−−−−−−−−−−−−−−−−NNNNNNNNNNNIIIIIIIIIIINNNNNNNNNIIIIIIIIIHHHHHHHHHHH...........HHHHHHHHH.......P..PPPPPPPPPPtttttttttttPPPPPPPPPttttttt2111t111t111175426310901759838246____________________00101001100001010100008794722425821221161111011111101011111166666665410228228222
Figure 2. Genome Comparisons of the Sphingomonas Isolates Obtained from Patients at the NIH Clinical Center.
Panel A shows antibiotic susceptibility patterns for the sphingomonas isolates obtained from patients. Panel B shows the genetic related-
ness of the isolates on the basis of average nucleotide identity, with purple squares indicating the greatest similarity. Isolates are named
according to species (Sphingomonas koreensis [SK] or other sphingomonas species [S]), location (NIH), patient number (Pt), and date
(month [MM] and year [YY]). Sphingomonas isolates from 2016 are highlighted in red.
Historical Investigation of S. koreensis mentary Appendix). These isolates shared simi-
at the NIH Clinical Center lar levels of nucleotide identity with one another
and with isolates from 2016 (>99.8% average
Eight additional S. koreensis clinical isolates from nucleotide identity), which supports the persis-
the NIH Clinical Center were identified that tence of a reservoir (Fig. 2). When the isolates
dated back to 2006, a year after the opening of obtained from patients were compared with the
a new hospital building (Fig. S2 in the Supple-
n engl j med 379;26 nejm.org December 27, 2018 2533
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
The new england journal of medicine
Table 1. Sphingomonas koreensis Genome Sequences Compared with the Reference Genome.*
Isolate Source and Sequence No. of Isolates Mean Percent Aligned Average No. of SNVs
12 (Range)† (Range)‡
Patients at the NIH Clinical Center 55
Chromosome 1 99.11 (98.76–99.26) 18.91 (11–41)
Plasmid 1 95.86 (93.94–96.98) 60.18 (0–661)§
2
Sinks at the NIH Clinical Center 98.94 (97.87–99.19) 20.13 (12–29)
Chromosome 96.41 (90.98–97.05) 6.53 (0–109)¶
Plasmid
84.82 33,754
CDC 34.12 1949
Chromosome
Plasmid 97.09 145
72.42 1658
Wadsworth Center
Chromosome 84.66 (84.63–84.69) 36,188 (36,188–36,188)
Plasmid 2.61 (2.10–3.13) 86 (81–92)
Korean environmental samples‖
Chromosome
Plasmid
* The S. koreensis isolate SK-NIH.Pt5_1016, obtained from a patient at the NIH Clinical Center, was sequenced with the
use of the PacBio system and was used as the reference genome (GenBank accession number, PRJNA354050). External
clinical isolates were obtained from the Centers for Disease Control and Prevention (CDC) and from the Wadsworth
Center, New York State Department of Health (Albany, NY), and environmental isolates were obtained from Korea.
† T he percent aligned indicates the percentage of nucleotides in the reference sequence that have at least one read aligned
with the sequenced isolate.
‡ S ingle-nucleotide variants (SNVs) were called with the use of Snippy (https://github.c om/tseemann/s nippy).
§ One plasmid sequence from an isolate obtained from a patient contained 661 SNVs. The remaining 11 plasmid sequences
contained 0 to 1 SNV.
¶ F our plasmid sequences obtained from sinks contained 27 to 109 SNVs. The remaining 51 plasmid sequences con-
tained 0 to 3 SNVs.
‖ Isolates of S. koreensis from Korea (KCTC_2882 and KCTC_2883) were obtained from the Korean Collection for Type
Cultures.
reference S. koreensis genome, they were found to late from the CDC (Atlanta). Although they were
have 11 to 41 chromosomal SNVs and 0 to 1 SNV genetically distinct, the isolate from the Wads
on the plasmid, except for one variant plasmid worth Center was the most similar to the iso-
that had 661 SNVs (Table 1). Over the course of a lates from the NIH Clinical Center, with 97%
decade, this represents a very low mutation rate, chromosomal alignment and 145 SNVs. The iso-
which is consistent with the ability of sphin- lates from Korea and the CDC were notably less
gomonas to survive in stasis in low-nutrient similar to isolates from the NIH Clinical Center,
environments.30,31 with approximately 84% chromosomal align-
ment with the reference genome and more than
Characterization of External S. koreensis 33,000 SNVs (Table 1). These external isolates
Isolates were resistant to multiple classes of antibiotics,
To determine whether these genetically related as were the isolates from the NIH Clinical Cen-
S. koreensis isolates were unique to the NIH Clini- ter (Fig. S1 in the Supplementary Appendix),
cal Center, four external S. koreensis isolates were which underscores the intrinsic resistance of
obtained and sequenced: two isolates from the S. koreensis to antibiotics.
original type strains from environmental sam-
ples from Korea; a clinical blood isolate from Genomic Analysis to Identify the Reservoir
the Wadsworth Center, New York State Depart- The phylogenetic tree, constructed on the basis
ment of Health (Albany, NY); and a clinical iso- of chromosomal SNVs in the core genome for
2534 n engl j med 379;26 nejm.org December 27, 2018
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
A Cluster of Sphingomonas koreensis Infections
SK−NIH.Env1_1016 SK−NIH.Pt6_1016
SK−NIH.Env8_1116
SK−NIH.Env36_0717
SK−NIH.Env12_0417
SK−NIH.Env9_0217
SK−NIH.Env4_1116
SK−NIH.Env23_0717
SK−NIH.Env21_0617
SK−NIH.Env38_0817
SK−NIH.Env31_0717
SK−NIH.Env14_0417
SK−NIH.Env24_0717
SK−NIH.Env34_0717
SK−NIH.Env22_0617
SK−NIH.Env32_0717
SK−NIH.Env7_1116
SK−NIH.Env27_0717
SK−NIH.Env26_0717
SK−NIH.Env35_0717
SK−NIH.Env29_0717
SK−NIH.Env13_0417
SK−NIH.Env37_0717
SK−NIH.Env33_0717
SK−NIH.Env19_0617
SK−NIH.Env30_0717
SK−NIH.Env28_0717
SK−NIH.Env25_0717
SK−NIH.Env5_1116
SK−NIH.Pt12_0112
SK−NIH.Env18_0517
SK−NIH.Env42_0817
SK−NIH.Pt7_0906
SK−NIH.Env56_1217
SK−NIH.Pt14_0612
SK−NIH.Env20_0617
SK−NIH.Env2_1116
SK−NIH.Pt16_1112
SK−NIH.Env43_0817
SK−NIH.Env10_0317
SK−NIH.Env44_0817
SK−NIH.Pt9_0208
SK−NIH.Pt11_0411
SK−NIH.Pt8_0208
SK−NIH.Pt5_1016
SK−NIH.Env53_1017
SK−NIH.Env45_1017
SK−NIH.Env48_1017
SK−NIH.Env49_1017
SK−NIH.Pt4_0816
SK−NIH.Env15_0417
SK−NIH.Env41_0817
SK−NIH.Pt10_0210
SK−NIH.Env57_1217
SK−NIH.Env52_1017
SK−NIH.Env51_1017
SK−NIH.Env50_1017
SK−NIH.Env47_1017
SK−NIH.Env46_1017
SK−NIH.Env55_1217
SK−NIH.Env54_1217
SK−NIH.Pt2_0716
SK−NIH.Env16_0517
SK−NIH.Env11_0417
SK−NIH.Env17_0517
SK−NIH.Env3_1116
SK−NIH.Env39_0817
SK−NIH.Env40_0817
2 SNVs
Figure 3. Phylogenetic Tree of the S. koreensis Isolates from the NIH Clinical Center.
The phylogenetic tree was constructed on the basis of single-nucleotide variants (SNVs) across the core genome.
Circles of the same color at the end of the nodes indicate environmental isolates (Env) from the same sink. Black
circles indicate that no other isolates were cultured from a given sink. The circles at the end of the nodes for isolates
obtained from patients are colored to match the isolates from the sink in the room in which the patient was staying.
The labels of isolates obtained from patients are highlighted in red, and isolates obtained from sinks are labeled in
black.
S. koreensis isolates obtained from patients, re- Center yielded few additional connections, which
vealed no clear structure and did not reflect the suggested a widely distributed reservoir within
timing of the identification of the isolates (Fig. the hospital (Fig. 3). Only the final isolate ob-
S3 in the Supplementary Appendix). Environ- tained from a patient (SK-NIH.Pt6_1016) was
mental S. koreensis isolates from the NIH Clinical genetically linked to the corresponding sink
n engl j med 379;26 nejm.org December 27, 2018 2535
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
The new england journal of medicine
isolate, which was collected 15 days later (five fixtures. Components from two sinks were im-
shared SNVs) (Fig. S4 in the Supplementary Ap- mersed in 71°C water baths for 20 minutes after
pendix) — a finding consistent with a direct culturing. Subsequent cultures of heated compo-
transmission event. nents were negative. Aerators were removed from
affected sinks. Additional remediation strategies
S. koreensis was cultured from water in patient are under consideration. No further S. koreensis
rooms but was not detected in the municipal infections have occurred since the augmentation
water entering the hospital, in water sampled of free chlorine concentrations and the adjust-
from the large pipes branching off the main ment of hot-water temperature in December 2016.
intake pipe, or in ice machines. To devise a re-
mediation strategy, we performed metagenomic Discussion
sequencing of primary environmental cultures In 2016, a cluster of sphingomonas infections
grown from the components of sinks and water sparked an epidemiologic investigation that iden-
pipes behind the walls of patient rooms (Fig. S5 tified 12 patients over 11 years who had been
in the Supplementary Appendix). We leveraged infected with genetically similar strains of
core genome SNVs to identify individual S. kore- S. koreensis, a rarely reported pathogen. Sink fau-
ensis strains and their relative abundance within cets and water from numerous patient rooms
each sample. Sinks were found to be colonized were positive for S. koreensis, which implicated
by multiple S. koreensis strains — a discovery that hospital plumbing infrastructure as a possible
would not have been made with conventional reservoir. In this study, genomic and metage-
single-colony isolation methods (Fig. 4). Metage- nomic techniques provided a higher-resolution
nomic analysis identified three distinct isolates understanding of this intermittent cluster and
from a single sink faucet; this faucet was re- revealed a pervasive reservoir in the water system
placed and resampled monthly (Fig. 4A and B). of the NIH Clinical Center.
Four months after replacement, two of the three
original S. koreensis strains were detected again, Whole-genome sequencing of 68 S. koreensis
which suggested recolonization of the new fau- isolates from the NIH Clinical Center (obtained
cet by a proximal reservoir. The third strain was from patients and the plumbing system) revealed a
not observed after the faucet was replaced, which genetically diverse population, a phenomenon in
suggested a localized colonization within the creasingly observed in microbial outbreaks.11,13,32,33
removed faucet. A single sink was found to be Understanding genetic diversity in an outbreak
colonized by two strains; one had been isolated is of value and can be leveraged to dissect com-
previously from this same sink, and the other plex issues, such as identification of the point
had been isolated previously from the adjoining source and patient-to-patient transmission.34 One
room (Fig. 4C and D). The two rooms shared a limitation of this study is that we could not per-
water pipe, which is consistent with a common fectly match any isolate from a sink in a patient
reservoir. Nine faucets that were positive for room to a clinical isolate. Matching isolates
S. koreensis were replaced. No further replace- would have strongly supported S. koreensis trans-
ments were made after serial cultures showed mission, though we would still have been unable
that six of the new faucets (67%) became recolo- to identify the precise modes of transmission
nized with S. koreensis within 5 to 90 days after from sinks to patients. In addition, we lack
replacement. metagenomic data for the initial isolates from
patients and sinks because we were not aware of
Investigation of S. koreensis Colonization strain diversity in early samples. Despite these
within Plumbing Fixtures limitations, a link was found between isolates
To assess the extent of colonization in sink com- from one patient and the corresponding room
ponents and to explore possible remediation faucet, which were collected 15 days apart. In
strategies, we disassembled and cultured three contrast, other paired patient–sink samples were
sinks that showed S. koreensis colonization in fau- acquired months and years apart. The heteroge-
cets. Nine components grew S. koreensis, including neity among S. koreensis isolates provided a valu-
horizontal sections of hot-water and cold-water able “genetic barcode” to parse subsequent
pipes immediately proximal to the sink, mixing metagenomic samples and to enable the inference
valves, aerators, faucets, and other plumbing
2536 n engl j med 379;26 nejm.org December 27, 2018
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
A Cluster of Sphingomonas koreensis Infections
A S. koreensis Strains in the Same Sink, According to Date of Collection B Three S. koreensis Isolates Collected from the Same Sink
100 November 2016 November 2016
75
50
25
0
Relative Abundance (%) 100 March 2017 March 2017
75
50
25
0
100 April 2017 April 2017
75
50
25
0
SK–NIH.Env5_1116 SK–NIH.Env10_0317 SK–NIH.Env11_0417
C S. koreensis Strains in Sinks in Two Adjoining Rooms D S. koreensis Isolates Collected from Sinks in Two Adjoining Rooms
100 February 2017Relative Abundance (%) Room 1 Room 2
75
50
25
0 SK–NIH.Env9_0217 SK–NIH.Env18_0517
WATER
Figure 4. Metagenomic Sequencing of Multiple S. koreensis Strains Cultured from Sinks in Patient Rooms at the NIH Clinical Center.
Panels A and B show the relative abundance of three S. koreensis strains (SK-NIH.Env5_1116, SK-NIH.Env10_0317, and SK-NIH.
Env11_0417) collected from the same sink in one patient room in November 2016, March 2017, and April 2017. The relative abundance
of each isolate is represented by the colored bars in Panel A. Relative abundance was estimated by calculation of the mean allele frequency
of SNVs that were unique to each isolate (dots). The sink was replaced after the first positive sample was obtained in November 2016
(indicated by the gray sink in Panel B), but S. koreensis was cultured again from this sink in March 2017 and in April 2017. The colored
rods in Panel B correspond to the three isolates in Panel A. Panels C and D show the relative abundance of two other S. koreensis strains
(SK-NIH.Env9_0217 and SK-NIH.Env18_0517) that were detected in the sinks of two adjoining patient rooms. The colored rods in Panel
D correspond to the two isolates in Panel C. The rods that appear outside the circle indicate individually cultured isolates. The gray rods
in Panels B and D represent isolates with unknown SNV profiles. The water in the two rooms in Panel D was supplied by the same pipes
(blue bars).
of a reservoir at the peripheries of the NIH sinks may pose to patients, from both potable
Clinical Center water-distribution system. water and splashback from the drain.38-43 The
steps taken in this study to prevent further
Outbreaks of sphingomonas have been re- S. koreensis infections within the NIH Clinical
ported previously.35-37 The ubiquity of this genus Center are applicable to many opportunistic
in ostensibly clean water makes it a possible haz- waterborne pathogens.
ard to immunocompromised patients. Previous
studies have underscored the threat that in-room Externally collected S. koreensis isolates and
n engl j med 379;26 nejm.org December 27, 2018 2537
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
The new england journal of medicine
banked NIH Clinical Center clinical isolates al- sessed more than 80 sequenced isolates and 49
lowed us to address two important questions: metagenomic samples. Whole-genome sequenc-
has the 2016 strain caused infections before, and ing allowed us to reach back more than a decade
is the strain unique to our hospital? Our results to a time shortly after a new hospital building
suggest that a single S. koreensis strain entered was occupied by patients and to identify the
the water system soon after construction of the onset of a sporadic clonal outbreak.
new NIH Clinical Center hospital building in
2004. Reports describe colonization of pipes with Supported by the NIH Intramural Research Programs, includ-
waterborne bacteria in newly constructed, unused ing the National Human Genome Research Institute, the NIH
facilities in which water has stagnated.44,45 The Clinical Center, and the National Institute of Allergy and Infec-
expansive genetic diversity within a small sam- tious Diseases.
pling window supports our hypothesis that this
strain disseminated throughout the hospital and Disclosure forms provided by the authors are available with
diversified at multiple distinct locations.46 the full text of this article at NEJM.org.
This study was a systematic genomic investi- We thank the clinical laboratory scientists of the NIH Micro-
gation to understand the dynamics of an indo- biology Service, Department of Laboratory Medicine and Clini-
lent outbreak within a single hospital, and it as- cal Center for their contributions to this study; and Dr. Daniel
Kastner and Dr. Matthew J. Arduino for their helpful discussions
and expert review of previous versions of the manuscript. The
computational resources of the NIH High-Performing Compu-
tation Biowulf Cluster (http://hpc.nih.g ov) were used for this
study.
References sols in patients with pulmonary disease paucimobilis: a persistent Gram-negative
1. Antibiotic/antimicrobial resistance: big- caused by NTM. J Clin Microbiol 2013;51: nosocomial infectious organism. J Hosp
gest threats and data. Atlanta: Centers for 3006-11. Infect 2010;7 5:1 53-7.
Disease Control and Prevention, 2013 9. Snitkin ES, Zelazny AM, Thomas PJ, 18. Tablan OC, Anderson LJ, Besser R, et
(http://www.cdc.gov/drugresistance/ et al. Tracking a hospital outbreak of car- al. Guidelines for preventing health-care–
threat-report-2013). bapenem-resistant Klebsiella pneumoniae associated pneumonia, 2003: recom-
2. Requirement to reduce Legionella risk with whole-genome sequencing. Sci Transl mendations of CDC and the Healthcare
in healthcare facility water systems to pre- Med 2012;4:1 48ra116. Infection Control Practices Advisory Com-
vent cases and outbreaks of legionnaires’ 10. Snitkin ES, Won S, Pirani A, et al. Inte- mittee. MMWR Recomm Rep 2004;53
disease (LD). Baltimore:Centers for Medi- grated genomic and interfacility patient- (RR-3):1 -36.
care & Medicaid Services, June 2, 2017 transfer data reveal the transmission 19. Marbjerg LH, Gaini S, Justesen US.
(https://www.cms.gov/Medicare/Provider pathways of multidrug-resistant Klebsiella First report of Sphingomonas koreensis
-Enrollment-and-Cert if icat ion/Sur vey pneumoniae in a regional outbreak. Sci as a human pathogen in a patient with
Cert if icat ionGenInfo/Downloads/ Transl Med 2017;9 (417):p ii:eaan0093. meningitis. J Clin Microbiol 2015;53:
Survey-a nd-Cert-Letter-17- 30. pdf ). 11. Coll F, Harrison EM, Toleman MS, 1028-30.
3. Conger NG, O’Connell RJ, Laurel VL, et al. Longitudinal genomic surveillance 20. Wallner J, Frei R, Burkhalter F. A rare
et al. Mycobacterium simae outbreak asso- of MRSA in the UK reveals transmission case of peritoneal dialysis-associated
ciated with a hospital water supply. Infect patterns in hospitals and the community. peritonitis with Sphingomonas koreensis.
Control Hosp Epidemiol 2004;2 5:1 050-5. Sci Transl Med 2017;9 (413):p ii:eaak9745. Perit Dial Int 2016;36:2 24-5.
4. Williams MM, Armbruster CR, Ardu- 12. Burckhardt F, Brion A, Lahm J, et al. 21. Dotson GA, NISC Comparative Se-
ino MJ. Plumbing of hospital premises is Confirming Legionnaires’ disease outbreak quencing Program, Dekker JP, et al. Draft
a reservoir for opportunistically pathogen- by genome-based method, Germany, 2012. genome sequence of a Klebsiella pneu-
ic microorganisms: a review. Biofouling Emerg Infect Dis 2016;2 2:1303-4. moniae carbapenemase-positive sequence
2013;29:1 47-62. 13. Gardy JL, Johnston JC, Ho Sui SJ, et al. type 111 Pseudomonas aeruginosa strain.
5. Osawa K, Shigemura K, Abe Y, et al. A Whole-genome sequencing and social- Genome Announc 2016;4(1):p ii:e01663-15.
case of nosocomial Legionella pneumonia network analysis of a tuberculosis out- 22. Conlan S, Lau AF, NISC Comparative
associated with a contaminated hospital break. N Engl J Med 2011;3 64:730-9. Sequencing Program, et al. Complete ge-
cooling tower. J Infect Chemother 2014; 14. Worby CJ, Lipsitch M, Hanage WP. nome sequence of a Klebsiella pneumoniae
20:68-70. Within-host bacterial diversity hinders ac- strain carrying blaNDM-1 on a multidrug
6. Aumeran C, Paillard C, Robin F, et al. curate reconstruction of transmission net- resistance plasmid. Genome Announc
Pseudomonas aeruginosa and Pseudomo- works from genomic distance data. PLoS 2016;4 (4):pii:e00664-16.
nas putida outbreak associated with con- Comput Biol 2014;1 0(3):e 1003549. 23. Bankevich A, Nurk S, Antipov D, et al.
taminated water outlets in an oncohae- 15. Nandy S, Dudeja M, Das AK, Tiwari R. SPAdes: a new genome assembly algo-
matology paediatric unit. J Hosp Infect Community acquired bacteremia by Sphin- rithm and its applications to single-cell
2007;65:47-53. gomonas paucimobilis: two rare case re- sequencing. J Comput Biol 2012;1 9:455-77.
7. Chand M, Lamagni T, Kranzer K, et al. ports. J Clin Diagn Res 2013;7:2 947-9. 24. Walker BJ, Abeel T, Shea T, et al. Pilon:
Insidious risk of severe Mycobacterium 16. Cheong HS, Wi YM, Moon SY, et al. an integrated tool for comprehensive mi-
chimaera infection in cardiac surgery pa- Clinical features and treatment outcomes crobial variant detection and genome as-
tients. Clin Infect Dis 2017;6 4:3 35-42. of infections caused by Sphingomonas sembly improvement. PLoS One 2014;
8. Thomson R, Tolson C, Carter R, Coul- paucimobilis. Infect Control Hosp Epide- 9(11):e 112963.
ter C, Huygens F, Hargreaves M. Isolation miol 2008;2 9:990-2. 25. Koren S, Walenz BP, Berlin K, Miller
of nontuberculous mycobacteria (NTM) 17. Ryan MP, Adley CC. Sphingomonas JR, Bergman NH, Phillippy AM. Canu:
from household water and shower aero-
2538 n engl j med 379;26 nejm.org December 27, 2018
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
A Cluster of Sphingomonas koreensis Infections
scalable and accurate long-read assembly GG, et al. Capturing the cloud of diversity ward. Antimicrob Agents Chemother 2015;
via adaptive k-mer weighting and repeat reveals complexity and heterogeneity of 59:714-6.
separation. Genome Res 2017;27:722-36. MRSA carriage, infection and transmis- 41. Umezawa K, Asai S, Ohshima T, et al.
26. Ondov BD, Treangen TJ, Melsted P, sion. Nat Commun 2015;6 :6560. Outbreak of drug-resistant Acinetobacter
et al. Mash: fast genome and metagenome 34. Correia AM, Ferreira JS, Borges V, et baumannii ST219 caused by oral care us-
distance estimation using MinHash. Ge- al. Probable person-to-person transmission ing tap water from contaminated hand
nome Biol 2016;17:1 32. of Legionnaires’ disease. N Engl J Med hygiene sinks as a reservoir. Am J Infect
27. R:a language and environment for 2016;374:497-8. Control 2015;4 3:1 249-51.
statistical computing. Vienna:R Founda- 35. Kilic A, Senses Z, Kurekci AE, et al. 42. Kotay S, Chai W, Guilford W, Barry K,
tion for Statistical Computing, 2017. Nosocomial outbreak of Sphingomonas Mathers AJ. Spread from the sink to the
28. Free chlorine testing. Atlanta: Centers paucimobilis bacteremia in a hemato/ patient: in situ study using green fluores-
for Disease Control and Prevention (https:// oncology unit. Jpn J Infect Dis 2007;60: cent protein (GFP)-expressing Escherichia
www.cdc.gov/safewater/chlorine-residual 394-6. coli to model bacterial dispersion from
-testing. html). 36. Meric M, Willke A, Kolayli F, Yavuz S, hand-washing sink-trap reservoirs Appl
29. Sehulster L, Chinn RY, CDC, HICPAC. Vahaboglu H. Water-borne Sphingomo- Environ Microbiol 2017;83(8):pii:e03327-
Guidelines for environmental infection nas paucimobilis epidemic in an intensive 16.
control in health-care facilities: recom- care unit. J Infect 2009;58:2 53-5. 43. De Geyter D, Blommaert L, Verbraeken
mendations of CDC and the Healthcare 37. Mutlu M, Bayramoglu G, Yilmaz G, N, et al. The sink as a potential source of
Infection Control Practices Advisory Com- Saygin B, Aslan Y. Outbreak of Sphingomo- transmission of carbapenemase-produc-
mittee (HICPAC). MMWR Recomm Rep nas paucimobilis septicemia in a neonatal ing Enterobacteriaceae in the intensive care
2003;52(RR-10):1 -42. intensive care unit. Indian Pediatr 2011; unit. Antimicrob Resist Infect Control 2017;
30. La Duc MT, Dekas A, Osman S, Moissl 48:723-5. 6:2 4.
C, Newcombe D, Venkateswaran K. Isola- 38. Hota S, Hirji Z, Stockton K, et al. Out- 44. Francois Watkins LK, Toews KE, Har-
tion and characterization of bacteria ca- break of multidrug-resistant Pseudomo- ris AM, et al. Lessons from an outbreak of
pable of tolerating the extreme conditions nas aeruginosa colonization and infection Legionnaires’ disease on a hematology-
of clean room environments. Appl Envi- secondary to imperfect intensive care unit oncology unit. Infect Control Hosp Epide-
ron Microbiol 2007;7 3:2 600-11. room design. Infect Control Hosp Epide- miol 2017;38:306-13.
31. Novikova N, De Boever P, Poddubko S, miol 2009;30:2 5-33. 45. Stout JE, Brennen C, Muder RR. Le-
et al. Survey of environmental biocontam- 39. Lowe C, Willey B, O’Shaughnessy A, gionnaires’ disease in a newly constructed
ination on board the International Space et al. Outbreak of extended-spectrum long-term care facility. J Am Geriatr Soc
Station. Res Microbiol 2006;157:5 -12. β-lactamase-producing Klebsiella oxytoca 2000;48:1589-92.
32. Stoesser N, Sheppard AE, Moore CE, infections associated with contaminated 46. McElroy KE, Hui JG, Woo JK, et al.
et al. Extensive within-host diversity in fecal handwashing sinks. Emerg Infect Dis 2012; Strain-specific parallel evolution drives
ly carried extended-spectrum-beta-lacta- 18:1 242-7. short-term diversification during Pseudo-
mase-producing Escherichia coli isolates: 40. Leitner E, Zarfel G, Luxner J, et al. monas aeruginosa biofilm formation.
implications for transmission analyses. Contaminated handwashing sinks as the Proc Natl Acad Sci U S A 2014;1 11:E1419-
J Clin Microbiol 2015;53:2122-31. source of a clonal outbreak of KPC-2-pro- E1427.
33. Paterson GK, Harrison EM, Murray ducing Klebsiella oxytoca on a hematology Copyright © 2018 Massachusetts Medical Society.
my nejm in the journal online
Individual subscribers can store articles and searches using a feature
on the Journal’s website (NEJM.org) called “My Account.”
Each article and search result links to this feature. Users can create
personal folders and move articles into them for convenient retrieval later.
n engl j med 379;26 nejm.org December 27, 2018 2539
The New England Journal of Medicine
Downloaded from nejm.org on March 27, 2022. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
Surveillance and outbreak report
Investigation using whole genome sequencing of
a prolonged restaurant outbreak of Salmonella
Typhimurium linked to the building drainage system,
England, February 2015 to March 2016
John Mair-Jenkins1,2,3, Roberta Borges-Stewart⁴, Caroline Harbour⁵, Judith Cox-Rogers⁵, Tim Dallman⁶, Philip Ashton⁶, Robert
Johnston⁷, Deborah Modha⁸, Philip Monk⁴, Richard Puleston3,9
1. Field Epidemiology Training Programme, Public Health England, United Kingdom
2. European Programme for Intervention Epidemiology Training (EPIET), European Centre for Disease Prevention and Control
(ECDC), Stockholm, Sweden
3. Field Epidemiology Service, National Infection Service, Public Health England, United Kingdom
4. East Midlands Health Protection Team, Public Health England, United Kingdom
5. Environmental Health, Blaby District Council, Blaby, United Kingdom
6. Gastrointestinal Bacteria Reference Unit, National Infection Service, Public Health England, United Kingdom
7. Food Water and Environment Laboratory, National Infection Service, Public Health England, United Kingdom
8. Clinical Microbiology, Leicester Royal Infirmary, University Hospitals of Leicester NHS Trust, United Kingdom
9. University of Nottingham, School of Medicine, Division of Epidemiology and Public Health, United Kingdom
Correspondence: John Mair-Jenkins ([email protected])
Citation style for this article:
Mair-Jenkins John, Borges-Stewart Roberta, Harbour Caroline, Cox-Rogers Judith, Dallman Tim, Ashton Philip, Johnston Robert, Modha Deborah, Monk Philip,
Puleston Richard. Investigation using whole genome sequencing of a prolonged restaurant outbreak of Salmonella Typhimurium linked to the building drainage
system, England, February 2015 to March 2016. Euro Surveill. 2017;22(49):pii=17-00037. https://doi.org/10.2807/1560-7917.ES.2017.22.49.17-00037
Article submitted on 11 Jan 2017 / accepted on 04 Jun 2017 / published on 07 Dec 2017
Following notification of a Salmonella enterica serovar Introduction
Typhimurium gastroenteritis outbreak, we identified
82 cases linked to a restaurant with symptom onset It is estimated that over 38,000 community cases
from 12 February 2015 to 8 March 2016. Seventy-two of salmonellosis occur annually within the United
cases had an isolate matching the nationally unique Kingdom (UK) [1,2]. Salmonellosis often results from
whole genome sequencing profile (single nucleo- consumption of contaminated food or water [3], how-
tide polymorphism (SNP) address: 1.1.1.124.395.395). ever, transmission via asymptomatic shedding by food
Interviews established exposure to the restaurant handlers and exposure to contaminated environments
and subsequent case–control analysis identified an where conditions are favourable for pathogen survival
association with eating carvery buffet food (adjusted have also been implicated [3,4]. Here we report the
odds ratios (AOR): 20.9; 95% confidence inter- findings of an investigation of an outbreak of salmonel-
val (CI): 2.2 – ∞). Environmental inspections, food/ losis where the environment was pivotal in continued
water testing, and a food trace-back investigation transmission.
were inconclusive. Repeated cycles of cleaning were
undertaken, including hydrogen peroxide fogging, The event
however, transmission continued. After 7 months of
investigation, environmental swabbing identified 106 On 7 March 2015, Public Health England (PHE) East
isolates from kitchen surfaces and restaurant drains Midlands was alerted by the clinical microbiology lab-
matching the outbreak profile. We found structural oratory of a local hospital to 21 cases of Salmonella
faults with the drainage system and hypothesised enterica serovar Typhimurium gastroenteritis, with
that a reservoir of bacteria in drain biofilm and under- onset in February 2015. Seven cases in this initial phase
floor flooded areas may have sustained this outbreak. of the outbreak required hospitalisation. Following this
Ineffective drain water-traps (U-bends) may have also notification we suspected there was a community out-
contributed by allowing transmission of contaminated break of S. Typhimurium; investigations and attempts
aerosols into the kitchen environment. These findings to control the outbreak followed.
suggest that routine swabbing of sink drain points and Hypothesis-generating interviews at the outset of the
inspection of drainage systems should be considered investigation identified that several cases had eaten
in future outbreak scenarios. at the same restaurant during the incubation period
for their illness. Descriptive epidemiological analyses
www.eurosurveillance.org including subsequent cases pointed to the restaurant
being the likely source. This popular, purpose (newly)
1
Figure 1
Reported date of symptom onset of cases of a Salmonella outbreak and people not meeting case definitions with matching
whole genome sequencing of clinical isolates, United Kingdom, Week 7 (February) 2015–Week 14 (March) 2016 (n = 102a)
24 Notfication of outbreak, initiation of environmental Non case (n = 25)
22 investigations Possible case (n = 9)
20 Confirmed staff case (n = 2)
18 Confirmed case (n = 66)
16
Number of cases 14
12
10 WGS positive isolates from
environmental samples
8
6
Cases linked by WGS
4 Case-control study
2
0
7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 1 3 5 7 9 11 13
2015 2016
Week number
WGS: whole genome sequencing.
a Onset dates were not available for two confirmed asymptomatic staff cases, two other confirmed cases, one possible case, and six non
cases with matching WGS results (n = 11).
built restaurant had opened only 18 months before 2015 where we defined confirmed cases as UK resi-
the outbreak. The restaurant offered a full table-ser- dents with S. Typhimurium infection (with or without
vice menu, self-service salad bar and hot self-service gastroenteritis) with an isolate matching the nation-
carvery buffet serving roasted meats (turkey, beef, ally unique WGS outbreak profile (five single nucle-
gammon and pork at weekends) and vegetables and otide polymorphisms (SNP) single linkage cluster
condiments. Despite interventions to control the initial 1.1.1.124.395.395) [6] with symptom onset or a positive
outbreak, cases continued to emerge followed by a pro- sample taken within 12 days of visiting the restaurant.
longed period of transmission until 2016. The evolution Possible cases were defined as having gastroenteritis
of the investigation into this community outbreak and (diarrhoea – three loose stools within 24 hours – or
subsequent control measures is described, with spe- any two of abdominal pain, fever or nausea) within 12
cific reference to the use of whole genome sequencing days of visiting the restaurant [7,8].
(WGS) to link isolates and the role of the drains in con- As part of the outbreak investigations a case–control
tinued pathogen transmission. study was undertaken, to understand the associa-
tion between consuming restaurant food or drink and
Methods illness, with the assumption that a food associated
Epidemiological investigations source was the most likely vector. Cases were recruited
and were requested to nominate people who had eaten
Case finding used information from existing cases, with them as controls (1:1 ratio). No other criteria were
local primary care practitioners and pre-existing sur- set for control selection. After data collection any con-
veillance systems (statutory disease and routine labo- trols who met case definitions were reassigned as
ratory notifications, including routine WGS) [5]. Active cases. We asked about exposure to food and drink and
case finding using restaurant booking information was potential confounding exposures: immunosuppression,
not possible as most customers did not pre-book. recent use of antibiotics or antacids, age and sex. We
Hypothesis-generating interviews and descriptive calculated univariate odds ratios (OR), and adjusted
analyses informed a case–control study in March ORs (aOR) using multivariable exact logistic regression.
2 www.eurosurveillance.org
Figure 2
Maximum likelihood phylogenetic tree of clinical (red), sewer (blue) and other environmental samples matching the SNP
address 1.1.1.124.395.395 of a Salmonella outbreak strain, United Kingdom, February 2015–March 2016 (n = 220 outbreak
sequencesa)
SNP: single nt polymorphism.
a 114 sequences from clinical isolates (cases and non-cases), this includes sequences where there were multiple samples from cases. 106
sequences from environmental isolates are shown.
An asterisk (*) denotes the position of a clinical isolate in the tree. The first two asterisks at the base of the tree denote the root of the tree
and nearest neighbouring sequence.
Due to continued case occurrences after the implemen- Clinical microbiological investigations
tation of outbreak control measures (as informed by
the findings of the case–control study), further inves- Stool samples were collected for culture and charac-
tigations ensued, including increased scrutiny of the terisation following clinical investigations and volun-
environment. tary sampling of restaurant staff in March 2015 and
again following further case occurrences in November
www.eurosurveillance.org 2015. Positive isolates were submitted to the PHE
3
Table 1
Results of univariate analysis of food items eaten at a restaurant, implicated in a Salmonella outbreak, United Kingdom,
March 2015 (n = 30 persons investigated)
Food eaten Total Cases % Total Controls % Crude OR (95% CI) Exact p value
20 Exposed 100 10 Exposed 50
Carvery 19 10 5 30 Inda (4.21–Ind) 0.002
Any sharing platter 20 20 0 10 3 20 0 (0.00–0.56) 0.05
Side dish 20 0 0 10 2 10 0 (0.00–0.89) 0.1
Any burger 20 0 0 10 1 10 0 (0.00–Ind) 0.3
Any pie 19 0 0 10 1 10 0 (0.00–Ind) 0.3
Any fish 19 0 0 10 1 10 0 (0.00–Ind) 0.3
Any starter 20 0 0 10 1 30 0 (0.00–Ind) 0.3
Any pudding 20 0 15 10 3 90 0.41 (0.04–3.97) 0.3
Any drink 3 90 9 1 (0.02–21.72)
18 1
CI: confidence interval; ind: indeterminate; OR: odds ratio.
No cases or controls ate items from the following menu sections: jacket potatoes, light bites, salads, steaks, hero dishes, classic dishes.
a Exact logistic regression, median unbiased estimate univariate OR: 21.37 (95% CI: 2.52– ∞).
Gastrointestinal Bacteria Reference Unit (GBRU) for that could provide an ongoing intermittent exposure
serological confirmation, phage typing and multilo- which had not previously been investigated. Repeated
cus variable-number tandem repeat analysis (MLVA), cycles of cleaning were undertaken at the restaurant,
until April 2015. Extracted DNA from isolates was also including deep cleans using hydrogen peroxide fog-
sequenced using Nextera library preparation (Illumina, ging, however these still failed to control the inci-
Inc. San Diego, California, United States) on an Illumina dent. Further investigations including a Water Fittings
HiSeq 2500 machine. High quality Illumina reads were Inspection [11], sewer sampling using a non-validated
mapped to the S. Typhimurium reference genome method similar to a Moore swab [12] and an inspection
(GenBank: AE006468) as previously described [9]. Core of the restaurant drainage systems were undertaken.
genome positions that had a high quality SNP (> 90% The drains were visually inspected and microbio-
consensus, minimum depth 10×, genotype quality ≥ 30) logical sampling of the kitchen floor drains and sink
in at least one strain were extracted and RaxMLv8.17 drains was undertaken. Expert advice was sought on
(Stamatakis, 2014) used to derive the maximum like- the movement of bio-aerosols along drainage systems
lihood phylogeny of the isolates under the general [13]. Drainage systems were inspected internally using
time-reversible (GTR)-CAT model of evolution. Single remote cameras, swabbing was conducted and smoke
linkage SNP clustering was performed [9]. FASTQ reads testing was used to establish air flows and the integrity
from all sequences in this study can be found at the of the drainage. Repeat inspections and environmental
PHE Pathogens BioProject at the National Center for swabbing were undertaken after remedial works.
Biotechnology Information (Accession PRJNA248792). Environmental samples were submitted to PHE Food,
Isolates within the five SNP single linkage cluster Water and Environmental Microbiology Laboratories.
1.1.1.124.395.395 were defined as part of the outbreak. These were tested for Salmonella sp. using routine
detection methods for surface swabs and food [14].
Environmental investigations and site visits Testing methods were adapted for non-routine samples
including kitchen cloths and spray bottles. Isolates
Environmental Health Officers (EHOs) carried out a from positive samples were referred to GBRU for WGS
routine inspection at the restaurant and reviewed staff and SNP analysis allowing comparison with clinical
sickness records when the outbreak was first identi- samples.
fied. Surface swabs using Dacron tipped swabs and
sponge swabs, along with food and water samples Results
were collected using standard techniques [10]. Repeat
site visits and inspections continued throughout the Eighty-two cases (72 confirmed and 10 possible)
outbreak. A food trace-back investigation was con- reported visiting the restaurant. The onset of symp-
ducted to identify the suppliers of meat, eggs, stuff- toms (mainly diarrhoea) ranged from 12 February 2015
ing and gravy to establish if local suppliers were being to 8 March 2016 with apparent point source outbreaks
used or whether larger national/international suppliers in February and June 2015, followed by a prolonged
had links to other reported cases or outbreaks of S. period of transmission (Figure 1). Four cases (all con-
Typhimurium. firmed) worked in the restaurant kitchen or served
In November 2015, following failure to control the out- food at the restaurant, two of whom were asympto-
break, we considered potential environmental sources matic (identified through staff sampling).
4 www.eurosurveillance.org
Table 2 for 16 isolates, 58 isolates were untypable, and one
Final exact logistic regression multivariable model of isolate was not phage-typed. A single MLVA profile
carvery food items eaten at a restaurant implicated in was identified (3–14–9-0–0211) in 34 cases (one addi-
a Salmonella outbreak, United Kingdom, March 2015 tional case had a single locus variant). This profile had
(n = 29a) been associated with human cases and pigs in several
regions of England (data not shown). In addition WGS
Exposure AOR P value was available for 96% (72/75) of isolates with the first
results available 7 days after the start of the investi-
Carvery (95% confidence interval) < 0.007 gation, providing increased discrimination over MLVA
Sex (male) 20.9 (2.2–∞) > 0.7 results.
Recent antibiotic use 2.7 (0.2–148.6) > 0.99 All the isolates clustered within a nationally unique
1.0 (0.7–63.6) five SNP cluster (which is further referred to as the
outbreak clade), 71% (53/72) of isolates were iden-
AOR: adjusted odds ratio. tical across their core genome with no SNP differ-
a One case had no record of antibiotic use and was excluded from ences observed (outbreak profile). The remaining 19
case isolates varied from the most frequent genotype
the model (n = 29), model p < 0.005. by between one and five SNPs. An overview of the
sequences associated with this nationally unique out-
Cases had a median age of 28 years (range: break profile is shown in the dendrogram (Figure 2)
5 months–85 years), 65% (53/82) were female and with sequences from environmental sampling, and the
99% (81/82) lived in the same county as the restau- additional 31 sequences identified from people during
rant. Symptoms were mainly self-limiting although 19 case finding who did not meet case definitions (i.e.
cases were hospitalised (1 to intensive care with inva- non cases). Eight of 13 sequences from non-cases that
sive disease). We also identified 31 people during case reported not visiting the restaurant had no SNP differ-
finding whose isolates matched the nationally unique ence from the outbreak sequence. Three sub-lineages
WGS outbreak clade but reported not visiting the res- (Figure 2, marked A–C) which emerged from November
taurant, and had no apparent link to cases or takeaway 2015 were apparent. Prior to this, genetic variation had
food (n = 13) or declined interview (n = 18). These peo- been observed, but distinct sub-lineages had not been
ple did not meet the case definitions, but otherwise did identified, with the exception of three cases in March
not appear different from cases; median age: 28 years 2015 which appeared distinct from the main outbreak
(range: 10 months – 85 years), 52% (16/31) female, with lineage (Figure 2, marked D).
similar symptoms and onset dates, however 19% (6/31)
lived in other areas of England or Northern Ireland. Environmental investigations
Case–control study All routine environmental inspections at the restaurant
had been satisfactory before the outbreak. Eighteen
At the time of the case–control study, 44 cases reported inspections were carried out during the investigation;
having eaten at the restaurant. Twenty (17 confirmed, 3 until November 2015 these were inconclusive and
possible) responded to the case–control questionnaire therefore the restaurant remained open. There were
(response rate: 20/44). Cases only nominated 10 con- no reports of staff illness before the outbreak. Carvery
trols; failing to achieve a 1:1 ratio. food was not available for sampling due to the high
A range of foods, including uncooked salad items were food turnover, however, raw meat and fresh water sam-
investigated in the case–control study (Table 1). Eating ples were taken at the beginning of each cluster and in
food from the carvery was the only significant (p < 0.01) early 2016 (n = 21, all negative). The food trace-back
harmful exposure identified by univariate analysis investigations were inconclusive and did not identify
(OR: 21.37; 95% confidence interval (CI): 2.52 – ∞). any suppliers linked with other cases or outbreaks.
All cases had consumed carvery food. Multivariable Environmental sampling in the kitchen, restaurant, staff
analysis identified sex and recent antibiotic use, as rooms, toilets and flats above the restaurant was con-
confounders. The adjusted odds of becoming a case ducted throughout the investigation, with 354 samples
remained over 20 times greater after eating any carvery collected (Table 3). These were negative throughout the
food (aOR: 20.9; 95% CI: 2.2 – ∞, Table 2). Effect modi- early stages of the investigation, but from November
fication was not identified between covariates in the 2015 onwards S. Typhimurium matching the outbreak
final model. These results were confirmed by sensitiv- clade was isolated from 106 environmental samples
ity analyses (data not shown). Throughout the inves- (Table 3). Isolates from three sewer swabs matched
tigation, 94% (60/64) of cases with a known food the outbreak profile. From November 2015 samples
history reported eating food from the carvery; no cases taken from cleaning materials, the pot wash area and
reported eating salads. wash hand basins were also found to have isolates
that matched the outbreak profile. From January 2016
Clinical microbiological investigations further areas of the kitchen were implicated along with
drains (Table 3). The majority of environmental isolates
S. Typhimurium was isolated in 75 case samples (72
confirmed, 3 possible); stool samples were not avail- 5
able for seven cases. Phage-type DT193 was identified
www.eurosurveillance.org
Table 3
Summary of environmental samples (n = 375) with isolates implicated in a Salmonella outbreak (n = 106) by whole genome
sequencing, United Kingdom, March 2015–May 2016
Sample date Water samples Food samples Environmental samples WGS positive isolates within Positive sample source
(n) (n) (n) five SNP clade(n)
Mar 2015 1 0 26 0 NA
Jun 2015 0 1 16 0 NA
Sep 2015 1 2 27 0 NA
19 3 Sinks/wash hand basins
Nov 2015 0 0 7 Cleaning materials
67 3a Sewer swabs
Dec 2015 0 0 5 Kitchen cloths
47 2b Sinks/wash hand basins
Jan 2016 6 9 2b Cleaning materials
42 Surface and deep drain
Feb 2016 1 0 41 15 swabs
0 59 Kitchen cloths
Mar 2016 0 0 10 4 Sinks/wash hand basins
Apr 2016 0 0 354 12a Meat sink waste trap
May 2016 0 12 4 Floor swabs
Total 9 7 Kitchen surface swabs
10 Surface and deep drain
swabs
19 Sinks/wash hand basins
Surface and deep drain
1 swabs
NA
12 NA
NA
0
0
106
NA: not applicable; SNP: single nucleotide polymorphism; WGS: whole genome sequencing.
a Three isolates matched outbreak sequence (0 SNP difference).
b One isolate matched outbreak sequence (0 SNP difference).
fell into sub-lineages A–C, however several deep drain Interventions
swabs were identical to the outbreak sequence (Figure
2). The restaurant responded to the outbreak by rotat-
Mapping and visual inspection of the drainage sys- ing staff away from carvery duties in case they were
tems identified significant issues. Water filled traps unsuspecting carriers, retraining staff on hygiene pro-
(u-bends) designed to prevent foul air flow from the cedures, facilitating voluntary staff sampling, installing
drainage system into the building had failed and and monitoring additional hand washbasins, moving
smoke testing revealed some ineffective drain seals, raw meat storage outside of the kitchen, and conduct-
potentially allowing contaminated bio-aerosol to be ing deep cleans of the indoor environment, which over
disseminated into the kitchen. One sink drain was not the outbreak increased in frequency and progressed to
connected to any drainage system with waste water include weekly steam cleaning and hydrogen peroxide
pooling under the floor. Other larger drains had failed fogging. Cameras were also installed to monitor the
after leaking waste-water washed away the support- carvery buffet, following concerns that deliberate con-
ing substrate forming a cavity under the kitchen area. tamination may have occurred.
It transpired at that point that drainage water had, on The restaurant voluntarily closed for 2 days for clean-
occasion, risen into the kitchen area, although this had ing following positive environmental swab results in
not been previously reported. Substantial remedial November 2015 and again for several weeks after fur-
works were undertaken, however, these were found to ther positive sampling to allow deep cleaning, kitchen
have failed on re-inspection and so these drains were refitting, and restaurant refurbishment. Remedial
later decommissioned. work was undertaken on the drains in February 2016.
Capping of all kitchen floor drains and hydrogen per-
6 oxide fogging followed in March 2016. This appeared
to resolve issues of contamination. Throughout the
www.eurosurveillance.org
investigation EHOs and PHE staff remained in close kitchen environment if this hypothesis is correct. Direct
contact with the restaurant and provided advice. transmission from asymptomatic staff could also not be
ruled out as not all staff complied with the first round of
Discussion screening. We could not rule out the effect of potential
lapses in cleaning and kitchen hygiene practices that
Our investigation identified a prolonged outbreak of S. may also have sustained this outbreak. Eating at busy
Typhimurium linked to a restaurant where 82 cases weekend periods appeared to be linked to cases in the
(72 confirmed, 10 possible) had eaten food. There was descriptive analysis, suggesting working practices may
strong epidemiological evidence of a link to the res- have been less rigorously applied during busy periods,
taurant early in the investigation, supported by timely even after staff received additional hygiene training
WGS results which provided a highly discriminatory during the outbreak.
microbiological link between cases and confirmed an Many of the isolates sampled from drains belonged
initial point source outbreak. In total, 72 case isolates to one of several genetic sub-lineages which
and 106 environmental isolates from the restaurant had evolved from the main outbreak lineage.
clustered into a unique 5 SNP cluster with 61 isolates Several Salmonella isolates sampled from the drains
(53 from cases and eight from environmental samples) also matched the main outbreak sequence supporting
identical at the core genome level. However, the envi- the hypothesis that the drains were potentially an
ronmental link with the restaurant was not established important reservoir. The kitchen drains and sewer
until 7 months after the start of the outbreak when system were unlikely to be the original source and may
WGS positive isolates were obtained from the kitchen have become contaminated by asymptomatic staff, or a
and drainage system. diner [20]. It is also possible that a one-off contamina-
We found the drains had failed in several places and tion event from raw or undercooked food was respon-
hypothesised that a reservoir of bacteria in biofilm sible for seeding this outbreak as raw meats such as
[15] and flooded areas in underfloor cavities may have pork cannot be guaranteed to be Salmonella free,
sustained this outbreak, after repeated environmen- although our food trace-back did not find a supply
tal cleaning failed. Drainage problems in one area of chain problem.
the kitchen led to liquid from the drains seeping into We were unable to identify sources of infection or trans-
the kitchen suggesting a contamination pathway. We mission routes for the 31 people infected with the out-
found isolates matching the outbreak strain on kitchen break strain (13 reported no restaurant exposure). We
cloths, swabs from kitchen sinks, and pot wash areas considered that some of these people may have been
suggesting contact with sinks may have provided a unknowing secondary cases, given the national unique
second contamination pathway. We also identified WGS results, which had not been observed before or
ineffective drain water-traps potentially allowing the since this outbreak, and also that asymptomatic car-
movement of contaminated bio-aerosols [13]. Smoke riage was observed during the outbreak, however no
tests demonstrated the potential for dissemination of epidemiological links to outbreak cases with restau-
foul air into the kitchen. rant exposure were identified.
While aerosolised drain contamination has not been
previously described in S. Typhimurium outbreaks, Strengths and limitations
there is supporting evidence for this transmission
pathway for Gram-negative bacteria in hospital drain- A key strength of this investigation was the use of
age systems (modelled using Pseudomonas putida) routine WGS which strengthened the epidemiological
[13]. Bio-aerosol transmission in building drains was evidence of a point-source outbreak enabling rapid
implicated in a large outbreak of severe acute respira- implementation of control measures before a holiday
tory syndrome suggesting this as a feasible hypothesis weekend in March 2015. However, there were also a
[16]. Previous outbreak investigations have also iden- number of limitations. Early cases noted their suspi-
tified the prolonged survival of food-borne pathogens cions of the restaurant on social media which may have
such as Listeria monocytogenes in drainage systems of introduced recall bias in our case–control study, inflat-
food production plants [15,17]. Sinks and drains have ing measures of association. The analytical study had a
also been linked with prolonged bacterial outbreaks in poor response rate and a small sample size. This possi-
hospital settings [18,19]. Occult environmental contam- bly introduced type II error for other food exposures or
ination can therefore be difficult to identify and control potential confounders which may partially explain why
in diverse settings. other foods were not identified as potential sources of
There was strong epidemiological evidence that eating infection even though contamination in the kitchen was
carvery food was associated with illness, suggesting possible. Our analysis plan excluded protective food
it may have been an important transmission vehicle. exposures potentially increasing residual confounding
However, there was no other evidence to confirm con- in our final model.
tamination of food or utensils, and temperature logs Other information biases may have been present in this
suggested food was cooked appropriately therefore investigation as many cases were identified after the
the cooked food must have been contaminated in the hypothesis had been established, potentially leading
www.eurosurveillance.org 7
to over ascertainment of exposure at the restaurant. epidemiological evidence used to demonstrate a need
However, the investigation did not identify a common for actions to prevent further cases.
source other than the restaurant, no other outbreaks
with the unique profile have been identified and since Acknowledgements
the drainage issues have been resolved, no further We are grateful to all restaurant staff and customers who
cases have been identified. aided this investigation along with Srilaxmi Degala, Lesley
Evidence of environmental contamination was lacking Larkin, Dr Jeremy Hawker (Public Health England), Deb
before November 2015. This was possibly due to the Morgan, Drazenka Tubin-Delic, Tracy Bishop (Food Standards
lower density and frequency of sampling or the tech- Agency) for contributions to the Outbreak Control Team. We
niques used, although it is unclear if these or other fac- thank Dr Sam Bracebridge (Public Health England) for her cri-
tors resulted in inconclusive findings. tique of this manuscript.
Implications for public health Conflict of interest
None declared.
Salmonella outbreaks are commonly linked to food,
restaurants and water supplies [21-26]. Biofilm may Authors’ contributions
harbour Salmonella sp. in drains and long-term Incident Director: PM. Conceived and designed the case-
environmental contamination is possible [3], but rarely control study protocol: JMJ, RP. Data collection, processing,
reported nor is the potential for bio-aerosol related analysis or interpretation: JMJ, RBS, PM, CH, JCR, TD, PA, RJ,
contamination. Our findings suggest greater considera- DM, RP. Drafting the manuscript: JMJ, RBS, TD, PM, RP.
tion should be given to undertaking drain swabbing at
an early stage of restaurant and food related outbreak References
investigations. This will enable identification of simi-
larities between environmental and clinical isolates 1. Roberts JA, Cumberland P, Sockett PN, Wheeler J, Rodrigues
that may have previously not been possible before the LC, Sethi D, et al. Infectious Intestinal Disease Study Executive.
routine use of highly discriminatory WGS. The study of infectious intestinal disease in England: socio-
This outbreak was unexpected in a newly built restau- economic impact. Epidemiol Infect. 2003;130(1):1-11. https://
rant managed by a national chain. Defective drains doi.org/10.1017/S0950268802007690 PMID: 12613740
may have been a one-off incident. However, repeated
failures of drains and continued contamination of 2. Tam CC, Rodrigues LC, Viviani L, Dodds JP, Evans MR, Hunter
the kitchen identified here suggests the design and/ PR, et al. IID2 Study Executive Committee. Longitudinal study
or installation of drainage systems was suboptimal. of infectious intestinal disease in the UK (IID2 study): incidence
Increased inspections in the building process may be in the community and presenting to general practice. Gut.
required. Consideration should be given to drainage 2012;61(1):69-77. https://doi.org/10.1136/gut.2011.238386
systems which facilitate inspection of water-traps or PMID: 21708822
use of multiple traps should be considered to reduce
the impact of failure. 3. Silva C, Calva E, Maloy S. One Health and Food-Borne Disease:
Salmonella Transmission between Humans, Animals, and
Conclusions Plants. Microbiol Spectr. 2014;2(1):OH-0020-2013. 10.1128/
microbiolspec.OH-0020-2013 PMID: 26082128
This outbreak of a nationally unique strain
of S. Typhimurium was linked to a single restaurant. 4. Holman EJ, Allen KS, Holguin JR, Torno M, Lachica M. A
The defective drainage system in the restaurant may community outbreak of Salmonella enterica serotype
have acted as an environmental reservoir and dissemi- Typhimurium associated with an asymptomatic food handler
nation of bio-aerosol from the drains into the kitchen in two local restaurants. J Environ Health. 2014;77(2):18-20.
was plausible. This linked with possible lapses in PMID: 25226780
kitchen hygiene may have enabled intermittent con-
tamination of food during the period of the outbreak. 5. Ashton PM, Nair S, Peters TM, Bale JA, Powell DG, Painset A,
While the original source remains unclear, this pro- et al. Salmonella Whole Genome Sequencing Implementation
tracted outbreak was controlled after remedial work on Group. Identification of Salmonella for public health
the drains, a kitchen and restaurant refurbishment and surveillance using whole genome sequencing. PeerJ.
a deep clean involving hydrogen peroxide fogging. 2016;4:e1752. https://doi.org/10.7717/peerj.1752 PMID:
Public health professionals should consider drainage 27069781
systems and bio-aerosols as potential sources in any
outbreak of salmonellosis and environmental investi- 6. Ashton P, Nair S, Peters T, Tewolde R, Day M, Doumith M,
gations should include swabbing drains early in out- et al. Revolutionising Public Health Reference Microbiology
break scenarios. Investigators should work towards using Whole Genome Sequencing: Salmonella as an exemplar.
accessing timely WGS analysis, which in this inves- bioRxiv. 2015.
tigation was essential for case finding, establishing
a single most probable source and underpinning the 7. Brooks JT, Matyas BT, Fontana J, DeGroot MA, Beuchat LR,
Hoekstra M, et al. An outbreak of Salmonella serotype
8 Typhimurium infections with an unusually long incubation
period. Foodborne Pathog Dis. 2012;9(3):245-8. https://doi.
org/10.1089/fpd.2011.0992 PMID: 22283668
8. Ethelberg S, Lisby M, Torpdahl M, Sørensen G, Neimann J,
Rasmussen P, et al. Prolonged restaurant-associated outbreak
of multidrug-resistant Salmonella Typhimurium among
patients from several European countries. Clin Microbiol
Infect. 2004;10(10):904-10. https://doi.org/10.1111/j.1469-
0691.2004.00990.x PMID: 15373885
9. Ashton PM, Peters T, Ameh L, McAleer R, Petrie S, Nair S, et al.
Whole Genome Sequencing for the Retrospective Investigation
of an Outbreak of Salmonella Typhimurium DT 8. PLoS Curr.
2015;7:ecurrents.outbreaks.2c05a47d292f376afc5a6fcdd8a7a
3b6. https://doi.org/10.1371/currents.outbreaks.2c05a47d292
f376afc5a6fcdd8a7a3b6 PMID: 25713745
10. Public Health England Microbiology Services. Detection and
enumeration of bacteria in swabs and other environmental
samples. Version 2. London: Public Health England; 2014.
www.eurosurveillance.org
11. The Water Supply (Water Fittings) Regulations. 1999. Statutory License and copyright
Instrument: 1999 No. 1148 (1999). Available from: http://www. This is an open-access article distributed under the terms of
dwi.gov.uk/stakeholders/legislation/ws(fittings)regs1999.pdf the Creative Commons Attribution (CC BY 4.0) Licence. You
may share and adapt the material, but must give appropriate
12. Moore B, Perry EL, Chard ST. A survey by the sewage credit to the source, provide a link to the licence, and indi-
swab method of latent enteric infection in an urban area. cate if changes were made.
J Hyg (Lond). 1952;50(2):137-56. https://doi.org/10.1017/
S0022172400019501 PMID: 14938529 This article is copyright of the authors, 2017.
13. Gormley M, Aspray TJ, Kelly DA, Rodriguez-Gil C. Pathogen
cross-transmission via building sanitary plumbing systems
in a full scale pilot test-rig. PLoS One. 2017;12(2):e0171556.
https://doi.org/10.1371/journal.pone.0171556 PMID: 28187135
14. Public Health England Microbiology Services. Detection of
Salmonella species, Microbiology Services Food Water and
Environmental Microbiology, Standard Method. Version 2.
London: Public Health England; 2014.
15. Coughlan LM, Cotter PD, Hill C, Alvarez-Ordóñez A. New
Weapons to Fight Old Enemies: Novel Strategies for the
(Bio)control of Bacterial Biofilms in the Food Industry.
Front Microbiol. 2016;7:1641. https://doi.org/10.3389/
fmicb.2016.01641 PMID: 27803696
16. McKinney KR, Gong YY, Lewis TG. Environmental transmission
of SARS at Amoy Gardens. J Environ Health. 2006;68(9):26-30,
quiz 51-2. PMID: 16696450
17. Berrang ME, Meinersmann RJ, Frank JF, Ladely SR.
Colonization of a newly constructed commercial chicken
further processing plant with Listeria monocytogenes. J Food
Prot. 2010;73(2):286-91. https://doi.org/10.4315/0362-
028X-73.2.286 PMID: 20132673
18. Vergara-López S, Domínguez MC, Conejo MC, Pascual Á,
Rodríguez-Baño J. Wastewater drainage system as an occult
reservoir in a protracted clonal outbreak due to metallo-β-
lactamase-producing Klebsiella oxytoca. Clin Microbiol Infect.
2013;19(11):E490-8. https://doi.org/10.1111/1469-0691.12288
PMID: 23829434
19. Hota S, Hirji Z, Stockton K, Lemieux C, Dedier H, Wolfaardt
G, et al. Outbreak of multidrug-resistant Pseudomonas
aeruginosa colonization and infection secondary to imperfect
intensive care unit room design. Infect Control Hosp Epidemiol.
2009;30(1):25-33. https://doi.org/10.1086/592700 PMID:
19046054
20. Sánchez-Vargas FM, Abu-El-Haija MA, Gómez-Duarte
OG. Salmonella infections: an update on epidemiology,
management, and prevention. Travel Med Infect Dis.
2011;9(6):263-77. https://doi.org/10.1016/j.tmaid.2011.11.001
PMID: 22118951
21. Moffatt CRM, Combs BG, Mwanri L, Holland R, Delroy B,
Cameron S, et al. An outbreak of Salmonella Typhimurium
phage type 64 gastroenteritis linked to catered luncheons in
Adelaide, South Australia, June 2005. Commun Dis Intell Q Rep.
2006;30(4):443-8. PMID: 17330386
22. Ethelberg S, Sørensen G, Kristensen B, Christensen K, Krusell
L, Hempel-Jørgensen A, et al. Outbreak with multi-resistant
Salmonella Typhimurium DT104 linked to carpaccio, Denmark,
2005. Epidemiol Infect. 2007;135(6):900-7. https://doi.
org/10.1017/S0950268807008047 PMID: 17335629
23. Silk BJ, Foltz JL, Ngamsnga K, Brown E, Muñoz MG, Hampton
LM, et al. Legionnaires’ disease case-finding algorithm, attack
rates, and risk factors during a residential outbreak among
older adults: an environmental and cohort study. BMC Infect
Dis. 2013;13(1):291. https://doi.org/10.1186/1471-2334-13-291
PMID: 23806063
24. Ethelberg S, Lisby M, Torpdahl M, Sørensen G, Neimann J,
Rasmussen P, et al. Prolonged restaurant-associated outbreak
of multidrug-resistant Salmonella Typhimurium among
patients from several European countries. Clin Microbiol
Infect. 2004;10(10):904-10. https://doi.org/10.1111/j.1469-
0691.2004.00990.x PMID: 15373885
25. Ailes E, Budge P, Shankar M, Collier S, Brinton W, Cronquist
A, et al. Economic and health impacts associated with a
Salmonella Typhimurium drinking water outbreak-Alamosa, CO,
2008. PLoS One. 2013;8(3):e57439. https://doi.org/10.1371/
journal.pone.0057439 PMID: 23526942
26. O’Reilly CE, Bowen AB, Perez NE, Sarisky JP, Shepherd CA,
Miller MD, et al. Outbreak Working Group. A waterborne
outbreak of gastroenteritis with multiple etiologies among
resort island visitors and residents: Ohio, 2004. Clin Infect Dis.
2007;44(4):506-12. https://doi.org/10.1086/511043 PMID:
17243052
www.eurosurveillance.org 9
pathogens
Review
Water as a Source of Antimicrobial Resistance and
Healthcare-Associated Infections
Claire Hayward 1,*, Kirstin E. Ross 1 , Melissa H. Brown 2 and Harriet Whiley 1
1 Environmental Health, College of Science and Engineering, Flinders University, Adelaide,
South Australia 5042, Australia; kirstin.ross@flinders.edu.au (K.E.R.); harriet.whiley@flinders.edu.au (H.W.)
2 College of Science and Engineering, Flinders University, Adelaide, South Australia 5042, Australia;
melissa.brown@flinders.edu.au
* Correspondence: hayw0077@flinders.edu.au; Tel.: +61-87221-8585
Received: 27 July 2020; Accepted: 14 August 2020; Published: 18 August 2020
Abstract: Healthcare-associated infections (HAIs) are one of the most common patient complications,
affecting 7% of patients in developed countries each year. The rise of antimicrobial resistant (AMR)
bacteria has been identified as one of the biggest global health challenges, resulting in an estimated
23,000 deaths in the US annually. Environmental reservoirs for AMR bacteria such as bed rails,
light switches and doorknobs have been identified in the past and addressed with infection prevention
guidelines. However, water and water-related devices are often overlooked as potential sources
of HAI outbreaks. This systematic review examines the role of water and water-related devices in
the transmission of AMR bacteria responsible for HAIs, discussing common waterborne devices,
pathogens, and surveillance strategies. AMR strains of previously described waterborne pathogens
including Pseudomonas aeruginosa, Mycobacterium spp., and Legionella spp. were commonly isolated.
However, methicillin-resistant Staphylococcus aureus and carbapenem-resistant Enterobacteriaceae
that are not typically associated with water were also isolated. Biofilms were identified as a hot spot
for the dissemination of genes responsible for survival functions. A limitation identified was a lack
of consistency between environmental screening scope, isolation methodology, and antimicrobial
resistance characterization. Broad universal environmental surveillance guidelines must be developed
and adopted to monitor AMR pathogens, allowing prediction of future threats before waterborne
infection outbreaks occur.
Keywords: antibiotic resistance; antimicrobial resistance; water; waterborne outbreak; healthcare
associated infection; biofilm
1. Introduction
Healthcare-associated infections (HAIs) are defined as infections caused as a direct or indirect
result of an individual receiving healthcare [1]. This may occur in hospitals, aged care facilities,
dental clinics and long-term care facilities [2]. The United States (US) Centers for Disease Control and
Prevention (CDC) have estimated that 1 in 25 hospital patients are diagnosed with a HAI each year [3].
Additionally, there are over 4 million HAIs in Europe, 1.7 million in the US and 165,000 in Australia
annually [4]. HAIs result in unnecessary morbidity and mortality with estimates from the US indicating
HAIs are responsible for approximately 99,000 unnecessary deaths every year [4]. Hospital patients
and aged care residents are especially vulnerable to infection due to their potentially compromised
immune systems [5]. HAIs are commonly associated with catheters, surgical sites and ventilators [6],
where the causative organisms may originate from the patient’s own microbial flora, other patients,
staff or from the healthcare facilities physical environment [5]. The US CDC have identified a
number of causative agents that pose serious threats to hospitalized patients including Acinetobacter
Pathogens 2020, 9, 667; doi:10.3390/pathogens9080667 www.mdpi.com/journal/pathogens
Pathogens 2020, 9, 667 2 of 21
spp., influenza, Klebsiella spp., methicillin-resistant Staphylococcus aureus (MRSA), Clostridium difficile,
Pseudomonas aeruginosa, non-tuberculous mycobacteria (NTM) and norovirus [7]. The significance and
severity of HAIs are increasing due to the rise in antimicrobial resistance and emergence of multidrug
resistance (MDR) [8]. Thus, treatment for patients suffering HAIs resistant to traditional antibiotic
therapies is more precarious, costly and, in the worst case scenario, unsuccessful [6]. The increase
in antimicrobial resistance is driven, in part, by the inappropriate use of antibiotics and ineffective
disinfectant protocols [9]. Understanding potential environmental reservoirs of infectious bacterial
species is needed to develop and implement effective infection control [1]. Strategies for the prevention
of person-to-person transmission are well defined, including disinfection procedures of dry surface
fomites such as bed rails, doorknobs and light switches [1,10–12]. However, there are limited studies
investigating the role of environmental microorganisms, including waterborne pathogens such as
Legionella spp., P. aeruginosa and Mycobacterium spp. [13–27]. It has been estimated that 20% of
nosocomial pneumonias are caused by waterborne P. aeruginosa in the US, resulting in a conservative
annual mortality of approximately 1400 individuals [28]. An outbreak of L. pneumophila infection in the
neonatal unit of a private hospital was linked to a cold-mist humidifier filled with contaminated tap
water, resulting in nine infections and three deaths [29]. Transmission of these waterborne pathogens
may occur via water related devices such as showers, drinking fountains, bathtubs, dental units,
ice machine, humidifiers, sinks and toilets [27]. Notably, approximately 80% of chronic and recurrent
microorganism infections are caused by biofilms [30], which are communities of microorganisms,
providing protection from adverse environmental conditions and antimicrobial agents [30].
This systematic review examined the role of water in the transmission of AMR pathogens that
are responsible for HAIs. Common waterborne devices, pathogens, and surveillance strategies are
discussed. A greater understanding of the ecological niche of these pathogens is needed to develop
improved management strategies for the prevention of waterborne HAIs.
2. Results
Two thousand, two hundred, and one papers were retrieved from SCOPUS and Web of Science
using the search terms identified (Figure 1). After applying the inclusion and exclusion criteria
described in Figure 1, a total of 88 papers were included for review. These were further divided such
that 21 papers (presented in Table 1) described studies specifically investigating the presence of AMR
bacteria in water and water-related devices including tap faucets, drains, showers, and baths. A further
67 papers that did not specifically investigate water but included some water sampling are presented
in the Table S1. These include clinical outbreak investigations and other studies screening a range of
environmental sources within healthcare facilities.
Pathogens 2020, 9, 667
Table 1. Summary of reports and studies identifying antimicrobial resistant
Study Site Reservoir Organism Country * Bacterial Isolation
Hospital Water Greece * Methods ♦
Legionella spp.
Hospital Water Greece ISO 11731 (filtration
Burkholderia cepacia untreated, heat and
Pseudomonas stutzeri
acid treatments)
Chryseobacterium plated on GVPC
meningosepticum
Stenotrophomonas maltophilia agar
Enterobacter cloacae
Acinetobacter baumannii Membrane filtratio
Escherichia coli and plated on
Proteus mirabilis
Alcaligenes xylosoxidans m-endo medium
Pseudomonas aeruginosa and cetrimide aga
Pseudomonas putida
Serratia liquefaciens
Moraxella osloensis
Serratia plymuthica
Hospital Water Acinetobacter haemolyticus Brazil MPN, APHA 2000
B. cepacia plated on
Pseudomonas aeruginosa MacConkey agar
P. stutzeri
Hospital Hot water Legionella pneumophila Italy Italian guidelines
system for prevention and
control of
Legionellosis
3 of 21
bacterial species within healthcare water sources and water-related devices.
n Antimicrobial Antimicrobial Characteristics Additional Reference
Methods † Comments × [31]
Five strains displayed low-level
n, resistance to CIP and ERY SGs 1–15
d identified.
Antibiotics
E-test strips
tested:
CIP, ERY
S. maltophila isolate resistance:
37% resistant to CAZ
58% resistant to FEP
on Agar dilution 100% resistant to IPM Antibiotics [32]
ar E. coli isolates: tested:
55% resistant to TIC AMK, CAZ, CIP,
P. mirabilis, P. putida, S. liquefaciens, FEP, IPM, TET,
TIC, SXT, TOB
P. stutzeri and S. plymuthica
exhibited resistance to tetracycline
19% of the total enterobacteria and
35% of the total non-fermenting
isolates were MDR
0 B. cepacian isolates showed Antibiotics [33]
Disc diffusion resistance to 10/11 antibiotics tested: [34]
P. aeruginosa isolates showed
r resistance to 11/11 antibiotics AMK, CAZ,
A. haemolyticus isolates showed CCHL, CIP, FEP,
resistance to 11/11 antibiotics GEN, IPM, TET,
TMP, TOB, TZP
P. stutzeri isolates showed
resistance to 7/11 antibiotics
MIC values of L. pneumophila SG 1 Antibiotics
s were higher than non-SG 1 isolates tested:
d VITEK-2 for AZI, CIP, LEV, MOX, and TIG AZM, CIP, CLR,
No difference in MIC values CTX, DOX, ERY,
between SGs for CEF, CLA, DOX, LVX, MXF, RIF,
ERY, and RIF TGC
Pathogens 2020, 9, 667
Study Site Reservoir Organism Country * Table 1
Legionella spp.
Bacterial Isolation
Methods ♦
Hospital Water Turkey Culture methods
system
Hospital Water L. pneumophila Spain UNE-ISO
11731:2007
Acinetobacter spp. (filtration:
Aeromonas spp. untreated, acid and
Citrobacter spp. heat treatments)
Enterobacter spp. plated on GVPC
Escherichia coli
Klebsiella oxytoca agar
Hospital Water Klebsiella Turkey Membrane filtratio
pneumoniaeLeclercia and inoculated in
MacConkey broth
adocarboxylata and MacConkey
Pseudomonas spp. agar
Hospital Water Serratia spp. France Membrane filtratio
P. aeruginosa
4 of 21
1. Cont. Antimicrobial Characteristics Additional Reference
n Antimicrobial Comments × [35]
MICs:
Methods † Greatest MIC to CLR Antibiotics [36]
tested:
s Broth dilution E-test strips:
Greatest average MIC resistance AZM, CIP, CLR,
d E-test strips, LVX, RIF
Disc diffusion from CIP and DOX
Lowest average MIC resistance Antibiotics
tested:
from AMC and AZT
Disc diffusion: E-test strips:
AMC, AZM,
Greatest average disc inhibition CIP, CTX, DOX,
from AZT and AMC ERY, LVX, MXF
Disc diffusion:
Lowest average disc inhibition AMC, AZM,
from SXT and RIF CIP, CTX, ERY,
FOX, LVX, MXF,
RIF, SXT
on E. coli isolates: Antibiotics
1 isolate resistant to CRO tested:
n Disc diffusion 5 isolates resistant to AMP [37]
h PCR 1 isolate resistant to PIP AMC, AMK,
AMP, CAZ, CEF,
Other species: CHL, CIP, CRO,
3 Pseudomonas spp. isolates FEP, FOX, GEN,
showed resistance to CAZ, IMP IPM, MEM, PIP,
TET, SXT, TZP
and GEN
on Disc diffusion Copper tolerant isolates. Antibiotics [38]
tested:
AMK, ATM,
CAZ, CIP, FEP,
FOF, IPM, MEM,
TOB, TZP
Pathogens 2020, 9, 667
Study Site Reservoir Organism Country * Table 1
P. aeruginosa India
Hospital Water Bacterial Isolation
Methods ♦
Membrane filtratio
Plated on R2A aga
immediately and o
either cetrimide,
Columbia + 5%
horse blood or R2A
after 14 days
Hospital Water P. aeruginosa Tanzania Water sample
inoculated directly
in malachite-green
broth then
subcultured on
blood and cetrimid
agar
Hospital Water Sphingomonadacae spp. Portugal * Membrane filtratio
Dental and plated on R2A
chair GSP, Pseudomonas
isolation and
tergitol-7 agar
5 of 21
1. Cont. Antimicrobial Characteristics Additional Reference
Comments ×
n Antimicrobial
Methods † All isolates showed resistance to Antibiotics [39]
TET and PEN tested:
on.
ar 2 isolates resistant to STR NET, OFX, PEN,
on 4 isolates resistant to NET STR, TET
Disc diffusion 5 isolates showed MDR
A
y Resistance (% of isolates): Antibiotics [40]
n ETP (2.6%); IPM (2.6%); TZP tested: [41]
(2.6%); TOB (5.1%); GEN (12.8%);
VITEK-2 CIP (15.4%); PIP (18%); FOF AMK, ATM,
CAZ, CIP, CST,
de (61.5%); ATM (100%) ETP, FEP, FOF,
GEN, IPM,
MEM, PIP, TOB,
TZP
Two hospitals
sampled; one
received water
from a deep
drilled well and
the other from
Lake Victoria
Hospital taps resistance (% of
isolates):
on ATB PSE EU TIM (2%); CIP (11%); MEM (17%); Antibiotics
A, system CAZ (21%); FEP (26%); TSU (30%); tested:
s TIC (36%); TOB (36%); LVX (42%);
FOS (42%); PIP (49%); TZP (36%); CAZ, CIP, CST,
FEP, FOF, GEN,
CST (94%)
Dental chair resistance (% of IPM, LVX,
MEM, PIP, TIC,
isolates): TIM, TOB, TSU,
TZP (17%); CAZ (17%); MEM
(17%); TOB (17%); TSU (17%); FEP TZP
(33%); GEN (33%); CIP (33%); TIC
(50%); PIC (67%); COL (83%)
Pathogens 2020, 9, 667
Study Site Reservoir Organism Country * Table 1
USA Bacterial Isolation
Medical Drain Achromobacter spp. Italy
centre Acinetobacter anitratus Methods ♦
Taps
Hospital Shower Acinetobacter lwoffi Drains swabbed
Residential Drinking Aeromonas spp. and plated on
care home fountain
Enterobacter agglomerans deoxycholate agar
Enterobacter cloacae biplate with GEN
Flavobacterium spp.
Moraxella spp. and AMK
Pseudomonas acidovorans UNI EN ISO
P. aeruginosa 16266:2008.
Membrane filtratio
Pseudomonas spp. and plated on
Pseudomonas cepacia Pseudomonas aga
Pseudomonas fluorescens
Pseudomonas putida with CN
supplement
P. stutzeri
Stenotrophomonas maltophila
P. aeruginosa
Hospital Water Legionella spp. Poland Culture methods
Sanatorium
Hospital Shower Erythrobacter spp. USA Biofilm removed
head Mycobacterium spp. from inner surface
Novosphingobium spp. and resuspended t
Sphingomonas spp. be plated on R2A
agar
6 of 21
1. Cont. Antimicrobial Characteristics Additional Reference
Comments ×
n Antimicrobial
Methods †
r Selective media Resistance (% of isolates): Antibiotics [42]
N AMK (77%); GEN (88%) tested:
AMK, GEN
on Disc diffusion 7.72% resistant to imipenem. Antibiotics
PCR 13.2% resistant to >1 antibiotic tested:
[43]
ar DNA AMK, ATM, [44]
sequencing CAZ, CIP, DOR, [45]
FEP, GEN, IPM,
s E-test strips L. pneumophila SG2-14 isolated
from one sanatorium showed LVX, MEM,
es High-throughput NET, PIP, TIC,
to sequencing resistance to AZM TIM, TOB, TZP
A
Resistance genes found: Antibiotics
aac2ib tested:
aac2ic
aph3ic AZM, CIP, RIF
baca
bL2b N/A
ceob
mfpa
Pathogens 2020, 9, 667 Organism Country * Table 1
Study Site Reservoir P. aeruginosa
Hospital Tap water Bacterial Isolation
Methods ♦
France * Hospital
standard—culture
method
Hospital Haemodialysis Enterococci spp. Greece Membrane filtratio
water
Tap water
Hospital Water P. aeruginosa France Membrane filtratio
and plated on
cetrimide agar
Hospital Sink U-bend P. aeruginosa France U-bend content
collected and
centrifuged pellet
was streaked on
cetrimide agar
7 of 21
1. Cont. Antimicrobial Characteristics Additional Reference
Comments × [46]
n Antimicrobial 7 isolates have Opr-mediated [47]
Methods † resistance to IPM Antibiotics [48]
tested:
e Disc diffusion Resistance (% of isolates): [49]
PGFE RIF (43%) CAZ, IPM, PIP
STR (60%) Samples taken
on Agar diffusion before and after
1 isolate resistant to ERY ICU move for
on
Disc diffusion P. aeruginosa resistant to chlorine comparison
disinfection treatment
t Disc diffusion Antibiotics
Strains: tested:
ST1725 (2 MDR isolates)
ST539 (100% resistant to IMI) AMC, AMP, CIP,
ST1416 (2 MDR isolates) ERY, GEN, RIF,
ST540 (1 MDR isolate) STR, TMP, VAN
STI11 (100% resistant to IPM,
Antibiotics
9 MDR isolates) tested:
ST622 (7 MDR isolates)
ST520 (100% resistant to IPM, AMK, CAZ,
CTX, FOF, GEN,
1 MDR isolate) IPM, OFX, CIP,
RIF, TIM, TOB
Antibiotics
tested:
AMK, CAZ, CIP,
FEP, GEN, IPM,
MEM, TIC, TOB,
TZP
Pathogens 2020, 9, 667 Organism Country * Table 1
Study Site Reservoir
Bacterial Isolation
Methods ♦
Hospital Tap water P. aeruginosa Italy Membrane filtratio
P. fluorescens and placed on
Ralstonia picketti cetrimide agar
S. maltophila
Hospital Bathtub Citrobacter diversus Zambia Swabs of bathtub
Tap water Citrobacter freundiii and cultured on
Enterobacter aerogenes
agar
E. cloacae
E. coli
K. pneumoniae
Pantoea agglomerans
P. aeruginosa
Serratia marcescens
Staphylococcus aureus
* In countries where the study location was not specified in the article, it was assumed that t
Public Health Association, APHA; Glycine Vancomycin Polymyxin Cycloheximide agar, G
Spanish Organization for Standardization, UNE ISO. † Abbreviations: BioMerieux suscep
PGFE; BioMerieux identification and antibiotic susceptibility testing instrument, VITEK-2
minimum inhibitory concentration, MIC; methicillin-resistant Staphylococcus aureus, MRSA; se
amoxicillin-clavulanic acid, AMC; ampicillin, AMP; azithromycin, AZZM; aztreonam, AZ
ceftriaxone, CRO; cephalothin, CEF; chloramphenicol, CHL; ciprofloxacin, CIP; clarithromyc
ERY; fosfomycin, FOF; fusidic acid, FA; gentamicin, GEN; imipenem, IPM; levofloxacin, LV
NET; ofloxacin, OFX; penicillin, PEN; piperacillin, PIP; piperacillin-tazobactam, TZP; rifam
TIM; tigecycline, TGC; tobramycin, TOB; trimethoprim, TMP; trimethoprim-sulfamethoxazo
phosphonium chloride, TTPC; didecyldimethylammonium chloride, DDAC; 2,2-dibromo-3-
(hydroxymethyl)phosphonium sulfate, THPS; sodium hypochlorite, NaOCl; benzalkonium c
8 of 21
1. Cont. Antimicrobial Characteristics Additional Reference
n Antimicrobial Comments × [50]
P. aeruginosa:
Methods † 17 strains non-MDR Antibiotics
tested:
on 4 MDR
ATB PSE 5 3 XDR AMK, AMP +
S. maltophila: SUL, CAZ, CIP,
1 strain non-MDR CST, FEP, FOF,
8 strains MDR
P. fluorescens: GEN, IPM,
1 MDR strain MEM, SXT,
TIM, TOB, TZP
b MRSA found on bathtubs Comparison of [51]
PCR clinical isolates
collected at the
same time
the country of origin was denoted by the country of the authors. ♦ Abbreviations: American
GVPC; International Organization for Standardization, ISO; most probable number, MPN;
ptibility test, ATB-PSE-EU; polymerase chain reaction, PCR; pulse gel field electrophoresis,
2. × Abbreviations: extended-spectrum beta-lactamase, ESBL; multidrug resistant, MDR;
erogroup, SG; extensively drug resistant, XDR. Antimicrobial abbreviations: AMK, amikacin;
ZM; aztreonam, ATM; cefepime, FEP; cefotaxime, CTX; cefoxitin, FOX; ceftazidime, CAZ;
cin, CLR; colistin, CST; doripenem, DOR; doxycycline, DOX; ertapenem, ETP; erythromycin,
LVX; meropenem, MEM; methicillin, MET; moxifloxacin, MXF; neomycin, NEO; netilmicin,
mpin, RIF; streptomycin, STR; tetracycline, TET; ticarcillin, TIC; ticarcillin-clavulanic acid,
ole, SXT; vancomycin, VAN; sulbactam, SUL; methylisothiazolinone, MIT; tributyl tetradecyl
-nitrilopropionamide, DBNPA; hydrogen peroxide + silver nitrate, H2O2 + AgNO3; tetrakis
chloride, BZK; cotrimoxazole, TSU; mupirocin, MUP.
Pathogens 2020, 9, 667 9 of 21
2.1. Study Sites
Of the 21 papers that specifically investigated the presence of pathogens associated with HAIs in
water sources (Table 1), 15 studies were from Europe, 3 from North and South America, 2 from Africa
and 1 from Asia. Seventeen studies sampled water sources from hospitals, one from a residential
care home, one from dental chair units, one from a medical center and another from a sanatorium.
AMR bacterial species were found in potable water samples (15 studies), followed by showers (2 studies)
and building water distribution systems (2 studies), sinks (1 study), baths (1 study), haemodialysis
water (1 study) and drains (1 study).
In those clinical outbreak investigations and studies examining a range of environmental sources
(Table S1), there were 27 reports from Europe, 22 from Asia, 11 from the Americas, 6 from Africa
and 1 published from Oceania. Of these studies, 30/67 found AMR bacterial contamination within a
water source, including tap water, hydrotherapy pool water, nasogastric water, and incubator water.
Taps and tap components such as aeration grids, tap handles, and hands-free taps had AMR bacterial
contamination in 18/67 studies. Sink and sink components such as drain holes, sink surfaces, drainpipe
leaks and sink traps were found to have multidrug resistant (MDR) bacterial contamination resistant
to two or more antimicrobials in 45/67 studies (Table S1). Shower components such as the shower
hoses, showerhead and outlets were contaminated with AMR bacteria in 11/67 studies. Baths were
found to have MDR bacterial contamination in 4 studies and bath toys were identified as a source of
contamination in 1 study [14,16,52–54].
2.2. Identified Pathogens Associated with HAIs
Seven of the studies used culture-based techniques to investigate the bacterial diversity in the water
sources in healthcare facilities. The pathogens identified are detailed in Table 1 and include Achromobacter
spp., Acinetobacter spp., Acinetobacter anitratus, Acinetobacter baumannii, Acinetobacter haemolyticus,
Acinetobacter lwoffi., Aeromonas spp., Alcaligenes xylosoxidans, Burkholderia cepacia, Chryseobacterium
meningosepticum, Citrobacter spp., Citrobacter diversus, Citrobacter freundii, Enterobacter spp., Enterobacter
aerogenes, Enterobacter agglomerans, Enterobacter cloacae, Erythrobacter spp., Escherichia coli, Flavobacterium
spp., Klebsiella oxytoca, Klebsiella pneumoniae, Leclercia adocarboxylata, Moraxella spp., Moraxella osloensis,
Mycobacterium spp., Novosphingobium spp., Pantoea agglomerans, Proteus mirabilis, Pseudomonas spp.,
Pseudomonas acidovorans, P. aeruginosa, Pseudomonas cepacia, Pseudomonas fluorescens, Pseudomonas putida,
Pseudomonas stutzeri, Ralstonia picketti, Serratia spp., Serratia liquefaciens, Serratia marcescens, Serratia
plymuthica, Sphingomonas spp., S. aureus, and Stenotrophomonas maltophila.
Fourteen studies investigated the presence of one specific bacterial species or genus that may
cause HAIs in water and water-related devices. Of these, seven papers investigated P. aeruginosa
exclusively, three investigated Legionella spp. and two papers focused specifically on L. pneumophila.
One paper focused on Enterococci spp. and another focused on Sphingomonadacae spp. (Table 1).
Twenty studies undertook comprehensive environmental bacterial screens of the study sites. These
studies included additional pathogens such as Acidovorax spp., Acinetobacter johnsonii, Aeromonas caviae,
Aeromonas hydrophila, Alkaligenes faecalis, Bosea spp., Chryseobacterium spp., Chryseobacterium indologenes,
Elizabethkingia meningoseptica, Enterobacter asburiae, Enterococci spp., Klebsiella ozenae, Methylobacterium
spp., Mycobacterium chelonae, Pantoea calida, Proteus spp., Proteus vulgaris, Providencia stuartii, Raoultella
ornithinolytica, Raoultella planticola, Sphingomonas paucimobilis, Staphylococcus citrus, Staphylococcus
epidermidis and Staphylococcus spp., as shown in Table S1. However, due to the design of some studies,
it was not always clear whether these bacterial species were isolated from the water samples taken or
from other environmental sources.
Thirty-five of 67 (Table S1) investigated bacterial clinical outbreaks in one or more
healthcare facilities identified contamination of water and/or a water related device as the likely
source of transmission via strain comparison. This included HAI outbreaks of Achromobacter
bacteraemia, Achromobacter denitrificans, Achromobacter xylosoxidans, Acinetobacter bereziniae, A. hydrophila,
carbapenem-resistant Enterobacteriaceae (CRE), carbapenem-resistant E. coli, Citrobacter amalonaticus,
Pathogens 2020, 9, 667 10 of 21
C. freundii, Collinsella aerofaciens, Comamonas testosterone, E. cloacae complex, Klebsiella spp., Pseudomonas
medocina, Pseudomonas nitroreducens, Pseudomonas oleovorans and P. putida. A surveillance review of
waterborne diseases in the US from 2013 to 2014 found that there were 42 outbreaks from drinking
water, resulting in 13 deaths all caused by Legionella spp. [55].
Nine studies compared clinical bacterial isolates and environmental isolates, including those
from water samples for molecular epidemiology in non-outbreak settings. These studies included
bacterial species such as Aeromonas spp., Burkholderia spp., Klebsiella quasipneumoniae, P. aeruginosa and
S. maltophila, as shown in Table S1.
2.3. Antimicrobial Resistance of Identified Strains
Several AMR pathogens of concern, as classified by the US CDC, were identified by studies included
in this review (Table 1 and Table S1). Specifically, three studies detected CRE, one from a plumbing
fixture, one from a water sample and one sample site was unspecified [21,52,56]. MDR P. aeruginosa
strains were also found in 12 studies, most commonly from potable water samples (7 studies),
sinks (3 studies) and faucets (2 studies) [14–16,33,39,40,43,49,50,57–59]. Eight studies reported AMR
Acinetobacter spp. of which five reported MDR isolates and one study identified the resistance genes tetG,
ermX and ermF in bacteria within a biofilm sample [32,37,42]. Additionally, the resistance gene OXA-23
was found in A. baumannii sampled from hospital water which has been linked to β-lactam antibiotic
resistance [60]. Specific genetic elements such as Opr protein-mediated resistance to fluoroquinolone
antibiotics was also found in P. aeruginosa isolates [37]. MRSA was detected in every bathroom sink tap
that was tested in a UK hospital. However, it is unclear which antibiotics this specific environmental
isolate was resistant to [61]. Sixteen studies that investigated water and water-related devices found
bacterial isolates that were resistant to two or more of the antibiotics that were tested (Table 1).
One study investigating P. aeruginosa, P. stutzeri, B. cepacian and A. haemolyticus in hospital water
samples found that all isolates were resistant to seven or more of the 11 antibiotics that were tested,
including amikacin, ceftazidime, chloramphenicol, ciprofloxacin, cefepime, gentamicin, imipenem,
tetracycline, trimethoprim, tobramycin and piperacillin-tazobactam [33]. One study into the presence
of L. pneumophila in a hospital hot water system found that the minimum inhibitory concentration (MIC)
values were higher in serogroup 1 isolates compared to non-serogroup 1 isolates for the antibiotics
azithromycin, ciprofloxacin, levofloxacin, moxalactam and tigecycline [34]. Resistance to β-lactamase
inhibitors such as tazobactam and clavulanic acid was identified in K. oxytoca, P. calida, R. ornithinolytica
and P. aeruginosa isolated from hospital sinks, drains, shower heads, water and aerators [15,25,62].
Biofilm samples taken from hospital shower heads contained Erythrobacter spp., Mycobacterium spp.,
Novosphingobium spp. and Sphingomonas spp. isolates that carried the resistance genes aac2Ib, aac2Ic,
aph3Ic, bacA, bL2b, ceoB and mfpA that have been linked to biofilm formation, virulence, peroxide
resistance, DNA repair, antibiotic resistance, and antigenic variation traits [45].
2.4. Detection Methods
There was significant variation in the methods used for detecting bacterial species from the
environment. Fifteen studies (Table 1) examined water using culture techniques. Specifically, eleven
studies performed membrane filtration followed by plating onto selective agar media, nine of these
studies used 0.45 µm pore diameter filters and two did not specify (Table 1). Of those that specifically
investigated Legionella spp., two studies referenced the International Organization for Standardization
(ISO) 11731—water quality enumeration of Legionella [31,36]. One study investigating L. pneumophila
followed Italian guidelines for prevention and the control of legionellosis [34] and two studies used other
culturing techniques [35,44]. Of the studies investigating P. aeruginosa, four papers used membrane
filtration methods followed by plating onto selective media such as R2A, cetrimide, and Columbia with
horse blood, one of which referenced the ISO 16266:2008—detection and enumeration of P. aeruginosa
specifically [38,39,43,48]. One paper alternatively inoculated malachite-green broth with the individual
environmental water sample and subcultured onto cetrimide agar to isolate P. aeruginosa [40]. Bacterial
Pathogens 2020, 9, 667 11 of 21
species from biofilm and swab samples taken from water-related devices were isolated using a variety
of methods including direct inoculation onto cetrimide, MacConkey, tegritol-7, or deoxycholate
agar, and centrifugation to resuspend a pellet for inoculation onto selective agar, as shown in
Table 1 [33,37,42,45,49]. Five studies used additional methods such as polymerase chain reaction (PCR),
matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF), VITEK-2, multiplex PCR and
16S gene sequencing to identify isolated bacterial species (Table S1) [63–65].
2.5. Antimicrobial Resistance Characterization Methods
A range of methods were used to determine the antimicrobial resistance characteristics of isolated
strains. Seventy-one of 88 studies (Table 1 and Table S1) used traditional microbiological methods
including disc diffusion (56 studies), agar dilution (4 studies), broth microdilution (5 studies) and
E-test strips (6 studies). Other approaches for characterizing antimicrobial resistance included PCR
(17 studies) and comparison to known AMR strains using VITEK-2 system (5 studies), pulse field
gel electrophoresis (PFGE) (3 studies), microscan (2 studies), microarray (1 study) and multilocus
sequencing typing (MLST) (1 study).
Comparing the antimicrobial resistance is challenging due to the varying approaches used in the
different studies. A joint initiative by the European CDC and US CDC provided definitions for the
terms MDR and XDR to standardize international terminology. To facilitate these definitions, lists of
antimicrobial categories and breakpoints were developed from the Clinical Laboratory Standards
Institute (CLSI), the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the
United States Food and Drug Administration (FDA). MDR was defined as acquired non-susceptibility
to at least one agent in three or more antimicrobial categories. XDR was defined as non-susceptibility
to at least one agent in all but two antimicrobial categories [66]. The terms MDR and extensively drug
resistant (XDR) were used by five studies and the terminology has been reported as stated in the papers
(Table 1 and Table S1); however, it was unclear what specific antibiotics the isolates were resistant
to [32,39,49,50,58]. Of the studies detailed in Table S1, 20/67 studies reported the environmental isolates
as a whole data set rather than describing the phenotypes of each individual strain.
3. Discussion
3.1. Water as a Source of HAIs
Water sources and water-related devices are often contaminated with pathogens responsible
for HAIs. This may occur when microorganisms survive treatment protocols or via end point
contamination [67]. The design of a hospital or healthcare facility’s water system can influence
the risk of microbial contamination [68]. Complex infrastructure may have points of heat transfer
and stagnation which can promote biofilm formation, microbial growth and the rise or transfer of
antimicrobial resistance [69]. The CDC Antibiotic Resistance Threats Report estimated that there are
more than 2.8 million AMR infections each year in the US resulting in approximately 35,000 deaths [8].
This review identified that water and water-related devices play a significant role in the transmission
of AMR HAIs with subsequently an economic and health imperative to improve the control of hospital
and healthcare water sources.
This review identified a range of waterborne pathogens present in the potable water supply and
plumbing surfaces (such as drains and tap faucets). However, pathogens not typically considered
waterborne were also detected, including S. aureus, Moraxella spp. and E. aerogenes [32,42,51,54,70–73].
For example, AMR pathogens of concern, extended-spectrum beta-lactamase-producing
Enterobacteriaceae and MRSA, were located in a hospital sink bowl, hospital bathroom sink taps
and a hospital bathtub [51,61,74]. This raises the hypothesis that end point contamination may be
occurring from patient-to-water source. A study examining the influence of contaminated splash backs
when handwashing in twenty faucet/sinks in hospital intensive care units found that the faucet spouts
were more contaminated than the sink bowl and drains. Flawed sink design such as shallow bowls
Pathogens 2020, 9, 667 12 of 21
enable splashing contaminated sink contents onto patient care items, healthcare workers hands and
the patients’ broader environment [75].
Numerous approaches are taken to ensure a facility’s potable water supply is suitable for human
use and consumption. The Healthcare Infection Control Practices Advisory Committee (HICPAC)
has published guidelines to prevent the growth of bacterial species such as Legionella spp. [69].
This includes recommendations such as maintaining adequate water pressure, temperature and
preventing stagnation. Some older healthcare facilities, built prior to such guidelines, often have
plumbing infrastructure that doesn’t meet these requirements. If infrastructure recommendations
can’t be met, additional measures such as chlorine treatment, copper-silver ionization or ultraviolet
light can be used to ensure water quality [76]. As municipal water passes through the distribution
network, the amount of residual disinfection agent can vary. If the facility is far away from the point
of disinfection, the water the building receives may have disinfectant levels lower than the effective
concentration [69]. The success of disinfection approaches may also be impacted by resistant species.
For example, copper resistant P. aeruginosa was isolated from a French water system and tap aeration
grids, and hydrogen peroxide and silver nitrate resistant Legionella spp. were isolated from a hospital’s
water supply [57,77]. Future work is needed to inform and improve HAI guidelines regarding the use
of water and prevent the spread of AMR pathogens.
3.2. Biofilm Formation and Antimicrobial Resistance
Biofilms are secure, often heterogeneous, communities of microorganisms which colonize and
grow on surfaces of medical implants, plumbing infrastructure and on patients [30]. They are comprised
of dense microbial populations immobilized by an extracellular matrix comprised of bacterial secreted
polymers such as exopolysaccharides (EPS), extracellular DNA and proteins [30]. Recently, point of use
filters have been implemented in healthcare facilities as an additional form of protection from bacteria
present in the water supply [78]. Even though P. aeruginosa and Legionella spp. were eliminated from
taps in an intensive care unit in Hungary when point of use filters were installed, decreasing cases
of infection to zero [79], they have been found to facilitate biofilm formation inside the filter when
not maintained correctly, directly affecting the bacterial load in the water over time [78,80]. Within
hospital water distribution systems and plumbing fixtures, biofilms provide a source of nutrients and
protection from disinfection processes [30]. Biofilm growth is promoted in areas of low flow rate and
stagnation which allows for bacterial attachment to the infrastructure surface [81].
The metabolic activity of the bacterial biofilm communities is different compared to planktonic
bacteria, such as increased rates of EPS production, activation or inhibition of genes associated with
biofilm formation and decreased growth rate [30]. The role of EPS has been linked to conferring
tolerance to aminoglycosides by quenching their activity via a diffusion reaction inhibition [82].
An outbreak strain of aminoglycoside resistant P. aeruginosa was found on a contaminated bath toy in
an Australian hospital [16]. Biofilm production confers protection to the microorganism communities
from harmful pH, osmolarity, nutrient scarcity and shear forces [30]. Bacteria in biofilms are also more
resistant to antimicrobial exposure by blocking the access of antibiotics, increasing the resistance by up to
1000-fold when compared to planktonic bacteria [45]. Once a biofilm community has reached maturation,
species such as L. pneumophila may enter a viable non-culturable (VBNC) stationary phase as a way of
surviving antibiotic stress [30,83]. Recent data suggests that hot water flushing and chlorination are not
effective in eliminating Legionella spp. from plumbing systems over long periods of time [76,84]. This
may be due to in part to bacterial species such as Legionella spp. being intracellular parasites of free living
amoeba, resulting in conferred protection from disinfection by techniques when phagocytized [76].
One of the predominant mechanisms for acquiring antimicrobial resistance is uptake of resistance
genes by horizontal gene transfer (HGT) [82]. The high cell density and presence of genetic elements
from a highly heterogeneous community promotes this transfer via mechanisms such as conjugation,
transformation or transduction [82]. Antimicrobial resistance may also be acquired via a mutation
event in a bacterial chromosome [85]. Once the resistance mutation has stabilized in a generation,
Pathogens 2020, 9, 667 13 of 21
it will be directly transmitted to all descendant cells by mitosis [86]. This process is known as
vertical transmission. Under antimicrobial stress, resistance may arise via a combination of both HGT
and vertical transmission. These genetic elements may enhance antimicrobial defense strategies by
restricting drug entry via modifications to the cell wall, pumping the drug out of the cell, enzymatic
degradation of the drug or deleting or decreasing the affinity of the involved target [87]. Exposure to
chlorine can also stimulate the expression of efflux pumps and drug resistance operons, as well as
induce mutations in some genes leading to increased antimicrobial resistance [45]. Some antibiotic gene
profiles observed in hospital shower hose metagenomes have been reported to be triggered by biocide
exposure [45,88,89]. These include commonly used antibiotics such as chloramphenicol, kanamycin,
and penicillin. Species such as Mycobacterium spp. are commonly found in biofilm communities [45].
This may be because of physiochemical properties such as plumbing pipes being galvanized or made
of copper, the disinfectant use and low organic carbon content of the water selectively favoring the
growth of some Mycobacterium spp. [45]. When exposed to stress conditions, Mycobacterium spp. can
modify the cell membrane fatty acid composition producing an altered permeability to biocide and
antibiotic compounds [45,90,91]. The biofilm-forming capacity of pathogens such as P. aeruginosa,
Mycobacterium spp. and S. maltophilia can promote the attachment of other pathogens such as Salmonella
spp., Campylobacter spp. and S. aureus that are typically found in the wider hospital environment [92].
3.3. Detection Methods
3.3.1. Outbreak Investigations
Environmental screening typically takes place in response to an outbreak rather than as routine
sampling, which leads to inconsistencies between the types of samples taken, isolation methods and
antimicrobial resistance reporting. Thirty-five of 88 papers included in this review explored clinical
outbreaks and sampled water and/or water related devices as a part of the investigation (Table S1).
In contrast, 20/88 papers conducted broad screens of the facilities’ environment in a non-outbreak
setting. The Australian Guidelines for the Prevention and Control of Infection in Healthcare suggest
that environmental testing should be carried out to identify risk factors [1]. However, it is not clear
what sampling techniques are to be used and which samples should be taken [1]. Similarly, in the UK,
there is guidance available from The National Specifications for Cleanliness in the NHS for monitoring
the hospital environment. However, there was no indication of microbiological screening [93,94].
The absence of a standard approach for when environmental sampling should occur and what samples
should be taken limits data comparisons that can be made and potentially overlooks reservoirs such as
water and water-related devices.
3.3.2. Pathogen Detection from Environmental Sources
International standards have been published for the processing of environmental water samples
for organisms such as Legionella spp., P. aeruginosa and E. coli. However, of the publications reviewed
in this study, only three referenced a specific ISO standard [31,36,43]. There was significant variation
between sampling techniques and selective growth media used in publications that investigated
water-related surfaces such as tap faucets and drain holes [32,33,39–41,43,47,48,50,51]. Traditional
microbial culturing techniques used for waterborne pathogens such as Legionella spp. has presented
challenges for some environmental samples as VBNC cells and result in false negative results [83,95].
Furthermore, environmental waterborne pathogens often adapt to environments that are nutrient
poor, which may be difficult to culture on nutrient-rich media types. Using nutritionally reduced
media types such as R2A agar for longer incubation periods (14–28 days) may enhance the recovery
of chlorine damaged and stressed bacteria [76]. Environmental water samples are often passed
through membrane filters to concentrate and isolate any bacterial cells present in the sample. The pore
diameter in these membrane filters typically ranges from 01 to 0.45 µm depending on the intended
use [96]. The size, shape and biovolume of bacteria may influence the filterability of a sample and
Pathogens 2020, 9, 667 14 of 21
potentially lead to inaccurate findings, particularly if multiple species of bacteria are being investigated
using the one pore diameter [96]. Alternative molecular techniques for bacterial detection such as
qPCR and whole-genome sequencing (WGS) have been employed by 27 studies included in this
review [13,14,17,18,20,25,26,34,37,40,43,45,46,51,52,56,57,59,60,63–65,97–101]. Molecular techniques
have significant advantages such as rapid turnaround times and detection of non-culturable cells [102].
However, limitations such as environmental inhibitors and potential overestimation of bacterial
presence due to the amplification of non-viable cells needs to be considered [103]. For some bacteria,
PCR-based techniques have been developed to differentiate viable cells from dead cells. For example,
ethidium monoazide bromide viability staining can be used in conjunction with qPCR to enumerate
viable cells (such as L. pneumophila) [102]. In order to implement effective surveillance programs,
detailed and consistent sampling techniques and detection methods are essential.
3.3.3. Characterizing AMR
International standards for antimicrobial susceptibility testing have been jointly published by
the Clinical and Laboratory Standards Institute and the European Centre for Disease Prevention and
Control and the US CDC [104]. These standards include antibiotics to be tested against species that
have commonly been associated with HAIs including Acinetobacter spp., P. aeruginosa, and S. aureus as
well as breakpoints to determine an isolate’s resistance to each antibiotic. Irrespectively, the reporting
of resistant species remains inconsistent. When papers report the resistance profiles of an AMR isolate
using differing units such as µg/mL or mg/mL MICs, percentage of isolates resistant or as specific
resistance genes, the comparisons that can be made between studies are limited to broad comments
rather than quantifiable data trends. 7 of 21
Pathogens 2020, 9, x FOR PEER REVIEW
Identification Records identified through
database searching
(n = 2201)
Records after duplicates removed
(n = 378)
Screening Records screened (n = 1823) Records excluded
(n = 1580)
Eligibility Full-text articles assessed for
eligibility (n = 243) Full-text articles
excluded, with reasons
(n = 157)
Studies included in qualitative
synthesis (n = 88)
Figure 1. Flow diagram presenting the search strategies used, based on the PRISMA statement reporting
guideFrleiipgnuoerrsteifno1gr. gsFuyloisdwteelmidniaeatsgicrfoalmritseyprsarteteusmernaettiricneglvitieethrwaetsus[re1ea0rrc5eh]v.iestwrast[e1g0i5e]s. used, based on the PRISMA statement
Pathogens 2020, 9, 667 15 of 21
4. Materials and Methods
This systematic literature review is based on an adapted version of the PRISMA statement [105],
presented in Figure 1. This tool is an evidence-based system for evaluating and reporting evidence.
A systematic search of the SCOPUS and Web of Science databases was performed, and all literature
published prior to 2020 was included. Keywords used in this search are presented in Table 2.
A detailed search strategy was established to ensure a comprehensive literature review of all identified
antimicrobial resistance bacteria in healthcare water environments was achieved.
Table 2. Complete search strategy and all keywords used to identify relevant literature.
Search Terms Employed to Identify Relevant Literature
“antibiotic resistance *”OR “antimicrobial resistance *” OR disinfectant * OR AMR
AND
Water OR potable OR drinking OR taps OR faucet OR bath OR shower OR drain OR bathroom OR sink
AND
Hospital OR healthcare OR “aged care” OR ICU OR “intensive care unit” OR nosocomial OR HCAI OR
“healthcare acquired infection” OR HAI OR “hospital acquired infection” OR “hospital associated infection”
OR “healthcare associated infection”
‘*’ Indicates wildcard symbol used to when variations of the search term may be possible.
All titles and abstracts of published literature were manually reviewed to ensure that they reported
antimicrobial resistant bacteria to the genus level. The paper must also have reported this presence
in a healthcare setting water source or water-related device. Papers were excluded if they were not
written in English, reviews, reports of human clinical infection with no mention of a contributing
water source, laboratory setting experiments and wastewater investigations. All relevant papers had
key points taken and recorded including the study site, water source, country, species of organism,
isolation method used, antimicrobial method used, and relevant characteristics.
5. Conclusions
Although environmental reservoirs such as dry surface fomites have been identified as potential
sources of HAIs, water and water-related devices are often overlooked. Understanding the role
that water and water-related devices play as reservoirs for AMR bacteria is imperative to prevent
transmission pathways that may cause HAIs. Water sources contaminated with AMR pathogens
provide unique environments for the dissemination of antimicrobial resistance genes that are often
unaffected by commonly employed disinfection strategies. Sinks, tap faucets, drains, bathtubs,
drinking water fountains, aeration grids, showers and haemodialysis water have all been identified as
contaminated with one or more species of AMR bacteria capable of causing HAIs. Broad universal
environmental surveillance guidelines must be developed, including sampling locations, methodology
and resistance reporting, to monitor resistant pathogens and predict future threats before infection
outbreaks occur. By understanding how water and water related devices may harbor AMR species,
better environmental controls can be implemented to significantly reduce the rates of waterborne HAIs.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-0817/9/8/667/s1,
Table S1: Summary of reports and studies identifying AMR bacterial species within healthcare water sources and
water-related devices during outbreak investigation, environmental screening, and molecular epidemiology.
Author Contributions: C.H., H.W., K.E.R. and M.H.B. designed and participated in review design. C.R.H. drafted
and edited the manuscript. H.W., K.E.R. and M.H.B. corrected and contributed to the manusciript. All authors
have read and agree to the publishsed version of the manuscript.
Funding: This research receieved no external funding.
Conflicts of Interest: The authors declare no conflict of interest.
Pathogens 2020, 9, 667 16 of 21
References
1. National Health and Medical Research Council. Australian Guidelines for the Prevention and Control of Infection
in Healthcare; National Health and Medical Research Council: Canberra, ACT, Australia, 2019.
2. Haque, M.; Sartelli, M.; McKimm, J.; Abu Bakar, M. Health care-associated infections—An overview.
Infect. Drug Resist. 2018, 11, 2321–2333. [CrossRef]
3. Centers for Disease Control and Prevention. Healthcare-Associated Infections (HAIs). Available online:
https://www.cdc.gov/winnablebattles/report/HAIs.html (accessed on 2 June 2020).
4. World Health Organization. Report on the Burden of Endemic Health Care-Associated Infection Worldwide.
Available online: https://apps.who.int/iris/handle/10665/80135 (accessed on 25 July 2020).
5. Spelman, D.W. Hospital-acquired infections. Med. J. Aust. 2002, 176, 286–291. [CrossRef]
6. Centers for Disease Control and Prevention. Types of Healthcare-Associated Infection. Available online:
https://www.cdc.gov/hai/infectiontypes.html (accessed on 2 June 2020).
7. Centers for Disease Control and Prevention. Diseases and Organisms in Healthcare Settings. Available
online: https://www.cdc.gov/hai/organisms/organisms.html (accessed on 2 June 2020).
8. Centers for Disease Control and Prevention. Antibiotic Resistance Threats in the United States; U.S. Department
of Health and Human Services: Atlanta, GA, USA, 2019.
9. Australian Government. What causes AMR? Available online: https://www.amr.gov.au/about-amr/what-
causes-amr (accessed on 25 July 2020).
10. Otter, J.A.; Yezli, S.; French, G.L. The role played by contaminated surfaces in the transmission of nosocomial
pathogens. Infect. Control Hosp. Epidemiol. 2011, 32, 687–699. [CrossRef]
11. Sexton, T.; Clarke, P.; O’Neill, E.; Dillane, T.; Humphreys, H. Environmental reservoirs of methicillin-resistant
Staphylococcus aureus in isolation rooms: Correlation with patient isolates and implications for hospital
hygiene. J. Hosp. Infect. 2006, 62, 187–194. [CrossRef]
12. Boyce, J.M.; Potter-Bynoe, G.; Chenevert, C.; King, T. Environmental contamination due to methicillin-resistant
Staphylococcus aureus possible infection control implications. Infect. Control Hosp. Epidemiol. 1997, 18, 622–627.
[CrossRef]
13. Amoureux, L.; Riedweg, K.; Chapuis, A.; Bador, J.; Siebor, E.; Péchinot, A.; Chrétien, M.L.; de Curraize, C.;
Neuwirth, C. Nosocomial infections with IMP-19-producing Pseudomonas aeruginosa linked to contaminated
sinks, France. Emerg. Infect. Dis. 2017, 23, 304–307. [CrossRef]
14. Berrouane, Y.F.; McNutt, L.A.; Buschelman, B.J.; Rhomberg, P.R.; Sanford, M.D.; Hollis, R.J.; Pfaller, M.A.;
Herwaldt, L.A. Outbreak of severe Pseudomonas aeruginosa infections caused by a contaminated drain in a
Whirlpool Bathtub. Clin. Infect. Dis. 2000, 31, 1331–1337. [CrossRef]
15. Bert, F.; Maubec, E.; Bruneau, B.; Berry, P.; Lambert-Zechovsky, N. Multi-resistant Pseudomonas aeruginosa
outbreak associated with contaminated tap water in neurosurgery intensive care unit. J. Hosp. Infect. 1998,
39, 53–62. [CrossRef]
16. Buttery, J.P.; Alabaster, S.J.; Heine, R.G.; Scott, S.M.; Crutchfield, R.A.; Garland, S.M. Multi-resistant
Pseudomonas aeruginosa outbreak in a pediatric oncology ward related to bath toys. Pediatr. Infect. Dis. J. 1998,
17, 509–513. [CrossRef]
17. Corona-Nakamura, A.L.; Miranda-Novales, M.G.; Leaños-Miranda, B.; Portillo-Gómez, L.;
Hernández-Chávez, A.; Anthor-Rendón, J.; Aguilar-Benavides, S. Epidemiologic study of Pseudomonas
aeruginosa in critical patients and reservoirs. Arch. Med. Res. 2001, 32, 238–242. [CrossRef]
18. Crespo, M.P.; Woodford, N.; Sinclair, A.; Kaufmann, M.E.; Turton, J.; Glover, J.; Velez, J.D.; Castañeda, C.R.;
Recalde, M.; Livermore, D.M. Outbreak of carbapenem-resistant Pseudomonas aeruginosa producing VIM-8, a
novel metallo-β-lactamase, in a tertiary care center in Cali, Colombia. J. Clin. Microbiol. 2004, 42, 5094–5101.
[CrossRef] [PubMed]
19. Dwivedi, M.; Mishra, A.; Singh, R.K.; Azim, A.K.; Baronia, A.; Prasad, K.N. Nosocomial cross-transmission
of Pseudomonas aeruginosa between patients in a tertiary intensive care unit. Indian J. Pathol. Microbiol. 2009,
52, 509–513. [CrossRef]
20. Fanci, R.; Bartolozzi, B.; Sergi, S.; Casalone, E.; Pecile, P.; Cecconi, D.; Mannino, R.; Donnarumma, F.;
Leon, A.G.; Guidi, S.; et al. Molecular epidemiological investigation of an outbreak of Pseudomonas aeruginosa
infection in an SCT unit. Bone Marrow Transplant. 2009, 43, 335–338. [CrossRef] [PubMed]
Pathogens 2020, 9, 667 17 of 21
21. Kohlenberg, A.; Weitzel-Kage, D.; van der Linden, P.; Sohr, D.; Vögeler, S.; Kola, A.; Halle, E.; Rüden, H.;
Weist, K. Outbreak of carbapenem-resistant Pseudomonas aeruginosa infection in a surgical intensive care unit.
J. Hosp. Infect. 2010, 74, 350–357. [CrossRef] [PubMed]
22. Lyytikäinen, O.; Golovanova, V.; Kolho, E.; Ruutu, P.; Sivonen, A.; Tiittanen, L.; Hakanen, M.; Vuopio-varkila, J.
Outbreak caused by tobramycin-resistant Pseudomonas aeruginosa in a bone marrow transplantation unit.
Scand. J. Infect. Dis. 2001, 33, 445–449. [CrossRef]
23. Soto, L.E.; Bobadilla, M.; Villalobos, Y.; Sifuentes, J.; Avelar, J.; Arrieta, M.; de Leon, S.P. Post-surgical nasal
cellulitis outbreak due to Mycobacterium chelonae. J. Hosp. Infect. 1991, 19, 99–106. [CrossRef]
24. Stjärne Aspelund, A.; Sjöström, K.; Olsson Liljequist, B.; Mörgelin, M.; Melander, E.; Påhlman, L.I. Acetic acid
as a decontamination method for sink drains in a nosocomial outbreak of metallo-β-lactamase-producing
Pseudomonas aeruginosa. J. Hosp. Infect. 2016, 94, 13–20. [CrossRef]
25. Wendel, A.F.; Ressina, S.; Kolbe-Busch, S.; Pfeffer, K.; MacKenzie, C.R. Species diversity of environmental
GIM-1-producing bacteria collected during a long-term outbreak. Appl. Environ. Microbiol. 2016, 82,
3605–3610. [CrossRef]
26. Witney, A.A.; Gould, K.A.; Pope, C.F.; Bolt, F.; Stoker, N.G.; Cubbon, M.D.; Bradley, C.R.; Fraise, A.;
Breathnach, A.S.; Butcher, P.D.; et al. Genome sequencing and characterization of an extensively drug-resistant
sequence type 111 serotype O12 hospital outbreak strain of Pseudomonas aeruginosa. Clin. Microbiol. Infect.
2014, 20, O609–O618. [CrossRef]
27. Kanamori, H.; Weber, D.J.; Rutala, W.A. Healthcare outbreaks associated with a water reservoir and infection
prevention strategies. Clin. Infect. Dis. 2016, 62, 1423–1435. [CrossRef]
28. Anaissie, E.J.; Penzak, S.R.; Dignani, M.C. The hospital water supply as a source of nosocomial infections:
A plea for action. Arch. Intern. Med. 2002, 162, 1483–1492. [CrossRef] [PubMed]
29. Yiallouros, P.K.; Papadouri, T.; Karaoli, C.; Papamichael, E.; Zeniou, M.; Pieridou-Bagatzouni, D.;
Papageorgiou, G.T.; Pissarides, N.; Harrison, T.G.; Hadjidemetriou, A. First outbreak of nosocomial
Legionella infection in term neonates caused by a cold mist ultrasonic humidifier. Clin. Infect. Dis. 2013, 57,
48–56. [CrossRef]
30. Sharma, D.; Misba, L.; Khan, A.U. Antibiotics versus biofilm: An emerging battleground in microbial
communities. Antimicrob. Resist. Infect. Control 2019, 8, 76. [CrossRef] [PubMed]
31. Alexandropoulou, I.; Parasidis, T.; Konstantinidis, T.; Panopoulou, M.; Constantinidis, T.C. A proactive
environmental approach for preventing legionellosis in infants: Water sampling and antibiotic resistance
monitoring, a 3-years survey program. Healthcare 2019, 7, 39. [CrossRef] [PubMed]
32. Arvanitidou, M.; Vayona, A.; Spanakis, N.; Tsakris, A. Occurrence and antimicrobial resistance of Gram
negative bacteria isolated in haemodialysis water and dialysate of renal units: Results of a Greek multicentre
study. J. Appl. Microbiol. 2003, 95, 180–185. [CrossRef]
33. Borges, C.R.M.; Lascowski, K.M.S.; Filho, N.R.; Pelayo, J.S. Microbiological quality of water and dialysate in
a haemodialysis unit in Ponta Grossa-PR, Brazil. J. Appl. Microbiol. 2007, 103, 1791–1797. [CrossRef]
34. De Giglio, O.; Napoli, C.; Lovero, G.; Diella, G.; Rutigliano, S.; Caggiano, G.; Montagna, M.T. Antibiotic
susceptibility of Legionella pneumophila strains isolated from hospital water systems in Southern Italy.
Environ. Res. 2015, 142, 586–590. [CrossRef]
35. Erdogan, H.; Can, F.; Demirbilek, M.; Timurkaynak, F.; Arslan, H. In vitro activity of antimicrobial agents
against Legionella isolated from environmental water systems: First results from Turkey. Environ. Monit. Assess.
2010, 171, 487–491. [CrossRef]
36. Graells, T.; Hernández-García, M.; Pérez-Jové, J.; Guy, L.; Padilla, E. Legionella pneumophila recurrently
isolated in a Spanish hospital: Two years of antimicrobial resistance surveillance. Environ. Res. 2018, 166,
638–646. [CrossRef]
37. Gümüs¸, D.; Küçüker, M.A. EHEC, EPEC, and ETEC strains and their antibiotic resistance in drinking and tap
water samples. Clin. Lab. 2015, 61, 113–121. [CrossRef]
38. Jeanvoine, A.; Meunier, A.; Puja, H.; Bertrand, X.; Valot, B.; Hocquet, D. Contamination of a hospital plumbing
system by persister cells of a copper-tolerant high-risk clone of Pseudomonas aeruginosa. Water Res. 2019, 157,
579–586. [CrossRef] [PubMed]
39. Mathias, A.J.; Madhuravani, M.A.; Kavitha, B. Incidence of Pseudomonas aeruginosa and other indicator
bacteria in drinking water supplies to hospitals of Bangalore City. Indian J. Microbiol. 2002, 42, 83–85.
Pathogens 2020, 9, 667 18 of 21
40. Moremi, N.; Claus, H.; Vogel, U.; Mshana, S.E. Surveillance of surgical site infections by Pseudomonas
aeruginosa and strain characterization in Tanzanian hospitals does not provide proof for a role of hospital
water plumbing systems in transmission. Antimicrob. Resist. Infect. Control 2017, 6. [CrossRef] [PubMed]
41. Narciso-da-Rocha, C.; Vaz-Moreira, I.; Manaia, C.M. Genotypic diversity and antibiotic resistance in
Sphingomonadaceae isolated from hospital tap water. Sci. Total Environ. 2014, 466, 127–135. [CrossRef]
42. Perryman, F.A.; Flournoy, D.J. Prevalence of gentamicin- and amikacin-resistant bacteria in sink drains.
J. Clin. Microbiol. 1980, 12, 79–83. [CrossRef]
43. Schiavano, G.F.; Carloni, E.; Andreoni, F.; Magi, S.; Chironna, M.; Brandi, G.; Amagliani, G. Prevalence and
antibiotic resistance of Pseudomonas aeruginosa in water samples in central Italy and molecular characterization
of oprD in imipenem resistant isolates. PLoS ONE 2017, 12, e0189172. [CrossRef]
44. Sikora, A.; Gładysz, I.; Kozioł-Montewka, M.; Wójtowicz-Bobin, M.; Stan´ czak, T.; Matuszewska, R.;
Krogulska, B. Assessment of antibiotic susceptibility of Legionella pneumophila isolated from water systems in
Poland. Ann. Agric. Environ. Med. 2017, 24, 66–69. [CrossRef]
45. Soto-Giron, M.J.; Rodriguez-R, L.M.; Luo, C.W.; Elk, M.; Ryu, H.; Hoelle, J.; Domingo, J.W.S.;
Konstantinidis, K.T. Biofilms on hospital shower hoses: Characterization and implications for nosocomial
infections. Appl. Environ. Microbiol. 2016, 82, 2872–2883. [CrossRef]
46. Tran-Dinh, A.; Neuher, C.; Amara, M.; Nebot, N.; Trache, G.; Breton, N.; Zuber, B.; Cavelot, S.; Pangon, B.;
Bedos, J.P.; et al. Impact of intensive care unit relocation and role of tap water on an outbreak of Pseudomonas
aeruginosa expressing OprD-mediated resistance to imipenem. J. Hosp. Infect. 2018, 100, E105–E114.
[CrossRef]
47. Tsakris, A. Recovery of high-level streptomycin-resistant enterococci from hemodialysis water and dialysate
in 85 greek renal units. Infect. Control Hosp. Epidemiol. 1999, 20, 686–689. [CrossRef]
48. van der Mee-Marquet, N.; Bloc, D.; Briand, L.; Besnier, J.M.; Quentin, R. Non-touch fittings in hospitals:
A procedure to eradicate Pseudomonas aeruginosa contamination. J. Hosp. Infect. 2005, 60, 235–239. [CrossRef]
[PubMed]
49. Varin, A.; Valot, B.; Cholley, P.; Morel, C.; Thouverez, M.; Hocquet, D.; Bertrand, X. High prevalence and
moderate diversity of Pseudomonas aeruginosa in the U-bends of high-risk units in hospital. Int. J. Hyg.
Environ. Health 2017, 220, 880–885. [CrossRef] [PubMed]
50. Vincenti, S.; Quaranta, G.; De Meo, C.; Bruno, S.; Ficarra, M.G.; Carovillano, S.; Ricciardi, W.; Laurenti, P.
Non-fermentative gram-negative bacteria in hospital tap water and water used for haemodialysis and
bronchoscope flushing: Prevalence and distribution of antibiotic resistant strains. Sci. Total Environ. 2014,
499, 47–54. [CrossRef] [PubMed]
51. Ziwa, M.; Jovic, G.; Ngwisha, C.L.T.; Molnar, J.A.; Kwenda, G.; Samutela, M.; Mulowa, M.; Kalumbi, M.M.
Common hydrotherapy practices and the prevalence of burn wound bacterial colonisation at the University
Teaching Hospital in Lusaka, Zambia. Burns 2019, 45, 983–989. [CrossRef] [PubMed]
52. Decraene, V.; Phan, H.T.T.; George, R.; Wyllie, D.H.; Akinremi, O.; Aiken, Z.; Cleary, P.; Dodgson, A.;
Pankhurst, L.; Crook, D.W.; et al. A large, refractory nosocomial outbreak of Klebsiella pneumoniae
carbapenemase-producing Escherichia coli demonstrates carbapenemase gene outbreaks involving sink
sites require novel approaches to infection control. Antimicrob. Agents Chemoth. 2018, 62. [CrossRef]
[PubMed]
53. Kilic, A.; Senses, Z.; Kurekci, A.E.; Aydogan, H.; Sener, K.; Kismet, E.; Basustaoglu, A.C. Nosocomial outbreak
of Sphingomonas paucimobilis bacteremia in a Hemato/Oncology unit. Jpn. J. Infect. Dis. 2007, 60, 394–396.
54. Layton, M.C.; Patterson, J.E.; Patterson, J.E.; Perez, M.; Heald, P. An outbreak of mupirocin-resistant
Staphylococcus aureus on a dermatology ward associated with an environmental reservoir. Infect. Control
Hosp. Epidemiol. 1993, 14, 369–375. [CrossRef]
55. Benedict, K.M.; Reses, H.; Vigar, M.; Roth, D.M.; Roberts, V.A.; Mattioli, M.; Cooley, L.A.; Hilborn, E.D.;
Wade, T.J.; Fullerton, K.E.; et al. Surveillance for waterborne disease outbreaks associated with drinking
water—United States, 2013-2014. Morb. Mortal. Wkl. Rep. 2017, 66, 1216–1221. [CrossRef]
56. Zheng, R.; Zhang, Q.; Guo, Y.; Feng, Y.; Liu, L.; Zhang, A.; Zhao, Y.; Yang, X.; Xia, X. Outbreak of
plasmid-mediated NDM-1-producing Klebsiella pneumoniae ST105 among neonatal patients in Yunnan, China.
Ann. Clin. Microbiol. Antimicrob. 2016, 15. [CrossRef]
Pathogens 2020, 9, 667 19 of 21
57. Deredjian, A.; Colinon, C.; Brothier, E.; Favre-Bonte, S.; Cournoyer, B.; Nazaret, S. Antibiotic and metal
resistance among hospital and outdoor strains of Pseudomonas aeruginosa. Res. Microbiol. 2011, 162, 689–700.
[CrossRef]
58. Hu, P.W.; Chen, J.Y.; Chen, Y.H.; Zhou, T.; Xu, X.; Pei, X. Molecular epidemiology, resistance, and virulence
properties of Pseudomonas aeruginosa cross-colonization clonal isolates in the non-outbreak setting.
Infect. Genet. Evol. 2017, 55, 288–296. [CrossRef] [PubMed]
59. Adwan, G.M.; Abu-Hijleh, A.A.; Bsharat, N.M. Antibiotic resistance and type III exotoxin encoding genes of
Pseudomonas. aeruginosa isolates from environmental and clinical sources in Northern West Bank in Palestine.
Jordan J. Biol. Sci. 2019, 12, 191–196.
60. Shamsizadeh, Z.; Nikaeen, M.; Esfahani, B.N.; Mirhoseini, S.H.; Hatamzadeh, M.; Hassanzadeh, A. Detection
of antibiotic resistant Acinetobacter baumannii in various hospital environments: Potential sources for
transmission of Acinetobacter infections. Environ. Health Prev. Med. 2017, 22. [CrossRef]
61. French, G.L.; Otter, J.A.; Shannon, K.P.; Adams, N.M.T.; Watling, D.; Parks, M.J. Tackling contamination
of the hospital environment by methicillin-resistant Staphylococcus aureus (MRSA): A comparison between
conventional terminal cleaning and hydrogen peroxide vapour decontamination. J. Hosp. Infect. 2004, 57,
31–37. [CrossRef]
62. Muzslay, M.; Moore, G.; Alhussaini, N.; Wilson, A.P.R. ESBL-producing Gram-negative organisms in the
healthcare environment as a source of genetic material for resistance in human infections. J. Hosp. Infect.
2017, 95, 59–64. [CrossRef]
63. Tofteland, S.; Naseer, U.; Lislevand, J.H.; Sundsfjord, A.; Samuelsen, Ø. A long-term low-frequency hospital
outbreak of kpc-producing Klebsiella pneumoniae involving intergenus plasmid diffusion and a persisting
environmental reservoir. PLoS ONE 2013, 8, e059015. [CrossRef]
64. Mathers, A.J.; Crook, D.; Vaughan, A.; Barry, K.E.; Vegesana, K.; Stoesser, N.; Parikh, H.I.; Sebra, R.; Kotay, S.;
Sarah Walker, A.; et al. Klebsiella quasipneumoniae provides a window into carbapenemase gene transfer,
plasmid rearrangements, and patient interactions with the hospital environment. Antimicrob. Agents Chemoth.
2019, 63. [CrossRef]
65. Hugon, E.; Marchandin, H.; Poirée, M.; Fosse, T.; Sirvent, N. Achromobacter bacteraemia outbreak in a
paediatric onco-haematology department related to strain with high surviving ability in contaminated
disinfectant atomizers. J. Hosp. Infect. 2015, 89, 116–122. [CrossRef]
66. Magiorakos, A.P.; Srinivasan, A.; Carey, R.B.; Carmeli, Y.; Falagas, M.E.; Giske, C.G.; Harbarth, S.;
Hindler, J.F.; Kahlmeter, G.; Olsson-Liljequist, B.; et al. Multidrug-resistant, extensively drug-resistant
and pandrug-resistant bacteria: An international expert proposal for interim standard definitions for
acquired resistance. Clin. Microbiol. Infect. 2012, 18, 268–281. [CrossRef]
67. Propato, M.; Uber, J.G. Vulnerability of water distribution systems to pathogen intrusion: how effective is a
disinfectant residual? Environ. Sci. Technol. 2004, 38, 3713–3722. [CrossRef]
68. Department for Health and Aging. Control of Legionella in Manufactured Water Systems in South Australia;
Government of South Australia: Adelaide, SA, Australia, 2013.
69. EnHealth. Guidelines for Legionella Control in the Operation and Maintenance of Water Distribution Systems in
Health and Aged Care Facilities; Australian Government: Canberra, ACT, Australia, 2015.
70. Hayanga, A.; Okello, A.; Hussein, R.; Nyong’o, A. Experience with methicillin resistant Staphylococcus aureus
at the Nairobi Hospital. East Afr. Med. J. 1997, 74, 203–204. [PubMed]
71. Nagoba, B.S.; Deshmukh, S.R.; Husain, R.A.; Gomashe, A.V.; Wadher, B.J. Bacteriological analysis of various
environmental sources in a rural hospital. Indian J. Med. Sci. 1997, 51, 465–469. [PubMed]
72. Squeri, R.; Grilo, O.C.; La Fauci, V. Surveillance and evidence of contamination in hospital environment from
meticillin and vancomycin-resistant microbial agents. J. Prev. Med. Hyg. 2012, 53, 143–145. [PubMed]
73. Sserwadda, I.; Lukenge, M.; Mwambi, B.; Mboowa, G.; Walusimbi, A.; Segujja, F. Microbial contaminants
isolated from items and work surfaces in the post-operative ward at Kawolo general hospital, Uganda.
BMC Infect. Dis. 2018, 18. [CrossRef]
74. Sexton, J.D.; Tanner, B.D.; Maxwell, S.L.; Gerba, C.P. Reduction in the microbial load on high-touch surfaces
in hospital rooms by treatment with a portable saturated steam vapor disinfection system. Am. J. Infect.
Control 2011, 39, 655–662. [CrossRef]
75. VanderElzen, K.; Zhen, H.; Shuman, E.; Valyko, A. The hidden truth in the faucets: A quality improvement
project and splash study of hospital sinks. Am. J. Infect. Control 2019, 47, S26. [CrossRef]
Pathogens 2020, 9, 667 20 of 21
76. Cervia, J.S.; Ortolano, G.A.; Canonica, F.P. Hospital tap water: A reservoir of risk for health care-associated
infection. Infect. Dis. Clin. Pract. 2008, 16, 349–353.
77. Garcia, M.T.; Baladron, B.; Gil, V.; Tarancon, M.L.; Vilasau, A.; Ibanez, A.; Elola, C.; Pelaz, C. Persistence
of chlorine-sensitive Legionella pneumophila in hyperchlorinated installations. J. Appl. Microbiol. 2008, 105,
837–847. [CrossRef]
78. Bielefeldt, A.R.; Kowalski, K.; Summers, R.S. Bacterial treatment effectiveness of point-of-use ceramic water
filters. Water Res. 2009, 43, 3559–3565. [CrossRef]
79. Barna, Z.; Antmann, K.; Pászti, J.; Bánfi, R.; Kádár, M.; Szax, A.; Németh, M.; Szego˝, E.; Vargha, M. Infection
control by point-of-use water filtration in an intensive care unit—A Hungarian case study. J. Water Health
2014, 12, 858–867. [CrossRef]
80. Su, F.; Luo, M.; Zhang, F.; Li, P.; Lou, K.; Xing, X. Performance of microbiological control by a point-of-use
filter system for drinking water purification. J. Environ. Sci. 2009, 21, 1237–1246. [CrossRef]
81. Toyofuku, M.; Inaba, T.; Kiyokawa, T.; Obana, N.; Yawata, Y.; Nomura, N. Environmental factors that shape
biofilm formation. Biosci. Biotechnol. Biochem. 2016, 80, 7–12. [CrossRef] [PubMed]
82. Abe, K.; Nomura, N.; Suzuki, S. Biofilms: Hot spots of horizontal gene transfer (HGT) in aquatic environments,
with a focus on a new HGT mechanism. FEMS Microbiol. Ecol. 2020, 96. [CrossRef]
83. Whiley, H. Legionella risk management and control in potable water systems: Argument for the abolishment
of routine testing. Int. J. Environ. Res. Public Health 2016, 14, 12. [CrossRef]
84. Mouchtouri, V.; Velonakis, E.; Hadjichristodoulou, C. Thermal disinfection of hotels, hospitals, and athletic
venues hot water distribution systems contaminated by Legionella species. Am. J. Infect. Control 2007, 35,
623–627. [CrossRef]
85. Vaz-Moreira, I.; Nunes, O.C.; Manaia, C.M. Diversity and antibiotic resistance in Pseudomonas spp. from
drinking water. Sci. Total Environ. 2012, 426, 366–374. [CrossRef]
86. Coculescu, B.-I. Antimicrobial resistance induced by genetic changes. J. Med. Life 2009, 2, 114–123.
87. Centers for Disease Control and Prevention. Antibiotic/Antimicrobial Resistance (AR/AMR). Available
online: https://www.cdc.gov/drugresistance/about/how-resistance-happens.html (accessed on 24 June 2020).
88. Karumathil, D.P.; Yin, H.B.; Kollanoor-Johny, A.; Venkitanarayanan, K. Effect of chlorine exposure on the
survival and antibiotic gene expression of multidrug resistant Acinetobacter baumannii in water. Int. J. Environ.
Res. Public Health 2014, 11, 1844–1854. [CrossRef]
89. Huang, J.J.; Hu, H.Y.; Tang, F.; Li, Y.; Lu, S.Q.; Lu, Y. Inactivation and reactivation of antibiotic-resistant
bacteria by chlorination in secondary effluents of a municipal wastewater treatment plant. Water Res. 2011,
45, 2775–2781. [CrossRef] [PubMed]
90. Steed, K.A.; Falkinham, J.O. Effect of growth in biofilms on chlorine susceptibility of Mycobacterium avium
and Mycobacterium intracellulare. Appl. Environ. Microbiol. 2006, 72, 4007–4011. [CrossRef] [PubMed]
91. Armstrong, J.L.; Calomiris, J.J.; Seidler, R.J. Selection of antibiotic-resistant standard plate count bacteria
during water treatment. Appl. Environ. Microbiol. 1982, 44, 308–316.
92. Maes, S.; Vackier, T.; Nguyen Huu, S.; Heyndrickx, M.; Steenackers, H.; Sampers, I.; Raes, K.; Verplaetse, A.;
De Reu, K. Occurrence and characterisation of biofilms in drinking water systems of broiler houses.
BMC Microbiol. 2019, 19, 77. [CrossRef]
93. Rawlinson, S.; Ciric, L.; Cloutman-Green, E. How to carry out microbiological sampling of healthcare
environment surfaces? A review of current evidence. J. Hosp. Infect. 2019, 103, 363–374. [CrossRef]
94. National Patient Safety Agency. The National Specifications for Cleanliness in the NHS: A Framework for Setting
and Measuring Performance Outcomes; National Health Service: London, UK, 2007.
95. Shih, H.-Y.; Lin, Y.E. Caution on interpretation of Legionella results obtained using real-time PCR for
environmental water samples. Appl. Environ. Microbiol. 2006, 72, 6859. [CrossRef]
96. Wang, Y.; Hammes, F.; Düggelin, M.; Egli, T. Influence of Size, Shape, and Flexibility on Bacterial Passage
through Micropore Membrane Filters. Environ. Sci. Technol. 2008, 42, 6749–6754. [CrossRef]
97. Clarivet, B.; Grau, D.; Jumas-Bilak, E.; Jean-Pierre, H.; Pantel, A.; Parer, S.; Lotthé, A. Persisting transmission
of carbapenemase-producing Klebsiella pneumoniae due to an environmental reservoir in a university hospital,
France, 2012 to 2014. Eurosurveillance 2016, 21. [CrossRef]
98. Gilbert, Y.; Veillette, M.; Duchaine, C. Airborne bacteria and antibiotic resistance genes in hospital rooms.
Aerobiologia 2010, 26, 185–194. [CrossRef]
Pathogens 2020, 9, 667 21 of 21
99. Jiang, X.B.; Wang, D.P.; Wang, Y.X.; Yan, H.; Shi, L.; Zhou, L.J. Occurrence of antimicrobial resistance genes sul
and dfrA12 in hospital environmental isolates of Elizabethkingia meningoseptica. World J. Microbiol. Biotechnol.
2012, 28, 3097–3102. [CrossRef]
100. Ng, D.H.L.; Marimuthu, K.; Lee, J.J.; Khong, W.X.; Ng, O.T.; Zhang, W.; Poh, B.F.; Rao, P.; Raj, M.D.R.;
Ang, B.; et al. Environmental colonization and onward clonal transmission of carbapenem-resistant
Acinetobacter baumannii (CRAB) in a medical intensive care unit: The case for environmental hygiene.
Antimicrob. Resist. Infect. Control 2018, 7. [CrossRef]
101. Umezawa, K.; Asai, S.; Ohshima, T.; Iwashita, H.; Ohashi, M.; Sasaki, M.; Kaneko, A.; Inokuchi, S.;
Miyachi, H. Outbreak of drug-resistant Acinetobacter baumannii ST219 caused by oral care using tap water
from contaminated hand hygiene sinks as a reservoir. Am. J. Infect. Control 2015, 43, 1249–1251. [CrossRef]
102. Yaradou, D.F.; Hallier-Soulier, S.; Moreau, S.; Poty, F.; Hillion, Y.; Reyrolle, M.; André, J.; Festoc, G.; Delabre, K.;
Vandenesch, F.; et al. Integrated real-time PCR for detection and monitoring of Legionella pneumophila in
water systems. Appl. Environ. Microbiol. 2007, 73, 1452–1456. [CrossRef] [PubMed]
103. Delgado-Viscogliosi, P.; Solignac, L.; Delattre, J.-M. Viability PCR, a culture-independent method for
rapid and selective quantification of viable Legionella pneumophila cells in environmental water samples.
Appl. Environ. Microbiol. 2009, 75, 3502. [CrossRef] [PubMed]
104. CLSI. Performance Standards for Antimicrobial Suspecptibilty Testing; Clinical Laboratory Standards Institute:
Annapolis Junction, MD, USA, 2017.
105. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G.; The, P.G. Preferred reporting items for systematic reviews
and meta-analyses: The PRISMA statement. PLoS Med. 2009, 6, e1000097. [CrossRef]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/341575165
Prevalence of E. coli, Salmonella, and Listeria spp. as potential pathogens: A
comparative study for biofilm of sink drain environment
Article · May 2020 READS
DOI: 10.1111/jfs.12816 161
CITATIONS Gamila El-Taweel
National Research Center, Egypt
9 51 PUBLICATIONS 466 CITATIONS
3 authors, including: SEE PROFILE
Mohamed Azab El-Liethy
National Research Center, Egypt
65 PUBLICATIONS 335 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Metal recovery using bioelectrochemical system: An update of current progress View project
water treatment View project
All content following this page was uploaded by Bahaa Hemdan on 22 May 2020.
The user has requested enhancement of the downloaded file.
Received: 13 March 2020 Revised: 14 April 2020 Accepted: 27 April 2020
DOI: 10.1111/jfs.12816
ORIGINAL ARTICLE
Prevalence of E. coli, Salmonella, and Listeria spp. as potential
pathogens: A comparative study for biofilm of sink drain
environment
Mohamed Azab El-Liethy | Bahaa A. Hemdan | Gamila E. El-Taweel
Environmental Microbiology Laboratory, Abstract
Water Pollution Research Department,
National Research Centre, Giza, Egypt Since knowledge and understanding of waterborne pathogens and their diseases are
well illuminated, a few research publications on the prevalence of pathogenic micro-
Correspondence organisms in various household sink drain pipes are often not extensively examined.
Mohamed Azab El-Liethy, Environmental Therefore, this study aims to (a) assess and monitor the densities of the bacterial
Microbiology Laboratory, Water Pollution community in the different natural biofilm that grow on plastic pipelines, (b) to detect
Research Department, National Research Escherichia coli, Salmonella, and Listeria spp. from natural biofilm samples that are col-
Centre, El.Behouth St., P.O.Box 12262, Dokki, lected from the kitchen (n = 30), bathroom (n = 10), laboratories (n = 13), and hospital
Giza, Egypt. (n = 8) sink drainage pipes. Three bacterial species selected were assessed using a
Email: [email protected] culture-dependent approach followed by verification of isolates using both BIOLOG
GEN III and polymerase chain reaction. The estimated number of each bacterium was
Funding information 122 isolates, while 60, 20, 26, and 16 isolates were obtained from the natural biofilm
Science and Technology Development Fund samples, kitchen, bathroom, laboratories, and hospital, respectively. As for the tests,
in all types of biofilm samples, the overall bacterial counts at low temperature (22 C)
were higher than those at high temperature (37 C). Meanwhile, E. coli had the most
significant number of bacterial microorganisms compared to the other two patho-
gens. Additionally, the most massive cell densities of E. coli, Salmonella, and Listeria
species were discovered in the biofilm collected from the kitchen, then the hospital.
Statistically, the results reveal that there is a positive correlation (p ≥ .0001) with sig-
nificance between the sources of biofilm. This work certainly makes the potential of
household sink drain pipes for reservoir contagious pathogens more explicitly notice-
able. Such knowledge would also be beneficial for prospective consideration of the
threat to human public health and the environment.
1 | INTRODUCTION storage tanks, showerheads, and sink drainage pipes (Elias &
Banin, 2012; Moore et al., 2002). The biofilm microbial cells have
Although the enlarged reports regarding the occurrence of bacterial behaviors that differ from the free cells. Whereas the majority of
biofilms in domestic sink drain pipes, the microbial community compo- microbiomes have adhered to the inner surface of water pipes as ses-
sition of this environment is less documented until now (Hemdan, sile cells, and about 5% are free in the water column (Hemdan, Sedik,
El-Liethy, ElMahdy, & EL-Taweel, 2019). Many studies have identified Kamel, & El-Taweel, 2015; Liu et al., 2014). Bacteria can form biofilms
the potential health risks of biofilm microbial contamination (Awoke, and colonize for an extended period in kitchens and bathroom sink,
Kassa, & Teshager, 2019; Mcbain et al., 2003). Biofilm usually develop where people may be exposed to a wide range of bacterial pathogens
on hydrated surfaces, such as drinking water distribution systems, and become more reliable for infection (Sinclair & Gerba, 2011). The
J Food Saf. 2020;e12816. wileyonlinelibrary.com/journal/jfs © 2020 Wiley Periodicals LLC. 1 of 10
https://doi.org/10.1111/jfs.12816
2 of 10 EL-LIETHY ET AL.
contact ways of these bacterial pathogens carried out either by the studies have examined bacterial biofilms, including E. coli, Salmo-
direct way like handling, preparing and eating food or by the indirect nella, and Listeria species in potable water delivery system pipelines
way like, contact with surfaces that carry a considerable number of and their simulated prototypes (Pizarro, Vargas, Pastén, &
bacterial pathogens that come from different sources, including Calle, 2014; Hemdan et al., 2015, 2016). Since the microbial popula-
humans, food and aerosolized water (Medrano-Felix et al., 2011). tion diversity of these potential bacterial pathogens in the natural
Domestic sink drain biofilms consider a favorable environment for biofilm of different sink drainage pipes are not extensively investi-
potentially pathogenic bacteria (Winder & Bonheyo, 2015). In addition gated primarily in lab and hospital drainage pipes. The present
to these, it is well-known that sink drains in the hospital considered a research aims to investigate the bacterial densities of biofilms that
significant source of pathogens (Hemdan et al., 2019; Mcgeer are full-grown on dissimilar drainage pipes and also to study the
et al., 1990). Sink drainage pipes of hospitals are made of different occurrence of E. coli, Salmonella, and Listeria species amid natural
solid surfaces that are suitable substrates for biofilm formation biofilms of kitchen, bathroom, laboratories, and hospital sink drain-
(Niquette, Servais, & Savoir, 2000). Moreover, Sink drainage pipes in age pipes. In addition to this, their isolates were confirmed by both
laboratories contain different types of bacterial pathogens involved BIOLOG and conventional polymerase chain reaction (PCR).
in biofilm formation (Hemdan et al., 2019). Heterotrophic bacteria
are commonly enumerated to verify the number of cultivable bacte- 2 | MATERIALS AND METHODS
ria in the biofilm by scraping this biofilm from the inner surface of
pipes (Gagnon et al. 2005). Escherichia coli and Salmonella species 2.1 | Collection and preparation of the natural
are gram-negative, rod shape, motile, and belong to the family biofilm samples
Enterobacteriaceae (Singleton, 1999). E. coli is a normal inhabitant
in the intestine of warm blooded animals. E. coli is usually used as an The kitchen (n = 30) and bathroom (n = 10) sink drainage pipes have
indicator for fecal pollution in water and foodstuff (Feng, Weagant, been obtained from household areas in the Helwan city, 30 km south
Grant, & Brukhardt, 2002). The presence of E. coli in water and food of Cairo, Egypt. Whereas the pipelines for laboratory sink drainage
is indicated for the probability presence of enteric pathogens such were obtained from the National Research Centre in the Dokki region,
as Salmonella spp. and hepatitis A virus (Odonkor & Ampofo, 2013) Giza Governrate, Egypt, the laboratories (n = 13), and hospital (n = 8).
and also E. coli includes some pathogenic serotypes containing viru- The collected biofilm ages varied from 1 to 5 years (The period that
lence genes (El-Shatoury, El-Leithy, Abou-Zeid, El-Taweel, & El- extend from installing sink drain pipelines to removing them), and all
Senousy, 2015). E. coli has been determined in biofilm samples of samples were deposited in the icebox and conveyed to the laboratory
drinking water and domestic drainage pipes (Juhna et al., 2007; for microbiological analysis instantly, according to American Public
Mcbain et al., 2003). Salmonella species are widely distributed in Health Association (APHA) (2017). All procedures in the collection
farmhouse wastes, human wastes, and polluted fecal matter. Salmo- and processing of biofilm samples were undertaken throughout opti-
nella is frequently detected in large numbers in raw wastewater mal conditions. The existing biofilm samples were extracted using dis-
(103–104 CFU/L) (Davidson, White, & Surette, 2008; El-Lathy, El- posable cotton buds to remove 10 cm2 from the inner layer of the
Taweel, El-Sonosy, Samhan, & Moussa, 2009). Salmonella is able to plastic pipes. The swab was dipped into a test tube containing 10 ml
form biofilm on plastics, glass, and stainless steel (Hemdan, El- of saline solution and homogenized for 5 min with a vortex trouble-
Liethy, Eissam, Kamel, & El-Taweel, 2016; Momba & Kaleni, 2002). maker (Zhou, Zhang, & Li, 2009).
Some studies on the biofilm formation process have confirmed that
E. coli, and Salmonella in addition to many other species of the 2.2 | Determination of total bacterial counts in
Enterobacteriaceae family, generate cellulose as a vital constituent biofilm samples
of the bacterial extracellular matrix (EPS) and its formation is essen-
tial for the prolong the survival of these bacteria in the surroundings In biofilm samples, total heterotrophic bacterial numbers were exe-
for an extended period (Lasa, Del Pozo, Penadés, & Leiva, 2005). cuted in accordance with APHA (2017) to evaluate the existing bacte-
Listeria species are gram-positive, facultative anaerobic, nonspore- rial flora at both 37 C and 22 C. The temperature at 37 C was
forming, and commonly distributed in different environments and chosen for referring to the presence of enteric bacterial pathogens.
can be found in freshwater, soil, animal fecal matter, and sewage Moreover, temperature at 22 C was chosen for referring to the pres-
(Pagadala et al., 2012). At present the genus Listeria contains 20 spe- ence of natural bacterial flora in the environment. The suspensions of
cies. Two out of 20 Listeria species namely L. monocytogenes and existing heterogeneous cells were serially diluted in a sterile saline
L. ivanovii are considered pathogens (Leclercq et al., 2019; Orsi & solution to acquire the most appropriate dilution. Specific bacterial
Wiedmann, 2016). L. monocytogenes is the third pathogen among numbers on plate count agar (BD DifcoTM) were determined using
microbes causing foodborne deaths in the United States (Scallan the pour plate approach. The quantities for total bacterial cells were
et al., 2011). Besides, Listeria species can be trapped on various evaluated and displayed in CFU/cm2 using an automated colonization
materials, like those of plastic surfaces, polyethylene, leather, stain- monitor (Stuart, FL).
less steel, and glass, and it can be easily damaged on the substratum
once attached (Pan, Breidt, & Kathariou, 2006). Several previous
EL-LIETHY ET AL. 3 of 10
2.3 | Determination and confirmation of potential 2.3.3 | Confirmation of bacterial isolates
pathogens in biofilm samples using PCR
2.3.1 | Determination of E. coli, Salmonella, and Bacterial preparation and DNA extraction
Listeria species The preserved bacterial isolates were suspended on the TSB medium
and then the tubes were kept in the incubator at 37 C for 18–24 hr.
In the natural biofilm samples, the target possible bacterial patho- According to Kapperud, Vardund, Skjerve, Hornes, and Michael-
gens were evaluated using a culture-dependent approach on specific sen (1993), bacterial DNA extraction was implemented with some
agar media. A 100 μl of the sufficient biofilm suspension was trans- adjustments. Approximately, 100 ml of the bacterial isolate was cen-
ported to the HiChrome ECC agar media to evaluate the cell viabil- trifuged at 12,500 rpm for 5 min at 7 C. The obtained pellets by cen-
ities of E. coli. The inoculated plate was incubated at 37 C for 24 hr. trifugation were suspended in 50 μl of 1X PCR buffer containing
The colonies of E. coli appeared in color blue/purple. HiCrome 0.2 mg of proteinase K/ml. For the lysis of the bacterial cell wall, the
improved salmonella agar was used to enumerate the Salmonella spe- suspension was incubated at 37 C for 1 hr. The suspension was then
cies. After optimum incubation, the plates at 37 C for 24–48 hr, the heated for 10 min and then centrifuged for 5 min at 4 C at
standard morphological characterizing of Salmonella species colonies 12,500 rpm. The aqueous phase is used to achieve PCR. The consis-
seemed in light pink and pink to red color. Besides, the HiCrome tency of the obtained DNA was analyzed using Nanodrop
Listeria agar base, modified supplemented HiCrome Listeria selective (NanoDropTM2000/2000C) spectrophotometers) to test their absor-
supplement was being used to count the biofilm cells of L. mono- bance at 260 and 280 nm. According to Lucena-Aguilar et al. (2016),
cytogenes, L. ivanovii and L. innocua. The inoculated plates have also the suitable array of the removed DNA for PCR is amid 1.6–1.8 ng/μl.
been kept in the incubator for 24–48 hr at 37 C. The typical colonies
of Listeria turned up in bluish-green color. All the specific agar media 2.3.4 | The PCR amplification
used for chemometrics were supplied from HiMedia, India. Biofilm
accumulation was expressed in CFU/10 cm2 in all of the experi- The PCR amplification was carried out in separate PCR for the
ments. For further validation using BIOLOG and PCR, two bacterial selected microorganism isolates. All the primers and PCR conditions
isolates from each sample were held at −20 C in tryptic soya broth used in this study are given in Table 1. The Primers used were prop-
(TSB) (Oxoid, UK) with 10% glycerol after bacterial growth at 37 C erly manufactured by Macrogen Co. (Soul, Republic of Korea). In this
for 18–24 hr. study, the primers URL-301 and URR-432 were used for amplification
of E. coli isolates and these primers targeting regulatory region of uidA
2.3.2 | Confirmation and metabolic fingerprint of structural gene in E. coli (Bej et al., 1991). Nevertheless, the primers
bacterial isolates using BIOLOG SAL-1F and SAL-2R were selected from conserved sequences within
a 2·3 kb randomly cloned DNA fragment from the Salmonella Typ-
In particular, for confirmatory testing, 366 of the stored bacterial iso- himurium chromosome (Aabo et al., 1993). On the other hand, the
lates were entered using BIOLOG and PCR and 122 isolates from primers S1F and S1R were selected for the amplification of the 23S
each of the species E. coli, Salmonella, and Listeria. Confirmations were rRNA gene in Listeria species (Paillard et al., 2003). The PCR reaction
rendered of presumed bacterial isolates and their biochemical meta- mixture of the bacterial isolate was carried out in 20 μl total volume
data using BIOLOG GEN III (Biolog Inc.) according to El-Liethy, as follows: 4 μl of 5x FIREPol Master Mix Ready to Load with
Hemdan, and El-Taweel (2018). On a microplate, each bacterial isolate 12.5 mM MgCl2, 0.5 μl (10 pmol) from each primer, and 2.5 μl from
was checked for 71 styles of carbon and 23 chemical issues. Every DNA extracted isolate. E. coli ATCC 25922, S. enterica Typhimurium
storage of bacterial isolates in TSB (Oxoid, UK), E. coli, salmonella, and ATCC 14028, L. monocytogenes ATCC 25152 were used as positive
Listeria species are suspended. The inoculated tubes have already controls. Agarose gel electrophoresis assessed the amplified materials.
been incubated for 24 hr at 37 C. After that the isolates were Gels were stained with ethidium bromide (0.005%, wt/vol), and use of
streaked onto the plate of tryptic soya agar (TSA) (Oxoid, UK) then 100 bp ladder and UVP BioDoc-it Imaging System was used to image
the plates were incubated for 18–24 hr at 37 C. Using a sterile the gel under UV transilluminator.
removable inoculator swab, a colony was assembled and inoculated
into 10 ml of inoculated fluid A (IF-A). The organism-containing IF-A 2.4 | Statistical analysis
was allocated to 96 microplate wells (100 μl per well). The microplates
were incubated at 37 C for 18–24 hr. The reading for each microplate The statistical study was conducted using GraphPad Prism 5.0. A two-
has been taken mechanically by the computerized MicroStation sys- way difference test (ANOVA) and a student t test were implemented
tem (Biolog Inc.) with the fingerprint data, which were previously fed to determine the values between viable cells using incubation
into the software.
4 of 10 EL-LIETHY ET AL.
T A B L E 1 Primer sets and PCR conditions used for bacterial isolates confirmation
Bacterial Primer name Primer sequence (50 to 30) TmoC PCR conditions (cycle) Product References
isolates URL-301 TGTTACGTCCTGTAGAAAGCCC 62.1 94 C for 1 min, 55 C for size (bp) Bej, Dicesare, Haff,
E. coli URR-432 AAAACTGCCTGGCACAGCAATT 60.3 153
SAL-1F GTA GAAATTCCCAGCGGGTAC TG 64.7 1 min and 72 C. (30x) and Atlas (1991)
Salmonella spp. 438
SAL-2R GTATCCATCTAGCCAACC ATT GC 62.9 95 C for 30 s, 60 C for 1 min Aabo, Rasmussen,
Listeria Spp. S1F AGT CGG ATA GTA TCC TTA C 53 and 72 C for 1.5 min. 460 Rossen,
(40×) Sorensen, and
S1R GGCTCTAACTACTTGTAG GC 58.4 Olsen (1993)
94 C for 1 min, 60/55 C for
1 min and 72 C for 1 min Paillard et al. (2003)
(35×)
Abbreviation: PCR, polymerase chain reaction.
temperatures of 22 C and 37 C. As well, the relationship between
three studied bacterial microorganisms is already verified.
3 | RESULTS AND DISCUSSION
3.1 | Determination of total bacterial counts in the
biofilm samples
Biofilm formation may be present in any wet area inhabited by natural F I G U R E 1 Log10 counts of the total bacteria at 37 C and 22 C in
microbes, including pathogenic bacteria. They can be harmful to the the natural biofilm cells harvested from different drainage pipes. Two-
health of individuals who are exposed to microhabitats, including bath- way analysis of variance states *** means that there is a high
room, kitchen, hospital, and laboratories sink drainage pipes, and so correlation (p ≤ .001) with significance between the counts of viable
on. (Chikere & Azubuike, 2014). Therefore, the total bacterial counts at cells at both 22 C and 37 C
both 22 C and 37 C are an essential parameter to determine the bac-
terial loads in the natural biofilm samples (Walker et al., 2000). In the Bates, Caporaso, Lauber, & Leff, 2013). Also, the most observed results
present study, the total bacterial counts at 22 C were slightly higher were that the kitchen sink which plays a role in carrying the large num-
than that at 37 C in all biofilm samples (Figure 1). This could be rev- ber of E. coli and Enterobacter spp., whereas, toilet area showed little
ealed that the lower temperature (22 C) showed the abundance of evidence of contamination with organisms of fecal sources (Hemdan
naturally occurring bacterial flora. In contrast, at high temperature et al., 2019). In this study, the average counts of total bacteria in all
(37 C), numerous bacterial pathogens, and members of collected natural biofilm samples were ranged between 104 and 108
Enterobacteriaceae that has direct effect on human health might be colony forming unit (CFU/cm2). In another study carried by Långmark,
present (Gensberger, Gössl, Antonielli, Sessitsch, & Kosti c, 2015). Storey, Ashbolt, and Stenström (2005) found that the concentrations
Moreover, the highest counts at 22 C may be due to the presence of of total cultivable bacteria in well-established biofilm varied from 10 to
autochthonous flora that already exists in water where the water tem- 106 CFU/cm2. On the other hand, the total bacterial counts in polypro-
perature is usually between 20–22 C, this temperature considered the pylene pipe materials biofilm ranged between 2.1 × 105 and
optimum for the microbial survival (Korhonen & Martikaine, 1991). In 5.5 × 107 CFU/cm2 (Rogers, Garrison, Grober, Hillis, & Franke, 1994).
the present work, the highest bacterial counts were observed in bio- In the present study, it was found that high statistical correlation with
film samples collected from kitchen sink drainage pipes followed by significance (p ≤ .001) was observed between the bacterial counts at
the hospital biofilm samples (Figure 1). The kitchen sink is considered 37 and 22 C (Figure 1).
as a shelter for huge number of microbial pathogens, while the toilet
showed little evidence of contamination with pathogenic microorgan-
isms and microbes of fecal origin (Josephson, Rubino, & Pepper, 1997).
Bacteria are capable of inhabiting the kitchen surfaces, and the direct
exposure of these microorganisms between humans and the kitchen
environment might have a direct outcome on human health (Flores,
The words you are searching are inside this book. To get more targeted content, please make full-text search by clicking here.
Sewage system as a source of Germs and bacterial outbreak-new studies
Discover the best professional documents and content resources in AnyFlip Document Base.
Search