physicalean- research 2022

Infection Control & Hospital Epidemiology (2018), 39, 1307–1315
doi:10.1017/ice.2018.235

Original Article

Sink traps as the source of transmission of OXA-48–producing
Serratia marcescens in an intensive care unit

Gili Regev-Yochay MD, MS1,2, Gill Smollan MD3, Ilana Tal RN1, Nani Pinas Zade RN1, Yael Haviv MD4,
Valery Nudelman RN4, Ohad Gal-Mor PhD2,5, Hanaa Jaber BsC1, Eyal Zimlichman MD6, Nati Keller MD3,7
and Galia Rahav MD2,8

1Infection Prevention and Control Unit, Sheba Medical Center, Tel-Hashomer, Israel, 2Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel,
3Microbiology Laboratory, Sheba Medical Center, Tel-Hashomer, Israel, 4Intensive Care Unit, Sheba Medical Center, Tel-Hashomer, Israel, 5Infectious Diseases
Research Laboratory, Sheba Medical Center, Tel-Hashomer, Israel, 6Management, Sheba Medical Center, Tel-Hashomer, Israel,, 7Ariel University, Ariel, Israel, and
8Infectious Disease Unit, Sheba Medical Center, Tel-Hashomer, Israel

Abstract
Background: Carbapenemase-producing Enterobacteriaceae (CPE) outbreaks are mostly attributed to patient-to-patient transmission via
healthcare workers.
Objective: We describe successful containment of a prolonged OXA-48–producing S. marcescens outbreak after recognizing the sink traps
as the source of transmission.
Methods: The Sheba Medical Center intensive care unit (ICU), contains 16 single-bed, semi-closed rooms. Active CPE surveillance
includes twice-weekly rectal screening of all patients. A case was defined as a patient detected with OXA-48 CPE >72 hours after admission.
A root-cause analysis was used to investigate the outbreak. All samples were inoculated on chrom-agar CRE, and carbapenemase genes
were detected using commercial molecular Xpert-Carba-R. Environmental and patient S. marcescens isolates were characterized
using PFGE.
Results: From January 2016 to May 2017, 32 OXA-48 CPE cases were detected, and 81% of these were S. marcescens. A single clone was the
cause of all but the first 2 cases. The common factor in all cases was the use of relatively large amounts of tap water. The outbreak clone was
detected in 2 sink outlets and 16 sink traps. In addition to routine strict infection control measures, measures taken to contain the outbreak
included (1) various sink decontamination efforts, which eliminated the bacteria from the sink drains only temporarily and (2) educational
intervention that engaged the ICU team and lead to high adherence to ‘sink-contamination prevention guidelines.’ No additional cases were
detected for 12 months.
Conclusions: Despite persistence of the outbreak clones in the environmental reservoir for 1 year, the outbreak was rapidly and successfully
contained. Addressing sink traps as hidden reservoirs played a major role in the intervention.

(Received 23 May 2018; accepted 20 July 2018)

A nationwide outbreak of carbapenem-resistant Enterobacter- OXA-48 enzymes, of the class D carbapenemases, were initi-
iaceae (CRE) emerged in Israel in 2006, with nosocomial spread ally identified in Turkey2 in K. pneumoniae and have recently
of Klebsiella pneumoniae carbapenemase (KPC)-producing K. emerged as a predominant carbapenemase type in various
pneumoniae. A national intervention was implemented in all Enterobacteriaceae in other countries.3 The first cases of OXA-
acute-care hospitals including active surveillance of high-risk -48–producing CPE in Israel were described in patients coming
patients and carrier isolation and cohorting. With these efforts, from the Palestinian Authority4,5 and from Syrian wounded
the outbreak was partially contained,1 yet carbapenemase- refugees treated in Israeli hospitals.6
producing Enterobacteriaceae (CPE) remain still widespread in
most of the large medical centers in Israel. Moreover, new types of Carbapenemase-producing Enterobacteriaceae can produce a
CPE have emerged, including NDM-1– and OXA-48–producing broad spectrum of infections; they associated with high mortality
Enterobacteriaceae. and very limited therapeutic options. Particularly troubling are
Serratia and Proteus, which are innately resistant to colistin, one
Author for correspondence: Gili Regev-Yochay, Infection Prevention and Control of the last-resort therapeutic options. Patient-to-patient cross
Unit, Sheba Medical Center, Ramat Gan, Israel. E-mail: [email protected] transmission is the main mechanism of spread of multidrug-
resistant (MDR) Enterobacteriaceae nosocomial outbreaks.
PREVIOUS PRESENTATION: These data were partially presented at the 28th Colonized patients are considered the most important reservoir,
European Society of Clinical Microbiology and Infectious Diseases Conference and infection control breaks in hand hygiene via healthcare
(ECCMID) on April XX, 2018, in Madrid, Spain workers is considered the predominant transmission chain.7
Environmental reservoirs have not been frequently reported as
Cite this article: Regev-Yochay G, et al. (2018). Sink traps as the source of the source for such outbreaks.
transmission of OXA-48–producing Serratia marcescens in an intensive care unit.
Infection Control & Hospital Epidemiology 2018, 39, 1307–1315. doi: 10.1017/ice.2018.235

© 2018 by The Society for Healthcare Epidemiology of America. All rights reserved.

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

1308 Gili Regev-Yochay et al

Herein, we describe a prolonged ICU S. marcescens outbreak, Outbreak investigation
at an ICU of a large tertiary-care medical center, which was
successfully contained by a combined intervention after recog- During October 2016, after 6 patients with OXA-48–producing S.
nizing the sink traps as the source of transmission. marcescens were detected, enhanced control measures were
undertaken, including increased hand hygiene observations as
Methods well as educational sessions. Thorough cleaning of all surfaces and
medical devices with 1,000 PPM sodium hypochlorite and qua-
Setting and CPE surveillance and isolation policy ternary ammonium, accordingly, was carried out. The number of
ICU patients was cut from 16 to 8. The outbreak was partially
The Sheba Medical Center (SMC) is a tertiary-care academic contained, but a few weeks later, during March 2017, new
medical center in central Israel, affiliated with the Tel Aviv acquisitions of OXA-48–producing S. marcescens were again
University; it has 1,600 beds for both acute-care and rehabilitation identified.
patients, with 96,800 annual admissions to the acute-care hospi-
tal. The general ICU consists of 16 active beds for medical and At this point, we took a new investigational approach by
surgical patients, with ~470 admissions per year. The mean ICU engaging the ICU team in the process. We conducted 2 work-
hospitalization duration is 9.6 days. The patient-to-nurse ratio is shops (8 hours each) for the whole ICU team, including physi-
2:1. Each bed is in a semi-closed space separated by glass walls on cians, nurses, cleaning team, pharmacist, and social worker. The
2 sides. The water supply and the wastewater removal system in workshop included discussion groups that investigated potential
the ICU consists of 22 sinks, including a sink in each patient unit infection control breaks that could explain the outbreak using a
at the entrance to the room, 4 additional sinks in the general root-cause analysis with the fishbone visual approach (http://
space, and 2 sinks in the staff toilet and dressing rooms. www.siliconfareast.com/ishikawa.htm).

Active surveillance CPE screening of all ICU patients has been Environmental sampling
in place since 2007 and includes admission as well as twice-
weekly rectal screening. All high-risk patients are placed in pre- Initial environmental sampling included swabs from beds, ven-
emptive isolation pending admission surveillance culture results. tilators, mattresses, care carts, high-touch surfaces (eg, monitor
Once a patient is detected as a CPE carrier, he is moved to a and computer keyboards), IVAC pumps, several solutions, fau-
designated cohort space with dedicated nursing staff and cets, and water samples. Following the root-cause analysis, the
equipment. water environment was suspected to be the source of transmis-
sion, and further environmental sampling included water sam-
Case definition and patient data ples, faucets, water pipes, sink surfaces, sink outlets, sink traps,
and drain systems.
A case was defined as an ICU hospitalized patient with OXA-48–
producing Enterobacteriaceae isolated either from surveillance Results
cultures or from clinical cultures obtained at the SMC since 2016.
Data collected from the electronic medical files included date of Since 2007, when routine screening for CRE began in our hos-
admission, date of CPE acquisition, source of admission (home, pital, and until 2016, 294 ICU patients acquired CRE, of whom
nursing home, etc), admission diagnosis, comorbidities, length of ~70% had Klebsiella pneumoniae and only a single case
stay (LOS) until and after CPE acquisition, type of clinical had Serratia marcescens. Routine determination of the
infection, and outcome. carbapenemase-producing genes began in 2013. Nearly all CPE
isolates carried the blaKPC gene, and only 6 isolates carried
Microbiological methods blaOXA-48 gene, 5 of which were K. pneumoniae and 1 was
Citrobacter freundii.
Clinical isolates were identified following Clinical and Laboratory
Standards Institute (CLSI) guidelines. Antibiotic susceptibility Outbreak description
was determined using the disk diffusion method. Minimum
inhibitory concentrations (MICs) of carbapenems were deter- From January 2016 until May 2017, 658 patients were hospita-
mined using the Etest (AB Biodisk). The MIC break points were lized at the ICU, of whom 41 patients (6.4%) acquired CPE. Of
defined according to CLSI guidelines. Surveillance rectal samples these, 31 (76%) acquired OXA-48–producing Enterobacteriaceae
and environmental samples were obtained using Copan Amies and 27 (87%) of these were S. marcescens. Five patients with
sterile transport swabs (Copan Diagnostics, Murrieta, CA). The OXA-48–producing S. marcescens acquired an additional
rectal swab samples were streaked onto Chromagar KPC plates OXA-48–producing Enterobacteriaceae. One patient acquired
(Hy Laboratories, Rehovot, Israel) and incubated overnight at 35° S. marcescens that harbored both blaOXA-48 and blaKPC genes.
C in ambient air. Suspicious colonies were identified using Three additional patients were infected with KPC-producing
matrix-assisted laser desorption/ionization time-of-flight mass S. marcescens. Patient characterizations of these 34 patients are
spectroscopy (MALDI-TOF MS). Carbapenemase genes were presented in Table 1. Of 34 cases, 30 were initially detected
identified by polymerase chain reaction (PCR) using the Xpert by routine rectal screening; 11 patients developed clinical
Carba-R cartridge (GeneXpert, Cepheid, Sunnyvale, CA). infections; and 3 mortality cases were attributed to the infection.
Altogether 12 patients died, for an in-hospital mortality rate of
Molecular characterization of the outbreak and environmental 35%. Figure 1 presents the temporal relationship between the
isolates was performed using pulsed-field gel electrophoresis outbreak cases, and Figure 2 presents all CPE acquisitions in the
(PFGE),8 where XbaI-digested Serratia DNA embedded in agar- ICU during the outbreak year.
ose plugs were subjected to PFGE analysis at 14°C using CHEF-
DR III system (Bio-Rad, Hercules, CA).

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

Infection Control & Hospital Epidemiology 1309

Table 1. Clinical Characteristics of Patients With OXA-48 Carbapenemase–Producing Enterobacteriaceae (CPE)

LOS LOS

Before, Site of Site of PFGE After, Antibiotic
da Detection Infection, (d)b
Sex Age Comorbidities Bacteria Type dc,d Outcome Susceptibilityd

1 F 59 Sepsis, Perforation of colon 7 RS Lung and E. coli- OXA48 180 Death Amikacin, tobramycin

blood (7 d)

2 M 61 Motor vehicle accident 7 RS S. marcescens - A 4 Alive
OXA48

3 F 51 Suicide attempt 26 RS C. freundii- 1 Alive
OXA48

4 M 38 Cerebrovascular accident 20 RS Lung (5 d) S. marcescens- B 32 Alive Cotrimoxazole, ceftazidime,
OXA48 azidothymidine, amikacin

5 F 42 Lymphoma, head tumor, 24 RS S. marcescens- C 31 Death
hydrocephalus OXA48

6 M 64 Burns 24 RS Blood (21d) S. marcescens- C”‘ 60 Alive Cotrimoxazole, amikacin
KPC

7 M 45 Motor vehicle accident 21 RS S. marcescens- C 24 Alive
OXA48

8 M 42 Intracranial hemorrhage due 39 RS E. coli- OXA48 NA 80 Alive
to trauma

9 F 74 COPD, SCAP severe 17 RS S. marcescens- … 111 Death
OXA48

10 f 34 Brain edema pregnancy 20 RS Blood (17 d) S. marcescens- C”‘ 22 Death Azidothymidine, amikacin,
OXA48 cotrimoxazole

11 M 78 Craniotomy due to ethmoid 21 RS S. marcescens C 8 Death
sinus cancer

12 f 79 Bladder cancer, E. faecalis 17 RS S. marcescens- F 5 Death
endocarditis, septic shock KPC

13 M 55 Cirrhosis, encephalopathy, 22 RS Blood (5 d) S. marcescens- C 8 Death
bleeding varices OXA48

14 M 26 Multiple trauma 14 RS S. marcescens- C' 7 Alive
fall OXA48

15 M 36 Multiple trauma 14 L K. pneumoniae- NA 5 Alive Cotrimoxazole, ceftazidime,
B + RS gentamycin, amikacin,
Blood (1 d) OXA48 ... ciprofloxacin,
azidothymidine
S. marcescens-

OXA48

16 M 70 Necrotizing pancreatitis, 58 B Blood S. marcescens- ... 16 Death Cotrimoxazole, tobramycin,
sepsis OXA-48 K. OXA48 amikacin, ciprofloxacin
pneumoniae

17 M 50 Multiple trauma 17 RS Enterobacter sp.- NA 15 Alive
OXA48

18 F 39 Schizophrenia, multiple 20 RS S. marcescens- - 55 Alive
trauma RS
OXA48 NA

K.

pneumoniae-

OXA48

19 F 65 Metastatic breast cancer, 27 RS S. marcescens- C 0 Alive
septic shock RS
OXA48 NA

K.

pneumoniae-

OXA48

20 M 50 B-cell lymphoma, septic shock 27 RS S. marcescens- C 10 Death
OXA48

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

1310 Gili Regev-Yochay et al

Table 1. (Continued )

LOS LOS

Before, Site of Site of PFGE After, Antibiotic
da Detection Infection, (d)b Bacteria
Sex Age Comorbidities Type dc,d Outcome Susceptibilityd
Blood (2 d) S. marcescens-
21 M 35 Lymphoma, ulcerative colitis, 7 Bile abscess (90 d) OXA48 C”‘ 25 Alive Cotrimoxazole, amikacin
necrotic bowel s/p
laparotomy

22 F 52 Metastatic cancer 29 RS S. marcescens- ... 36 Death
OXA48

23 M 72 Ruptured aortic aneurysm 40 RS S. marcescens- ... 9 Alive
OXA48

24 F 37 Necrotizing fascitis 9 RS S. marcescens- C 17 Alive
OXA48

25 F 40 S. pyogenes, toxic shock 23 RS S. marcescens- C 3 Alive
syndrome related to birth OXA48

26 M 72 Neck abscess 17 RS S. marcescens- C 45 Alive

OXA48 NA

Enterobacter-

OXA48

27 M 21 Motor vehicle accident, anoxic 3 RS Blood (74 d) K. pneumoniae- NA > 365 Alive Gentamycin
brain OXA48 + KPC

28 M 78 Lymphoma, myelodysplastic 14 RS S. marcescens- C 15 Death
syndrome, colon cancer OXA48

29 F 59 Urosepsis, metastatic breast 12 RS S. marcescens- C 34 Alive
cancer OXA48

30 M 50 Multiple trauma 16 RS S. marcescens- ... 4 Death
KPC

31 M 63 Gastrointestinal bleeding, 13 L K. pneumoniae- NA 27 Alive Amikacin, tigecycline,
OXA48 + KPC C colistin, cotrimoxazole
gastric cancer, anoxic brain Wound S. marcescens-
OXA48 + KPC

32 M 42 Fall, multiple trauma 17 RS S. marcescens- ... Alive
OXA48

33 F 69 MS, septic shock due to renal 14 RS S. marcescens- C 18 Alive
abscess OXA48

34 M 21 Multiple trauma gunshot 14 RS Blood (90 d) & S. marcescens- C 9 Alive Cotrimoxazole, gentamycin,
wounds amikacin, ceftazidime,
RS (7 d) urine (150 d) OXA48 NA azidothymidine, colistin,
ciprofloxacin, ceftazidime
RS (19 d) K. NA

pneumoniae-

OXA48

C. koseri-

OXA48

Note. LOS, length of stay; PFGE, pulsed-field gel electrophoresis; RS, rectal swab; COPD, chronic obstructive pulmonary disease NA, nonapplicable (not a Serratia isolate); SCAP, severe

community acquired pneumonia; MS, multiple sclerosis.
aLength of stay (days) from admission until CPE was initially detected.
bNo. of days after initial CPE detection.
cLength of stay (days) from CPE detection until hospital discharge.
dAntibiotic susceptibility of clinical strains.

Results of the outbreak investigation OXA-48–producing S. marcescens in many sink traps and sink
outlets (Table 2).
Our initial investigation, including environmental sampling and
consultations with the Israeli national infection control unit, did Typing S. marcescens isolates
not identify any potential source. The root-cause analysis inves-
tigation revealed that significant use of tap water was a common Of the 30 S. marcescens patient isolates, 26 were available and
factor in all cases. For most ICU procedures in which water is were characterized by PFGE. Apart from the first 2 cases, all
needed, sterile bottled water is used and tap water is used only for OXA-48–producing S. marcescens isolates belonged to a single
cleaning patients suffering from diarrhea, or rarely when flushing clone (defined ‘C’). Notably, 1 of the KPC-producing S. marcescens
gastric tubes. Sampling of the water environment revealed

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

Infection Control & Hospital Epidemiology 1311

Fig. 1. Synoptic curve of patients colonized and/or infected with OXA-48–producing carbapenemase-producing Enterobacteriaceae (CPE) as well as Klebsiella pneumoniae
carbapenemase (KPC)-producing S. marcescens. Black lines denote the duration of hospitalization in the intensive car unity ICU. Dashed lines denote hospitalization outside
the ICU that had an unfavorable outcome (mortality or additional CPE acquisition). Note. X denotes the discharge day; red X denotes mortality; triangle denotes acquisition of
OXA-48–producing carbapenemase; diamond denotes KPC-producing carbapenemase; red denotes S. marcescens; orange denotes Citerobacter sp; blue denotes E. coli; yellow
denotes Enterobacter sp; and green denotes K. pneumoniae.

as well as the KPC + OXA-48–producing isolate were non- 85% in the intervention versus preintervention periods (Table 2,
distinguishable (clone ‘C’). The other available KPC-producing S. Figure 3). Notably, different sinks were contaminated each week,
marcescens isolate was distinct (defined ‘F’) (Table 1). and the staff complained of the odor of acetic acid. On August 3,
2017, this intervention was discontinued. All sink traps were
Of 101 S. marcescens isolates from the sinks, 57 were available replaced again, and weekly sink-trap sampling continued.
for PFGE. All S. marcescens sink isolates belonged to 2 clones; 38
(67%) were nondistinguishable from the major outbreak clone Following Wolf et al,10 a single ‘self-disinfecting’ sink trap
(‘C’). Most of these (55%) harbored both blaOXA48 and blaKPC (Medizinische Hygiene-Siphone BIOREC MoveoMed, Dresden,
genes as detected by PCR, while in 29% only blaOXA48 was Germany) was purchased (~US$2,300 or ~2000 Euro) and
detected, in 11% only blaKPC was detected, and in 2 isolates (5%) installed in room 12, where a KPC-producing K. pneumoniae had
blaVIM was detected together with blaOXA48 + /-blaKPC. The sec- been detected 1 week earlier (Figure 4). This sink trap presumably
ond clone detected in the sinks was nondistinguishable from the prevents the formation of a biofilm by means of permanent
KPC-producing S. marcescens of patient 12 (clone ‘F’). Indeed, physical disinfection (heating and ultrasound), electromagnetic
most of these isolates (63%) harbored only blaKPC, but 7 isolates cleansing, and antibacterial coating. In the following weeks, the
(37%), harbored blaOXA48 + blaKPC. same bacteria continuously grew in the sink outlet despite the
experimental sink trap. One month after installation of this sink
Control and prevention measures trap, a patient in room 11 acquired a genetically similar KPC-
producing K. pneumoniae. This sink trap was therefore replaced
Following the identification of the sinks, and particularly the sink back to a standard sink trap. Currently, a strict daily cleaning
traps, as the source of transmission, we carried out 2 main protocol with 1,000 ppm hypochlorite is being used, and once-
measures: (1) sink-trap decontamination efforts and (2) an edu- monthly sink-trap surveillance is carried out. The current average
cational intervention enhancing specific infection control mea- proportion of contaminated sinks is ~15% (Table 2).
sures and focusing on the sink as a source of transmission. Within
a few weeks, bed occupancy returned to full, with 15–16 patients Educational intervention
consistently hospitalized in the ICU.
A bundled intervention to decrease CPE transmission has been
Sink-trap decontamination efforts implemented in our institution for a decade as described above.
Hand hygiene compliance has been persistently ~80% in the pre-
Following the identification of the sink traps as a source, all sink outbreak year. To further engage the ICU team, we conducted 2
traps were replaced, and the water supply was treated according workshops, as described above. Following the first workshop, we
to the Prevention and Control of Legionella Infection Protocol identified the sink traps as the source of transmission. The second
(PCLIP). This intervention consisted of heating and hyper- workshop focused on measures to avoid further transmission
chlorination of the main water tank and terminal points for from contaminated sinks. Strict guidelines on sink use were
12 hours with free residual chlorine (20–30 mg/L). Three days enforced: (1) using sinks only for hand washing and only when
after this procedure, no CPE was detected in any sink trap. necessary, (2) prohibiting contamination of sinks with clinical
However, 10 days later, the outbreak clone was again detected in waste disposal, and (3) prohibiting storage of materials around
14 of 16 sink traps (Figure 3). the sinks. Adherence to these measures was continuously
encouraged by the infection control unit as well as by the ICU
Following Stjarne et al,9 sink traps were treated with 250 mL team itself, with update presentations, observations, and 10-
25% acetic acid once weekly for 10 consecutive weeks. Sink traps minute ‘time-out’ sessions.
were also sampled weekly before treatment. This treatment sig-
nificantly reduced the number of contaminated sinks: 28% versus

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

1312 Gili Regev-Yochay et al

Table 2. Environmental Samples Table 2. (Continued )

Environmental Sources Result No. Positive Isolates Environmental Sources Result No. Positive Isolates
Sampled Sampled K. pneumoniae-NDM 1

Preintervention period (Nov 2016–Apr 2017)

Ventilators Kluyvera accorbat-KPC 2

Expiratory valve, internal Negative Intervention period 2: only strict sink cleaning (Aug 2017–Apr 2018)
and external filters
Sink traps (n = 289; 16% S. marcescens- 16
positive) OXA48 + KPC
Beds

Bed rails, mattresses Negative S. marcescens- OXA48 10

Ultrasound S. marcescens- KPC 7

Ultrasound probes, Negative S. marcescens- OXA48 2
screen, handles, probe + VIM + KPC
stations
K. pneumoniae- KPC 5

Patient cooling device Enterobacter sp- KPC 6

External machine and Negative Sink outlets (n = 311; 11% S. marcescens- OXA48 7
water outlets
positive) + KPC

Additional equipment S. marcescens- OXA48 5

Computer keyboard, Negative S. marcescens- KPC 3
isolation equipment
stands S. marcescens- OXA48 1
+ VIM + KPC
Dialysis equipment

Drainage system, water Negative K. pneumoniae- KPC 14
outlet
K. pneumoniae- NDM1 1

Sink environment Enterobacter sp- KPC 4

Tap water Negative

Faucets Negative To further increase awareness and transparency, the number
of days without CPE acquisition were posted daily on a centrally
Faucet filters Negative located infection control board. On day 100 without CPE
acquisition, a celebration was organized with the medical center
Sink outlets (n = 5) S. marcescens- OXA48 5 director, the infection control team, and the ICU teams.
+ KPC
Intervention outcome and follow-up
Sink traps (n = 20; 85% S. marcescens- OXA48 14
The last CPE S. marcescens detected in ICU patients, and in
positive) + KPC general in our medical center, occurred on May 7, 2017, 1 week
after the identification of the sink traps as a source of transmis-
S. marcescens-OXA48 2 sion of CPE S. marcescens. Despite continuous growth of
carbapenemase-producing S. marcescens in some of the sink traps
S. marcescens- KPC 1 in the ICU for >11 months, the transmission ceased. Since May
2017, a few CPE-infected patients have been hospitalized in the
Central pipe trap (n = 1) S. marcescens- OXA48 1 ICU, but the only 2 CPE acquisitions documented in the ICU
were KPC-producing K. pneumoniae, which was probably
Intervention period 1: Acetic acid (May 2017–Aug 2017) transmitted from the sink in which the MoveoMed siphon was
installed.
Sink traps (n = 141; 26% S. marcescens- OXA48 19
positive) + KPC Discussion

S. marcescens- OXA48 6 We describe an OXA-48–producing S. marcescens outbreak in an
ICU of a large tertiary-care medical center, and we have
S. marcescens- OXA48 2 demonstrated that the source of transmission was the sink traps.
+ VIM We show that complete elimination of CPE from the sink traps
was not successful despite various decontamination approaches,
S. marcescens- OXA48 1
+ VIM + KPC

S. marcescens- KPC 1

Enterobacter sp-KPC 4

Enterobacter sp- 1
KPC + VIM + NDM

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

Infection Control & Hospital Epidemiology 1313

Fig. 2. Carbapenemase-producing Enterobacteriaceae (CPE) acquisitions in the intensive care unit (ICU) during 2016–2017.

Fig. 3. Results of the weekly surveillance of sink traps in the ICU and the concurrent sink-trap interventions.

even after nearly 1 year without hospitalized colonized patients as recognized as a potential source of carbapenem-resistant organ-
well as several replacements of the sink traps. The outbreak, ism (CRO) transmission.11,12. Most reports on CRO outbreaks
however, was rapidly contained by a combined intervention: related to a water reservoir describe Pseudomonas spp or Acine-
conducting an educational intervention, engaging the team with tobacter baumannii outbreaks,9,12–14 with only a few CPE out-
focus on the proper use of sinks in the ICU, and reducing the breaks directly related to water reservoirs, including a KPC-2
environmental contamination load. producing K. oxytoca,15 IMP-4-producing Enterobacter cloacae,16
KPC-2–producing K. pneumoniae,17 and recently, OXA-48–pro-
Environmental transmission has been considered to have a ducing C. freundii.18
negligible role in CPE outbreaks and CPE transmission, while
patient-to-patient transmission via the hands of healthcare The Serratia species is known to survive and grow in moist
workers has been considered the major route of transmission.7 environments. Several Serratia outbreaks have been associated with
However, the hospital water environment is increasingly contaminated liquids including saline bottles,19 chlorhexidine,20–22

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

1314 Gili Regev-Yochay et al

and IV pain control fluids.23 However, the sink environment as a changes required with regard to the underappreciated con-
potential source has been suggested only rarely. Ivady et al24 reported tamination risk of the sink surrounding.
2 consecutive S. marcescens outbreaks in a NICU where the epidemic
strain was detected in 1 culture of a tap water sample between the 2 It is difficult to measure the role of each of the intervention
outbreaks, despite many negative environmental cultures obtained arms. However, we believe that the team compliance with the new
during the outbreaks. Kotsanas et al25 reported the persistence of S. strict sink-use guidelines played a major role in the containment.
marcescens in several sink traps of an ICU during a CPE outbreak of Engaging teams to a culture change in the field of infection
various species, where 4 of 11 clinical cases were due to IMP-4– control is extremely complicated, particularly in an ICU, where
producing S. marcescens. the treatment of severely acute patients is a high-pressure envir-
onment for healthcare workers.
Sink traps have been recently suggested to be a hidden source
of gram-negative rod outbreaks in general and CRO in parti- Our study has several limitations. First, we used a 2-arm
cular.10,18,25–27 The mechanism of transmission from sink traps to intervention, and it is impossible to determine the particular role
patients is probably that running water drives aerosol con- of each arm. Second, the molecular characterization of the isolates
taminated with bacteria from the biofilm of the sink trap to the was conducted using PFGE. Further analysis using whole-genome
sink surrounding. Using a fluorescent marker, it was demon- sequencing could further clarify the evolution of CPE in sink
strated that drain contents splashed at least 1 meter from the sink traps. Last, while we demonstrate the failure of the self-
during hand washing.28,29 Furthermore, we observed sterile disinfecting sink trap (Moveo), we installed only a single device.
materials and devices that were about to be inserted into the
patients that were misplaced in the very near surrounding of the Currently, there seems to be no complete solution for pre-
sinks in the ICU. venting CRO transmission from sink traps to patients. Patient
room architecture and sink designs should be carefully conceived.
The source for the initial OXA-48–producing S. marcescens in Eliminating sinks in ICU rooms has been suggested,32 but this
our medical center is unclear. Despite rigorous screening for CPE, would raise many difficulties because hand washing is frequently
OXA-48–producing S. marcescens was never detected in our necessary, even in the era of alcohol-based hand hygiene, mainly
medical center during at least 6 years before the outbreak. A few when handling patients with C. difficile infections. A technical
cases of KPC-producing S. marcescens were previously identified plumbing solution, a chemical solution, or an architectural
in our ICU as well as a single case of OXA-48–producing solution should be sought. Meanwhile, healthcare teams should
K. pneumoniae. As intra- and interpatient dissemination of be educated to understand the important role of this hidden
blaOXA-48 from one species to another has been previously reservoir: the sink trap.
reported.30 Many of the carbapenemase-producing S. marcescens
found in the sink traps in our ICU harbored both blaKPC and Acknowledgments. We thank the ICU team for their cooperation, especially
blaOXA-48, suggesting evolution of multiple CPE in the sink-trap Dr Nir Shimony and Dr David Livingstone for their deep involvement. We
microbiome. Flushing fluids containing antibiotics, including thank the Sheba logistic team, particularly the plumbers, Arie Hikrey and Avi
colistin, into the sinks could have further promoted selection of Simchi, for their assistance. We thank the Infection Prevention and Control
resistant bacteria, particularly Serratia, which are innately resis- team for their dedicated work, and Aylana Reiss-Mandel for her assistance in
tant to colistin. the laboratory work, as well as Ronen Gellstein, Raya Bornstein, and Iris Stein
from Magen Haim for their continuous support.
The persistence of CPE, and particularly the epidemic strain,
for many months after the last infected patient was discharged Financial support. No financial support was provided relevant to this article.
highlights the importance of this underappreciated environmental
reservoir and suggests that the plumbing system is persistently Conflicts of interest. All authors report no conflicts of interest relevant to
contaminated, even after intensive decontamination efforts this article.
including PCIP treatment, weekly treatment with acetic acid, and
repeated sink-trap replacement. Indeed, Weingarten et al31 References
showed that wastewater pipes are a vast resilient reservoir of
CPOs; they found common plasmid backbones in both envir- 1. Schwaber MJ, Carmeli Y. An ongoing national intervention to contain the
onmental areas and in the patient population. The various spread of carbapenem-resistant Enterobacteriaceae. Clin Infect Dis
infection control strategies to contain outbreaks related to hos- 2014;58:697–703.
pital water environment were recently summarized,12 but none of
the strategies reviewed (including bleach, hydrogen peroxide, UV 2. Poirel L, Heritier C, Tolun V, Nordmann P. Emergence of oxacillinase-
light, or sink and drainage system replacement) was particularly mediated resistance to imipenem in Klebsiella pneumoniae. Antimicrob
successful, similar to our experience. Especially concerning was Agent Chemother 2004;48:15–22.
the failure of the new and expensive Moveoshiphon technology in
our experience. Our result is somewhat unlike that of Mathers 3. Mataseje LF, Boyd DA, Fuller J, et al. Characterization of OXA-48–like
et al,27 who reported a successful bundled intervention including carbapenemase producers in Canada, 2011–2014. J Antimicrob Chemother
hopper covers and use of the Moveosiphon. However, they con- 2017. doi: 10.1093/jac/dkx462.
cluded that their ability to assess the efficacy of these self-
disinfecting sink traps was limited and requires further 4. Adler A, Solter E, Masarwa S, et al. Epidemiological and microbiological
evaluation. characteristics of an outbreak caused by OXA-48–producing Enterobac-
teriaceae in a neonatal intensive care unit in Jerusalem, Israel. J Clin
While our sink-trap decontamination efforts were not com- Microbiol 2013; 51:2926–2930.
pletely successful, they did reduce the number of contaminated
sinks and, thus, the overall level of environmental contamination. 5. Goren MG, Chmelnitsky I, Carmeli Y, Navon-Venezia S. Plasmid-
The second arm of our intervention was an intensive effort to encoded OXA-48 carbapenemase in Escherichia coli from Israel. J
educate the healthcare teams and to engage them in behavioral Antimicrob Chemother 2011; 66:672–673.

6. Lerner A, Solter E, Rachi E, et al. Detection and characterization of
carbapenemase-producing Enterobacteriaceae in wounded Syrian patients
admitted to hospitals in northern Israel. Eur J Clin Microbiol Infect Dis
2016;35:149–154.

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

Infection Control & Hospital Epidemiology 1315

7. Friedman ND, Carmeli Y, Walton AL, Schwaber MJ. Carbapenem- 20. Morillo A, Torres MJ, Alonso Salas MT, Conde M, Aznar J. Implication of
resistant Enterobacteriaceae: a strategic roadmap for infection control. a national outbreak of Serratia marcescens associated with a contaminated
Infect Control Hosp Epidemiol 2017;38:580–594. solution of chlorhexidine in a paediatric hospital. An Pediatr (Barc) Mar
2018;88:171–172.
8. Ribot EM, Fair MA, Gautom R, et al. Standardization of pulsed-field gel
electrophoresis protocols for the subtyping of Escherichia coli O157:H7, 21. Merino JL, Bouarich H, Pita MJ, et al. Serratia marcescens bacteraemia
Salmonella, and Shigella for PulseNet. Foodborne Pathogen Dis 2006;3:59–67. outbreak in haemodialysis patients with tunnelled catheters due to
colonisation of antiseptic solution. Experience at 4 hospitals. Nefrologia
9. Stjarne Aspelund A, Sjostrom K, Olsson Liljequist B, Morgelin M, 2016;36:667–673.
Melander E, Pahlman LI. Acetic acid as a decontamination method for
sink drains in a nosocomial outbreak of metallo-beta-lactamase– 22. de Frutos M, Lopez-Urrutia L, Dominguez-Gil M, et al. Serratia
producing Pseudomonas aeruginosa. J Hosp Infect 2016;94:13–20. marcescens outbreak due to contaminated 2% aqueous chlorhexidine.
Enferm Infec Microbiol Clin 2017;35:624–629.
10. Wolf I, Bergervoet PW, Sebens FW, van den Oever HL, Savelkoul PH, van
der Zwet WC. The sink as a correctable source of extended-spectrum 23. Chiang PC, Wu TL, Kuo AJ, et al. Outbreak of Serratia marcescens
beta-lactamase contamination for patients in the intensive care unit. J postsurgical bloodstream infection due to contaminated intravenous pain
Hosp Infect 2014;87:126–130. control fluids. Int J Infect Dis 2013;17:e718–e722.

11. Kanamori H, Weber DJ, Rutala WA. Healthcare outbreaks associated with 24. Ivady B, Szabo D, Damjanova I, Pataki M, Szabo M, Kenesei E. Recurrent
a water reservoir and infection prevention strategies. Clin Infect Dis 2016; outbreaks of Serratia marcescens among neonates and infants at a
62:1423–1435. pediatric department: an outbreak analysis. Infection 2014;42:891–898.

12. Kizny Gordon AE, Mathers AJ, Cheong EYL, et al. The hospital water 25. Kotsanas D, Wijesooriya WR, Korman TM, et al. “Down the drain”:
environment as a reservoir for carbapenem-resistant organisms causing carbapenem-resistant bacteria in intensive care unit patients and
hospital-acquired infections—a systematic review of the literature. Clin handwashing sinks. Med J Austral 2013;198:267–269.
Infect Dis 2017;64:1435–1444.
26. Leung GH, Gray TJ, Cheong EY, Haertsch P, Gottlieb T. Persistence of
13. Lalancette C, Charron D, Laferriere C, et al. Hospital drains as reservoirs related bla-IMP-4 metallo-beta-lactamase–producing Enterobacteriaceae
of Pseudomonas aeruginosa: multiple-locus variable-number of tandem from clinical and environmental specimens within a burns unit in
repeats analysis genotypes recovered from faucets, sink surfaces and Australia—a six-year retrospective study. Antimicrob Resist Infect Control
patients. Pathogens. 2017;6(3). doi: 10.3390/pathogens6030036. 2013;2:35.

14. Salm F, Deja M, Gastmeier P, et al. Prolonged outbreak of clonal MDR 27. Mathers AJ, Vegesana K, German Mesner I, et al. Intensive care unit
Pseudomonas aeruginosa on an intensive care unit: contaminated sinks wastewater interventions to prevent transmission of multi-species
and contamination of ultra-filtrate bags as possible route of transmission? Klebsiella pneumoniae carbapenemase (KPC) producing organisms. Clin
Antimicrob Resist Infect Control 2016; 5: 53. Infect Dis 2018;67:171–178.

15. Leitner E, Zarfel G, Luxner J, et al. Contaminated handwashing sinks as 28. Hota S, Hirji Z, Stockton K, et al. Outbreak of multidrug-resistant
the source of a clonal outbreak of KPC-2–producing Klebsiella oxytoca on Pseudomonas aeruginosa colonization and infection secondary to
a hematology ward. Antimicrob Agent Chemother 2015;59:714–716. imperfect intensive care unit room design. Infect Control Hosp Epidemiol
2009; 30:25–33.
16. Betteridge T, Merlino J, Natoli J, Cheong EY, Gottlieb T, Stokes HW.
Plasmids and bacterial strains mediating multidrug-resistant hospital- 29. Kotay S, Chai W, Guilford W, Barry K, Mathers AJ. Spread from the sink
acquired infections are coresidents of the hospital environment. Microb to the patient: in situ study using green fluorescent protein (GFP)-
Drug Resist 2013; 19:104–109. expressing Escherichia coli to model bacterial dispersion from hand-
washing sink-trap reservoirs. Appl Envir Microbiol 2017;83(8):e03327–16.
17. Tofteland S, Naseer U, Lislevand JH, Sundsfjord A, Samuelsen O. A long-
term low-frequency hospital outbreak of KPC-producing Klebsiella 30. Arana DM, Saez D, Garcia-Hierro P, et al. Concurrent interspecies and
pneumoniae involving intergenus plasmid diffusion and a persisting clonal dissemination of OXA-48 carbapenemase. Clin Microbiol Infect
environmental reservoir. PloS One 2013;8:e59015. 2015;21:148:e141–e144.

18. De Geyter D, Blommaert L, Verbraeken N, et al. The sink as a potential 31. Weingarten RA, Johnson RC, Conlan S, et al. Genomic analysis of
source of transmission of carbapenemase-producing Enterobacteriaceae in hospital plumbing reveals diverse reservoir of bacterial plasmids
the intensive care unit. Antimicrob Resist Infect Control 2017;6:24. conferring carbapenem resistance. mBio 2018;9:e02011–17.

19. Us E, Kutlu HH, Tekeli A, Ocal D, Cirpan S, Memikoglu KO. Wound 32. Hopman J, Tostmann A, Wertheim H, et al. Reduced rate of intensive
and soft-tissue infections of Serratia marcescens in patients receiving care unit acquired gram-negative bacilli after removal of sinks and
wound care: a healthcare-associated outbreak. Am J Infect Control introduction of ‘water-free’ patient care. Antimicrob Resist Infect Control
2017;45:443–447. 2017;6:59.

Downloaded from https://www.cambridge.org/core. 26 Mar 2022 at 18:14:08, subject to the Cambridge Core terms of use.

J Infect Chemother 22 (2016) 450e455

Contents lists available at ScienceDirect

Journal of Infection and Chemotherapy

journal homepage: http://www.elsevier.com/locate/jic

Original article

Sources of sporadic Pseudomonas aeruginosa colonizations/infections
in surgical ICUs: Association with contaminated sink trap

ZhaoYan Zhou a, BiJie Hu b, *, XiaoDong Gao b, Rong Bao a, Min Chen c, HuaYin Li a

a Clinical Microbiology Laboratory, Department of Respiratory Disease, Zhongshan Hospital of Fudan University, Shanghai, China
b Department of Infection Control, Zhongshan Hospital of Fudan University, Shanghai, China
c Laboratory of Microbiology, Shanghai Municipal Center for Disease Control & Prevention, Shanghai, China

article info abstract

Article history: Background: Many studies have reported the hospital outbreaks of Pseudomonas aeruginosa due to cross-
Received 24 November 2015 contamination between patients and water fittings, but the importance of water fittings as sources of
Received in revised form sporadic P. aeruginosa colonizations/infections remains ambiguous.
1 March 2016 Aim: To investigate the sources of sporadic P. aeruginosa colonizations/infections in a clinical intensive
Accepted 31 March 2016 surveillance, and further analysis the potential of sink trap for P. aeruginosa transmission in intensive care
Available online 14 May 2016 units (ICUs).
Methods: Patients monitoring and targeted environmental screening for P. aeruginosa was performed
Keywords: prospectively over a 27-week period, in absence of recognized outbreak, in two surgical intensive care
Intensive care unit units (SICUs). All isolates were genotyped by Pulsed field gel electrophoresis analysis.
Pseudomonas aeruginosa Findings: 18.9% (46/244) of water fitting samples harbored P. aeruginosa, and active screening samples
Sink from 9.2% (55/595) of hospitalized patients carried with P. aeruginosa. According to genotype results,
PFGE approximately 50% of P. areuginosa colonizations/infections of patients were of exogenous origin. 64.7%
Waterborne infections (11/17) of exogenous sourced cases were associated with contaminated sink traps. There was a significant
correlation between the incidence of exogenous colonization/infection and the prevalence of
P. areuginosa in water fitting in SICU-2 (rs ¼ 0.972; p ¼ 0.014). Furthermore, P. areuginosa from sink trap
possessed a higher level of resistance to multi-antibiotics as opposed to cross-transmission from other
patients.
Conclusion: Water fitting especially sink trap act as an important role in sporadic P. aeruginosa trans-
mission in SICU patients. This report highlights the necessity of identification of potential environmental
reservoirs, such as sinks, for control of infections of environmentally hardy multi-resistant P. areuginosa.

© 2016 Japanese Society of Chemotherapy and The Japanese Association for Infectious Diseases.
Published by Elsevier Ltd. All rights reserved.

1. Introduction Pathways of P. aeruginosa infections in ICUs include endogenous
infection from the patient's own oropharyngeal and intestinal flora,
Pseudomonas aeruginosa is a ubiquitous pathogen prevalent in and exogenous infection from the hands of the personnel, or from
humid environments that is responsible for severe hospital in- neighboring patients and environment. Contaminated water fit-
fections. It is associated with high morbidity and mortality in tings, including drinking water, mineral bottled water, electronic
immunocompromised hosts and in intensive care units (ICUs) pa- faucet, water outlet, sink and waste-water system, have been re-
tients [1]. Its remarkable ability is to adapt to adverse condition ported as the sources of many outbreaks of P. aeruginosa infection in
with the presence of antiseptic substances and limited nutrition ICUs [4e9]. It is found that only seven publications provided
(e.g. water) [2,3]. plausible evidence of a link between a water fittings acting as a
reservoir for P. aeruginosa and infection/colonization in patient [10].
* Corresponding author. Department of Infection Control, Zhongshan Hospital of Most of the literature is descriptive and based on outbreak reports.
Fudan University, 180 Fenglin Road, Shanghai, China. Tel.: þ86 021 64041990 2307; Furthermore, several studies suggest that exogenous transmission
fax: þ86 021 64038770. is uncommon during non-outbreak period [11e13]. Water fitting
seems to play a smaller role in non-epidemic situations than
E-mail address: [email protected] (B. Hu).

http://dx.doi.org/10.1016/j.jiac.2016.03.016
1341-321X/© 2016 Japanese Society of Chemotherapy and The Japanese Association for Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Downloaded for Anonymous User (n/a) at Region Hovedstad from ClinicalKey.com by Elsevier on May 17, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

Z. Zhou et al. / J Infect Chemother 22 (2016) 450e455 451

expected by many operational hospital hygiene teams [14]. 2.2. Study design
Whereas, other studies have shown it play a key role as source of
colonizations/infections using molecular typing [15,16]. Knowledge This study has been approved by the ethics committee. The
of the accurately sources and patterns of P. aeruginosa coloniza- approval number is 2009-64. During the study, patients hospital-
tions/infections are crucial for designing optimal infection control ized in SICUs were screened for the presence of P. aeruginosa on
and prevention strategies. Thus, we prospectively monitor patients admission and weekly thereafter by rectal or respiratory swab.
(clinical samples and active screening samples), water fittings (tap Patients stayed in SICUs less than 72 h were excluded. In case of any
water, tap outlet, and sink trap), hands of healthcare workers suspicious infections, diagnostic specimens were collected from all
(HCWs), and environmental surfaces for contamination with possible infected sites. We monitored the environmental surfaces
P. aeruginosa in two adult surgical intensive care units (SICUs) to and hands of HCWs only if P. aeruginosa were isolated from diag-
elucidate the sources of sporadic P. aeruginosa colonizations/ nostic samples. Hands of HCWs were sampled only from nursing
infections. staff working directly with patients. Seven locations were selected
as environmental surface sampling sites, including the surface of
The importance of tap water as the source of spread of counter, bed rail, bed control, instrumental panel, infusion pump,
P. aeruginosa has been accepted by many specialists [17,18]. The medical chart, and stethoscope. At 4-weekly interval, samples from
clinical impact of sink traps as reservoirs of P. aeruginosa has not yet all cold tap water, tap inner surface, and sink trap were obtained.
been fully explored. Some studies have reported outbreaks of
2.3. Microbiological analysis
multidrug-resistant bacteria such as extended-spectrum b-lacta-
One hundred milliliters of cold tap water was collected into
mase-producing Klebsiella pneumoniae linked contaminated sink sterile bottle immediately after opening the tap. Samples were
traps [19,20]. Furthermore, sink traps can act as cradles to the subsequently passed through sterile filter membranes (pore size:
emergence of bacteria armed with abilities to resist multiple anti-
biotics [21]. Therefore, an in-depth epidemiological study is also 0.45 mm; Millipore, USA). The membranes were placed on cetrimide
necessary to investigate whether sink traps play an important role
in P. aeruginosa transmission in ICU. agar plates (Oxiod, UK) at 37 C and examined for growth of col-
onies after 24 h and 48 h. The swab samples, isolated from tap inner
2. Materials and methods surface, sink trap, active screening samples, HCWs' hand samples
and environmental surface, were directly inoculated on cetrimide
2.1. Setting agar plates. P. aeruginosa was identified by routine methods such as
growth characteristics and biochemical fermentation. Sensitivity
This study was performed in two adult surgical ICUs of a 1600- testing was performed using disk diffusion methodology. Antibi-
bed university-affiliated tertiary care hospital during a 27-week otics tested included carbapenems (imipenem and/or meropenem),
period, from March 2011 to October 2011. No recognized out- cefazidime, cefepime, amikacin, piperacillin-tazobactam, piper-
breaks were observed during the study. The SICUs had two distinct acillin, cefoperazone-sulbactan, levofloxacin and aztreonam.
water distribution networks and waste water systems. Shown in
Fig. 1, there were 29 beds (SICU-1) and 10 beds (SICU-2) in an open 2.4. Molecular typing
room harboring 4 taps and 4 sinks. The other water sites were
located in the treatment room and waste room. All environmental and clinical isolates were genotyped by
Pulsed field gel electrophoresis (PFGE) analysis of SpeI restricted

Fig. 1. The layout of two adult surgical ICUs. The number in circles was the serial number of water fittings (tap and sink).

Downloaded for Anonymous User (n/a) at Region Hovedstad from ClinicalKey.com by Elsevier on May 17, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

452 Z. Zhou et al. / J Infect Chemother 22 (2016) 450e455

genomic DNA [22]. PFGE was performed using a CHEF-DR III fluid, drainage fluid and blood. Moreover, active screening samples
apparatus (Bio-Rad, USA) setting at 6.0 v/cm, with a linear ramping from 9.2% (55/595) of hospitalized patients harbored P. aeruginosa
from 2.19 s to 30.82 s for 18 h. The PFGE-SpeI profiles were con- (Table 1). In total, 93 strains collected from patients were analyzed:
verted into a matrix of discrete characters and analyzed by a 58 screening samples from 55 patients and 35 diagnostic samples
parsimony method. Strains showing identical SpeI PFGE profiles from 32 patients.
were defined as being individuals of a same clone.
3.2. Environmental investigation
2.5. Epidemiological definitions and classification of cases
Totally, 456 environmental samples were collected from both
A case was defined as the isolation of a given genotype of SICUs (244 to water fittings, 189 to environmental surfaces and 23
P. aeruginosa from a diagnostic specimen obtained during a pa- to hands of HCWs) (Table 2). Samples positive for P. aeruginosa were
tient's stay in a SICU [23]. Thus the isolates from an individual 18.6% (18/97) of the tap water, 22.1% (21/95) of the sink traps and
patient having two different PFGE patterns (genotypes) yielded two 13.5% (7/52) of the tap outlets, respectively. The positive site rate of
distinct cases. Cases were classified into four categories according sink traps (92.3%, 12/13) was higher than that of tap water (53.8%, 7/
to molecular typing data: (A) identical to isolates from water fit- 13). Almost half of taps sites (tap water and tap outlets) were
tings; (B) identical to at least one other patient sample and/or negative to P. aeruginosa (tap1, tap2 and tap3 in SICU-1; tap1, tap2
environmental surface sample, but not found in water fitting; (C) and tap7 in SICU-2), while all sink traps were detected with
identical to rectum and/or throat sample of the same patient; and P. aeruginosa except sink 1 in SICU-2. No strains were isolated from
(D) unique genotype. Cases in categories A and B were considered the hands of HCWs. Only 5 strains were detected from two bed
as possibly exogenous transmission, and cases in category C and D rails, one counter, one instrument panel, and one medical chart. In
as possibly endogenous transmission. total, 53 strains from environment were collected and analyzed.
Two samples taken from sink traps contained two strains.
2.6. Statistical analysis
3.3. Sink trap as a continuous exogenous source of P. aeruginosa
The data were analyzed with SPSS version 16.0. Percentage of transmission
P. aeruginosa carriage and water fitting contamination of two SICUs
PFGE typing distributed the 146 isolates into 89 different ge-
were compared using the c2 test, a p value of less than 0.05 was notypes: 23 genotypes recovered from 35 patient strains (14 to
SICU-1 and 9 to SICU-2), 53 genotypes from 58 active screening
considered statistically significant. The correlation coefficient of isolates, and 24 from 53 hospital environmental isolates. The
incidence of exogenous colonization/infection and the percentage strains from active screening samples of patients showed a rela-
of water fitting contamination with P. aeruginosa was calculated by tively greater genetic diversity than those from environment and
using two-tailed spearman's analysis. clinical patients.

3. Results According to epidemiological and molecular data, thirty-four
cases were divided into four categories (Table 1). One patient was
3.1. Study participants found to harbor three different strains (genotypes). Our data
showed that 17.6% (6/34), 32.4% (11/34), and 50.0% (17/34) of
During the 27-week period, from 595 patients stayed in SICUs colonization/infection with P. aeruginosa were original from other
more than 72 h (467 to SICU-1 and 128 to SICU-2) with a total of patients, water fitting, and endogenous flora, respectively. In SICU-
5674 days of hospitalization (4401 to SICU-1 and 1273 to SICU-2), 2, a significant correlation between numbers of exogenous cases
thirty-two patients (21 to SICU-1 and 11 to SICU-2) presented at per 1000 admissions and the percentage of water fitting contami-
least one clinical diagnosis sample positive for P. aeruginosa. The nation at two-monthly intervals was observed (rs ¼ 0.972;
patients with P. aeruginosa at admission were not counted in. The p ¼ 0.014). It is apparent that water fitting was an important part of
overall incidence of colonization/infection was 5.6 per 1000 exogenous transmission. Furthermore, 64.7% (11/17) of exogenous
patient-days of hospitalization (4.8 to SICU-1 and 8.6 to SICU-2). sourced cases were associated with contaminated sink traps.
Most of patients were suffered from pneumonia (n ¼ 8) or surgi- Whereas, no strains (genotypes) recovered from tap water were
cal wound infection (n ¼ 9). Thirty strains were isolated from identical to that from patients. Fig. 2 showed the spatio-temporal
sputum samples. The others were isolated from secretion, pleural

Table 1
Epidemiological data of patients with Pseudomonas aeruginosa stayed in surgical intensive care units (SICUs) more than 72 h (2011.3e2011.10).

Epidemiological data SICU-1 SICU-2 Total

Number of admissions stayed in SICUs more than 72 h 467 128 595
Patient-days of hospitalization 4401 1273 5674
Number of patients colonization/infection with P. aeruginosa 21 11 32
Number of patients colonization/infection with P. aeruginosa per 1000 admissions 45.0 85.9 53.8
Number of patients colonization/infection with P. aeruginosa per 1000 patient-days of hospitalization 4.8 8.6 5.6
Percentage of hospitalized patients carried with P. aeruginosa (%) 7.7 14.8 9.2
Percentage of water fitting contamination with P. aeruginosa (%) 19.6 17.0 18.4
Categories of cases (number of cases per 1000 admissions)a
Category A 23.6 0 18.5
Category B 4.3 31.3 11.8
Category C 4.3 15.6 6.7
Category D 17.1 39.1 21.8

a Category A, identical to water fitting; Category B, identical to at least one other patient and/or environmental surface sample, not found in water fitting; Category C,
identical to active screening sample of the same patient; Category D, unique genotypes.

Downloaded for Anonymous User (n/a) at Region Hovedstad from ClinicalKey.com by Elsevier on May 17, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

Z. Zhou et al. / J Infect Chemother 22 (2016) 450e455 453

Table 2 Positive rates (%)
Positive rates of Pseudomonas aeruginosa isolated from environment and patients (2011.3e2011.10). 11.2
18.6
Source of samples Number of samples Number of positive samples 13.5
22.1
Environmental samples 456 51 0
Tap water 97 18 2.6
Tap outlets 52 9.2
Sink traps 95 7 N/A
Hands of staff 23 21
Environmental surfaces 189
595 0
Screening samples N/A 5
Diagnostic samples 55
32

N/A: Not Applicable.

Treatment The number of Patient beds 5 9 A3 3
9 3
room 5 E 3
5 56
6 9 8
5
8B 5 D3
F 55
5 5

9C 9 5

9

11

16

24

25 G5

26 H 2 5

I 53

28 J 3 K 3

5 M5

30 L 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Weeks

Fig. 2. Illustration of the spatio-temporal dynamics among patients with exogenous sourced P. aeruginosa in SICU-1. Number in circles: genotype from patients; Number in
rhombuses: genotype from active screening samples (rectum); Number in starts: genotype from sink traps; Number in triangles: genotype from environmental surfaces; Horizontal
line: patient's individual stay.

dynamics among patients with exogenous sourced P. aeruginosa in Especially, patient isolates associated with sink trap showed more
SICU-1. Thirteen patients with 6 different genotypes were included. resistant to antibiotics than patient-to-patient transmission strains
Three genotypes recovered from patients were identical to that (the percentage of carbapenem-resistant P. aeruginosa: 81.8% vs.
from sink trap samples (genotype 3, genotype 5 and genotype 9). 16.7%).
Among these cases, cross contaminations between the patients J
and A, K, D can be excluded since no overlapping hospitalization 4. Discussion
period occurred. It seemed that some vehicles such as sink trap can
be contaminated and consecutively contribute to spread of P. aeruginosa is one of the most important aetiological agents of
P. aeruginosa. Four strains found on the environmental surface had nosocomial infections. Several studies have investigated the sour-
the same genotype with strain from patients. ces of colonizations/infections of ICU patients by P. aeruginosa, and
showed that cross-contamination between patients and water
Differences in antibiotic resistance levels were observed within fitting could explain several nosocomial outbreaks [4e9]. Strains
the isolates from diagnostic, screening, and environmental sam- involved in these outbreaks are considered as transient colonizers
ples. The percentage of carbapenem-resistant P. aeruginosa of of the hospital units. Rarely, a P. aeruginosa population of a unit is
diagnostic samples (45.7%, 16/35) was higher than that of screening analyzed outside the scope of an epidemic outbreak. Here, we
samples (3.4%, 2/58) and environmental samples (15.1%, 8/53). All initiated a long-term clinical intensive study regarding sources and
strains from water-related samples were sensitive to all antibiotics.
Some sink-related strains showed the multi-resistance profiles.

Downloaded for Anonymous User (n/a) at Region Hovedstad from ClinicalKey.com by Elsevier on May 17, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

454 Z. Zhou et al. / J Infect Chemother 22 (2016) 450e455

patterns of P. aeruginosa colonizations/infections in two surgical reduce by almost 50% if these measures have been done. Finally,
ICUs patients. routine admission screening and targeted environmental moni-
toring in specific wards seems to be efficiently, and the rate of water
P. aeruginosa outbreaks sourced from sinks were observed as fitting contamination can be used as an index to place more
long ago as the early 1980s [24]. Early reports implicated the sinks emphasis on infection control procedures.
may be reservoirs for large number of highly resistant P. aeruginosa
but are rarely the source of organisms colonizing patients in ICU Conflicts of interest statement
[25]. Since the 2000s, tap water contaminated with P. aeruginosa
had been recognized as a major and continuous source for trans- None declared.
mission. Recently, several studies reported the outbreaks of MDRP
in hospitals, which highlight the potential of hospital waste system Funding sources
to act as reservoir of MDRP [9,19,26]. To our knowledge, the present
study is the first to investigate the sink trap as an important source Hu Bijie received grant support from Special Fund for Health-
of colonization/infection of P. aeruginosa over a long period of time. scientific Research in the Public Interest: Research & application
for the prevention & control of nosocomial infections coursed by
According our data, water fittings were found to contain a long- multi-drug resistant bacteria (No. 201002021).
standing population of P. aeruginosa, with strains were detected in
18.6% of tap water samples and 22.1% of sink trap samples. The Acknowledgments
positive site rate of sink traps (92.3%, 12/13) was higher than that of
tap water (53.8%, 7/13). It suggested that the extent of P. aeruginosa We gratefully thank the nurses and medical staff of two SICUs,
colonization was more seriously for sink trap than for tap water. who cooperated fully with this study. And we also thank Dr. Liang
Moreover, approximately 50% of P. areuginosa colonizations/in- Qin and Dr. Lili Tao for their valuable suggestions and the review of
fections of ICU patients were of exogenous origin, the data in line manuscript.
with the results of Agodi et al. [27]. 64.7% (11/17) of exogenous
sourced cases were associated with contaminated sink traps. References
Whereas, no strains (genotypes) recovered from tap water were
identical to that from patients. It also showed the positive corre- [1] Rosenthal VD, Hu BJ, Maki DG, Mehta Y, Apisarnthanarak A, et al. International
lation between the incidence of exogenous colonization/infection nosocomial infection control consortium (INICC) report, data summary of 36
and the prevalence of P. areuginosa in water fitting in SICU-2 countries, for 2004-2009. Am J Infect Control 2012;40:396e407.
(rs ¼ 0.972; p ¼ 0.014). Shown as the spatio-temporal dynamics
among patients with exogenous sourced P. aeruginosa in SICU-1, the [2] Anaissie EJ, Penzak SR, Dignani MC. The hospital water supply as a source of
pattern of infection was small clusters with long infection-free in- nosocomial infections: a plea for action. Arch Intern Med 2002;162:1483e92.
tervals, suggesting that person-to-person spread did not play a
major role in the infection. Together these findings implicated that [3] Fujitani S, Sun HY, Yu VL, Weingarten JA. Pneumonia due to Pseudomonas
water fitting especially sink trap can be an important and contin- aeruginosa: part I: epidemiology, clinical diagnosis, and source. Chest
uous source of exogenous colonizations/infections. P. aeruginosa 2011;139:909e19.
transmission from sink trap to patients may be initiated by
splashing back the organisms residing down the trap onto staff [4] Wong V, Levi K, Baddal B, Turton J, Boswell TC. Spread of Pseudomonas flu-
hands or patient areas when staffs wash hands or something is orescens due to contaminated drinking water in a bone marrow transplant
poured into the basin. unit. J Clin Microbiol 2011;49:2093e6.

Interestingly, clinical isolates associated with the sink trap (as [5] Naze F, Jouen E, Randriamahazo RT, Simac C, Laurent P, Bleriot A, et al.
opposed to cross transmissions from other patients) possessed a Pseudomonas aeruginosa outbreak linked to mineral water bottles in a
higher level of resistance to multi-antibiotics. Several factors neonatal intensive care unit: fast typing by use of high-resolution melting
probably enhanced the development of resistance. More bacteria analysis of a variable-number tandem-repeat locus. J Clin Microbiol 2010;48:
and more antibiotics are flushed down the sink due to the nature of 3146e52.
hospitals caring for patients that are ill and treated with antibiotics.
Thus, biofilm in building drains provided the cradle to the transfer [6] Yapicioglu H, Gokmen TG, Yildizdas D, Koksal F, Ozlu F, Lalw-Cekinmez E, et al.
of antibiotic resistance between different bacteria [28]. Pseudomonas aeruginosa infections due to electronic faucets in a neonatal
intensive care unit. J Paediatr Child Health 2012;48:430e4.
Several limitations of our study must be mentioned. Firstly, Only
34 cases were included in this observational study. A caseecontrol [7] Aumeran C, Paillard C, Robin F, Kanold J, Baud O, Bonnet R, et al. Pseudomonas
study or cohort study with abundance of cases should be per- aeruginosa and Pseudomonas putida outbreak associated with contaminated
formed in the future. Secondly, the limitation of sampling tech- water outlets in an oncohaematology paediatric unit. J Hosp Infect 2007;65:
nique and microbiological analysis of water fitting samples may 47e53.
underestimate the role of water fitting. Some studies typed up to
4e10 different colonies from each culture [29]. Just one to two [8] Hota S, Hirji Z, Stockton K, Lemieux C, Dedier H, Wolfaardt G, et al. Outbreak of
colonies from a culture were picked and genotyped in our study. multidrug-resistant Pseudomonas aeruginosa colonization and infection sec-
Therefore, more correlations between sink trap and patient isolates ondary to imperfect intensive care unit room design. Infect Control Hosp
might have been detected. At last, we did not analyze various Epidemiol 2009;30:25e33.
sources from equipment and might have underestimated the
number of exogenous source. [9] Breathnach AS, Cubbon MD, Karunaharan RN, Pope CF, Planche TD. Multidrug-
resistant Pseudomonas aeruginosa outbreaks in two hospitals: association with
In conclusion, sink trap can serve as a reservoir for waterborne contaminated hospital waste-water systems. J Hosp Infect 2012;82:19e24.
pathogens and act as the source of P. aeruginosa transmission in
ICUs patients under non-outbreak situations. If we find [10] Loveday HP, Wilson JA, Kerr K, Pitchers R, Walker JT, Browne J. Association
P. aeruginosa in the sink, we may find colonization/infection in between healthcare water systems and Pseudomonas aeruginosa infections: a
patients. Therefore, appropriate sinks disinfection or maintenance rapid systematic review. J Hosp Infect 2014;86:7e15.
should be adapted actively for preventing the P. aeruginosa infec-
tion. The incidence of P. aeruginosa colonization/infection would [11] Zoltanski J, Dul M, O'Riordan MA, Blumer J, Toltzis P. Low frequency of
endemic patient-to-patient transmission of antibiotic-resistant gram-negative
bacilli in a pediatric intensive care unit. Infect Control Hosp Epidemiol
2011;32:915e7.

[12] Bertrand X, Thouverez M, Talon D, Boillot A, Capellier G, Floriot C, et al.
Endemicity, molecular diversity and colonisation routes of Pseudomonas
aeruginosa in intensive care units. Intensive Care Med 2001;27:1263e8.

[13] Berthelot P, Grattard F, Mahul P, Pain P, Jospe R, Venet C, et al. Prospective
study of nosocomial colonization and infection due to Pseudomonas aerugi-
nosa in mechanically ventilated patients. Intensive Care Med 2001;27:
503e12.

[14] Cholley P, Thouverez M, Floret N, Bertrand X, Talon D. The role of water fit-
tings in intensive care rooms as reservoirs for the colonization of patients
with Pseudomonas aeruginosa. Intensive Care Med 2008;34:1428e33.

[15] Reuter S, Sigge A, Wiedeck H, Trautmann M. Analysis of transmission path-
ways of Pseudomonas aeruginosa between patients and tap water outlets. Crit
Care Med 2002;30:2222e8.

Downloaded for Anonymous User (n/a) at Region Hovedstad from ClinicalKey.com by Elsevier on May 17, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

Z. Zhou et al. / J Infect Chemother 22 (2016) 450e455 455

[16] Rogues AM, Boulestreau H, Lashe ras A, Boyer A, Gruson D, Merle C, et al. [23] Cuttelod M, Senn L, Terletskiy V, Nahimana I, Petignat C, Eggimann P, et al.
Contribution of tap water to patient colonisation with Pseudomonas aerugi- Molecular epidemiology of Pseudomonas aeruginosa in intensive care units
nosa in a medical intensive care unit. J Hosp Infect 2007;67:72e8. over a 10-year period (1998-2007). Clin Microbiol Infect 2011;17:57e62.

[17] Trautmann M, Michalsky T, Wiedeck H, Radosavl V, Ruhnke M. Tap water [24] Falkiner FR, Jacoby GA, Keane CT, Mccann SR. Amikacin, gentamicin and
colonization with Pseudomonas aeruginosa in a surgical intensive care unit tobramycin resistant Pseudomonas aeruginosa in a leukaemic ward: epide-
(ICU) and relation to Pseudomonas infections of ICU patients. Infect Control miology and genetic studies. J Hosp Infect 1982;3:253e61.
Hosp Epidemiol 2001;22:49e52.
[25] Levin MH, Olson B, Nathan C, Kabins SA, Weinstein RA. Pseudomonas in the
[18] Trautmann M, Bauer C, Schumann C, Hahn P, Hoher M, Haller M, et al. sinks in an intensive care unit: relation to patients. J Clin Pathol 1984;37:
Common RAPD pattern of Pseudomonas aeruginosa from patients and tap 424e7.
water in a medical intensive care unit. Int J Hyg Environ Health 2006;209:
325e31. [26] Kotsanas D, Wijesooriya WR, Korman TM, Gillespie EE, Wright L, Snook K,
et al. “Down the drain”: carbapenem-resistant bacteria in intensive care unit
[19] Roux D, Aubier B, Cochard H, Quentin R, Mee Marquet N. Contaminated sinks patients and hand-washing sinks. Med J Aust 2013;198:267e9.
in intensive care units: an underestimated source of extended-spectrum beta-
lactamase-producing Enterobacteriaceae in the patient environment. J Hosp [27] Agodi A, Barchitta M, Cipresso R, Giaquinta L, Romeo MA, Denaro C. Pseu-
Infect 2013;85:106e11. domonas aeruginosa carriage, colonization, and infection in ICU patients.
Intensive Care Med 2007;33:1155e61.
[20] Vergara Lopez S, Dominguez MC, Conejo MC, Pascual A, Rodriguez Bano J.
Wastewater drainage system as an occult reservoir in a protracted clonal [28] Savage VJ, Chopra I, O'Neill AJ. Staphylococcus aureus biofilms promote hori-
outbreak due to metallo-beta-lactamase-producing Klebsiella oxytoca. Clin zontal transfer of antibiotic resistance. Antimicrob Agents Chemother
Microbiol Infect 2013;19:E490e8. 2013;57:1968e70.

[21] Blom K. Drainage systems, an occluded source of sanitation related outbreaks. [29] Valle s J, Mariscal D, Corte s P, Coll P, Villagra A, Diaz E, et al. Patterns of
Arch Public Health 2015;73:8. colonization by Pseudomonas aeruginosa in intubated patients: a 3-year
prospective study of 1,607 isolates using pulsed-field gel electrophoresis with
[22] Ro€mling U, Wingender J, Müller H, Tümmler B. A major Pseudomonas aeru- implications for prevention of ventilator-associated pneumonia. Intensive
ginosa clone common to patients and aquatic habitats. Appl Environ Microbiol Care Med 2004;30:1768e75.
1994;60:1734e8.

Downloaded for Anonymous User (n/a) at Region Hovedstad from ClinicalKey.com by Elsevier on May 17, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

PUBLIC AND ENVIRONMENTAL
HEALTH MICROBIOLOGY

crossm

Spread from the Sink to the Patient: In
Situ Study Using Green Fluorescent
Protein (GFP)-Expressing Escherichia coli
To Model Bacterial Dispersion from
Hand-Washing Sink-Trap Reservoirs

Shireen Kotay,a Weidong Chai,a William Guilford,b Katie Barry,a Amy J. Mathersa,c Received 8 December 2016 Accepted 7
February 2017
Division of Infectious Diseases and International Health, Department of Medicine, University of Virginia Health Accepted manuscript posted online 24
System, Charlottesville, Virginia, USAa; Department of Biomedical Engineering, University of Virginia, February 2017
Charlottesville, Virginia, USAb; Clinical Microbiology, Department of Pathology, University of Virginia Health Citation Kotay S, Chai W, Guilford W, Barry K,
System, Charlottesville, Virginia, USAc Mathers AJ. 2017. Spread from the sink to the
patient: in situ study using green fluorescent
ABSTRACT There have been an increasing number of reports implicating Gamma- protein (GFP)-expressing Escherichia coli to
proteobacteria as often carrying genes of drug resistance from colonized sink traps model bacterial dispersion from hand-washing
to vulnerable hospitalized patients. However, the mechanism of transmission from sink-trap reservoirs. Appl Environ Microbiol
the wastewater of the sink P-trap to patients remains poorly understood. Herein we 83:e03327-16. https://doi.org/10.1128/
report the use of a designated hand-washing sink lab gallery to model dispersion of AEM.03327-16.
green fluorescent protein (GFP)-expressing Escherichia coli from sink wastewater to Editor Andrew J. McBain, University of
the surrounding environment. We found no dispersion of GFP-expressing E. coli di- Manchester
rectly from the P-trap to the sink basin or surrounding countertop with coincident Copyright © 2017 Kotay et al. This is an open-
water flow from a faucet. However, when the GFP-expressing E. coli cells were al- access article distributed under the terms of
lowed to mature in the P-trap under conditions similar to those in a hospital envi- the Creative Commons Attribution 4.0
ronment, a GFP-expressing E. coli-containing putative biofilm extended upward over International license.
7 days to reach the strainer. This subsequently resulted in droplet dispersion to the Address correspondence to Amy J. Mathers,
surrounding areas (Ͻ30 in.) during faucet operation. We also demonstrated that [email protected].
P-trap colonization could occur by retrograde transmission along a common pipe. S.K. and W.C. contributed equally to this article.
We postulate that the organisms mobilize up to the strainer from the P-trap, result-
ing in droplet dispersion rather than dispersion directly from the P-trap. This work aem.asm.org 1
helps to further define the mode of transmission of bacteria from a P-trap reservoir
to a vulnerable hospitalized patient.

IMPORTANCE Many recent reports demonstrate that sink drain pipes become
colonized with highly consequential multidrug-resistant bacteria, which then re-
sults in hospital-acquired infections. However, the mechanism of dispersal of
bacteria from the sink to patients has not been fully elucidated. Through estab-
lishment of a unique sink gallery, this work found that a staged mode of trans-
mission involving biofilm growth from the lower pipe to the sink strainer and
subsequent splatter to the bowl and surrounding area occurs rather than splat-
ter directly from the water in the lower pipe. We have also demonstrated that
bacterial transmission can occur via connections in wastewater plumbing to
neighboring sinks. This work helps to more clearly define the mechanism and
risk of transmission from a wastewater source to hospitalized patients in a world
with increasingly antibiotic-resistant bacteria that can thrive in wastewater envi-
ronments and cause infections in vulnerable patients.

KEYWORDS antibiotic resistance, biofilms, dispersion, hand washing, hospital
infections, premise plumbing, sink, transmission, wastewater

April 2017 Volume 83 Issue 8 e03327-16 Applied and Environmental Microbiology

Kotay et al. Applied and Environmental Microbiology
aem.asm.org 2
Despite early reports (1–5), the premise that hand-wash sink traps can act as
reservoirs of bacteria that cause nosocomial infections has been frequently over-
looked. There has recently been an alarming increase in sink-related outbreaks world-
wide, with many reports establishing an observational link (6–13). A sink often operates
as an open conduit to wastewater in a patient care area that is often in the same room
as the patient.

Health care establishments often invest in desperate interventions to deal with
nosocomial outbreaks. The preferred method for addressing most of the environment-
related transmission is to employ enhanced cleaning using chemical and physical
agents (14, 15). Unfortunately, routine approaches are inefficient in completely elimi-
nating drug-resistant Gammaproteobacteria in an inaccessible microbiologically active
area such as a sink trap (6, 16–20). The wet, humid, and relatively protected environ-
ment in a sink trap favors the formation of rich stable microbial communities (16, 21,
22). These communities will be exposed to liquids and waste that are discarded in a sink
and may include antimicrobials, discarded beverages, soap, presumably pathogenic
bacteria from health care workers’ hands, and other items. In short, sink traps could
serve as a breeding ground for opportunistic and highly antimicrobial-resistant bacteria
that cannot be easily cleaned or removed (23–28).

There are many reports of a genetic association between pathogens found in sink
traps and those found in patients (29, 30). However, surprisingly little work has been
done to understand the microscale transmission dynamics. It was previously demon-
strated using a suspension of fluorescent particles (Glo Germ; Glo Germ Co., Moab, UT)
that material injected into the P-trap gets dispersed around a hand-washing sink (6).
This result however has not been replicated hitherto in the follow-up studies. Disper-
sion has never been investigated with living organisms. Ultimately, many details remain
unaddressed surrounding the spread of Enterobacteriaceae in sink-trap wastewater
systems: (i) can organisms grow retrograde from the P-trap water to the sink strainer,
(ii) can organisms spread from one sink to another along the internal surfaces of pipes
with shared drainage systems, and (iii) which portion of a colonized drain pipe results
in dispersion into the sink bowl during a hand-washing event? We aim to better
understand the dispersion dynamics of Gammaproteobacteria living in the wastewater
of a sink strainer and P-trap into an area where patients and health care workers could
be exposed. To study this dynamic, we used a surrogate organism that could be easily
tracked while remaining in the Enterobacteriaceae family, where some of the most
concerning threats in antimicrobial resistance are developing (30).

RESULTS
Growth and colonization of GFP-expressing E. coli in the P-trap. In the first 14

days following the installation of the P-trap with established green fluorescent protein
(GFP)-expressing Escherichia coli and just water running from the faucet, GFP-
expressing E. coli was not detected in the tailpipe beyond 1.5 in. above the liquid level
in the P-trap. GFP-expressing E. coli, however, was found to be viable in the P-trap
without any nutrients added. A nutrient regimen was then instituted to understand the
influence of nutrients on mobility and upward growth. The addition of tryptic soy broth
(TSB) promoted GFP-expressing E. coli growth as early as day 1, with growth observed
in the tailpipe 2 in. above the liquid surface in the P-trap (Table 1). On day 7, the strainer
(ϳ8 in. above the liquid in the P-trap) was found to be colonized with GFP-expressing
E. coli. This translates to an average growth rate of 1 in./day along the length of the
tailpipe with the addition of nutrients and without faucet operation. GFP-expressing E.
coli was not detected in the faucet water.

Sink-to-sink transmission of bacteria. In these experiments, a flanking sink (sink 5)
was the only P-trap inoculated with GFP-expressing E. coli and therefore was the sole
source for transmission to the connected sinks. Starting with a lower inoculum con-
centration (103 CFU/ml) in sink 5, on day 7, GFP-expressing E. coli was detected in the
sink 2 and sink 3 P-traps (Fig. 1a). With inoculum concentrations of 106 CFU/ml and
Ͼ1010 CFU/ml in sink 5, all of the sink P-traps in the sink gallery with the exception of

April 2017 Volume 83 Issue 8 e03327-16

GFP-Expressing E. coli Sink-Trap Dispersion Applied and Environmental Microbiology
aem.asm.org 3
TABLE 1 Growth in the tailpipe connected to the P-trap colonized with GFP-expressing
E. coli biofilm

Presence of GFP-expressing E. coli on daya:

Sampling area 01234567

Strainer 8 in. above P-trap water ؊ ؊ ؊ ؊ ؊ ؊ ؊ ؉

Tailpipe ؊؊؊؊؉؉؉؉
6 in. above P-trap water ؊؊؊؉؉؉؉؉
4 in. above P-trap water ؊؉؉؉؉؉؉؉
2 in. above P-trap water

P-trap ؉؉؉؉؉؉؉؉

a“Ϫ” and “ϩ” denote the absence and presence of GFP-expressing E. coli, respectively.

sink 1 were found to be colonized with GFP-expressing E. coli after 7 days (Fig. 1b and
c). Faucet water and aerators tested negative for GFP-expressing E. coli. Irrespective of
the starting inoculum concentration, on day 7 the highest level of colonization was
recorded in the sink 3 P-trap. After day 7, when the nutrient regimen (described
previously) was followed for an additional 7 days in each of the sinks in the sink gallery
with an inoculum concentration of Ͼ1010 CFU/ml, GFP-expressing E. coli was detected
in the strainers of sinks 2 and 3 on day 14. This finding validated the upward growth
and growth rate in the tailpipe when nutrients were added. Nonfluorescent colonies
were occasionally observed in the P-trap water samples collected from the sinks, which
were subsequently identified to be Pseudomonas sp. or Stenotrophomonas maltophilia,
and fluorescent colonies were confirmed to be E. coli.

Dispersion of microspheres from sinks. In the first dispersion experiment, when
fluorescent microspheres were inoculated into the offset drain tailpiece only 4 in. below
the strainer, no microspheres were detected on the polyester sheets placed on the
counter space.

However, when the sink bowl was coated with the microspheres, polyester sheets
overlaid on the counter space captured the dispersed microspheres caused by the
faucet operation. Dispersion was observed on almost all zones of the sink counter space
(Fig. 2). Relatively higher levels of dispersion were observed along the major and minor
axes of the elliptical sink bowl (zones 2, 5, 6, 9, 11, and 12). Anterior corners of the sink
counter space (zones 4 and 7), which were most distant from the impact of water in the
sink bowl, received the lowest dispersion.

Dispersion of GFP-expressing E. coli from sinks. Initially the P-trap alone was
inoculated with GFP-expressing E. coli and carefully installed, keeping the tailpipe and
strainer free of GFP-expressing E. coli before operating the faucets. No fluorescent CFU
were observed on the plates placed on the counter or attached to the bowl surface
after faucet operation. Similarly, no fluorescent CFU were detected when GFP-
expressing E. coli was inoculated into the offset drain tailpiece only 4 in. below the
strainer. Interestingly, when there was conspicuous water backup over the strainer as
a result of a higher water flow rate from the faucet than the drainage rate from the
P-trap, dispersal was detected on the plates attached to the bowl surface.

The dispersion pattern recorded when the sink bowl was coated with GFP-
expressing E. coli was comparable to the pattern recorded when the sink bowl was
coated with fluorescent microspheres (Fig. 2). Dispersion was significantly higher along
the axes (zones 6, 9, 11, and 12) and lower at the corners of the sink counter space
(zones 4, 7, and 10).

In contrast, dispersion of GFP-expressing E. coli caused by the faucet operation was
much more extensive when the strainer was allowed to be colonized with GFP-
expressing E. coli prior to the dispersion experiment. In addition to the sink counter
space, we measured dispersion to the sink bowl, faucet, faucet handles, splatter shields,
and the extended counter surface. Dispersion of GFP-expressing E. coli was highest on
the plates attached to the sink bowl (Fig. 3b). Further, dispersion was greater along the

April 2017 Volume 83 Issue 8 e03327-16

Kotay et al. Applied and Environmental Microbiology

FIG 1 GFP-expressing E. coli detected in the P-traps attached to each of the sinks on day 0 (black bars) aem.asm.org 4
and day 7 (gray bars) using (a) 103, (b) 106, and (c) 1010 CFU/ml as the starting inoculum concentrations
in sink 5.

minor axis of the sink bowl (Fig. 3b, zones B3, B4, and B10) than along the major axis
of the sink bowl, associated with a shorter distance from the strike point of the faucet
water to the bowl along this axis. The next highest CFU count from the dispersal was
recorded on the counter area near the faucets (Fig. 3a, zones 12 and 11). A similar
pattern of higher dispersion near the faucets and lower dispersion at the corners of the

April 2017 Volume 83 Issue 8 e03327-16

GFP-Expressing E. coli Sink-Trap Dispersion Applied and Environmental Microbiology

FIG 2 Dispersion of microspheres (gray bars) and GFP-expressing E. coli (black bars) on the area aem.asm.org 5
surrounding the sink when the sink bowl was coated. The x axis represents the designated zones of the
sink counter.

counter space (Fig. 3a, zones 4, 7, and 10) was also observed using microspheres.
Dispersion was also recorded in other zones of the counter space, on the Plexiglas
splatter shields, faucets, faucet handles, and extended surface (Fig. 3c). There were no
GFP-expressing E. coli CFU recorded on plates placed beyond 30 in. from the strainer,
demarcating the range of dispersion under these experimental conditions.

Table 2 gives a summary of the total distribution loads recorded using fluorescent
microspheres and GFP-expressing E. coli across each experiment. The loads of disper-
sion on the sink counter were comparable when the sink bowl was coated with
microspheres or GFP-expressing E. coli before the faucet operation. Although the
dispersion load on the sink counter was lower when the sink strainer was colonized, it
is interesting to note that the sink bowl received the highest dispersion.

DISCUSSION
To mimic dispersion in a hospital setting, we first investigated whether GFP-

expressing E. coli would establish consistent colonization in a sink trap as many other
Gammaproteobacteria implicated in nosocomial outbreaks have done (6, 28). Many
recent reports demonstrate that P-traps become colonized with highly consequential
Gammaproteobacteria, which then results in nosocomial transmission (29, 31, 32). The
retained water in a sink P-trap is present to provide a water barrier to prevent
off-gassing of sewer smell, but it may inadvertently provide favorable conditions for
pathogenic and opportunistic antibiotic-resistant microorganisms to survive and de-
velop resilient biofilms (3, 33). However, the mechanism of dispersal of the bacteria in
the P-trap to patients or the surrounding health care area had not been fully elucidated.
We began with the hypothesis that the bacteria originate from the P-trap via droplet
creation when the water from the faucet hits the P-trap water, thus contaminating the
sink bowl and the surrounding area. The finding supporting this theory had been
previously reported using Glo Germ particles (6). However, in the present study with
careful attention to avoid strainer and tail piece contamination, the dispersal directly
from the sink P-trap with either microspheres or GFP-expressing E. coli could not be
reproduced as previously reported (6).

Rather this work demonstrates a different, more staged mode of transmission from
a P-trap reservoir to the sink and surrounding environment. GFP-expressing E. coli in
the P-trap alone was sustained for 14 days but did not grow or mobilize up the tailpipe
to the strainer with just intermittent water exposure. However, when nutrients were

April 2017 Volume 83 Issue 8 e03327-16

Kotay et al. Applied and Environmental Microbiology

FIG 3 Dispersion of GFP-expressing E. coli on the area surrounding the sink when the strainer, tailpipe, aem.asm.org 6
and P-trap were colonized. (a) Sink counter; (b) sink bowl; (c) other surrounding areas. The x axis
represents the designated zones of the sink counter.

April 2017 Volume 83 Issue 8 e03327-16

GFP-Expressing E. coli Sink-Trap Dispersion Applied and Environmental Microbiology

TABLE 2 Comparison of dispersion loads across different experiments

Dispersion load

Dispersion expt Sink counter Sink bowl Faucets and Splatter shields
(>30 in.) faucet handles
Microspheres (microspheres/cm2) NAa NA
Offset drain inoculated with microspheres 0 NA NA NA
Sink bowl coated with microspheres 206 Ϯ 10 NA
0 0
GFP-expressing E. coli (CFU/cm2) 0 NA 0 NA
P-trap inoculated with GFP-expressing E. coli NA NA NA
Offset drain inoculated with GFP-expressing E. coli 0 342 Ϯ 17 NA 3Ϯ1
Sink bowl coated with GFP-expressing E. coli 232 Ϯ 17 17 Ϯ 3
Strainer colonized with GFP-expressing E. coli 171 Ϯ 15

aNA, not applicable.

subsequently added to the system, the organisms rapidly grew up the tailpipe to the aem.asm.org 7
strainer at approximately an inch per day. In a real-world setting, motility of bacteria
inside the tailpipe is restricted to relatively sporadic and brief wetting events in which
swimming is an opportunity to colonize new surfaces. It is assumed that once estab-
lished, the biofilm promotes the upward growth of GFP-expressing E. coli in the tailpipe
at an accelerated rate. The nutrient regimen that promoted colonization in our model
reflects our and others’ observations of items commonly disposed of in hospital sinks
(intravenous fluids, feeding supplements, and leftover beverages) (5, 32).

Transmission of bacteria between sinks via a common pipe was a key finding in this
study as this highlights the concept that premise plumbing may be a more continuous
system with shared microbiology than a single isolated sink. The sink gallery used in
this study provided a unique in situ advantage to investigate sink-to-sink transmission
of bacteria through common drains. The two possible mechanisms for P-trap strainers
becoming colonized are seeding of organisms from above and retrograde spread of
organisms along common pipes in a hospital wastewater infrastructure. Here we
demonstrate that it is possible for GFP-expressing E. coli to contaminate adjacent
P-traps with just time and water given a standard U.S. code piping rise of 0.25 ft/ft.
Sink-to-sink or retrograde transmission may explain the recurrence of pathogen colo-
nization following intervention strategies like disinfection or replacement of plumbing
(23). Sink 3 was lowest on the slope in the drain line (see Fig. 4) with arguably the most
opportunity for reflux and retrograde wetting. Sink 1, on the other hand, was farthest
away from the source (sink 5), and its P-trap had the greatest incline in the drain line
connecting the sinks, which could perhaps contribute to the reasons there was no
GFP-expressing E. coli colonization detected in it after 7 days. There has been more
investigation about microbiologic dynamics of infectious viral particles such as those of
severe acute respiratory syndrome (SARS) and Ebola viruses through premise plumbing
systems (34–36). However, the microbiology, sustainability, and dynamics might be
very different, although the backflow and inoculation issues could have some parallels
when comparing viruses to bacteria. As Enterobacteriaceae can either multiply or
remain viable for long periods of time in biofilms coating the interior of P-traps and the
connected plumbing, it may not be sustainable to target any intervention limited to a
single isolated sink as a source of a particular pathogen.

Data from different dispersal experiments suggest that although P-traps can act as
the source or the reservoir of pathogens, the physical presence of the organism in the
sink bowl or colonization of the strainer is necessary for the dispersal to occur.
Colonization of strainers or drains reported in earlier studies (7, 10, 13, 24, 37) was
perhaps a result of ascending biofilm growth from the P-trap to the strainer or
introduction through contaminated fluids. Many of the studies used swab samples,
which likely sampled the strainer rather than P-trap water (17, 20). Once the strainer
was colonized, the water from the faucet resulted in GFP-expressing E. coli dispersion
in the bowl and to the surrounding surfaces of up to 30 in. The range of dispersal

April 2017 Volume 83 Issue 8 e03327-16

Kotay et al. Applied and Environmental Microbiology

FIG 4 Layout of the sink gallery comprising the 5 sink modules and the associated plumbing. aem.asm.org 8

recorded in this study was comparable to that reported earlier (6). Greater dispersal
near the faucet may be attributed to the specific designs of the sink bowl and faucet
in this study, which determine the contact angle of water impact. This is an important
finding since many sinks in hospitals are similar in design, with faucet handles repre-
senting a high-touch surface for the sink users (38). It can also be concluded from the
dispersion experiments that secondary and successive dispersals would likely increase
the degree and the scope of dispersion.

There are several limitations to this work. First the use of similar sink bowls across these
sinks only examines the dispersion pattern of this particular sink design. Similarly the
sink-to-sink transmission may not be applicable to all wastewater plumbing systems as the
fixtures on the pipe are very close together, unlike most layouts in health care settings.
However, we speculate that transmission could occur on larger systems over greater
time scales, especially if heavy nutrient and contamination loads were also included.
GFP-expressing E. coli is a laboratory surrogate, and the putative biofilms established in
the short time frame of our experiments are unlikely to be as complex or stable as
biofilms developed in a hospital wastewater system over many years. However, to
address the monomicrobial dominance of the GFP-expressing E. coli added to the
system, we kept the system open, and other environmental organisms were able to
cocolonize in an attempt to mimic the hospital system. Another limitation was the need
to add nutrients to the drain to ensure rapid and robust colonization. We are not clear
how widespread the practice of disposing of dextrose-containing intravenous fluids or
leftover beverages in the hand-wash sinks is; however, we have observed this practice,
and anecdotally it appears to be relatively common in the United States. We also did
not completely characterize the droplet sizes, nor do we demonstrate air sampling to
understand if the dispersion is only droplet or if there are also aerosols that contain
GFP-expressing E. coli. This would require additional testing and is planned as future
work.

In summary, this work for the first time better models the mechanisms of spread of
multidrug-resistant pathogens arising from the sink drain and infecting patients. Drop-
let dispersion from the P-trap does not happen directly. Rather it is a multistage
process: dispersal originates from the strainer and/or the bowl after growth of the
biofilm up from the microbial reservoir of the P-trap. We also demonstrate sink-to-sink
transmission via a common sanitary pipe. This work could have implications for patient
safety, infection control, and interventions as well as the design of future hospital
plumbing systems to eliminate this mode of transmission to vulnerable hospitalized
patients.

MATERIALS AND METHODS

Sink gallery design. A dedicated sink gallery was set up to simulate hospital hand-washing sinks.
The gallery was comprised of five sink modules assembled next to each other (Fig. 4). The five hand-wash
sink stations were identical in bowl designs and dimensions and were modeled from the most common
intensive care unit hand-washing sink type in the acute care hospital at the University of Virginia Medical
Center. Partitions made of 24-in.-high Plexiglas sheet were installed between the sinks to prevent splatter
and cross contamination. Each sink module was built with Corian integrated sink/countertops without an
overflow and fitted with an 8-in. centerset 2-handle gooseneck faucet (Elkay, Oak Brook, IL). The drain line

April 2017 Volume 83 Issue 8 e03327-16

GFP-Expressing E. coli Sink-Trap Dispersion Applied and Environmental Microbiology

FIG 5 (a) Parts of the sink drain line: 1, faucet and handles; 2, sink counter; 3, strainer; 4, tailpipe; 5, sampling ports; aem.asm.org 9
6, trap arm; 7, P-trap. (b and c) Schematic of the nutrient regimen (b) and offset drain tailpiece used for dispersion
experiments (c).

under each sink was comprised of a flat-top fixed strainer (drain size, 2 in. by 3 in.), 17-gauge
(1.47-mm-thick) 8- to 10-in.-long tailpipe, P-trap, and trap arms of 0.25-in. outside diameter (o.d.)
(Dearborn Brass-Oatey, Cleveland, OH). All of the fixtures were made of brass with chrome plating. Each
of the sink P-traps was connected to a 3-in. common cast-iron pipe sloping into a T-joint leading into the
building sanitary line located behind sink 3 (Fig. 4).

Inoculation, growth, and establishment of GFP-expressing E. coli in sink P-traps. For the
GFP-expressing E. coli strain (ATCC 25922GFP), the green fluorescent protein (GFP) gene is contained on
a plasmid that also contains an ampicillin resistance gene. A single isolated colony of GFP-expressing E.
coli grown from a Ϫ80°C stock was inoculated into 5 ml tryptic soy broth (TSB) (Becton, Dickinson and
Company, Sparks, MD) containing 100 ␮g/ml ampicillin (ATCC medium 2855). The inoculum concentra-
tion and method varied for each experiment. For establishment of GFP-expressing E. coli in sink P-traps,
new autoclaved P-traps were filled with 100 ml 0.1ϫ TSB and inoculated with ϳ103 CFU/ml GFP-
expressing E. coli. Following inoculation, both ends of the P-traps were covered with perforated Parafilm
(Bemis, Inc., Oshkosh, WI) and allowed to incubate at room temperature (22 Ϯ 2°C) for 14 days to
facilitate adherent bacterial growth. The medium in the P-trap was decanted and replaced with fresh
0.1ϫ TSB every 48 h. An aliquot of decanted medium and a swab sample from the inner surface of the
P-trap were plated on tryptic soy agar (Becton, Dickinson and Company, Sparks, MD) plates containing
100 ␮g/ml ampicillin (TSA) to monitor the growth of GFP-expressing E. coli in the P-traps. TSA plates were
incubated overnight at 37°C, and CFU fluorescing under UV light were enumerated. All preparatory
culturing of GFP-expressing E. coli took place in a separate room from the sink gallery to avoid
unintentional contamination.

Installation of P-traps colonized with GFP-expressing E.coli. After the 14-day incubation, P-traps
were fastened into the plumbing of the sinks (Fig. 5a). The remainder of the drain line was either
autoclaved (strainer, tailpipe, and trap arms) prior to installation or surface disinfected (sink bowl,
countertop, and faucets) with Caviwipes-1 (Meterx Research, Romulus, MI), maintaining at least 1 min of
contact time. After the P-trap was installed, a daily regimen was followed in which 25 ml of TSB followed
by 25 ml of 0.9% NaCl solution (saline) were added in the ratio 1:3 via the strainer (Fig. 5b) to mimic the
potential nutrient exposure in the hospital.

Sampling and enumeration of GFP-expressing E. coli. To monitor the growth of GFP-expressing
E. coli in the plumbing, sampling ports were drilled along the length of the tailpiece (between the P-trap
and the strainer) and the trap arm (between the P-trap and the common line). These holes were fitted
with size 00 silicone stoppers (Cole-Parmer, Vernon Hills, IL) (Fig. 5a). Sterile cotton swabs (Covidien,
Mansfield, MA) presoaked in saline were inserted through these sampling ports, and samples were
collected by turning the swab in a circular motion on the inner surface (ϳ20 cm2) of the tailpipes. Sample
swabs were pulse-vortexed in 3 ml saline, and serial dilutions were plated on TSA. The strainer, faucet
aerator, and bowl surface were sampled with presoaked swabs and processed as described earlier.

Sink-to-sink transmission of bacteria. To investigate sink-to-sink transmission of bacteria, a distal
sink (sink 5) (Fig. 4) was fitted with a P-trap inoculated with GFP-expressing E. coli. The effects of different
inoculum concentrations of GFP-expressing E. coli—103, 106, and Ͼ1010 CFU/ml (colonized for 14

April 2017 Volume 83 Issue 8 e03327-16

Kotay et al. Applied and Environmental Microbiology

FIG 6 (a) Layout of the zones of the sink counter, bowl, and extension surface designated to monitor aem.asm.org 10
droplet dispersion and (b) layout of the TSA plates used for GFP-expressing E. coli droplet dispersion on
the surfaces surrounding the sink.

days)—were investigated. Identification to the species level of fluorescent and nonfluorescent colonies
identified from mixed pipe cultures was performed using a matrix-assisted laser desorption-ionization
(MALDI)–time of flight (MALDI-TOF) mass spectrometer (Vitek-MS; bioMérieux, Durham, NC). The waste-
water paths of sinks 1 to 4 were either autoclaved (strainer, tailpipe, P-traps, and trap arms) prior to
installation or surface disinfected (sink bowl, countertop, and faucets) with Caviwipes-1 (Meterx Research,
Romulus, MI). Faucets on each of the five sinks were turned on simultaneously for 1 min, supplying water
at a flow rate of 8 liters/min, once every 24 h for 7 days. No additional feed to any of the sinks was added
during this period of 7 days. On days 0 and 7, P-traps on each of the five sinks were unfastened, and swab
samples from the P-trap were collected and processed as described earlier.

Dispersion measured using fluorescent microspheres. Fluoresbrite YO carboxylate microspheres
(Polysciences, Inc.) which had a 1-␮m diameter and maximum excitation and emission of 529 nm and 546
nm, respectively, were used as a tracer in the preliminary experiments to understand droplet dispersion
from the hand-wash sinks.

To test whether microspheres could be dispersed from below the sink strainer, a 1-ml suspension of
microspheres (ϳ1010 particles) was injected through a strainer attached to a Hert 4.5-in. offset drain
tailpiece typically used for wheelchair-accessible sinks (American Standard, model 7723018.002) (Fig. 5c).
The vertical distance between the strainer and microsphere suspension injected into the tailpipe was ϳ4
in. Counter space around the sink bowl was thoroughly wiped with alcohol wipes (Covidien Webcol 6818;
Kendall), and polyester sheets precut to appropriate shapes were placed on the counter to cover the
entire sink counter and labeled according to position (Fig. 6a). The faucet was turned on for 5 min at a
water flow rate of 1.8 to 3.0 liters/min. Polyester sheets were harvested and immediately analyzed using
a ChemiDoc MP system (Bio-Rad Laboratories, Inc.) with an exposure time of 5 s. Fluorescent micro-
spheres were enumerated from the digital micrographs using the Image Lab Software (Bio-Rad Labo-
ratories, Inc.).

To test whether microspheres could be dispersed from the surface of the sink bowl, the sink bowl was
evenly coated with a 20-ml microsphere suspension (ϳ1010 particles/ml) using a disposable swab (Sage
Products, Inc., Cary, IL), and the dispersion experiment was repeated following the protocol described
above. To ascertain there was no nonspecific background fluorescence in the sink and/or the water from
the faucet, a control using the same protocol but without the fluorescent microspheres was performed
before each experiment.

April 2017 Volume 83 Issue 8 e03327-16

GFP-Expressing E. coli Sink-Trap Dispersion Applied and Environmental Microbiology

Dispersion measured using GFP-expressing E.coli. Dispersion using GFP-expressing E. coli was
investigated in three experiments. To test whether live organisms in the P-trap could be dispersed by
running water, ϳ1010 CFU/ml GFP-expressing E. coli in saline was added to an autoclaved P-trap and
fitted into the drain line that was preautoclaved (strainer, tailpipe, and trap arms). Similarly, to test
whether live organisms could be dispersed from the tailpieces of wheelchair-accessible sinks, a suspen-
sion of ϳ1010 CFU/ml GFP-expressing E. coli was added with a syringe through the strainer into the Hert
4.5-ft offset drain tailpiece (Fig. 5c). Just as in the microsphere dispersion experiment, the vertical
distance between the strainer and GFP-expressing E. coli suspension injected into the tailpipe was ϳ4 in.

We next tested whether live organisms from the surface of the sink bowl could be dispersed by
running water. The sink bowl surface was evenly coated with an approximately 20-ml suspension of 1010
CFU/ml GFP-expressing E. coli.

Finally, to mimic all of these conditions, a P-trap colonized with GFP-expressing E. coli (for 14 days)
was installed, and a nutrient regimen (Fig. 5b) was followed for 14 days to intentionally promote the
GFP-expressing E. coli colonization in the attached tailpipe and strainer. On day 15, the dispersion
experiment was performed.

Before each of the GFP-expressing E. coli dispersion experiments, the counter space was thoroughly
disinfected with Caviwipes-1. TSA plates were then positioned on the sink counter surrounding the bowl
and an extension platform (Fig. 6b). Additional plates were attached to the sink bowl, faucets, Plexiglas
partitions, and faucet handles using adhesive tape. TSA plates were also placed 3 m away from the sink
as negative controls. The faucet was turned on for 5 min with a water flow rate of 1.8 to 3.0 liters/min.
Lids of the TSA plates were removed only during faucet operation. Swab samples from the faucet
aerators before and after operation were collected and plated on TSA. Prior to each dispersion
experiment, 50 ml water from the faucet was also collected, and aliquots were plated to assess for the
presence of GFP-expressing E. coli in source water and ensure cross contamination of GFP-expressing E.
coli had not occurred. A control dispersion experiment was also performed using the same protocol prior
to GFP-expressing E. coli inoculation in each case. Dispersion per defined area (CFU per square
centimeter) was deduced by dividing the CFU counts in the TSA plate by the surface area of the TSA
plate.

REFERENCES patient environment. J Hosp Infect 85:106 –111. https://doi.org/10.1016/
j.jhin.2013.07.006.
1. Edmonds P, Suskind RR, Macmillan BG, Holder IA. 1972. Epidemiology of 12. Knoester M, de Boer M, Maarleveld J, Claas E, Bernards A, Jonge E, van
Pseudomonas-aeruginosa in a burns hospital—surveillance by a com- Dissel J, Veldkamp K. 2014. An integrated approach to control a pro-
bined typing system. Appl Microbiol 24:219 –225. longed outbreak of multidrug-resistant Pseudomonas aeruginosa in an
intensive care unit. Clin Microbiol Infect 20:O207–O215. https://doi.org/
2. Perryman F, Flournoy D. 1980. Prevalence of gentamicin-resistant and 10.1111/1469-0691.12372.
amikacin-resistant bacteria in sink drains. J Clin Microbiol 12:79 – 83. 13. Leitner E, Zarfel G, Luxner J, Herzog K, Pekard-Amenitsch S, Hoenigl M,
Valentin T, Feierl G, Grisold AJ, Hogenauer C, Sill H, Krause R, Zollner-
3. Levin M, Olson B, Nathan C, Kabins S, Weinstein R. 1984. Pseudomonas Schwetz I. 2015. Contaminated handwashing sinks as the source of a
in the sinks in an intensive-care unit—relation to patients. J Clin Pathol clonal outbreak of KPC-2-producing Klebsiella oxytoca on a hematology
37:424 – 427. https://doi.org/10.1136/jcp.37.4.424. ward. Antimicrob Agents Chemother 59:714 –716. https://doi.org/10
.1128/AAC.04306-14.
4. French G, Shannon K, Simmons N. 1996. Hospital outbreak of Klebsiella 14. Carling PC, Bartley JM. 2010. Evaluating hygienic cleaning in health care
pneumoniae resistant to broad-spectrum cephalosporins and beta- settings: what you do not know can harm your patients. Am J Infect
lactam– beta-lactamase inhibitor combinations by hyperproduction of Control 38:S41–S50. https://doi.org/10.1016/j.ajic.2010.03.004.
SHV-5 beta-lactamase. J Clin Microbiol 34:358 –363. 15. Weber D, Rutala W, Miller M, Huslage K, Sickbert-Bennett E. 2010. Role of
hospital surfaces in the transmission of emerging health care-associated
5. Rutala WA, Weber DJ. 1997. Water as a reservoir of nosocomial patho- pathogens: Norovirus, Clostridium difficile, and Acinetobacter species. Am
gens. Infect Control Hosp Epidemiol 18:609 – 616. J Infect Control 38:S25–S33. https://doi.org/10.1016/j.ajic.2010.04.196.
16. Dancer SJ. 2014. Controlling hospital-acquired infection: focus on the
6. Hota S, Hirji Z, Stockton K, Lemieux C, Dedier H, Wolfaardt G, Gardam M. role of the environment and new technologies for decontamination. Clin
2009. Outbreak of multidrug-resistant Pseudomonas aeruginosa coloni- Microbiol Rev 27:665– 690. https://doi.org/10.1128/CMR.00020-14.
zation and infection secondary to imperfect intensive care unit room 17. Vergara-Lopez S, Dominguez M, Conejo M, Pascual A, Rodriguez-Bano J.
design. Infect Control Hosp Epidemiol 30:25–33. https://doi.org/10.1086/ 2013. Wastewater drainage system as an occult reservoir in a protracted
592700. clonal outbreak due to metallo-beta-lactamase-producing Klebsiella oxy-
toca. Clin Microbiol Infect 19:E490 –E498. https://doi.org/10.1111/1469
7. Hong K, Oh H, Song J, Lim J, Kang D, Son I, Park J, Kim E, Lee H, Choi E. -0691.12288.
2012. Investigation and control of an outbreak of imipenem-resistant 18. Chapuis A, Amoureux L, Bador J, Gavalas A, Siebor E, Chretien ML, Caillot
Acinetobacter baumannii infection in a pediatric intensive care unit. D, Janin M, de Curraize C, Neuwirth C. 2016. Outbreak of extended-
Pediatr Infect Dis J 31:685– 690. https://doi.org/10.1097/INF spectrum beta-lactamase producing Enterobacter cloacae with high MICs
.0b013e318256f3e6. of quaternary ammonium compounds in a hematology ward associated
with contaminated sinks. Front Microbiol 7:1070. https://doi.org/10
8. Lowe C, Willey B, O’Shaughnessy A, Lee W, Lum M, Pike K, Larocque C, .3389/fmicb.2016.01070.
Dedier H, Dales L, Moore C, McGeer A. 2012. Outbreak of extended- 19. Clarivet B, Grau D, Jumas-Bilak E, Jean-Pierre H, Pantel A, Parer S, Lotthe
spectrum beta-lactamase-producing Klebsiella oxytoca infections associ- A. 28 April 2016. Persisting transmission of carbapenemase-producing
ated with contaminated handwashing sinks. Emerg Infect Dis 18: Klebsiella pneumoniae due to an environmental reservoir in a university
1242–1247. https://doi.org/10.3201/eid1808.111268. hospital, France, 2012 to 2014. Euro Surveill https://doi.org/10.2807/
1560-7917.ES.2016.21.17.30213.
9. Starlander G, Melhus A. 2012. Minor outbreak of extended-spectrum 20. Wolf I, Bergervoet P, Sebens F, van den Oever H, Savelkoul P, van der
beta-lactamase-producing Klebsiella pneumoniae in an intensive care
unit due to a contaminated sink. J Hosp Infect 82:122–124. https://doi aem.asm.org 11
.org/10.1016/j.jhin.2012.07.004.

10. Kotsanas D, Wijesooriya W, Korman T, Gillespie E, Wright L, Snook K,
Williams N, Bell J, Li H, Stuart R. 2013. “Down the drain”: carbapenem-
resistant bacteria in intensive care unit patients and handwashing sinks.
Med J Aust 198:267–269. https://doi.org/10.5694/mja12.11757.

11. Roux D, Aubier B, Cochard H, Quentin R, van der Mee-Marquet N. 2013.
Contaminated sinks in intensive care units: an underestimated source of
extended-spectrum beta-lactamase-producing Enterobacteriaceae in the

April 2017 Volume 83 Issue 8 e03327-16

Kotay et al. Applied and Environmental Microbiology

Zwet W. 2014. The sink as a correctable source of extended-spectrum 29. Decker B, Palmore T. 2013. The role of water in healthcare-associated
beta-lactamase contamination for patients in the intensive care unit. infections. Curr Opin Infect Dis 26:345–351. https://doi.org/10.1097/QCO
J Hosp Infect 87:126 –130. https://doi.org/10.1016/j.jhin.2014.02.013. .0b013e3283630adf.
21. Vickery K, Deva A, Jacombs A, Allan J, Valente P, Gosbell I. 2012. Presence
of biofilm containing viable multiresistant organisms despite terminal 30. Centers for Disease Control and Prevention. 2013. Antibiotic resistance
cleaning on clinical surfaces in an intensive care unit. J Hosp Infect threats in the United States. Centers for Disease Control and Prevention,
80:52–55. https://doi.org/10.1016/j.jhin.2011.07.007. Office of Infectious Disease, Atlanta, GA.
22. Dancer SJ. 2016. Dos and don’ts for hospital cleaning. Curr Opin Infect
Dis 29:415– 423. https://doi.org/10.1097/QCO.0000000000000289. 31. Ferranti G, Marchesi I, Favale M, Borella P, Bargellini A. 2014. Aetiology,
23. La Forgia C, Franke J, Hacek D, Thomson R, Robicsek A, Peterson L. 2010. source and prevention of waterborne healthcare-associated infections: a
Management of a multidrug-resistant Acinetobacter baumannii outbreak review. J Med Microbiol 63:1247–1259. https://doi.org/10.1099/jmm.0
in an intensive care unit using novel environmental disinfection: a .075713-0.
38-month report. Am J Infect Control 38:259 –263. https://doi.org/10
.1016/j.ajic.2009.07.012. 32. Kanamori H, Weber DJ, Rutala WA. 2016. Healthcare outbreaks associ-
24. Breathnach AS, Cubbon MD, Karunaharan RN, Pope CF, Planche TD. ated with a water reservoir and infection prevention strategies. Clin
2012. Multidrug-resistant Pseudomonas aeruginosa outbreaks in two Infect Dis 62:1423–1435. https://doi.org/10.1093/cid/ciw122.
hospitals: association with contaminated hospital waste-water sys-
tems. J Hosp Infect 82:19 –24. https://doi.org/10.1016/j.jhin.2012.06 33. McBain AJ, Bartolo RG, Catrenich CE, Charbonneau D, Ledder RG, Rickard
.007. AH, Symmons SA, Gilbert P. 2003. Microbial characterization of biofilms
25. Tofteland S, Naseer U, Lislevand JH, Sundsfjord A, Samuelsen O. 2013. A in domestic drains and the establishment of stable biofilm microcosms.
long-term low-frequency hospital outbreak of KPC-producing Klebsiella Appl Environ Microbiol 69:177–185. https://doi.org/10.1128/AEM.69.1
pneumoniae involving intergenus plasmid diffusion and a persisting .177-185.2003.
environmental reservoir. PLoS One 8:e59015. https://doi.org/10.1371/
journal.pone.0059015. 34. Li Y, Duan S, Yu IT, Wong TW. 2005. Multi-zone modeling of probable SARS
26. Stjarne Aspelund A, Sjostrom K, Olsson Liljequist B, Morgelin M, Me- virus transmission by airflow between flats in Block E, Amoy Gardens.
lander E, Pahlman LI. 2016. Acetic acid as a decontamination method for Indoor Air 15:96–111. https://doi.org/10.1111/j.1600-0668.2004.00318.x.
sink drains in a nosocomial outbreak of metallo-beta-lactamase-
producing Pseudomonas aeruginosa. J Hosp Infect 94:13–20. https://doi 35. Ye Y, Ellenberg RM, Graham KE, Wigginton KR. 2016. Survivability, par-
.org/10.1016/j.jhin.2016.05.009. titioning, and recovery of enveloped viruses in untreated municipal
27. Soothill J. 2016. Carbapenemase-bearing Klebsiella spp. in sink drains: wastewater. Environ Sci Technol 50:5077–5085. https://doi.org/10.1021/
investigation into the potential advantage of copper pipes. J Hosp Infect acs.est.6b00876.
93:152–154. https://doi.org/10.1016/j.jhin.2016.03.013.
28. Fusch C, Pogorzelski D, Main C, Meyer C, el Helou S, Mertz D. 2015. 36. Lin K, Marr LC. 10 February 2017. Aerosolization of Ebola virus surrogates
Self-disinfecting sink drains reduce the Pseudomonas aeruginosa biobur- in wastewater systems. Environ Sci Technol https://doi.org/10.1021/acs
den in a neonatal intensive care unit. Acta Paediatr 104:e344 – e349. .est.6b04846.
https://doi.org/10.1111/apa.13005.
37. Ambrogi V, Cavalie L, Mantion B, Ghiglia MJ, Cointault O, Dubois D, Prere
MF, Levitzki N, Kamar N, Malavaud S. 2016. Transmission of metallo-
beta-lactamase-producing Pseudomonas aeruginosa in a nephrology-
transplant intensive care unit with potential link to the environment. J
Hosp Infect 92:27–29. https://doi.org/10.1016/j.jhin.2015.09.007.

38. Devnani M, Kumar R, Sharma RK, Gupta AK. 2011. A survey of hand-
washing facilities in the outpatient department of a tertiary care teach-
ing hospital in India. J Infect Dev Ctries 5:114 –118.

April 2017 Volume 83 Issue 8 e03327-16 aem.asm.org 12

De Geyter et al. Antimicrobial Resistance and Infection Control (2017) 6:24
DOI 10.1186/s13756-017-0182-3

RESEARCH Open Access

The sink as a potential source of
transmission of carbapenemase-producing
Enterobacteriaceae in the intensive care
unit

Deborah De Geyter1* , Lieve Blommaert1, Nicole Verbraeken1, Mark Sevenois2, Luc Huyghens2, Helena Martini1,
Lieve Covens1, Denis Piérard1 and Ingrid Wybo1

Abstract

Background: Carbapenemase-producing Enterobacteriaceae (CPE) are emerging pathogens that represent a major
public health threat. In the University Hospital of Brussels, the incidence of new patients with CPE rose from 1 case
in 2010 to 35 cases in 2015. Between January and August 2015, five patients became infected/colonized with CPE
during their stay in the same room in the intensive care unit (ICU). Since the time period between those patients
was relatively short and the strains belonged to different species with different antibiograms and mechanisms of
resistance, the hypothesis was that the environment could be a possible source of transmission.

Methods and results: Environmental samples suggested that a contaminated sink was the source of the outbreak.
Besides other strains, Citrobacter freundii type OXA-48 was frequently isolated from patients and sinks. To investigate
the phylogenetic relationschip between those strains, pulsed-field gel electrophoresis was performed. The strains
isolated from patients and the sink in the implicated room were highly related and pointed to sink-to-patient
transmission. In total, 7 of 8 sinks in the isolation rooms of the ICU were found to be CPE contaminated. To control
the outbreak, the sinks and their plumbings were replaced by new ones with another structure, they were flushed
every morning with a glucoprotamin solution and routines regarding sink practices were improved leading to
discontinuation of the outbreak.

Conclusions: This outbreak highlights that hospital sink drains can accumulate strains with resistance genes and
become a potential source of CPE.

Keywords: Carbapenemase-producing Enterobacteriaceae, Hospital sinks, Outbreak, Intensive care unit, Citrobacter
freundii OXA-48, Transmission

Background confirmed cases by 2014. Patients in the intensive care
Carbapenemase-producing Enterobacteriaceae (CPE) rep- unit (ICU) comprised the majority of all colonized or
resent a major public threat in both the acute and chronic infected patients with CPE. Despite screening and en-
care sectors as well as in the community. Treatment of hanced cleaning, the incidence rose further in the first half
patients infected by CPE is a challenge since only a few of 2015 and could be linked to contaminated sinks.
drugs remain active against these strains [1]. Between
2010 and 2014, our healthcare facility was confronted with It is described that sink drains beneath washbasins in
a rising number of CPE positive patients, with 26 hospitals contain 106–1010 colony forming units (CFU)/
ml of bacteria of which approximately 103–105 CFU/ml
* Correspondence: [email protected] are Gram-negative (GN) rods, especially waterborne
1Department of Microbiology, Vrije Universiteit Brussel (VUB), Universitair bacteria [2]. These bacteria can colonize/infect the pa-
Ziekenhuis Brussel (UZ Brussel), Laarbeeklaan 101, 1090 Brussels, Belgium tient via different transmission routes. Contaminated
Full list of author information is available at the end of the article sinks have been implicated in several outbreaks. Kramer

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

De Geyter et al. Antimicrobial Resistance and Infection Control (2017) 6:24 Page 2 of 6

and colleagues described that sinks can be hidden reser- Development of a colonization/infection with CPE means
voirs generating large quantities of aerosols. They found that rectal screening samples were negative on admission.
that 100% of the sinks in a neonatal ICU were contami- An outbreak can be defined as at least two cases linked by
nated with GN rods [3]. Starlander and co-workers an epidemiological chain of transmission.
noticed that four patients became infected or colonized
by an ESBL producing Klebsiella pneumoniae strain CPE surveillance and isolation policy
during a seven month period on the neurological ICU. Patients in the ICU are screened rectally on admission;
Environmental sampling led to the conclusion that the on discharge and weekly for all patients hospitalized
plughole of the sink was the source of transmission [4]. more than one week. When a screening is considered as
The group of Lowe described an outbreak with an ESBL positive, the patient is isolated on contact precautions in
producing Klebsiella oxytoca. Cultures from hand wash- a single room with use of gloves and a disposable
ing sinks in the ICU unit yielded Klebsiella oxytoca with overcoat.
identical molecular patterns to cultures from the pa-
tients [5]. Wolf and colleagues discovered that mechan- Environmental cleaning policy
ically ventilated patients were sometimes colonized with The room has to be cleaned daily with Incidin® Plus
bacteria positive for ESBLs. During a five month obser- (0.5% glucoprotamin). At discharge, the room is cleaned
vation period, four patients became colonized with ESBL intensively and equipment such as gloves and hand
positive bacteria that were genetically identical to those alcohol are discarded. Periodic checks of the quality of
that had previously been isolated from the sink [6]. Roux terminal cleaning are performed with the Glowcheck®
and co-workers found that 31% of the 185 sinks in their (Hartmann, Heidenheim, Germany).
ICU were contaminated with ESBL positive Klebsiella
and Enterobacter species [7]. The group of Leitner pub- Microbiological methods
lished an article about contaminated handwashing sinks Environmental sampling and microbiological methods
as the source of a clonal outbreak with a resistant Klebsi- Swabs (eSwab, Copan, Brescia, Italy) from the plughole
ella oxytoca strain on a hematology ward [8]. of the sink (10–15 cm depth), the environment of the
sink and room (high-touch surfaces such as the bed,
The aim of this study was to describe an outbreak of monitor, door knob, the Velcro of the blood pressure de-
CPE in the ICUs of the UZ Brussels and to investigate vice, matrass and curtain) were taken at several time
whether the sink could be a possible source of transmis- points. After sampling, 2 ml of Fastidious Organisms
sion and in addition, whether these transmissions could Broth (FB, own preparation) was added to the eSwab
be avoided in the future by means of improving infection and incubated for 24 h on 35 °C. After one day, sink and
control measures and replacing the sinks by new ones. environmental specimens were investigated for the pres-
This study highlights the fact that sink drains can be ence of CPE on two separate chromogenic media: chro-
hidden reservoirs for CPE which is in general not known mID® CARBA and OXA-48 (bioMérieux, Marcy l’Etoile,
by health care workers and even by infection control France) 48 h after sampling (24 h incubation on 35 °C).
specialists. Identification of suspicious colonies was performed by
matrix-assisted laser desorption ionization–time of flight
Methods mass spectrometry (MALDI-TOF MS) using a Microflex
LT mass spectrometer with MALDI Biotyper 3.0 software
Setting and Reference Library 3.2.1.0 (Bruker DaltonikGmbH,
The University Hospital Brussels is a teaching hospital Bremen, Germany).
with more than 700 beds. There are four ICUs for
adults. Each unit contains six beds, two of them are Antibiotic susceptibility testing was performed by the
placed in a single room that can be closed (room one disk diffusion method and by using the interpretative
and six; room one has an anteroom). These two separate criteria of The European Committee on Antimicrobial
rooms are predominantly in use for patients who need Susceptibility Testing (EUCAST) combined with recom-
additional isolation precautions. Every unit has eight mendations of the National Reference Center (NRC) and
sinks: one for every patient, one in the anteroom and BAPCOC (Belgian Commission for the Coordination of
one placed centrally. In total there are thus 32 sinks the Antibiotic Policy). Colonies considered as suspicious
available in the ICUs. for the production of a carbapenemase need further mo-
lecular characterization as a confirmation step.
Case definitions
New cases of patients with CPE were defined as patients Air sampling experiments
(infected or colonized) identified in the ICUs in 2015 (not The bacterial aerosol was measured 10 cm above the sinks
known with a CPE before 2015 or not known in another during tap water running over 10 min and compared with
hospital) with a CPE positive culture from any site.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

De Geyter et al. Antimicrobial Resistance and Infection Control (2017) 6:24 Page 3 of 6

results without running tap water. The measurements initial screenings were negative, developed CPE carriage/
were performed with the MAS-100® Airsampler (Merck infection during their hospital stay on the ICU A in bed
Millipore, Darmstadt, Germany); programmed to measure 6. In Table 1, the isolates from the five patients who
100 L air per minute and containing MacConkey plates became positive for CPE on ICU A, room 6 are repre-
(bioMérieux, Marcy l’Etoile, France) specific for the detec- sented. The isolates belonged to different species with
tion of GN bacteria. These plates were incubated for 18– different antibiograms. The time period between the de-
24 h on 35 °C. Samples from the environment of the sink tection of these CPE positive patients was relatively
were also taken. Further identification and susceptibility small and since carbapenemase resistance determinants
testing was performed as described previously (in “Envir- are located on genetic mobile elements, the environment
onmental sampling and microbiological methods”). was suspected as a possible source of transmission.

Molecular characterization Environmental investigation
Swabs on dry surfaces taken were taken as well as swabs
A) Bacterial colonies suspicious for the production of from the sinks’ drain in room 6, ICU A. Two samples
carbapenemases became positive for CPE: the sinks’ drain and the blood
Molecular characterization was performed using the pressure cuff although a procedure for cleaning was in
Xpert Carba-R Assay on the GeneXpert® system place for the latter. The samples from other high-touch
(Cepheid, Sunnyvale, California, USA). Colonies surfaces were all negative for CPE, indicating that the
considered as suspicious for the production of cleaning was appropriate.
carbapenemases but with a negative result on the
GeneXpert®, were send to the NRC of CPE to To investigate whether other isolation rooms were also
exclude resistance. carrying CPE in their sinks, samples were taken from
the drains. Besides one, every isolation room carried
B) Genetic relatedness CPE in the sink. Several species and types of CPE were
To confirm genetic relatedness between isolates isolated: Klebsiella pneumoniae NDM, Klebsiella pneu-
from patients and environmental samples, pulsed- moniae KPC, Klebsiella pneumoniae OXA-48, Entero-
field gel electrophoresis (PFGE) was performed. bacter cloacae complex OXA-48, Citrobacter freundii
OXA-48, Citrobacter freundii OXA-48 + NDM, Klebsi-
Patients’ samples ella oxytoca/Raoultella species OXA-48, and Escherichia
Sampling in patients is part of routine daily practice. coli OXA-48, suggesting that resistance genes did accu-
Screening samples are taken via rectal swabs (eSwab, mulate in the drains as patients with CPE were prefera-
Copan, Brescia, Italy). For rectal screenings and other bly cared for in this rooms. The results also revealed
clinical samples, the same protocol as for environmental that in some rooms, the bacterial flora in the sink was
sampling is followed, except that they are not enriched probably the same as the bacteria cultured in the patient.
with FB. Between the end of July and the end of August 2015, 5
patients were positive for CPE that were probably the
Results same as in the sinks. Three of them were not known
with a CPE on admission and here it can be stated that
Epidemiological investigation the sink colonized/infected the patients. From those
Between 2010 and 2015, the number of new patients three patients, two of them had positive respiratory
with a CPE per year on the ICU rose from 1 to 21 and samples with CPE.
the most abundant type became the Klebsiella pneumo-
niae carbapenemase. In 2015, patients with CPE staying In 2015, Citrobacter freundii type OXA-48 was fre-
on the ICU contributed 67% of all cases with CPE. quently isolated from patients and sinks. To investigate the
Between January and August 2015, five patients whose relatedness between those strains, PGFE was performed.

Table 1 CPE carriage/infection on the ICU A, bed 6

Patient 1 (colonization) KPC, OXA-48 and NDM (2/2/2015). Different species

Patient 2 (infection site: endotracheal aspirate) KPC, OXA-48 and NDM (23/2/2015). Different species

Patient 3 (colonization) OXA-48 and NDM (18/5/2015). Different species

Patient 4 (infection site: abdominal abscess) OXA-48 (15/7/2015). Different species

Patient 5 (infection site: endotracheal aspirate) NDM and OXA-48 (30/8/2015). Different species

Five patients developed CPE carriage/infection during their hospital stay on the ICU A, room six between January and August 2015. They all had negative
screenings on admission. The isolates belonged to different species with different antibiograms. The time period between the detection of these CPE positive
patients was relatively small

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

De Geyter et al. Antimicrobial Resistance and Infection Control (2017) 6:24 Page 4 of 6

Citrobacter freundii CPE positive strains from patients Follow up
staying on the ICU isolated in 2013 and 2014 were also in- Between September and the end of December 2015,
cluded. These results showed that the strains isolated from swabs of the plugholes of the sinks were taken on a
patients and sinks in room 6 are highly related (Fig. 1). regular base in every room of the ICUs. Despite all the
efforts made, still 9 out of the 32 sinks (28%) were posi-
Infection control measures tive for CPE at the end of September 2015. During that
In the weeks and months following the previous results, time period, three patients had a sample showing CPE
the infection control team tried to get rid of the out- but two of them were already positive on admission. A
break by implementing new infection control measures third person had a respiratory infection with a CPE but
in addition to existing ones. Medical staff, nurses and in his case, it was not possible to know whether this in-
the cleaning personnel were informed about the out- fection was nosocomial or not since he was not screened
break, presentations were given, the extreme import- on admission.
ance of correct hand hygiene was highlighted and
measurements on intensively cleaning were performed. Because of the high number of siphons containing CPE
Since the blood pressure cuff yielded a CPE, it was de- even after replacement, the technical department was
cided to work with disposable material. There was an asked to replace all the whole sinks (not only the siphons)
agreement to replace all the siphons in the isolation by newly engineered ones with an open inlet. This makes
rooms and to culture the siphons in the other rooms. A cleaning easier and prevents the formation of biofilms.
daily disinfection of the sinks with Incidin® Plus (Micro-
tek, Zutphen, The Netherlands) was implemented. The In general: between the 1st of January and the 31st of
sinks were used for washing hands and medical instru- December 2015, 21 patients became colonized/infected
ments before disinfection but also to flush patient’s with a CPE on the ICUs; 8 of them probably due to a con-
substances such as dialysis removals containing antibi- taminated siphon. Since the replacement of the whole
otics and microorganisms. These actions can promote sinks at the end of 2015, only one patient became positive
biofilm formation and selection of resistant bacteria. with a CPE in March 2016 after he had been hospitalized
Therefore, it was decided to use the sinks only for on the ICU. At that moment, he was already hospitalized
“clean work”. For the removal of dialysis fluids, special for four weeks. The drain of the sink in the room where
waste containers were bought. the patient had stayed was positive with CPE, although it
was another species and strain. In this case, it was hard to
say whether the CPE was transmitted from the sink to the
patient or the other direction since the patient was not

Fig. 1 Phylogenetic relatedness between Citrobacter freundii CPE strains. On the left side of the figure, a dendogram is shown representing the
relationship between the strains in %. In blue, the patients’ and drain cultures from ICU A bed 6 are shown and in black, the patients’ and drain
cultures from ICU A bed 1are represented together with the date of isolation. The strains isolated from patients and sinks in room 6 are highly related

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

De Geyter et al. Antimicrobial Resistance and Infection Control (2017) 6:24 Page 5 of 6

screened on admission. The other sinks stayed negative disinfection, as waste bins and as water suppliers for
until the last day of the sampling and no other transmis- shaving men. Reprocessing of reusable medical devices
sions were noted. should be centralized, shaving can be done by means of
a separate wash basin and fluids have to be removed in
Airsampling experiments with two sinks showed that special containers that provide easy transport to the util-
it is possible that bacteria in the biofilm of the sink can ity room. Body fluids should not be flushed through the
get carried up into the air above the drain through aero- sink anymore. A second possibility is to build a room
sols when tap water is running, since we were able to with two separate sinks. One sink should be rigorously
pick up Gram-negative bacteria in the air and the envir- restricted to hand washing. The sinks used for waste dis-
onment on the sink after tap water was running. In the posal should be systematically considered as potentially
air, Stenotrophomonas maltophilia, Serratia marcescens, contaminated. In our ICUs, the distance between the
Pseudomonas oleovorans group and Pseudomonas putida sink and the patient’s bed is less than one meter. There
were recovered. On the taps, Serratia marcescens and are no guidelines about the minimum distance required
Pseudomonas fluorescens group were found and on the and it depends also on the sink’s architecture, but as
rims, Serratia marcesens CPE suspected and Stenotro- seen in literature, aerosols and splashes coming from the
phomas maltophila were cultured. plughole of the sink can be propagated up to 1 m from
the sink when the tap is turned on [11]. Therefore, we
Discussion suggest that the distance between a sink and a patient’s
The prevalence of infections with multidrug resistant bed should be at least two meter.
(MDR) GN bacteria such as CPE is increasing worldwide
[9] and there is probably a wider range of environmental In our ICU, there are two types of taps. The taps in
reservoirs for those bacteria compared with Gram posi- the isolation rooms are designed well with a distance
tive MDR bacteria [10]. between tap and inlet of 40 cm. In contrast, the taps in
the standard rooms are not ergonomically designed: the
Both the results in the literature and in this work show distance between the tap and the inlet of the wash basin
that the sink is an ideal moist reservoir for (waterborne) is too small (20 cm) which makes it possible to contam-
GN bacteria to survive. The fact that fluids, often inate the tap when water is running. In the future, we
containing antibiotics, were flushed through the drains, will replace them. Moreover, in the sinks of the standard
promotes the selection of resistant bacterial strains. Despite rooms, water from the tap is directed straight into the
the efforts made and the discontinuation of the outbreak outlet, allowing splash-back from the sink’s drain trap.
with CPE on ICU A room six, even after replacing the
complete sinks, we could find multiresistant Pseudomonas A German company brought a self-disinfecting siphon
species and Stenotrophomonas maltophilia. Both bacteria on the market (MoveoSiphon ST24, MoveoMed, Dresden,
can also colonize/infect patients and indeed, we noticed Germany). That device prevents the formation of a biofilm
that some patients showed positive respiratory samples in the sink by means of permanent physical disinfection,
with those species after a few days of hospitalization on the electromagnetic cleaning and antibacterial coating [12].
ICU (data not shown). However, we didn’t do molecular This siphon was tested during five months (July- November
characterization to prove this hypothesis. 2016) in the ICU A room 1 for the presence of GN bacteria.
During that entire period, we could not pick up any GN
Air sampling experiments taught us that the air above bacteria (data not published). However; there still need to
and the environment of the sink got contaminated with be investigated whether that siphon could really prevent
bacteria after tap water was running. Hands of health nosocomial transmissions in our ICUs and whether this
care workers can be contaminated via this way. This un- intervention will be cost-effective.
derlines the primordial importance of hand hygiene. A
limitation of the study is that although we could pick-up Conclusions
bacteria in the air and the environment of the sink after There can be concluded that the environment is an
water was running, we could not prove that these bac- important reservoir for MDR Gram-negative bacteria, as
teria really came from the biofilm of the siphon and we demonstrated in this CPE outbreak linked to contami-
were not able to pick up CPE in the air above the sink. nated sinks. Despite the efforts already made and the
This is possibly due to the limited time and availability discontinuation of the outbreak in ICU A, some sinks are
of the siphons for the measurements in order to not to date still contaminated with MDR strains. Therefore,
disturb the critical ill patients in the room. different disciplines need to sit together and decide about
how to change our “sink attitude” and how to make struc-
In the next years, a new ICU will be built in the UZ tural changes to the architecture of a patient’s room in a
Brussels. A few propositions according to the architec- way that it stays both practical for the health care workers
ture of the room and sinks can be made. First of all, it and economical.
was considered not to place sinks in the rooms. Sinks
were used to wash reusable medical devices before

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

De Geyter et al. Antimicrobial Resistance and Infection Control (2017) 6:24 Page 6 of 6

Abbreviations lactamase contamination for patients in the intensive care unit.
BAPCOC: Belgian commission for the coordination of the antibiotic policy; J Hosp Infect. 2014;87(2):126–30.
CFU: Colony forming units; CPE: Carbapenemase-producing 7. Roux D, Aubier B, Cochard H, Quentin R, van der Mee-Marquet N.
Enterobacteriaceae; ESBLs: Extended-spectrum β-lactamases; EUCAST: The Contaminated sinks in intensive care units: an underestimated source of
European committee on antimicrobial susceptibility testing; FB: Fastidious extended-spectrum beta-lactamase-producing Enterobacteriaceae in the
organisms Broth; GN: Gram-negative; ICU(s): Intensive care unit(s); MALDI-TOF patient environment. J Hosp Infect. 2013;85(2):106–11.
MS: Matrix assisted laser desorption/ionization-time of flight mass 8. Leitner E, Zarfel G, Luxner J, Herzog K, Pekard-Amenitsch S, Hoenigl M,
spectrometry; MBL: Metallo-β-lactamase; MDR: Multidrug resistant; NDM: New Valentin T, Feierl G, Grisold A, Högenauer C, Sill H, Krause R, Zollner-
Delhi metallo-β-lactamase; NRC: National reference center; OXA: Oxacillinase; Schwetzd I. Contaminated Handwashing Sinks as the Source of a Clonal
PFGE: Pulsed-field gel electrophoresis Outbreak of KPC-2-Producing Klebsiella oxytoca on a Hematology Ward.
AAC. 2015;59(1):714–6.
Acknowledgements 9. Hoxha A, Kärki T, Giambi C, Montano C, Sisto A, Bella A, D’Ancona F.
We would like to thank Hans Boterbergh and Filip van Impe for the technical Attributable mortality of carbapenem-resistant Klebsiella pneumoniae
help and knowledge in the placement of the new siphons. infections in a prospective matched cohort study in Italy, 2012–2013.
J Hosp Infect. 2016;92(1):61–6.
Funding 10. Wilson AP, Livermore DM, Otter JA, Warren RE, Jenks P, Enoch DA, Newsholme
Not applicable. W, Oppenheim B, Leanord A, McNulty C, Tanner G, Bennett S, Cann M, Bostock
J, Collins E, Peckitt S, Ritchie L, Fry C, Hawkey P. Prevention and control of
Availability of data and materials multi-drug-resistant Gram-negative bacteria: recommendations from a Joint
Not applicable. Working Party. J Hosp Infect. 2016;92 Suppl 1:S1–44.
11. Hota S, Hirji Z, Stockton K, Lemieux C, Dedier H, Wolfaardt G, Gardam MA.
Authors’ contributions Outbreak of multidrug-resistant Pseudomonas aeruginosa colonization and
DDG was the writer of this article and the main investigator of this work. LB, infection secondary to imperfect intensive care unit room design. Infect
NC and IW helped to set up the infection control measures and the study Control Hosp Epidemiol. 2009;30(1):25–33.
design. IW and DP corrected the draft of the manuscript. MS and LG helped 12. Sissoko B, Sütterlin R, Blaschke M, Flicker J, Schluttig A. Infektionreservoir
in the follow-up of the infection control measures in the ICU. HM performed Geruchsverschluss: Prävention nosokomialer Infektionen. Hyg Med. 2004;
the PFGE. LC conducted the identification and susceptibility testing of the 29(12):451–5.
strains. All authors read and approved the final manuscript.

Competing interests
The authors declare that they have no competing interests.

Consent for publication
Not applicable.

Ethics approval and consent to participate
This study was approved by the Ethical Medical Committee of the UZ
Brussels (number B.U.N. 143201526301).

Author details
1Department of Microbiology, Vrije Universiteit Brussel (VUB), Universitair
Ziekenhuis Brussel (UZ Brussel), Laarbeeklaan 101, 1090 Brussels, Belgium.
2Department of Intensive Care Unit, Vrije Universiteit Brussel (VUB),
Universitair Ziekenhuis Brussel (UZ Brussel), Laarbeeklaan 101, 1090 Brussels,
Belgium.

Received: 3 December 2016 Accepted: 13 February 2017

References Submit your next manuscript to BioMed Central
1. Doi Y, Paterson DL. Carbapenemase-producing Enterobacteriaceae. Semin and we will help you at every step:

Respir Crit Care Med. 2015;36(1):74–84. • We accept pre-submission inquiries
2. Döring G, Ulrich M, Müller W, Bitzer J, Schmidt-Koenig L, Münst L, Grupp H, • Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
Wolz C, Stern M, Botzenhart K. Generation of Pseudomonas aeruginosa • Convenient online submission
aerosols during handwashing from contaminated sink drains, transmission • Thorough peer review
to hands of hospital personnel, and its prevention by use of a new heating • Inclusion in PubMed and all major indexing services
device. Zentralbl Hyg Umweltmed. 1991;191(5–6):494–505. • Maximum visibility for your research
3. Kramer A, Daeschlein G, Niesytto C, Sissoko B, Sütterlin R, Blaschke M, Fusch
C. Contamination of sinks and emission of nosocomial gramnegative Submit your manuscript at
pathogens in a NICU - outing of a reservoir as risk factor for nosocomial www.biomedcentral.com/submit
colonization and infection. Umweltmed. Forsch. Prax. 2005;10(5):327.
4. StarlanderMelhus Å G, Melhus Å. Minor outbreak of extended-spectrum β-
lactamase-producing Klebsiella pneumoniae in an intensive care unit due to
a contaminated sink. J Hosp Infect. 2012;82(2):122–4.
5. Lowe C, Willey B, O’Shaughnessy A, Lee W, Lum M, Pike K, Larocque C,
Dedier H, Dales L, Moore C. Outbreak of extended-spectrum β-lactamase-
producing Klebsiella oxytoca infections associated with contaminated
handwashing sinks. Emerg Infect Dis. 2012;18(8):1242–7.
6. Wolf I, Bergervoet P, Sebens F, van den Oever H, Savelkoul P, van der
Zwet W. The sink as a correctable source of extended-spectrum β-

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions

Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access

control;

2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is

otherwise unlawful;

3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in

writing;

4. use bots or other automated methods to access the content or redirect messages

5. override any security feature or exclusionary protocol; or

6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal

content.

In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

[email protected]

ORIGINAL ARTICLE EPIDEMIOLOGY

Wastewater drainage system as an occult reservoir in a protracted clonal
outbreak due to metallo-b-lactamase-producing Klebsiella oxytoca

S. Vergara-Lo´ pez1, M. C. Dom ınguez2, M. C. Conejo3, A´ . Pascual3,4 and J. Rodr ıguez-Ban~o4,5
1) Internal Medicine Service, Hospital La Merced, 2) Laboratory of Microbiology, Hospital La Merced, Osuna, Seville, 3) Department of Microbiology, University of
Seville, 4) Infectious Diseases and Clinical Microbiology Unit, University Hospital Virgen Macarena and 5) Department of Medicine,
University of Seville, Seville, Spain

Abstract

We describe the epidemiology of a protracted nosocomial clonal outbreak due to multidrug-resistant IMP-8 producing Klebsiella oxytoca
(MDRKO) that was finally eradicated by removing an environmental reservoir. The outbreak occurred in the ICU of a Spanish hospital from
March 2009 to November 2011 and evolved over four waves. Forty-two patients were affected. First basic (active surveillance, contact
precautions and reinforcement of surface cleaning) and later additional control measures (nurse cohorting and establishment of a minimum
patient/nurse ratio) were implemented. Screening of ICU staff was repeatedly negative. Initial environmental cultures, including dry surfaces,
were also negative. The above measures temporarily controlled cross-transmission but failed to eradicate the epidemic MDRKO strain that
reappeared two weeks after the last colonized patients in waves 2 and 3 had been discharged. Therefore, an occult environmental reservoir
was suspected. Samples from the drainpipes and traps of a sink were positive; removal of the sink reduced the rate number but did not stop
new cases that clustered in a cubicle whose horizontal drainage system was connected with the eliminated sink. The elimination of the
horizontal drainage system finally eradicated the outbreak. In conclusion, damp environmental reservoirs (mainly sink drains, traps and the
horizontal drainage system) could explain why standard cross-transmission control measures failed to control the outbreak; such reservoirs
should be considered even when environmental cultures of surfaces are negative.

Keywords: Carbapenemase, environmental reservoir, IMP-8, Klebsiella oxytoca, outbreak
Original Submission: 22 December 2012; Revised Submission: 17 April 2013; Accepted: 27 May 2013
Editor: J.-M. Rolain
Article published online: 31 May 2013
Clin Microbiol Infect 2013; 19: E490–E498
10.1111/1469-0691.12288

Corresponding author: J. Rodr ıguez-Ba~no, Infectious Diseases and control measures [9] is often not enough to fully control the
Clinical Microbiology Unit, Hospital Universitario Virgen Macarena, outbreaks, so that many evolve over long periods of time or
Avda Dr. Fedriani 3, Sevilla 41009, Spain even spread to other healthcare centre [6,10]. Specifically, the
E-mail: [email protected] medical literature has not drawn enough attention to the
potential importance of environmental reservoirs during
Introduction complex outbreaks, because they have been considered less
important than the reservoir formed by colonized patients
Outbreaks caused by multidrug-resistant (MDR) Klebsiella spp. [5,7].
are a growing worldwide problem. Such outbreaks occur more
frequently (but not exclusively) in intensive care units (ICUs) We report a prolonged clonal outbreak of nosocomial
[1], have been associated with significant mortality [2] and are infection due to a multidrug-resistant strain of Klebsiella oxytoca
usually clonal [3–6], although different clones or species (MDRKO), which was previously characterized as the first
sharing epidemic plasmids have also been described [7,8]. Most IMP-8-producing Enterobacteriaceae in Spain [11], with the aim
importantly, the implementation of recommended infection of describing its epidemiological features and the control
measures implemented; we emphasize our finding that iden-
tifying and isolating an environmental reservoir was key for the
eradication of this outbreak.

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases

CMI Vergara-Lo´ pez et al. Sinks perpetuate an outbreak due to carbapenem-resistant Klebsiella oxytoca E491

Materials and Methods Patients, environmental and healthcare staff studies, and
actions taken
We followed the recommendations of the ORION statement Before the onset of the outbreak, all patients with >1 month
for reporting outbreaks [12]. stay in the ICU underwent active screening by rectal and
pharyngeal swabbing for colonization with multidrug-resistant
Setting Gram negative bacteria. As described below, an active
The study was conducted in La Merced Hospital, a 240-bed screening protocol including weekly and discharge rectal and
community public centre in Osuna, Seville (Spain). The pharyngeal swabbing (or tracheal aspirate if under mechanical
hospital has an eight-bed medical and surgical ICU (with ventilation) was started when the outbreak was detected.
three additional beds in an adjoining room to be occupied
if necessary), which receives ~350 admissions annually. During the outbreak, six main environmental investigations
The structure of the ICU is shown in Fig. 1. The water were carried out (Fig. 2 and Table 1). The earlier ones focused
supply and wastewater removal system comprised 11 sinks on dry surfaces, while later investigations tended to concen-
(labelled S1 to S11 in Fig. 1); each sink drained into trate on surfaces surrounding the patients and damp environ-
a wastepipe (W), labelled according to the number of ments. The methodology used is described in detail below.
the sink, except for: S6 and S7, which shared the same
drainpipe and, together with S5, drained into W5; S8 and The control measures undertaken during the outbreak are
S9, which drained into W7; and S10 and S11, which drained described in the Results section (the environmental measures
into W8. are summarised in Table 1). An additional routine intervention
carried out once a year, the Prevention and Control of Legionella
Patients Infection Protocol (PCLIP), which was non-specifically imple-
The investigation included all 42 patients colonized or infected mented because of this outbreak, is described here because of
by the epidemic strain of MDRKO over the outbreak, which its potential impact on the evolution of the outbreak. The PCLIP
evolved over four waves (Fig. 2). The index case was detected is applied once a year to treat water supply networks of all
in March 2009 and the last case in November 2011. Patients Spanish hospitals. In our centre, this is performed by hyper-
with MDRKO were considered to be infected if presenting chlorination of the main water tank for 3 h with free residual
with signs of active infection considered to have been caused chlorine (20–30 mg/L) and hyperchlorination of the terminal
by MDRKO, according to CDC criteria [13], and colonized points (1–2 mg/L) for 2 h.
otherwise.
As regards healthcare workers, ICU staff were screened
twice during the outbreak (Fig. 2): in wave 1, a pharyngeal
swab was collected, and in wave 3, pharyngeal and rectal swabs
were collected.

FIG. 1. Plan of the ICU. Grey square:
sink. Grey circle: wastepipe. Discontinous
line: drainpipe. S: sink. W: wastepipe.

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

E492 Clinical Microbiology and Infection, Volume 19 Number 11, November 2013 CMI

FIG. 2. Synoptic curve of patients colonized and/or infected due to multidrug-resistant Klebsiella oxytoca and control measures. ES (in red):
environmental study. SS (in green): staff study. PCLIP (in blue): Prevention and Control of Legionella Infection Protocol. Discontinuous purple arrow:
basic cross-transmission control measures. Continuous orange arrow: additional cross-transmission control measures. Circle with cross: admission
date of patient. Open circle: discharge date of case. Black circle: date of death of case patient. Grey square: date of first isolation of MDRKO.
Discontinous line: ICU stay. Black square: positive environmental culture. White square: negative environmental culture.

Antimicrobial therapy consumption environmental samples were studied. Screening samples were
With the purpose of investigating the consumption of the main seeded onto MacConkey agar plates (Difco, Detroit, MI, USA)
group of antibiotics in the ICU, they were grouped as follows: and chromID ESBL (BioM erieux, Marcy L’Etoile, France). From
third-generation cephalosporins (ceftriaxone, cefotaxime and September 2010, environmental samples were also cultured in
ceftazidime); fourth-generation cephalosporins (cefepime); thioglycolate broth (Difco). Identification and susceptibility
fosfomycin; piperacillin/tazobactam; carbapenems (imipenem); testing was carried out first with an automated system
glycopeptides (vancomycin); aminoglycosides (amikacin, tob- (MicroScanâ; Siemens Healthcare Diagnostics, West Sacra-
ramycin and gentamicin); tigecycline; and fluoroquinolones mento, CA, USA). The characterization of the first nine
(ciprofloxacin and levofloxacin). Consumption was measured outbreak isolates, obtained between March and August 2009,
using defined daily doses (DDD) [14] per 100 patient-days. was previously reported [11]. In summary, the isolates showed
intermediate susceptibility or resistance to all b-lactams tested
Microbiological studies (the MICs of imipenem, ertapenem and meropenem were
All MDRKO isolates (i. e. isolates showing resistance to 2 mg/L, 1–2 mg/L and 0.5–1 mg/L, respectively), and resis-
carbapenems) obtained from clinical, surveillance and/or tance to ciprofloxacin, trimethoprim-sulpamethoxazole and

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

CMI Vergara-Lo´ pez et al. Sinks perpetuate an outbreak due to carbapenem-resistant Klebsiella oxytoca E493

TABLE 1. Summary of environmental studies and actions

Environmental Wave month/year Areas investigated Result Actions
study

1 1 April/2009 Medical equipment, medical ventilators, Negative None

oxygen and air vents in walls, faucets and

fibreoptic bronchoscope.

2 2 November/2009 Enteral nutrition preparations, telephones, Negative None

tables, computer keyboards, sink surfaces,

monitors, portable medical ventilators,

electrocardiographs and portable

X-ray equipment.

3 3 September/2010 Environment surrounding a patient, including MBLKO isolated from a On 10 September 2010, all

instrument cases, tables, chairs, monitors urinary catheter and a stethoscopes were investigated.

and all sink basins. stethoscope around a case. None showed a positive culture

4 3 February/2011 All sink drainpipes and traps MBLKO isolated from On 25 February 2011, S6 and its

S6 drainpipes and trap drainage system were permanently

removed. Drainage system of

S7 was replaced

5 3 April/2011 Several surfaces of cubicle 5, the storage Negative None

area where S6 had been installed, all

sink traps and main wastepipes

6 4 October/2011 S5 (water from faucets, basin, overflow Negative S5 and S7 drainpipes connected

and drainage grille, drainpipes and traps) to W4. Installation of shut-off

valves in the drainpipe of every

sink to carry out biweekly chemical
cleaning with Biguanidâa

S5-6-7: sink numbers 5-6-7 in Figure 1. W4: wastepipe 4 in Figure 1.
aAlthough the environmental study was negative, a secondary environmental focus was suspected and it was decided to implement these measures (see text).

tobramycin; only fosfomycin, colistin and amikacin were active. MDRKO (Fig. 2); of these, 14 (33.3%) developed an infection
The isolates were closely related by pulsed-field gel electro- (Table 2). The crude mortality of colonized/infected patients
phoresis (PFGE), and were IMP-8 metallo-b-lactamase pro- was 44%. The outbreak evolved in four waves (Fig. 2 and
ducers and chromosomal OXY b-lactamase hyperproducers Table 2): from March to May 2009 (nine cases), from
[11]. K. oxytoca isolated after August 2009 was considered as September 2009 to April 2010 (17 cases), from September
belonging to the outbreak clone if it shared the same 2010 to April 2011 (12 cases), and from September to
susceptibility profile; PFGE was also performed on selected November 2011 (four cases). The incidence density of
isolates, including all environmental ones, for confirmation. colonization/infection due to MDRKO decreased during
Additionally, the susceptibility profiles of all K. oxytoca clinical successive waves from 1.91 (wave 1) to 1.24 (wave 2) and
isolates between February 2008 and February 2009 were 0.82 (wave 3) cases per 100 patient-days. The percentage of
retrospectively investigated. cases that were detected by means of a rectal swab increased
from 55.5% in wave 1 to 88.2% in wave 2 and 83.3% in wave 3
Statistical analysis (p 0.02). The average time between admission and acquisition
Categorical variables are expressed as percentages and of MDRKO was 8 days (IQR, 6–37), 16 days (12–27) and 14
continuous variables as medians (interquartile range, IQR). days (9–40) in waves 1, 2, and 3, respectively (p 0.22).
We performed a chi-square test for trend for the incidence
density of colonization/infection by the epidemic strain and for K. oxytoca isolates before the onset of the outbreak
the percentage of cases detected by means of a rectal swab. The bimonthly average number of K. oxytoca isolated in the
The Kruskal-Wallis test was used to analyse the time between whole hospital during 2008 was 2.08 cases, but during
admission and acquisition of MDRKO, over the first three January–February 2009, just before the outbreak started,
waves of the outbreak. Wave 4 was not included in the eight K. oxytoca strains were isolated from different patients.
comparisons because of the low number of cases that Of these, three were isolated from ICU patients; one of
occurred in this last wave. We also studied the trend in them, isolated from a patient who was admitted to the ICU
antimicrobial consumption before, during and after the from 10 December 2008 to 5 February 2009, was an
outbreak. Epi info version 3.5.1 was used. ESBL-producer but was fully susceptible to carbapenems.
These isolates were not available for further microbiological
Results studies.

Overall, 42 (6.4%) of the 660 patients admitted to the ICU Antimicrobial consumption
during the outbreak period were colonized/infected by As shown in Fig. 3, the use of fluoroquinolones, fosfomycin
and aminoglycosides significantly increased after the onset of

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

E494 Clinical Microbiology and Infection, Volume 19 Number 11, November 2013 CMI

TABLE 2. Main characteristics of the patients colonized and/or infected due to multidrug-resistant Klebsiella oxytoca

Case number/wave Cubicle Sex/Age Type of Date sample Type of Pattern of Discharge
admission (month/day/year) sample acquisition status

1/1a 8 M/53 Medical 03/13/09 TA Colonization Alive
2/1a
3/1a 7 M/66 Medical 03/16/09 TA Infection (VAN) Dead

4/1a 5 M/75 Medical 03/18/09 TA/RS Colonization Dead

5/1a 1 M/75 Surgical 03/24/09 RS Colonization Dead

6/1a 5 M/37 Medical 04/04/09 TA Infection (VAN) Dead

7/1a 3 M/42 Medical 04/05/09 TA Colonization Alive

8/1a 3 M/45 Medical 05/04/09 TA/RS Colonization Alive

9/1a 6 M/80 Medical 05/10/09 PS Colonization Alive

6 M/62 Medical 05/20/09 RS Colonization Alive

10/2 7 M/79 Surgical 09/07/09 B Infection (bacteraemia) Dead

11/2 8 M/47 Medical 09/29/09 TA Colonization Dead

12/2 7 M/63 Medical 10/02/09 TA/RS Infection (VAN) Alive

13/2 4 M/83 Medical 10/08/09 RS Colonization Dead

14/2 5 F/78 Medical 10/15/09 TA/RS Colonization Dead

15/2 6 F/71 Medical 10/27/09 RS Colonization Alive

16/2 1 F/41 Medical 11/03/09 RS Colonization Alive

17/2 5 M/67 Medical 11/10/09 TA/RS Infection (VAN) Alive

18/2 4 M/77 Medical 11/17/09 TA/RS/U Urine tract infection Dead

19/2 8 F/79 Surgical 11/17/09 RS Colonization Dead

20/2 8 F/52 Medical 12/02/09 TA/RS/B Infection (bacteraemia) Alive

21/2 8 F/80 Medical 12/03/09 TA/RS/B Infection (bacteraemia) Dead

22/2 1 F/50 Medical 12/16/09 TA/RS/B Infection (bacteraemia) Dead

23/2 2 F/61 Medical 01/26/10 RS Colonization Alive

24/2 3 M/74 Surgical 03/03/10 RS Colonization Alive

25/2 5 F/82 Medical 03/16/10 RS Colonization Dead

26/2 6 F/66 Medical 04/20/10 RS Colonization Alive

27/3 4 F/80 Medical 09/01/10 AS Colonization Alive

28/3 4 F/76 Medical 08/23/10 TA Infection (VAN) Dead

29/3 6 M/58 Surgical 08/27/10 RS Colonization Alive

30/3 1 M/75 Medical 10/06/10 RS Colonization Alive

31/3 2 M/79 Medical 10/26/10 RS Colonization Alive

32/3 1 F/59 Medical 12/09/10 RS Colonization Dead

33/3 5 M/68 Medical 01/12/11 TA/RS Infection (VAN) Dead

34/3 1 M/47 Medical 02/01/11 RS Colonization Alive

35/3 2 F/62 Medical 02/01/11 TA/RS Infection (VAN) Dead

36/3 7 M/70 Medical 02/22/11 RS Colonization Alive

37/3 5 M/57 Medical 03/15/11 RS Colonization Dead

38/3 5 M/76 Medical 04/19/11 RS Colonization Alive

39/4 5 F/76 Medical 09/03/11 Ascites Infection (peritonitis) Alive

40/4 3 F/69 Medical 09/20/11 TA Colonization Alive

41/4 7 F/63 Medical 09/23/11 TA/RS Colonization Alive

42/4 NA M/68 Surgical 11/04/11 B Infection (bacteraemia) Alive

M, man; F, female; TA, tracheal aspirate; RS, rectal swab; PS, pharyngeal swab; B, blood culture; U, urine culture; AS, axilla swab; VAN, ventilator-associated pneumonia; NA, not
applicable.
aPreviously published data [11].

FIG. 3. Consumption of antimicrobial
before, during and after the outbreak in the
intensive care unit, Hospital La Merced,
Osuna, Seville, Spain. DDD: defined daily
dose. *p <0.05. Aminoglycosides: amikacin,
tobramycin and gentamycin. Carbapenems:
imipenem. Fluoroquinolones: ciprofloxacin
and levofloxacin. Third-generation cephalo-
sporins: ceftriaxone, cefotaxime and ceftaz-
idime. Fourth-generation cephalosporins:
cefepime. Glycopeptides: vancomycin.

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

CMI Vergara-Lo´ pez et al. Sinks perpetuate an outbreak due to carbapenem-resistant Klebsiella oxytoca E495

the outbreak (the latter two were used to treat some patients painted, medications removed and fomites (medication vials,
infected with MDRKO). In contrast, the use of third-gener- gloves and containers) eliminated. On 4 June 2010, the annual
ation cephalosporins decreased. During the outbreak there PCLIP was carried out in the hospital. The outbreak was again
was a non-significant reduction in the consumption of carba- considered to be over.
penems and a non-significant increase in the consumption of
piperacillin-tazobatam and tigecycline. Wave 3. In late August 2010, after 4 months with no cases,
MDRKO was again isolated from the clinical sample of a
Description of the outbreak and control measures patient who had been admitted for only 4 days. It was noticed
Wave 1. On 19 March 2009, two MDRKO strains were that a patient, who was then in the Internal Medicine ward, had
isolated from the clinical samples of two patients admitted to previously spent a lengthy period in the ICU; this patient was
adjacent ICU cubicles (7 and 8). Immediately, an infection screened and detected as colonized, and thus was considered
control task force was formed. The basic control measures the probable index case for the third wave. All basic and
undertaken, which were repeated during later waves (Fig. 2), additional measures were again implemented. ICU staff were
included: (i) active screening of all patients admitted to the screened again (SS2) and all cultures were negative. During the
ICU; (ii) contact precautions in individual cubicles for colo- third environmental study (ES3), a urinary catheter removed
nized/infected patients; (iii) reinforcement of standard from a colonized patient and a stethoscope used with that
cross-transmission control measures; (iv) periodic educational patient yielded MDRKO. Because in February 2011 the
sessions; and (v) implementation of twice daily thorough outbreak was still out of control, a fourth environmental
cleaning. All surfaces except medical devices were cleaned with study (ES4), which included sinks drainpipes and traps, was
bleach (1:10 dilution of 5.25% sodium hypochlorite). Medical carried out. Only samples from S6 were positive, showing
devices were cleaned with Biguanidâ (first generation quater- countless colonies of MDRKO that were cultured from every
nary ammonium) at 1.6%. These disinfectants were tested in pipe, trap and drainage grille sample taken; samples from the
vitro at the used concentrations and showed inhibition of faucet or overflow grille were negative. Samples from the pipe
MDRKO growth (data not shown). The first nine isolates were connecting S6 and S7 were also positive. On 25 February 2011,
submitted to a reference laboratory (Hospital Universitario S6 and its drain system were permanently removed and the
Virgen Macarena, Seville) and the clonal nature of the outbreak drain system of S7 was replaced. However, another two
was shown [11]. All screening samples taken from staff (SS1) patients admitted to adjacent cubicle 5 acquired MDRKO in
and the environment (ES1) were negative (Fig. 2). The PCLIP March and April 2011 (18 and 53 days after the environmental
was carried out in the hospital on 5 June. In July 2009, the last intervention). A fifth environmental study (ES5) was carried
MDRKO-colonized patient still remaining in the ICU in this out, including surfaces in cubicle 5, the adjacent storage area
wave was transferred to another hospital; as no new cases had where S6 used to be and all the sink traps and main wastepipes
been detected by then, the outbreak was thought to have been of the unit (Table 1). The 18 samples taken were negative. No
eradicated. new cases emerged in the following months. On 17 June 2011,
the annual PCLIP was performed and the outbreak was once
Wave 2. In late August 2009, the last affected patient of wave 1 again considered eradicated.
was readmitted to the ICU. Although contact precautions
were implemented from admission, MDRKO was isolated Wave 4. On 3 September 2011, MDRKO was again isolated
from the blood culture of another patient in early September, from the clinical sample of a patient admitted to cubicle 5. All
and cross-transmission from the previous patients was patients admitted were screened again; two further colonized
suspected. All basic measures were immediately reactivated; cases were detected. A sixth environmental study (ES6) was
however, new cases emerged. In November 2009, expert undertaken, involving 11 samples taken from S5 (faucets,
external advisors from the Hospital Universitario Virgen surfaces, overflow hole, drainage grilles, drainpipes and trap);
Macarena recommended ‘additional measures’ that were all were negative for MDRKO. In spite of that and because of
implemented, including: nurse cohorting; establishment of the high suspicion of a hidden reservoir, in October 2011 the
minimum nurse/patient and auxiliary/patient ratios of 1:2 and infection-control task force decided to isolate W5, which S5
1:2.5, respectively; review of the clean/dirty circuit; and and S7 still drained into. Thus, the complete horizontal
reviewing the use of broad spectrum antibiotics. A second drainage system of S5 and S7 was replaced and connected up
environmental study (ES2) was performed, with negative to W4. Shut-off valves were also installed to each sink drainage
results. In April 2010, the last case in this wave was detected. system. Since then, a disinfection of the drainage system is
The ICU and the adjoining room were left empty, walls were performed twice a week using Biguanidâ at 1.6% for 30 min

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

E496 Clinical Microbiology and Infection, Volume 19 Number 11, November 2013 CMI

(through closing the valves), followed by opening the valves the first stage of outbreaks caused by MDR bacteria if such
and running hot water (70°C) for 5 min. On 4 November clones eventually acquire MDR genes [19]. We also hypoth-
2011, MDRKO was isolated from blood cultures in a patient esize that the gene codifying for the IMP-8 was introduced into
admitted to the surgical ward. This patient and the index case the ICU by a colonized patient.
for wave 4 who was discharged from the ICU on 25
September 2011, shared healthcare staff in the surgical ward; Colonized patients are the most important reservoir and
therefore it was assumed that he had acquired MDRKO during patient-to-patient cross-transmission is considered to be the
his stay in the surgical ward. Three and 6 months after the end main mechanism of spread in outbreaks of nosocomial
of the outbreak, unannounced transversal screening studies of infection caused by multidrug-resistant Enterobacteriaceae [9].
both patients and the environment were carried out, and were However, it is recognized that standard infection control
negative. Screening of patients with a >1 month ICU stay was measures are frequently not enough to eradicate outbreaks
resumed. No new strains of K. oxytoca have been detected in [4,6,20]. In our case, basic control measures failed and
the hospital, as of April 2013. additional measures including nurse cohorting (which was
effective in other similar outbreaks [4,6]) had to be imple-
Concerning the isolates not included in the earlier report mented. Nevertheless, the epidemic strain was not eradicated
[11], their susceptibility profiles were identical to the previous and the last waves occurred long after the last colonized
ones. blaIMP-8 was detected by PCR in all 17 isolates from wave patients from previous waves had been discharged. This
2, and they all showed an identical PFGE profile to the previously observed phenomenon [3,20] may raise the suspi-
epidemic strain from the first wave. Selected isolates from cion of an unrecognized human or environmental reservoir
waves 3 and 4 and all the environmental samples were studied that enabled the outbreak strain to survive in spite of the
for the presence of blaIMP-8 and molecular relatedness by PFGE preventive measures to stop cross-transmission. Nevertheless,
profile. Every strain studied carried blaIMP-8 and they showed staff and patient surveillance studies did not identify any human
the same PFGE profile as previous isolates. long-term carrier.

Discussion Identifying potential environmental reservoirs has often
been neglected, so that few authors to date have reported
We described one of the most prolonged nosocomial environmental studies when describing earlier outbreaks
outbreaks due to carbapenemase-producing Klebsiella spp., [1,3,4,6]. There are no standardized recommendations about
which was eradicated, that to our knowledge has been when, where and how environmental sampling should be
described to date [1,3–8,15–18]. Since the outbreak was due performed. Many of the environmental studies reported have
to a clonal strain in a small unit that rarely admits patients from in fact usually been carried out on dry surfaces [6].
other centres, it represented an excellent model for evaluating MDR-resistant Klebsiella pneumoniae has been recovered from
the complex evolving nature of its epidemiology. Our results beds and various medical devices [1,6] and even from
show that control measures aimed at preventing cross-trans- contaminated roll boards [16]. K. oxytoca has also been
mission were partially effective but were unable to definitively recovered from ventilator surface cultures [21]. We also
eradicate the outbreak strain; this, together with the epide- found the MDRKO outbreak strain on medical instruments in
miological data, strongly suggests the key role of an environ- the vicinity of an affected patient. However, all isolates taken
mental reservoir, at least during the later waves of the from dry surfaces could merely reflect breaks in standard
outbreak. control measures. A stable reservoir could also be established
in a moist environment where suitable conditions might favour
The outbreak strain was previously characterized as the the formation of microbial biofilms [20–24]. We found an
first IMP-8-producing Enterobacteriaceae in Spain [11]. We do environmental reservoir of this kind in the trap and pipes of S6
not know how the carbapenemase gene entered the hospital. during wave 3 (Figs 1 and 2). Because of the results of
Because the number of non-MDR K. oxytoca isolates dramat- environmental cultures and the association of the cases during
ically increased just before the outbreak, it may be hypothe- that wave with cubicle 5, we think that a change in the
sized that the epidemic clone was spreading before acquiring epidemiology of the outbreak occurred, evolving from a
the IMP-8 gene, which probably increased the chances of predominantly patient-related reservoir during the first wave
spreading in the context of high antibiotic use; unfortunately, (although we cannot discard water drain system involvement)
K. oxytoca strains isolated before the outbreak were not to an environmental one.
available for typing. This hypothesis follows the theory that
spread of ‘susceptible’ (or not MDR) clones may sometimes be Importantly, contaminated sinks and drainage systems are
becoming more frequently identified as relevant reservoirs of
MDR Gram-negative bacteria, including Acinetobacter baumannii

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

CMI Vergara-Lo´ pez et al. Sinks perpetuate an outbreak due to carbapenem-resistant Klebsiella oxytoca E497

[23], Pseudomonas aeruginosa [24] and recently also ESBL-pro- lusian Society of Infectious Diseases), 13th–15th December
ducing K. oxytoca [20]. Contaminated water (i.e. used for staff 2012, Seville (Spain).
hand washing, hygiene of patients or washing devices) is
drained through the sinks; biofilm-forming bacteria may form Transparency Declaration
stable reservoirs in the waste pipes and even in the semi-hor-
izontal drain pipes if they are not inclined enough or are This work was supported by the Ministerio de Econom ıa y
partially blocked. Water splashing from the faucets creates an Competitividad, Instituto de Salud Carlos III – co-financed by
aerosol effect from the sink’s drain, which may later contam- European Development Regional Fund ‘A way to achieve
inate the basin and surrounding surfaces [23,24]. In our Europe’ ERDF, Spanish Network for the Research in Infectious
outbreak, removing the S6 and S7 drainage system failed to Diseases (REIPI RD12/0015). The funders had no role in the
completely eradicate the outbreak strain; even though we had design, analysis or writing of the manuscript, or the decision to
eliminated the S6 reservoir, S5 remained connected to the S6 publish.
pipe and to W5 (Fig. 1), and new cases occurred in cubicle 5. It
was only when the horizontal drainage system to S5 was A . Pascual has been a consultant for Merck and Pfizer, has
removed and S5 and S7 were connected to W4 that the served as speaker for Astra-Zeneca, Merck and Pfizer and has
outbreak was finally brought under control. We think that S5 received research support from Merck and Pfizer. J.
was also a reservoir that could not be detected. Rodr ıguez-Ban~o has been a consultant for Merck, Pfizer and
Roche, has served as a speaker for Merck, Pfizer, Astra-Zeneca
Finally, an unintended action may have played an important and Astellas, and has received research support from Merck
role in the epidemiology of the outbreak. In the first three and Novartis. All other authors have no conflicts of interest to
waves, the emergence of new cases was stopped, coinciding declare.
with the application of the PCLIP, only to remerge a few
months later (Fig. 2). This would suggest that the PCLIP may References
have had a significant but not decisive impact on the
contaminated drainage systems, perhaps because it was unable 1. Souli M, Galani I, Antoniadou A et al. An outbreak of infection due to
to completely eradicate the bacterial biofilms. Chemical beta-lactamase Klebsiella pneumoniae carbapenemase 2-producing
disinfection of the drainage system has been previously K. pneumoniae in a Greek University Hospital: molecular characteriza-
reported to have had little impact on the bacterial burden in tion, epidemiology, and outcomes. Clin Infect Dis 2010; 50: 364–373.
attempts to solve similar problems [20,24].
2. Patel G, Huprikar S, Factor SH, Jenkins SG, Calfee DP. Outcomes of
In conclusion, the epidemiology of outbreaks due to carbapenem-resistant Klebsiella pneumoniae infection and the impact of
carbapenemase-producing Enterobacteriaceae may be complex antimicrobial and adjunctive therapies. Infect Control Hosp Epidemiol
and evolving; even when the main reservoirs are formed by 2008; 29: 1099–1106.
colonized patients, and cross-transmission is the principal
means of their spread, alternative reservoirs should be 3. Steinmann J, Kaase M, Gatermann S et al. Outbreak due to a Klebsiella
suspected if strictly applied traditional control measures are pneumoniae strain harbouring KPC-2 and VIM-1 in a German university
not efficacious. In these circumstances, a wet environmental hospital, July 2010 to January 2011. Euro Surveill 2011; 16: 19944.
reservoir (mainly the drain, trap and horizontal drainage
system of a sink) should be considered. Finally, initiatives 4. Kassis-Chikhani N, Saliba F, Carbonne A et al. Extended measures for
standardizing environmental investigations in healthcare set- controlling an outbreak of VIM-1 producing imipenem-resistant
tings should be encouraged. Klebsiella pneumoniae in a liver transplant centre in France, 2003–
2004. Euro Surveill 2010; 15: 19713.
Acknowledgements
5. Tokatlidou D, Tsivitanidou M, Pournaras S et al. Outbreak caused by a
We thank the staff from the Preventive Medicine, Laboratory, multidrug-resistant Klebsiella pneumoniae clone carrying blaVIM-12 in a
Pharmacy, Maintenance and Cleaning services as well as the university hospital. J Clin Microbiol 2008; 46: 1005–1008.
Direction Team and the entire UCI staff of La Merced
Hospital, for their efforts and contribution in tackling the 6. Borer A, Eskira S, Nativ R et al. A multifaceted intervention strategy
outbreak. for eradication of a hospital-wide outbreak caused by carbape-
nem-resistant Klebsiella pneumoniae in Southern Israel. Infect Control
The results of this paper have been presented as a poster Hosp Epidemiol 2011; 32: 1158–1165.
communication to the regional XIV congress of SAEI (Anda-
7. Tato M, Coque TM, Ru ız-Garbajosa P et al. Complex clonal and
plasmid epidemiology in the first outbreak of Enterobacteriaceae
infection involving VIM-1 metallo-beta-lactamase in Spain: toward
endemicity? Clin Infect Dis 2007; 45: 1171–1178.

8. Gaibani P, Ambretti S, Berlingeri A et al. Rapid increase of carbape-
nemase-producing Klebsiella pneumoniae strains in a large Italian
hospital: surveillance period 1 March–30 September 2010. Euro Surveill
2011; 16: 19800.

9. Centers for Disease Control and Prevention. Guidance for control of
infections with carbapenem-resistant or carbapenemase-producing

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

E498 Clinical Microbiology and Infection, Volume 19 Number 11, November 2013 CMI

Enterobacteriaceae in acute care facilities. MMWR Morb Mortal Wkly Rep acquisition among hospitalized adults and effect of acquisition on
2009; 58: 256–260. mortality. Antimicrob Agents Chemother 2008; 52: 1028–1033.
10. Bratu S, Landman D, Haag R et al. Rapid spread of carbapenem-resis- 18. Wiener-Well Y, Rudensky B, Yinnon AM et al. Carriage rate of
tant Klebsiella pneumoniae in New York City: a new threat to our carbapenem-resistant Klebsiella pneumoniae in hospitalised patients
antibiotic armamentarium. Arch Intern Med 2005; 165: 1430–1435. during a national outbreak. J Hosp Infect 2009; 74: 344–349.
11. Conejo MC, Dom ınguez MC, Lo pez-Cerero L, Serrano L, 19. Mart ınez JL, Baquero F. Interactions among strategies associated with
Rodr ıguez-Ba~no J, Pascual A. Isolation of multidrug-resistant Klebsiella bacterial infection: pathogenicity, epidemicity, and antibiotic resistance.
oxytoca carrying blaIMP-8, associated with OXY hyperproduction, in Clin Microbiol Rev 2002; 15: 647–679.
the intensive care unit of a community hospital in Spain. J Antimicrob 20. Lowe C, Willey B, O’Shaughnessy A et al. Outbreak of Extended-Spec-
Chemother 2010; 65: 1071–1073. trum b-Lactamase-producing Klebsiella oxytoca infections associated
12. Stone SP, Cooper BS, Kibbler CC et al. The ORION statement: with contaminated handwashing sinks. Emerg Infect Dis 2012; 18: 1242–
guidelines for transparent reporting of outbreak reports and interven- 1247.
tion studies of nosocomial infection. J Antimicrob Chemother 2007; 59: 21. Schulz-St€ubner S, Kniehl E. Transmission of extended-spectrum
833–840. b-lactamase Klebsiella oxytoca via the breathing circuit of a transport
13. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition ventilator: root cause analysis and infection control recommendations.
of health care-associated infection and criteria for specific types of Infect Control Hosp Epidemiol 2011; 32: 828–829.
infections in the acute care setting. Am J Infect Control 2008; 36: 309– 22. Khan AS, Dancer SJ, Humphreys H. Priorities in the prevention and
332. control of multidrug-resistant Enterobacteriaceae in hospitals. J Hosp
14. WHO Collaborating Centre for Drug Statistics Methodology. Guide- Infect 2012; 82: 85–93.
lines for ATC classification and DDD assignment 2013. Oslo, Norway: 23. La Forgia C, Franke J, Hacek DM, Thomson RB Jr, Robicsek A,
WHO Collaborating Centre for Drug Statistics Methodology, 2012. Peterson LR. Management of a multidrug-resistant Acinetobacter
15. Kochar S, Sheard T, Sharma R et al. Success of an infection control baumannii outbreak in an intensive care unit using novel environmental
program to reduce the spread of carbapenem-resistant Klebsiella disinfection: a 38-month report. Am J Infect Control 2010; 38: 259–263.
pneumoniae. Infect Control Hosp Epidemiol 2009; 30: 447–452. 24. Hota S, Hirji Z, Stockton K et al. Outbreak of multidrug-resistant
16. van’t Veen A, van der Zee A, Nelson J, Speelberg B, Kluytmans JA, Pseudomonas aeruginosa colonization and infection secondary to
Buiting AG. Outbreak of infection with a multiresistant Klebsiella imperfect intensive care unit room design. Infect Control Hosp Epidemiol
pneumoniae strain associated with contaminated roll boards in 2009; 30: 25–33.
operating rooms. J Clin Microbiol 2005; 43: 4961–4967.
17. Schwaber MJ, Klarfeld-Lidji S, Navon-Venezia S, Schwartz D, Leavitt A,
Carmeli Y. Predictors of carbapenem-resistant Klebsiella pneumoniae

ª2013 The Authors
Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI, 19, E490–E498

The Art and Science of Infusion Nursing

Understanding the Economic Impact of Health
Care-Associated Infections: A Cost Perspective
Analysis

R. Douglas Scott II, PhD Steven D. Culler, PhD Kimberly J. Rask, MD, PhD

ABSTRACT

The economic impacts from preventing health care-associated infections (HAIs) can differ for patients, health care provid-
ers, third-party payers, and all of society. Previous studies from the provider perspective have estimated an economic bur-
den of approximately $10 billion annually for HAIs. The impact of using a societal cost perspective has been illustrated by
modifying a previously published analysis to include the economic value of mortality risk reductions. The resulting costs to
society from HAIs exceed $200 billion annually. This article describes an alternative hospital accounting framework outlining
the cost of a quality model which can better incorporate the broader societal cost of HAIs into the provider perspective.
Key words: health care-associated infections, health care cost, regulatory impact analysis, value of statistical life

Measuring the cost of health care-associated infec- evidence on cost of HAI infection control and prevention
tions (HAIs) has been a challenge for health econ- efforts was not compelling because of the variety of study
omists and health services researchers, given designs and settings, statistical methods, and cost outcomes
 the lack of detailed hospital cost data needed to (length of stay vs costs) used.1 It also has been noted that
accurately reflect the economic value of the resources used many of the methods used in the measurement of the
in infection control. A 1995 review of the published research attributable cost of HAIs were inappropriate and resulted
on the cost of HAIs to hospitals found that the economic in erroneously high-cost estimates.2 A complicating factor
for any economic analysis of HAIs is that there are divergent
Author Affiliations: Centers for Disease Control and Prevention, perspectives on the cost impacts of HAIs and who pays for
Atlanta, Georgia (Dr Scott); and Emory University, Atlanta, Georgia them.
(Drs Culler and Rask).
To help those who work as health care providers get
R. Douglas Scott II, PhD, is an economist for the Division of a better understanding of the controversies surrounding
Healthcare Quality Promotion at the Centers for Disease Control the measurement of the financial and economic impact
and Prevention. His research focuses on the use of prevention resulting from HAIs, this article will (1) review the economic
effectiveness methods and regulatory impact analyses to evaluate theory underlying divergent cost perspectives; (2) present
programs to prevent health care-associated infections and improve cost estimates for hospital-onset HAIs from varying cost
patient safety. Steven D. Culler, PhD, is a health economist at the perspectives, including the economic burden to the health
Rollins School of Public Health at Emory University. He has pub- care system and excess payments made by insurers such as
lished numerous peer-reviewed articles explaining the variation in Medicare; (3) introduce the societal cost perspective and the
the use of health care services, estimating the incremental cost of use of the value per statistical life (VSL) as used by the federal
treating selected adverse events, and performing cost-effectiveness government for regulatory impact analysis; and (4) present
studies. Kimberly J. Rask, MD, PhD, is a primary care physician an alternative approach to hospital accounting practices that
and health economist, holding joint appointments in health policy broadens the hospital cost perspective to include the costs
and management and medicine at Emory University. Her publi- incurred by providers to avoid downstream patient harm.
cations focus on primary care practice, quality improvement, and
outcomes measurement. She also serves on national expert panels THE ANALYTICAL COST PERSPECTIVE
on value-based purchasing programs and quality measurement.
When a researcher begins an economic evaluation of a
The authors have no conflicts of interest to disclose. The findings health care intervention or policy, he or she must first specify
and conclusions in this report are those of the authors and do not the point of view on which the cost analysis is based. This
necessarily represent the official position of the CDC.

Corresponding Author: R. Douglas Scott II, PhD, Division of
Healthcare Quality Promotion, National Center for Zoonotic and
Emerging Infectious Diseases, Centers for Disease Control and
Prevention, 1600 Clifton Road MS A16, Atlanta, GA 30329-4027
([email protected]).

DOI: 10.1097/NAN.0000000000000313

VOLUME 42 | NUMBER 2 | MARCH/APRIL 2019 journalofinfusionnursing.com 61

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.

decision determines which benefits and costs will be included HAI COSTS FROM THE HEALTH CARE
in the analysis.3 There are 4 cost perspectives to choose from: SYSTEM PERSPECTIVE

1. The health care provider perspective measures the costs The majority of economic evaluations of hospital-onset
of the resources used to prevent and/or treat patients HAIs have been done from the cost perspective of a hospital
while in their care. or health care system and have measured direct medical
treatment costs only. Economic evidence from these studies
2. The patient perspective includes lost work days; out-of- is used to make the “business case” that HAI prevention will
pocket costs for care; pain and suffering; the effect on reduce hospital costs by reducing the incidence of HAIs and
family and/or other caregivers; long-term morbidity; and the associated resources used during the excess length of
mortality. stay.9,10 However, these studies are difficult to conduct using
traditional economic models of production given that hos-
3. The third-party payer perspective (such as Medicare) pital care is a nonstandard production process (eg, treating
considers the excess, or shortfall, in payments made to patients with the same diagnosis may require a different
reimburse providers for the cost of treatment for cov- combination of tests and treatments given other comorbidi-
ered patients. ties). Exacerbating the problem, the variety of cost account-
ing practices used by hospitals results in cost accounting
4. The societal perspective attempts to measure all of the data that measure costs associated with revenue-generat-
costs and benefits accruing to all members of society.4 ing departments (eg, oncology, cardiac care), but lacks the
A review of basic economic principles will help in under- details to measure treatment cost at the patient level.11-13
As a result, most studies of attributable HAI cost use epi-
standing why divergent cost perspectives in health care demiologic methods, usually retrospective cohort studies,
markets exist. The classic market structure taught in basic in which patients are grouped according to whether or not
economic theory courses is the “perfectly competitive mar- they have an HAI. However, the variety of statistical meth-
ket,” in which under the right conditions, buyers and sellers ods used to analyze cohort study data leads to even more
can freely interact to set a market price at which not only variability in HAI estimates of attributable costs.
will the goods sold (by sellers) be equal to the amount of
goods demanded (by buyers), but society will receive the Table 1 shows the results of a single-center study of
greatest value for the resources used.5 There are a number HAI costs for a large, urban teaching hospital where the
of characteristics needed for a perfectly competitive market HAI attributable cost, inclusive of all infection types, was
to function properly, which include: measured using several different estimation approach-
es including generalized linear regression, ordinary least
1. A large number of buyers and sellers willing to buy or sell squares (OLS) regression, OLS regression using a definition
the product at a certain price; of variable cost that included physician salaries and bedside
procedures; OLS models in which extremely small and large
2. Perfect information, ie, all consumers and producers (outlier) cost values had been restricted or winsorized; pro-
know the prices of all products and their value to the pensity scoring; and a 3-state proportional hazards model
consumer; estimation to measure attributable length of hospital stay
(LOS) as a result of an HAI and then multiplying LOS either
3. Homogeneous products, the products made by produc- by the average daily cost of patients with an HAI or the daily
ers, are perfect substitutes for one another; costs of patients overall.14 The different methods resulted
in estimates of attributable cost that ranged from $9,000 to
4. A well-defined system of property rights with no exter- $21,000. This somewhat troubling result shows how much
nalities (the costs and benefits of a transaction affect the selection of the estimation method can influence the
only the well-being of the buyer and seller); and amount of the estimated costs.

5. No barriers to entry or exit.5 Recent efforts to measure the direct medical costs or the
All of these characteristics are lacking in real-world health economic burden of HAIs to the health care system have
been based on literature reviews. Studies by Zimlichman
care markets, but the most critical missing element is perhaps et al15 and the nonpartisan and objective research organi-
perfect information. A classic paper by Kenneth Arrow6 noted zation NORC at the University of Chicago, sponsored by the
that there is “asymmetric” information between physicians Agency for Healthcare Research and Quality (AHRQ), have
and patients. This asymmetry exists because patients rely on produced estimates of the attributable costs of select HAIs
the recommendations of physicians who have the expertise that have traditionally been the focus of surveillance and
and qualifications to diagnose and treat disease. However, prevention efforts (Table 2).15,16 Adjusting Zimlichman and
physicians are able to influence both the level of services that colleagues’ estimates to 2015 dollars (to match the NORC
will be provided and the price patients must pay for it. In the study) using the Bureau of Labor Statistics producer price
economics literature, this is an example of the principal-agent index for general medical and surgical hospitals, there is
problem.7,8 This problem occurs when the physician (the
agent), who has more information, is able to make decisions
on behalf of the patient (or principal), who cannot be sure
the physician is acting in his or her interests. For example,
the physician is paid the same amount whether the physician
chooses an effective (eg, appropriately targeted antibiotic)
or an ineffective (eg, wrong-spectrum antibiotic) treatment.

62   Copyright © 2019 Infusion Nurses Society Journal of Infusion Nursing

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.

TABLE 1 The discussion above illustrates how direct medical
cost estimates for HAIs that are derived from literature
Attributable Cost of a Hospital- reviews are sensitive to the methods used to estimate
Onset Health Care-Associated attributable costs. Readers should be aware of these limita-
Infectiona tions when considering results from any economic burden
study. Additionally, studies of HAI burden of disease have
Method of Measurement Estimated Cost tended to focus on device- or procedure-related infections
per Infection because surveillance systems typically do not track non–
device- or non–procedure-related HAIs in acute care hospi-
Generalized linear regression model $20,888 tals.19 A recent study does provide burden estimates for all
HAI types, but because the study sample of hospitals was
OLS linear regression $19,917 small, there is significant variability in the estimates, adding
even more variability in the economic burden estimates.19
OLS linear regression: total cost minus MD and $18,615
procedures THIRD-PARTY PAYER PERSPECTIVE

Propensity score-matched comparison $19,251 When considering the cost perspective of a third-party
payer, it’s important to distinguish reimbursements from
LOS multiplied by mean HAI cost per day $19,344 direct medical costs. Economic evaluation methods, includ-
ing cost-benefit analysis and cost-effectiveness analysis,
OLS linear regression; 98% Winsorized $15,203 require that the monetary valuation of all inputs associated
with an intervention or program be based on the actual con-
LOS multiplied by mean non-HAI cost per day $15,149 sumption of resources used as program inputs.20 The direct
medical costs of medical treatments and policies reflect the
3S-PHM LOS multiplied by mean HAI cost per day $11,889 economic value of the resources used in the treatment of
patients, such as labor, procedures performed, number of
Quantile linear regression $11,662 laboratory tests, and more. Reimbursements are payments
made to hospitals by third parties, ie, insurance companies,
OLS linear regression; 95% Winsorized $11,299 on behalf of patients who are policy holders. The amount of
these payments has been negotiated between the hospital
3S-PHM LOS multiplied by mean non-HAI cost $9,310 and the insurer, either a private-sector company or a gov-
per day ernment program—ie, Medicare and Medicaid—for their
patients who have purchased coverage. These payments
Abbreviations: 3S PHM, 3-state proportional hazard model; HAI, health can be referred to as transfer payments because they do not
involve any additional consumption of medical resources.21,22
care-associated infection; LOS, length of hospital stay; MD, medical doctor; OLS,
Because transfer payments are not costs, the impacts
ordinary least squares. of HAIs on third-party payer reimbursements would not be
aInclusive of all infection types, both antibiotic susceptible or resistant, by meth- included in a cost evaluation taken from either the health
care provider or the societal cost perspective. However,
od of measurement. insurance companies have an incentive to minimize reim-
Data from Roberts et al.14 bursements—particularly Medicare and its Part A insur-
ance program, which covers hospitalizations—and they can
some agreement in the estimates for central line-associated
bloodstream infections (CLABSIs) and surgical site infections,
while there are differences in the estimates for ventilator-as-
sociated pneumonia, hospital-acquired antibiotic-associated
Clostridium difficile, and catheter-associated urinary tract
infections.17 Zimlichman et al15 estimated the annual eco-
nomic burden to be approximately $10.1 billion in 2015
dollars. While the AHRQ has yet to publish economic burden
estimates using the NORC attributable cost estimates, the
authors’ derivation of annual economic burden of these
select HAIs using the NORC attributable cost estimates,
along with 2014 burden estimates from the Partnership for
Patients program, suggests a burden of approximately $9.3
billion in 2015 dollars.18

TABLE 2

Estimates of Attributable HAI Cost Estimates From Literature Reviews

HAI Type Zimlichman et al15 NORC Report16

Catheter-associated urinary tract infections $924 $13,793

Central line-associated bloodstream infections $47,254 $48,108

Surgical site infections $21,438 $28,219

Ventilator-associated pneumonia $41,406 $47,238

Hospital-acquired antibiotic-associated Clostridium difficile $11,640 $17,260

Abbreviation: HAI, health care-associated infection; NORC, the nonpartisan and objective research organization NORC at the University of Chicago.

VOLUME 42  |  NUMBER 2  |  MARCH/APRIL 2019 j o u r n a l o f i n f u s i o n n u r s i n g . c o m    63

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.

conduct economic evaluations of HAI prevention programs TABLE 3
to determine any effect on their budgets. Researchers at
the Centers for Disease Control and Prevention (CDC) have Societal Cost of Hospital-Acquired
conducted a number of studies that take the third-par- Infectionsa
ty payer perspective, specifically Medicare. Results from
these studies have found that (1) the excess Medicare Category Societal Societal
reimbursement for a CLABSI in intensive care units (ICUs) Direct costs (billions, 2010) Low High
was $25,000; (2) the excess Medicare reimbursement for   Index hospitalization
a catheter-associated urinary tract infection was $8,500 in   Professional fees index hospitalization $24.8 $53.9
ICU patients and $1,500 for patients in non-ICU wards; and   Postdischarge outpatient $4.9 $13.2
(3) implementation of a multifaceted infection control and   Readmission postindex hospitalization $0.2 $0.2
antibiotic stewardship program to prevent C difficile is pre-   Professional fees readmission $3.4 $4.0
dicted to save $2.5 billion in Medicare reimbursements over   Postdischarge diagnosed infection $0.7 $1.0
a 5-year period.23-25 Consistent with the third-party payer  Subtotals $0.3 $1.7
cost perspective, cost impacts on patients (ie, additional Indirect costs (billions, 2010) $74.0
out-of-pocket costs, lost wages, productivity losses, etc) are   Lost wages, incapacitation ($149 a day) $34.3
ignored in these studies.   Lost future wages, premature death $3.9
$2.5 $68.7
SOCIETAL COST PERSPECTIVE AND VSL  ($685,225) $59.1 $72.6
Subtotals $61.6 $146.6
The societal cost perspective includes the cost and benefits Totals $95.9
to all members of society. It is challenging to measure the
economic impact of long-term sequelae, such as amputa- aData from Marchetti and Rossiter.26
tions or stays in long-term facilities; lost labor productivity;
premature death; intangible costs, such as lost leisure time (low), $9.3 million (central), and $14.2 million (high), based
or disability; and patient out-of-pocket costs. There are on 2014 dollars and income level.28 To calculate the economic
few clinical data on the magnitude of these outcomes for value of mortality risk reductions, we must also have an esti-
patients suffering an HAI. One attempt to measure HAI costs mate of the number of deaths that are attributable or the
from a societal perspective was carried out by Marchetti result of acquiring an HAI. Some studies estimate that, among
and Rossiter.26 They found that the annual direct medical patients with Acinetobacter infection, methicillin-resistant
costs of hospital HAIs ranged from $34 billion to $74 billion, Staphylococcus aureus infection, C difficile infection, and
while the additional social costs of HAIs, which included lost CLABSI, the proportion of mortality attributable to the infec-
wages for incapacitation and premature death, ranged from tion can range from 50% to 70%. There is limited evidence
$62 billion to $73 billion (Table 3). The value of lost wages on the proportion of mortality attributable to other HAIs
was derived from a survey of judgments in liability cases that are not associated with antibiotic resistance or medical
where the value of lost productivity averaged $685,225. devices or procedures.29-32 If the low and high VSL estimates
Adding in these costs increased the total societal cost esti- were substituted in Marchetti and Rossiter’s study26 and it
mate, which ranged from $96 billion to $147 billion a year. was conservatively assumed that the overall proportion of
mortality among patients with HAIs attributable to the infec-
As an alternative to using lost wages and liability judg- tion was 40% (ie, 39 595 of the 98 987 premature deaths
ments in valuing the cost of premature death, federal govern- noted in the study), the additional social benefits would
ment regulatory agencies have a long history of using the VSL range from $174 billion to $562 billion a year, increasing the
in regulatory impact analyses that affect human health and overall societal benefits, including direct costs, to a range of
mortality.27 The VSL is the value an individual will place on a $208 billion to $636 billion (Table 4). The economic value of
marginal change in the likelihood of their death. For example, mortality risk reductions is significantly larger—5 to 7 times
consider a population of 100 000 in which it has been deter- larger—than the reduced direct medical costs alone.
mined that each person would be willing to pay an average
of $50 to reduce their risk of dying by 1/100 000. Here, the A HOSPITAL COST MODEL THAT
VSL is equal to $50 ÷ (1/100 000) or $5 million, which is the INCORPORATES DOWNSTREAM PATIENT
estimated value that society would pay to reduce the risk of IMPACT
1 person dying.27 In practice, the VSL is measured in 2 ways:
(1) wage studies that examine wage differentials for jobs with The full social and economic impact of HAIs on patients
varying levels of job-related risks, or (2) survey methods in can be ignored by hospitals because hospitals are not
which respondents are asked what they would be willing to affected by many of the costs incurred by patients who
pay for changes in their risk of death.

The range of VSL estimates currently used by the
Department of Health and Human Services is $4.4 million

64   Copyright © 2019 Infusion Nurses Society Journal of Infusion Nursing

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.

TABLE 4

Direct Cost of Associated Hospital-Onset HAI and HHS VSL

Measure Number of Attributable Per Patient VSL and Cost Total Costs
Mortality Cases Estimates $174 billion–$562 billion

Value of mortality risk reductionsa 39 595 $4.4 million (low) $34.3 billion–$74 billion
$14.2 million (high) $208.3 billion–$636 billion

Direct medical costsb (from Table 3) 1 453 077 (low) $17,070 (low)
1 676 628 (high) $32,176 (high)

Total societal costs

Abbreviations: HAI, health care-associated infection; HHS, US Department of Health and Human Services; VSL, value of statistical life.
aData from Scott et al.23
bData from Marchetti and Rossiter.26

suffer an HAI. There is a managerial costing framework In this model, when an HAI occurs, the total cost is not
that can better incorporate the downstream economic only the treatment cost of an HAI (internal failure costs) but
consequences faced by patients who contract an HAI. The also the costs associated with appraisal (surveillance and
cost-of-quality (CoQ) model is a framework in which all of management) and prevention (infection control interven-
the resources used to achieve product quality are included, tions). Costs associated with insurance premiums and any
which in this case is quality of care in terms of preventable liability judgments paid by hospitals would also be included
HAIs. The CoQ model makes explicit that all costs associated (external failure costs). Hospitals with poor or excellent
with HAI prevention, including infection control programs, performance in preventing HAIs could also lose or gain mar-
hand hygiene protocols, environmental and housekeeping ket share as the result of the impact on their reputation,
services, and sterilization services, must be counted along which is another external failure cost. Figure 1 presents a
with the additional treatment costs of HAIs. The classical hypothetical distribution of daily hospital costs for patients
model for costing quality is the prevention-appraisal-fail- in the diagnosis-related group (DRG) for those with diabe-
ure (PAF) model.33-35 The model can be expressed more tes and complicating conditions (DRG 638) and an average
formally as (1) CoQ = appraisal costs + prevention costs length of stay of 5 days. Fixed costs (in blue) are charged on
+ internal failure costs + external failure costs. day 1 and day 5, including administrative fees (eg, creating

Figure 1 Hypothetical distribution of daily patient costs.

VOLUME 42  |  NUMBER 2  |  MARCH/APRIL 2019 j o u r n a l o f i n f u s i o n n u r s i n g . c o m    65

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.

Figure 2 Hypothetical distribution of daily hospital costs, including infection prevention and failure costs.

a patient’s record) and services, such as infection control example, the increased investment results in a decrease in
(for day 1), and discharge fees (day 5); daily hospital room average treatment costs on days 4 and 5 ($100 and $150,
and board costs (orange) and daily treatment costs (purple) respectively) and a savings of $250 in hospital room and
vary across the 5 days. Within this DRG patient group, it board, as the average length of stay decreases from 5 days
costs the hospital an average of $6,550 for each patient’s to 4.5 days (Figure 2). The total cost savings of $500, minus
care.36 The average daily cost of $1,310 ($6,550/5) is repre- the additional $250 in increased infection control costs,
sented by the horizontal line. results in an overall decrease in average patient cost of
$250, for a total of $6,300. Now all the average daily costs
In the CoQ model, activity-based accounting practices across the 5 days, which declines to $1,260 ($6,300/5) a
would identify the daily charges associated with infection day, are covered by the average daily DRG payment. In
prevention, including the infection control program, house- 2015, cases designated with DRG 638 received an average
keeping/environmental services, sterilization, and all other total payment of $6,593 (Medicare reimbursement and
prevention activities, as well as the failure costs associated other copayments from patients or other insurers). Before
with treatment of HAIs, insurance premiums, and repu- the investment in infection control, the hospital’s net reve-
tation costs. In Figure 2, the costs of infection prevention nue per patient was $43.36 In this example, the additional
activities plus failure costs are explicitly accounted for on expenditures in infection control resulted in an additional
a daily basis (in yellow). For day 1, fixed costs are now $250 in patient cost savings and increased net per patient
reduced by $500 because the costs related to infection revenue to $293.
control are now allocated on a per day basis ($100) across
the 5 days. To apply the CoQ model, most hospital accounting
systems would have to be redesigned. To start, an activ-
Now it is possible for the accounting system to assess ity-based accounting identity for infection control would
how increasing investments in infection control can pro- have to be created to assess the total cost of an infection
duce savings through reduced treatment costs and hospital control program. An activity-based account for infection
room and board costs. If the hospital increases its invest- control should include most of the budgets for environ-
ment in infection control appraisal and/or prevention pro- mental services (housekeeping); central services and
grams that are both effective and cost saving by 50% ($250 supply; laundry services; the cost of the infection control
or $50 a day), the impact would look like Figure 3. For this

66   Copyright © 2019 Infusion Nurses Society Journal of Infusion Nursing

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.

Figure 3 Hypothetical distribution of daily hospital costs, including new level of investment in infection prevention.

program; labor costs associated with HAI prevention edu- have reduced data needs and have been used to identify
cation, hand hygiene practices, and contact precautions; the actual cost of care for operational improvements,
any capital costs related to infection control, ie, number reducing costs while maintaining quality, and to inform
of airborne infection isolation rooms; insurance pre- reimbursement policy.39-41 In addition, the Centers for
miums and other risk management practices; and any Medicare and Medicaid Services’ (CMS’) efforts to move
other costs associated with materials and supplies used to bundled payments for selected episodes of care will
for current infection prevention interventions. Based encourage hospitals and other providers to enhance their
on the 2015 Medicare hospital cost reports, the annual cost accounting to understand the true resource cost of
budgets for housekeeping, laundry, and central services providing care for selected types of patients.
averaged $1,730,000, $433,000, and $2,370,000, respec-
tively.37 From a 2008 estimate, the average annual cost Since 2015, hospitals have been subject to financial
of a hospital infection control program, when adjusted to penalties from Medicare through the Hospital-Acquired
2015 dollars, is approximately $288,000.23,38 Adding these Condition (HAC) Reduction Program.42 Under this program,
expenditures, the annual cost of infection control for an Medicare reduces total inpatient reimbursements by 1% for
average hospital in 2015 is approximately $4.8 million. the lowest-performing 25% of hospitals, based on a set of
HAC measures that includes HAIs.43 In fiscal year 2016, 769
A challenge with activity-based costing is that the data hospitals were penalized an average of $560,000 under this
needs for a hospital’s accounting system are greater, and program.44 Given the potential benefits to patients, fami-
these systems are costlier to implement and maintain lies, the health system, and society from HAI prevention,
administratively. For example, to understand the labor the costs of improved accounting practices are trivial.
costs associated with hand hygiene, a hospital must mea-
sure the labor time needed to comply with the hospital’s CONCLUSION
hand hygiene guidelines across the spectrum of employees
required to comply. Any purchases of supplies and mate- Understanding the cost perspective used in economic studies
rials for infection control must be transferred from the of HAIs is critical to interpreting the results. Results from eco-
facilities and maintenance budget to the infection control nomic studies of HAI prevention programs using the hospital
activity. However, new modifications to the activity-based perspective have produced attributable cost estimates that
accounting model (ie, time-driven, activity-based costing)

VOLUME 42  |  NUMBER 2  |  MARCH/APRIL 2019 j o u r n a l o f i n f u s i o n n u r s i n g . c o m    67

Copyright © 2019 Infusion Nurses Society. Unauthorized reproduction of this article is prohibited.