Characterizing the ESBL-Producing Enterobacterales Bioburden in a Neonatal Unit Using Chromogenic Culture Media: A Feasible and Ecient Environmental Sampling Method

Infections due to extended spectrum beta-lactamase producing Enterobacterales (ESBL-PE) have emerged as the leading cause of sepsis among hospitalized neonates in Botswana and much of sub-Saharan Africa and south Asia. Yet, ESBL-PE reservoirs and transmission dynamics within the neonatal intensive care unit (NICU) environment are not well-understood. This study aimed to assess the eciency and feasibility of a chromogenic-culture-media-based environmental sampling approach to characterize the ESBL-PE bioburden within a NICU. A series of four surveys were conducted from Samples were collected on 3 occasions under semi-sterile technique using 1) ocked swabs & templates (at surfaces); 2) sterile syringe & tubing (water aspiration); and 3) structured swabbing techniques (hands & equipment). Swabs were transported in physiological saline-containing tubes, vortexed, and 10 µL was inoculated onto chromogenic-agar that was selective and differential for ESBL-PE (CHROMagar™ ESBL, Paris, France), streaking plates to isolate individual colonies. Bacterial colonies were quantied and phenotypically characterized using biochemical identication tests.


Introduction
In recent decades, infections due to multidrug resistant (MDR) enteric organisms have emerged as the leading causes of neonatal sepsis in sub-Saharan Africa and south Asia. (1)(2)(3) In Botswana, the most common causes of laboratory-con rmed bloodstream infection (BSI) among hospitalized neonates are Enterobacterales, most commonly Klebsiella species.(4) Over 80% of K. pneumoniae neonatal bloodstream isolates in this setting are reported to be extended spectrum beta-lactamase-producing Enterobacterales [1] (ESBL-PE).(4) Carbapenem-resistant Enterobacterales (CRE) are also emerging as a cause of neonatal sepsis in southern Africa. (5,6) Infections due to CRE and ESBL-PE are di cult to treat and confer a high mortality risk in neonates; over one in three neonates with an ESBL-PE bloodstream infection will die. (1,7) Hyper-endemic rates of neonatal sepsis in sub-Saharan Africa and south Asia are thought to be driven by overcrowded neonatal wards, barriers to effective hand hygiene, equipment re-use, and limited laboratory capacity to detect and respond to outbreaks. Infection prevention efforts are often thwarted because the ecology and transmission dynamics of these organisms within the neonatal intensive care unit (NICU) are not well understood. Enterobacterales are well suited to survive in moist and warm settings and emerging data suggest that damp reservoirs within the hospital environment, such as sink drains, washbasins, and oxygen humidi ers, could contribute to the acquisition of ESBL-PE among patients. (8-10) However, little is known about potential ESBL-PE reservoirs speci cally within NICUs and whether they contribute to neonatal colonization and disease.
Environmental surface sampling in healthcare facilities is warranted to identify reservoirs and vehicles for clinically important pathogens. (11) However, infection control teams in resource-limited settings often lack the conventional equipment (e.g. sponge sticks, extractors, etc.) and laboratory capacity to conduct a meaningful investigation. Moreover, environmental sampling data can be di cult to interpret and is often skewed toward recovery of Gram positive organisms with unclear clinical relevance. (12) Environmental sampling techniques can be classi ed based on: a) sampling device (e.g. swab vs. sponge), b) method of sample preparation (e.g. dry vs. pre-moistened), transport and storage method, and c) sample processing (e.g. extraction, enrichment, incubation). (13) Techniques which seek to identify speci c pathogens typically focus on demonstrating qualitative presence of organisms of interest, whereas quantitative data are needed to characterize microbial bioburden to target interventions to the reservoirs with the greatest burden of potentially pathogenic organisms. (13) Environmental sampling in NICUs has historically focused on qualitative identi cation of pathogens as part of an outbreak investigation or as part of speci c quality assurance purposes. (14) Reports of persistent reservoirs (as opposed to transient contamination) for ESBL-PE identi ed within a NICU are rare. (15) In this study, we aimed to assess the e ciency and feasibility of a chromogenic-culture-media-based environmental sampling approach using basic sample collection equipment in a NICU where infections with ESBL-PE are hyper-endemic. By deploying this technique over time on a range of samples from water, surfaces, hands and equipment in the hospital environment, we aimed to characterize the ESBL-PE bioburden, and identify reservoirs and vehicles which could then be targets for remediation and disinfection.
Footnote: [1] For clinical reports from Botswana, ESBL-production is inferred from extended-spectrum cephalosporin resistance detected on phenotypic antibiotic susceptibility testing.

Study design
This study consisted of a series of point prevalence surveys which took place on four separate occasions Study setting: The study was conducted in a 36-bed NICU within a 530-bed public tertiary referral hospital in Botswana where over 8000 deliveries occur annually. The neonatal unit is a one-storey block covering a total of 315 square meters, of which 87 square meters are patient care areas (average of 2.5 m 2 per patient-mother dyad). The most common diagnoses among patients in this NICU are prematurity-related complications, hypoxia-related injuries, and sepsis. Neonatal care is provided by approximately 16 healthcare workers per day shift, 4-6 of whom are nurses. Neonatal care includes: oxygen/ventilatory support, cardio-respiratory monitoring, enteral and parenteral nutrition, thermoregulation, transfusion, post-surgical care, phototherapy, and uid/electrolyte management. Because of the shortage of staff, enteral feeding (both oral and gavage) is administered by primarily by caregivers (mainly mothers). The unit's doors and windows open directly outdoors, which episodically results in entrance by ies and cockroaches, and has prompted the use of Light Emitting Diode (LED) insect traps.
Environmental Sampling Technique a. Sampling Devices: We used sampling equipment which was relatively inexpensive and easy to source using local vendors in our setting (Fig. 1). This included: 15 cm sterile nylon ocked swabs withsolid plastic handle and breakpoint (Puritan®; Cat. No. 25-3406-U, a 10 cm x 10 cm paper single-use paper sampling template for at surface sampling (Environmental Express®; Cat. No. EE-C1010), 15 ml plastic conical centrifuge tubes (SPL Life Sciences; Cat. No. 51015) for water samples and for transport of ocked swab samples, a 100 ml sterile syringe and tubing (CareJoy, Cat. No. 784121) for water aspiration, and nitrile or latex examination gloves for the person collecting the sample. b. Sample preparation & transport Prior to sample collection, ocked swabs were pre-moistened with 1 ml of normal saline at room temperature which had been inoculated into the conical tubes. After sample collection, water samples and ocked swab samples were placed in the conical tubes and transported in an ice-pack and cooler box to the laboratory, kept in a 4 o degrees Celsius until being processed within 24-48 hours.

c. Sample Collection Technique i. Water
Free-ow water samples were collected directly from taps after allowing water to run for 5 seconds. Water samples taken from plumbing traps were obtained by inserting a 4 mm diameter sterile plastic hose (alternatively, a naso-gastric tube can also be used) into the sink grid and water aspirated into a 100 ml sterile syringe (smaller syringes can also be used).
ii. Flat surfaces A 10 cm x 10 cm single-use paper sampling template was placed on a at surface and a pre-moistened ocked swab was streaked in a zig-zag motion in three different orientations, rotating the swab while streaking for a total of 20 seconds.
iii. Non-at surfaces A pre-moistened ocked swab was streaked in a zig-zag motion in and around the non-at surface, rotating the swab while streaking for a total of 20 seconds.

iv. Infant formula
A pre-moistened, sterile ocked swab was stirred into an opened powered infant formula tin for approximately ve seconds and placed into a conical tube.

v. Trapped insects
Insects were trapped in the immediate patient care environment using an LED light insect trap already in use in the unit to control insects. Insect corpses were lifted into the conical tubes using ocked swabs.

vi. Hands
Voluntary, anonymous swabbing of hands of healthcare workers and caregivers was carried out during three of the four sampling events. Healthcare workers and caregivers were asked to self-swab their hands using a standard technique illustrated in Fig. 2 prior to washing their hands with soap, water, and then drying and sanitizing with an alcohol-based hand sanitizer. Hand swabs were taken in the same manner after handwashing.

d. Sites sampled
We sampled a combination of water, high-touch surfaces, infant formula, medical equipment, and trapped insects recovered from the immediate patient care area (Fig. 3). Most locations were sampled on all four occasions, however some sampling sites were added and removed as the sampling strategy was re ned over time.
e. Sample processing and culturing Swabs and saline-containing tubes were shaken on a vortex mixer to elute the sampled organisms and directly inoculated onto chromogenic-agar-media (CHROMagar™ ESBL, Paris, France), which is selective and differential for ESBL-PE, using a 10 µL graduated inoculating loop, streaking the plates to isolate individual colonies. Bacterial colonies were quanti ed and phenotypically characterized using biochemical identi cation tests. Light bacterial growth was de ned as growth of 10-100 colony-forming units (CFUs), growth of 100-1000 CFUs was considered moderate growth, and > 1000 CFUs was considered heavy growth. For at surfaces, bacterial growth was categorized using these CFU breakpoints per 100 cm 2 .
f. Data analysis Data were analysed using basic descriptive statistics: crude numbers and frequencies using Microsoft Excel Software.

a. Results from environmental samples
Dense and consistent ESBL-PE contamination was detected inside plumbing traps, sink drain grids, and sink basins at all locations within the unit during all four sampling events ( Figure 4). Contamination was also consistently observed in certain high-touch surfaces (telephones, door handles, keyboards), while contamination was intermittently demonstrated on medical equipment (suction catheters, feeding tubes, and oxygen humidi ers) and feeding stations (including formula powder). Free-ow samples from tap water consistently demonstrated no growth.

b. Results from hand swabs
Hand contamination rates were higher among caregivers than among healthcare workers ( for Pseudomonas spp. Following these results, all participants were instructed to use alcohol-based hand-sanitizer following washing with soap and water. Of note, bacterial genera were recovered with roughly equal distribution over time, with the exception of Acinetobacter spp. which was not recovered at all during the rst sampling event in January, but from 16 and 21 diverse sampling sites during the 2 nd and 3 rd sampling events. The third sampling event corresponded with an increase in Acinetobacter infections hospital wide, as per microbiological blood culture surveillance.

c. Feasibility and E ciency Assessment
Most supplies were sourced using local vendors. To collect specimens, plate, and analyse 50 samples, we estimated a total expenditure of $269.40 United States Dollars and 13.5 cumulative work hours among all personnel. Culture results were available within 24-72 hours of collection. Although these results were analysed by a clinical microbiologist, the processing and culturing techniques were deemed to be appropriate for the level of a microbiology lab technician.

Discussion
Using basic environmental sampling and laboratory techniques aided by chromogenic culture media, we identi ed ESBL-PE reservoirs (sinks) and plausible transmission vehicles (medical equipment, infant formula, hands of caregivers/healthcare workers, insects) in this NICU environment. This strategy was a simple and cost-e cient method to assess ESBL-PE bioburden and may be feasible for use in other settings to support ongoing infection control assessments and outbreak investigations.
The identi cation of sinks as stable reservoirs for ESBL-PE in this setting was anticipated; the moist and warm environments created by plumbing traps provide ideal conditions for bio lm formation.(16) Furthermore, eradication of ESBL-PE from plumbing traps using conventional remediation techniques is challenging and therefore allows for long term survival of bacterial communities. (17) Additionally, the stable contamination of sink grids, basins, tap handles, and some backsplashes demonstrated in this study illustrates the retrograde model of bacterial dispersal from wastewater previously postulated by in situ studies.(16) The fact that some healthcare workers' hands demonstrated Pseudomonas spp. posthandwashing which were not present prior to handwashing is concerning for acquisition from the sink environment (although could be explained by variability of swab technique). Following these results, the importance of using alcohol-based hand-sanitizers after washing with soap and water was reemphasized to all staff and caregivers.
The recognition of sinks as environmental reservoirs for ESBL-PE has led some intensive care units (ICUs) to adopt "water free" care, moving sinks away from the immediate patient care environment, with corresponding declines in patient colonization rates for pathogens of interest.(18) In settings where these changes may not be feasible, interim measures targeting transmission vehicles are appropriate. In this study, we identi ed medical equipment, formula, insects, and hands of healthcare workers and caregivers as potential transmission vehicles that might move pathogenic organisms from a xed location to the patient. Infection control teams should engage with doctors, nurses, and cleaning staff to ensure shared medical equipment is adequately disinfected in between uses, according to equipment package inserts or best practices, such as the U.S. Centers for Disease Control and Prevention's "Environmental Cleaning in Healthcare Facilities in Resource-Limited Settings". (19) Preparation of shared powered infant formula sources should be carefully monitored and, if possible, carried out in accordance with World Health Organisation's guidelines for safe preparation, storage and handling of powdered infant formula. (20) We found that insects within the patient care environment were often colonized with potentially pathogenic organisms. While insect infestations are not typically reported as contributing to either endemic or outbreaks of healthcare-associated infections, the importance of vector control in the patient care environment cannot be over-emphasized. For example, cockroaches have been implicated in outbreaks of neonatal K. pneumoniae infections, (21) and ESBL-PE colonization of fruit ies is welldocumented. (22) In our study setting, LED-light insect traps have proved to be an effective measure, along with window screens, at reducing the presence of insects in the patient care environment.
There are several limitations to this study, both in its technical and conceptual approach. First, environmental sampling was conducted at a single centre, and ndings may not be generalizable to other units, depending on local climate, patient census, and cleaning practices. These point prevalence surveys were conducted in a ward that was actively used for patient care and the timing of sample collection had to adjust to the demands of patient care. Our sampling events were not coordinated with cleaning times, so we are unable to determine the extent to which current cleaning practices were effective in reducing the burden of contamination. We attempted to overcome this limitation by performing multiple sampling events and categorizing single contaminations as "transient" and repeated contaminations as "stable". Our use of "semi-sterile" techniques could have resulted in some false-positive results due to inadvertent contamination, but again we sought to counteract these limitations through the use of serial sampling.
Sampling bias may have in uenced our ndings on the prevalence of hand carriage of potential pathogens by caregivers and healthcare workers. For example, those con dent with their hand hygiene practices may have been more willing to participate, leading to an under-estimation of the true contamination rate.
This study took place in the midst of the Corona Virus Disease 2019 (COVID-19) pandemic, when heightened efforts to mitigate hospital transmission of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) were in place. In comparison to pre-pandemic practices, hand hygiene was enforced more strictly and soap and alcohol-based hand sanitizer were more readily available. These important measures may have contributed to an overall decreased bacterial bioburden, particularly on the hands of healthcare workers and caregivers.
We are unable to con rm that the identi ed environmental contamination contributed to concurrent clinical infections. In future studies, whole genome sequencing can help to establish true transmission pathways.
Despite these limitations, this simple environmental sampling technique might be a feasible way in which neonatal and other high-risk units facing hyper-endemic rates of ESBL-PE infections can better understand the nature of contamination and transmission dynamics in their unit. Further, it might help catalyse the implementation of infection control measures targeting de ned reservoirs and suspected transmission pathways. The temporal link between increased recovery of Acinetobacter spp. from the environment and increased incidence of Acinetobacter infections hospital-wide is an anecdote of the importance of timely and reliable environmental sampling as an important outbreak response tool. This sampling method may also be pivotal in trialling and measuring the impact of novel remediation and prevention strategies. For example, neonatal units who have identi ed sinks as stable ESBL-PE reservoirs should consider removing sinks and implementing "water-free" care for ICU patients,(18) placing sink covers, (23)  Availability of data and materials: All data generated or analysed during this study are included in this published article.
Competing interests: The authors declare that they have no competing interests.  Figure 1 We used sampling equipment which was relatively inexpensive and easy to source using local vendors in our setting (Figure 1).

Figure 2
Healthcare workers and caregivers were asked to self-swab their hands using a standard technique illustrated in Figure 2 prior to washing their hands with soap, water, and then drying and sanitizing with an alcohol-based hand sanitizer. Figure 3