MRSA Transmission in Intensive Care Units: Genomic Analysis of Patients, Their Environments, and Healthcare Workers

Abstract

Background

Methicillin-resistant Staphylococcus aureus (MRSA)—and now USA300 MRSA—is a significant intensive care unit (ICU) pathogen; healthcare worker (HCW) contamination may lead to patient cross-transmission.

Methods

From September 2015 to February 2016, to study the spread of MRSA, we enrolled HCWs in 4 adult ICUs caring for patients on MRSA contact precautions. Samples were collected from patient body sites and high-touch surfaces in patient rooms. HCW hands, gloves, and personal protective equipment were sampled pre/post-patient encounter. Whole genome sequencing (WGS) was used to compare isolates from patients, HCWs, and environment.

Results

There were 413 MRSA isolates sequenced (38% USA300, 52% USA100) from 66 patient encounters. Six of 66 HCWs were contaminated with MRSA prior to room entry. Isolates from a single patient encounter were typically either USA100 or USA300; in 8 (12%) encounters both USA300 and USA100 were isolated. WGS demonstrated that isolates from patients, HCWs, and environment often were genetically similar, although there was substantial between-encounter diversity. Strikingly, there were 5 USA100 and 1 USA300 clusters that contained similar strains (<22 single-nucleotide variants [SNVs], with most <10 SNVs) within the cluster despite coming from different encounters, suggesting intra- and inter-ICU spread of strains, that is, 4 of these genomic clusters were from encounters in the same ICU; 5 of 6 clusters occurred within 1 week.

Conclusions

We demonstrated frequent spread of MRSA USA300 and USA100 strains among patients, environment, and HCWs. WGS identified possible spread within and even between ICUs. Future analysis with detailed contact tracing in conjunction with genomic data may further elucidate pathways of MRSA spread and points for intervention.

We used genomic sequencing to understand the spread of methicillin-resistant Staphylococcus aureus (MRSA) among healthcare workers, patients, and intensive care units (ICUs). We demonstrated spread of MRSA strains among patients, environment, and healthcare workers. Genomic sequencing identified possible spread within and even between ICUs.

(See the Editorial Commentary by Azarian on pages 1888–90.)

Healthcare-associated infections (HAIs) are a significant cause of morbidity, mortality, and hospital costs [1, 2]. It is felt that the major pathway for transmission of potential pathogens in the hospital is by healthcare workers (HCWs) who contaminate their hands, clothing, or equipment during patient contact [3, 4]. Intensive care units (ICUs) provide multiple opportunities for HCW contamination and possible cross-transmission of pathogens due to the complexities of patients, colonization burden in units, and the level and urgency of care in ICUs [5–7]. To improve our ability to prevent spread of potential pathogens in healthcare settings, we need to better understand transmission dynamics.

HAIs due to multidrug resistant organisms (MDROs), such as methicillin-resistant Staphylococcus aureus (MRSA), are of particular concern [8, 9]. S. aureus—both MRSA and methicillin-susceptible S. aureus—account for the largest proportion of healthcare-associated infections [9]. By pulsed-field gel electrophoresis (PFGE), USA100 has been identified as the traditional hospital-associated MRSA strain in the United States.¹⁰ Although USA300 has emerged as an important strain of MRSA in the community in the past 20 years [10], USA300 has spread beyond the community and now is a significant cause of hospital-onset infections [11]. The complex interplay of strains of MRSA between community, healthcare, and ICU settings remains poorly understood.

Often, traditional molecular subtyping methods (eg, PFGE) have limited the ability to discriminate among endemic hospital strains such as MRSA—hampering analysis of MRSA spread in ICUs [12]. In contrast, whole genome sequencing (WGS) is able to provide very fine differentiation between MRSA strains [12]. Epidemiologic data can be integrated with these genomic data to infer possible transmission events. WGS has been used to understand possible instances of MRSA transmission and outbreaks within adult and neonatal ICUs [13, 14]. Such studies underscored the need for rich epidemiologic metadata to supplement genomic data to improve interpretation of genomic results.

Patients colonized or infected with MRSA during hospitalization are typically placed on contact precautions [15]. HCWs wear personal protective equipment (PPE) when caring for a patient on contact precautions to prevent self-contamination. However, there can be errors in the donning and doffing of PPE, and the impact of these errors on HCW contamination is not well established [16]. One hypothesized path of hospital transmission of MRSA is how HCWs who contaminate their hands, equipment, or clothes during patient contact. The objective of this study was to use WGS to improve our understanding of the origins of HCW contamination with MRSA during a patient encounter involving donning and doffing of PPE.

METHODS

Enrollment

Enrollment for this study occurred from September 2015 to February 2016 in 4 adult ICUs at Rush University Medical Center, a 720-bed tertiary care teaching hospital in Chicago, Illinois, USA. The current study focuses on MRSA and is one objective of a larger project [16], which sought to analyze HCW contamination with MDROs after caring for a patient on contact precautions. HCWs caring for adult patients who were placed on contact precautions for MRSA in the medical, surgical, coronary care, and neurologic ICUs were eligible for enrollment. In Illinois [17], active surveillance is mandated for all patients at admission to ICUs to detect MRSA colonization. At the time of the study, Rush University Medical Center patients were placed on contact precautions if patients had MRSA isolated from a surveillance nares screen or clinical culture.

HCWs were eligible for enrollment if they were caring for a patient on MRSA contact precautions. HCWs included physicians, nurses, patient-care technicians, respiratory therapists, and physical therapists. HCWs caring for a patient on contact precautions for historical MRSA but who did not have MRSA isolated for the index hospitalization were not eligible. For enrolled HCWs, activities and interactions of 1 HCW in the hospital room of 1 patient were observed; this was classified as an encounter. Because the focus of the larger study was the association between donning and doffing errors and HCW contamination, we enrolled each HCW only once. Patients could participate in up to 2 encounters. Daily patient bathing with chlorhexidine gluconate (CHG) occurred in all adult ICUs in this study.

Culturing Strategy

Samples were collected using nylon flocked swabs with liquid Amies media (ESwab^TM, Copan Diagnostics, Inc., Murrieta, CA, USA) from standardized areas of patient body sites, environmental room surfaces, and HCWs immediately before and after patient encounters as described previously [16]. If HCWs had rings, bracelets, or watches, swab cultures from around and under these items were obtained [18]. Swabs collected from HCWs before and after a patient encounter included gloves, gowns, clothes, and equipment (eg, stethoscope, phone) [16]. Environmental cultures of the patient’s room were obtained within 2 hours before a HCW-patient interaction. Contact plates (RODAC, Bectin Dickinson, Franklin Lakes, NJ, USA) with neutralizer were used in addition to nylon flocked swabs for environmental sampling in order to improve detection sensitivity [19].

Swab Processing

A selective broth enrichment method was used to improve sensitivity of MRSA detection [20], followed by subculture to chromogenic agar. Identification of MRSA was done as previously published and susceptibility testing was performed according to 2015 Clinical Laboratory Standards Institute guidelines [21].

Whole Genome Sequencing

Genomic DNA extracted from MRSA isolates was prepared for sequencing using a Nextera XT kit (Illumina, San Diego, CA, USA), according to manufacturer’s instructions. Libraries from each strain were sequenced to an average coverage of >100× using an Illumina NextSeq500 sequencer (paired-end 2 × 75 base reads). Comparative genomic analyses were performed using the software package SPANDx version 2.4 [22].

For comparative genome evaluations, the SNV matrix was imported into the software package MEGA7 [23] for phylogenetic analyses. Constructed neighbor-joining trees, with 100 bootstrap resamplings, were used to visualize clusters. Evolutionary distances are in units of the number of base differences per sequence. Clusters were defined by examining if isolates were within 40 SNVs of each other [24]. Pairwise SNV comparisons of USA100 and USA300 isolates were plotted to determine if the previously established 40 SNV cutoff was consistent with genomes in this study (Supplementary Figures 2 and 3). Based on this analysis and prior literature [24], putative transmission clusters were groups of isolates that were within 40 SNVs of each other but contained isolates from encounters involving at least 2 patients. Trees exported from MEGA were also visualized using the online Interactive Tree of Life (iTOL; v5.2) software package [25] and the iTOL editor for Excel. Library preparation and sequencing were performed by the University of Illinois at Chicago (UIC) Genome Research Core (GRC). Data were processed through the SPANDx pipeline at the UIC Research Informatics Core.

Integration of Epidemiologic and Genomic Data

Sample metadata (eg, isolate from patient/HCW/environment; episode number; care unit) were overlaid on genomic phylogenetic trees. For isolates found to be part of genomic clusters that spanned different patient encounters, additional chart review and epidemiologic evaluation were performed. Location of enrollment, time of enrollment, and room of encounter were integrated with genomic analysis.

Data Access

The genome sequence data from this study have been submitted to the NCBI Sequence Read Archive under Bioproject ID PRJNA595570.

The study was approved by the Rush University Medical Center Institutional Review Board.

RESULTS

Enrollment

From September 2015 through February 2016, we enrolled 66 HCWs (66 encounters) and 46 unique patients in contact precautions for MRSA. Most HCWs were nurses (70%) and physicians (18%); 6% each were respiratory therapists or patient-care technicians. Patients were in the medical (n = 25, 38%), surgical (n = 10, 15%), coronary care (n = 17, 26%), or neuro/neurosurgery (n = 14, 21%) ICU. Environmental cleaning was performed a median of 20.9 hours (range, 7.5–23.3) before study encounters. For 49 patients with chart-documented CHG bathing, the median time from bathing to enrollment was 13.6 hours (range 7.1, 23.8). Two encounters had only non-USA100 and non-USA300 strains and were not included in epidemiologic analyses.

Prevalence of MRSA Strains in the ICUs

There were 413 MRSA isolates that underwent WGS (Supplemental Figure 1). Of these, 214 (52%) mapped most closely (<621 SNVs, with most <343 SNVs) to a USA100 reference genome (CP029474 and CP029475) [26], 159 (38%) mapped most closely (<1064 SNVs, with most <140 SNVs) to a USA300 reference genome (NC_010079), and 40 (10%) were classified as non-USA300 and non-USA100 (Supplemental Table 1). While MRSA isolates within an encounter were typically found to be one USA type, 8 (12%) encounters contained >1 USA type. Of the 66 encounters, 34 contained only USA100 isolates, 17 contained only USA300 isolates, 8 contained USA300 and USA100 isolates, 4 contained USA100 and non-USA300/USA100 isolates, and 3 contained USA300 and non-USA300/USA100 isolates.

Level of Genomic Diversity of MRSA USA100 and USA300 Clusters Within and Between Patient Encounters

Of these 44 genomic clusters of MRSA, 25 (57%) mapped to the USA100 genome and 19 (43%) to the USA300 genome (Figures 1A, B) (Supplemental Table 2). Forty-two (95.5%) of these clusters contained strains that were all within 40 SNVs of other strains in that cluster. Two clusters contained isolates greater than 40 SNVs from each other, including USA 300 cluster of encounters 69 and 70 (1–59 SNVs among 11 genomes) and USA100 cluster of encounter 30 and 32 (single outlier with 49–50 SNVs to all other genomes, and 0–2 SNVs between the 5 remaining genomes). For 38 (86.4%) genome clusters, a cluster contained isolates only from a particular patient, that patient’s environment, and the HCW caring for that patient. In contrast, there was substantial genetic diversity between encounters, with isolates from different patient encounters being >50 SNVs from other isolates in 43 of 44 clusters (97.7%) A single set of clusters was <50 SNVs from each other—specifically, the USA100 cluster of encounters 47 and 89/91 (23–28 SNVs).

Figure 1.

A, Phylogenetic analysis of USA100 MRSA strains. Unrooted boot-strapped neighbor-joining tree is used to demonstrate relationships between USA100 MRSA strains. Metadata for each isolate include a sampling location (identified in name and by color and symbol in the innermost ring), an encounter number (identified by name and in color in the middle color gradient ring), and hospital location (identified by color in the outermost ring). Nodes supported by bootstrap (>70%) are indicated with black circles. Statistically supported clusters (see Supplementary Methods) composed of isolates from encounters from different healthcare workers, different patients, or different environments are highlighted in orange, and lettered according to Table 1. The tree is based on WGS analysis of USA100 MRSA isolates only, and the distance matrix is number of nucleotide variants between genomes. A total of 3598 positions were analyzed. The reference USA100 genome for the SPANDx analysis is highlighted in red. B, Phylogenetic analysis of USA300 MRSA strains. Unrooted boot-strapped neighbor-joining tree is used to demonstrate relationships between USA300 MRSA strains. Metadata for each isolate include a sampling location (identified in name and by color and symbol in the innermost ring), an encounter number (identified by name and in color in the middle color gradient ring) and hospital location (identified by color in the outermost ring). Nodes supported by bootstrap (>70%) are indicated with black circles. Statistically supported clusters (see Supplementary Methods) composed of isolates from encounters from different healthcare workers, different patients, or different environments are highlighted in orange, and lettered according to Table 1. The tree is based on WGS analysis of USA300 MRSA isolates only, and the distance matrix is number of nucleotide variants between genomes. A total of 3057 positions were analyzed. The reference USA300 genome for the SPANDx analysis is highlighted in red. Isolates 87 and 92 on the USA300 tree form a cluster. However, isolates from encounters 87, 88, and 90 also represent a cluster on the USA100 tree. Encounters 90 and 92 represent isolates from the same patient in the same patient room but with different healthcare workers. Abbreviations: CCU, coronary care unit; HCW, healthcare worker; MICU, medical intensive care unit; MRSA, methicillin-resistant Staphylococcus aureus; NeuroICU, neurologic intensive care unit; SICU, surgical intensive care unit; WGS, whole genome sequencing.

Identification of HCW Contamination Before Room Entry

We identified 6 (9%) HCWs who were contaminated with MRSA before they entered the patient room. Sites of contamination included hands (2), clothes (3), cell phone (2), Cisco phone (1), stethoscope (1), and patient sign-out list (1). Three of these HCWs were contaminated with USA300 and 3 with USA100 strains. For 4 of these HCWs, strains identified on HCWs were genetically similar (<12 SNVs) to strains isolated from the patient and that patient’s room. In a fifth encounter, no patient isolate was recovered, but the HCW isolate was within 4 SNVs of environmental samples from the patient’s room. These 5 HCWs may have become contaminated during an earlier interaction with that specific patient. For the other HCW who had USA300 isolated before room entry, the strains isolated from the patient or the room for that encounter were both USA100, suggesting this HCWs acquired the MRSA strain from an alternate source.

We previously reported [16] that HCWs who made multiple PPE doffing errors were more likely to have contaminated their clothes or hands with a potential pathogen following a patient interaction, especially when gloves were removed before gowns. Of the 6 HCWs who were identified as being contaminated before entering a patient room, 3 were observed to have errors in donning of gloves before entering the room, and 4 were observed to have PPE doffing errors after the patient encounter. One HCW was an attending physician, 1 was a respiratory therapist, and 4 were nurses. Although errors in donning of gloves and doffing PPE were observed among noncontaminated HCWs as well, the colonization status of patients who were seen by HCWs before our sampling is unknown and may have been a source of contamination.

Genomic Clusters of MRSA: Evidence of Potential for MRSA Spread Within and Between ICUs

Although most genomic clusters comprised isolates from only 1 encounter (or 2 if the same patient was involved), we identified 6 (13.6%) of 44 genomic clusters with genetically similar strains (all <22 SNVs, 5 of 6 with <10 SNVs) that suggested MRSA spread between patients, that is, for these 6 genomic clusters, isolates within a cluster were from different encounters that involved care for different patients (Supplemental Table 2). Isolates in these 6 clusters were from patients, environmental sites, and/or HCWs (Figure 1A, Figure 1B).

Five clusters contained an isolate from an HCW (nurses and a resident physician) (Table 1). For 4 clusters (B, D, E, and F), encounters within that particular cluster were in the same ICU and for the 2 remaining clusters (A and C) between ICUs (Figures 1 and 2). For 5 clusters the encounters within the cluster occurred within 1 week. For 1 cluster, isolates represent encounters involving the same hospital room enrolled 3 days apart with different HCWs and different patients. Four of these clusters were due to USA100 MRSA, 1 was due to USA300, and 1 contained both USA300 and USA100 strains (Figure 3).

Table 1.

Epidemiologic Descriptions of Hypothesized Transmission Clusters

Cluster	Encountersa	Corresponding Epidemiologic Data	Isolates in Cluster	Interpretation	SNV range
A	77	Encounters occurred within 1 week of each other; different units.	Patient, environmental (infusion pump panel, room phone, bed linen), healthcare worker equipment before and after encounter	May represent transmission via intermediaries. Encounters occurred in different units yet the healthcare worker in encounter 81 acquired a genetically similar strain as isolated in encounter 77. May suggest that source of glove contamination may have been a site in the environment or on portable equipment that was not cultured as part of the study.	0–9
	81		Healthcare worker glove after encounter (nurse)
B	55	Encounters occurred in the same unit, were within 1 week of each other, and involved the same single occupancy unit room (but with different patients).	Patient, environmental (bed rail, room phone, call button), healthcare worker glove after encounter (nurse)	May represent residual environmental contamination. There was likely residual environmental contamination on the bed rail from prior room occupant. This represents an opportunity for transmission to healthcare workers and patients.	0–6
	56		Environmental (bed rail)
Cb	87, 88	Encounters occurred within 1 week of each other; different units.	Patient, environmental (call button, bed linen)	May represent transmission via intermediaries. Encounters occurred in different units but given short time between isolation of genetically similar isolates, this cluster could represent cross-transmission via contaminated portable equipment or healthcare worker hand.	0–4
	90		Patient (tracheostomy site)
D	89	Encounters within 1 week of each other and occurred in the same unit but in different single occupancy unit rooms.	Healthcare worker clothes before encounter (nurse)	May represent opportunity for transmission. Healthcare worker in encounter 89 may have entered room for patient in encounter 91 previously and we didn’t capture this. However, our findings demonstrate residual contamination of a healthcare worker from one patient when they are caring for a separate patient.	0–1
	91		Patient, environmental (door handle inside room, vital sign monitor, call button, blood pressure cuff, bed linen, bed rail, over bed table, sink, healthcare worker glove after encounter (resident physician)
Ec	7	Encounters are all on the same unit over a 6-week time period.	Healthcare worker equipment after encounter (nurse)	May represent residual environmental contamination. Healthcare worker in encounter 7 at some point contaminated their equipment (possibly from another patient) and as healthcare workers and equipment can move from room to room, this could account for persistence of this MRSA strain in the ICU over a 6-week time period.	1–2
	17		Environmental (keyboard/mouse of portable computer in hospital room)
	41		Environmental (blood pressure cuff)
F	93, 95	Encounters occurred within 1 week of each other in the same unit but in different single occupancy unit rooms.	Patient, environmental (vital sign monitor, bed rail, door handle inside room, keyboard/mouse of portable computer in hospital room, toilet handle, infusion pump panel, over bed table, bed linen, call button, room phone, blood pressure cuff, sink), healthcare worker glove and ring after encounter (nurse)	May represent transmission. Unclear if isolate from patient’s hand in encounter 96 represents acquisition. Since the patients are on the same unit within a 1 week time frame, acquisition could have occurred via contaminated healthcare workers or portable environmental equipment. All other patient, environmental, and healthcare worker isolates for encounter 96 are genomically different than this cluster.	0–21
	96		Patient hands

Abbreviations: ICU, intensive care unit; MRSA, methicillin-resistant Staphylococcus aureus.

^aEncounters 87 and 88 represent isolates from the same patient in the same patient room but different healthcare workers. Encounters 93 and 95 represent isolates from the same patient in the same patient room but with different healthcare workers.

^bEncounters 87, 88, and 92 represent a cluster on the USA100 tree. However, an isolate from 87 (patient antecubital fossae) and an isolate from encounter 92 (patient trach) form a cluster on the USA300 tree. Encounters 90 and 92 represent isolates from the same patient in the same patient room but with different healthcare workers. These 2 clusters between 2 patients from 4 encounters could represent transmission with intermediaries.

^cAll other isolates for encounters 7 and 41 are part of separate clusters. Isolates from encounter 7 form a cluster on the USA300 tree and include samples from patient body sites and environmental sites in that patient’s room. Isolates from encounter 41 form a separate cluster on USA100 tree and include samples from patient body sites and environmental sites in that patient’s room.

Figure 2.

Hospital location and time of enrollment for hypothesized transmission clusters. Black indicates isolate is from the medical intensive care unit. Red indicates isolate is from the coronary care unit. Gray indicates isolate is from the neurologic intensive care unit. There were 6 clusters, each cluster is represented in a black rectangle. Letters on right refer to epidemiologic clusters identified in phylogenetic analysis.

Figure 3.

Heatmap of USA100 MRSA isolates collected from patients, healthcare workers, and the hospital environment in adult intensive care units. Pairwise genetic distance is shown for isolates belonging to the 5 USA100 genomic clusters containing isolates from more than 1 patient. Genomic clusters were identified as groups of isolates that are all within 40 SNVs of each other and contain isolates from encounters involving at least 2 patients. The heatmap indicates that in contrast to the small genetic distances among isolates from the same patient, or between different patients in the same cluster, there are large genetic distances between isolates from different clusters. Abbreviations: HCW, healthcare worker; MRSA, methicillin-resistant Staphylococcus aureus; SNV, single-nucleotide variants.

Persistent environmental contamination (eg, bed rail, blood pressure cuff, room phone, bed linen, infusion pump panel, call button, vital sign monitor screen, door handle, portable computer in hospital room) and HCW contamination (eg, clothes, stethoscope, phone) were identified as possible sources of spread within and between ICUs (Table 1).

Discussion

Within a particular patient encounter, we found that typically patient, environmental, and/or HCW isolates were genetically similar within that encounter, suggesting the spread of MRSA during that patient-healthcare worker interaction. In addition, 6 genome-defined clusters of highly related MRSA isolates from different patient encounters suggested spread of MRSA in and between ICUs. Persistent environmental contamination in patient rooms and HCW contamination were documented as possible sources of spread for MRSA in and between ICUs and could represent targets for intervention.

With comprehensive sampling for MRSA of all individuals entering 4 adult ICUs over a 6-month period, we found that 38% of patient encounters contained USA300 isolates and 52% contained USA100 isolates, the strain typically associated with hospital-onset infections [27]. Prior work has demonstrated that USA300 strains have entered healthcare facilities—acute care settings, hemodialysis centers, nursing homes, and long-term care facilities—and can even be a cause of nosocomial infections [11, 28–31]. Our findings further support the stronghold USA300 strains have created in acute care settings and, in our study, in various adult intensive care units. It remains unclear if USA300 and USA100 have differences in their ability to spread to the hospital environment, healthcare workers, and other patients.

Daily CHG bathing of ICU patients is 1 strategy often employed as a mechanism to attain “source control” and thus reduce risk for HCWs to contaminate themselves and the hospital environment following interactions with patients [32]. Although most patients in the study received a CHG bath before enrollment, compliance and thoroughness of bathing were not observed [33]. It is unclear if strategies such as CHG bathing may need to be more stringent in regularity and adherence of application with certain pathogens that more readily contaminate or colonize patient skin.

Using WGS we were able to detect MRSA contamination of HCW’s clothes, hands, or equipment before that HCW entered the patient room. In most cases, sequencing demonstrated similarity of strains isolated on the HCW to those identified on that patient or in the hospital room environment. Nonetheless, identifying contamination of HCW’s clothes, hands, and equipment before they even enter a patient room demonstrates opportunity for an HCW to transmit MRSA between patients. We did not follow HCWs over time, so we cannot comment on the number of opportunities these HCWs had to enter other patient rooms while contaminated with MRSA. However, as HCWs often care for more than 1 patient, this finding of contamination with MRSA on clothes and equipment highlights potential for MRSA spread in and even between ICUs.

Although PPE would be a way to ensure hands and clothes are covered before interacting with a patient, we previously demonstrated that errors in PPE donning and doffing occur and that these errors can be associated with HCW contamination [16]. Continued vigilance in hand hygiene, thorough environmental cleaning, and routine cleaning of HCW equipment that is portable and transported around the ICU, in conjunction with adherence to guidelines for PPE, are essential [15]. Our results highlight important opportunities where interventions could be targeted to effectively interrupt spread of MRSA in the ICU. Prior work by Armellino and colleagues [34] demonstrated that remote video auditing with feedback led to improved hand hygiene compliance; novel methodologies such as these could be adapted to help reduce HCW contamination with potential pathogens such as MRSA.

Our study has limitations. First, for each enrolled HCW we only observed movements and interactions for 1 patient encounter; because we did not capture HCW activities beyond this encounter, we cannot definitely define risk for MRSA transmission to other patients, particularly for those HCWs who were found to be contaminated with MRSA, that is, the risk we identified, although substantial, may be a minimum estimate. Second, we used a comprehensive sampling strategy, but cases of MRSA could have gone undetected at ICU admission and therefore would not have been included in the study. However, Illinois state law mandates nasal surveillance for MRSA at ICU admission [17]; so our study population is representative of individuals who would be identified as MRSA carriers in hospitals employing such a strategy. Third, we did not screen HCWs for nasal carriage of S. aureus. However, prior work by Price and colleagues [35] examined HCW carriage for S. aureus and did not identify it as a common source of transmission in the ICU. A single-ICU study by Dancer et al included culturing of staff hands in their genomic assessment of S. aureus transmission in the ICU but did not identify staff hands as a common mechanism for transmission [36]. Our work extends these studies by including culturing of HCW gowns, hands, and equipment. Finally, the hospital environment was not sampled a second time relative to the enrollment encounter, so we cannot comment if environmental sites became positive following a HCW-patient interaction.

In conclusion, for most patient interactions, MRSA isolates from HCWs, the patient, and hospital environment had similar genomes. In addition, our findings support prior work underscoring the widespread incursion of USA300 in healthcare settings [11], including ICUs. WGS, in conjunction with robust epidemiologic data, allowed identification of possible opportunities for spread of MRSA strains within and even between ICUs. Although USA100 strains were identified in most of our genome-defined clusters that demonstrated wider spread, it remains unclear if there are differential transmission dynamics between USA100 and USA300 in ICU settings. WGS allowed us to further move the needle in our understanding of MRSA transmission in an ICU and, in addition, highlighted important opportunities for infection control improvements to help interrupt nosocomial spread of MRSA. Future analysis with detailed contact tracing in conjunction with genomic data may further elucidate pathways of MRSA spread in the ICU.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank Karen Lolans for her laboratory assistance in this project. The authors also thank Dr. Zhengdeng Lei for assistance with data processing through the SPANDx pipeline.

Financial support. CDC EPICENTER GRANT Cooperative Agreement no. 1U54CK000161-05 (SUPP) (PI: R. A. W.)

Potential conflicts of interest. M. S. reports a grant from the NIH. M. H. reports being a co-investigator on research projects for which Sage, Molnlycke, and Clorox donated products at no charge. Neither M. H. nor the institution received product. R. W. reports participation in clinical studies where participating hospitals or nursing homes received contributed product from Sage Products Inc., Molnlycke, Clorox, Medline, or Bio-K+. Neither R. W. nor his hospital received product, funding, or payments. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.