Resistance Mechanisms, Epidemiology, and Approaches to Screening for Vancomycin-Resistant Enterococcus in the Health Care Setting

Abstract

Infections attributable to vancomycin-resistant Enterococcus (VRE) strains have become increasingly prevalent over the past decade. Prompt identification of colonized patients combined with effective multifaceted infection control practices can reduce the transmission of VRE and aid in the prevention of hospital-acquired infections (HAIs). Increasingly, the clinical microbiology laboratory is being asked to support infection control efforts through the early identification of potential patient or environmental reservoirs. This review discusses the factors that contribute to the rise of VRE as an important health care-associated pathogen, the utility of laboratory screening and various infection control strategies, and the available laboratory methods to identify VRE in clinical specimens.

INTRODUCTION

Hospital-acquired infections (HAIs) are a serious threat for patient care and carry a significant cost to hospitals, since treatment of these infections is no longer reimbursable. In addition, regulations requiring hospitals to report HAIs creates further pressure to reduce incidence rates. Screening patients at admission for methicillin-resistant Staphylococcus aureus (MRSA) has been a successful approach in reducing MRSA HAIs in some health care systems and may be a successful strategy for controlling other health care-associated pathogens, including Clostridium difficile, carbapenem-resistant Enterobacteriaceae (CRE), and vancomycin-resistant Enterococcus (VRE) (1). However, there is debate about the optimal approach to screening and infection control, which may differ between pathogens of interest.

Members of the genus Enterococcus are well-documented pathogens associated with various clinical manifestations, including bacteremia, infective endocarditis, intra-abdominal and pelvic infections, urinary tract infections, and, in rare cases, central nervous system infections (2–4). Infection with vancomycin-resistant Enterococcus is associated with an increased mortality rate, illustrated by a 2.5-fold increase in mortality for patients suffering from VRE bacteremia (5). Vancomycin resistance in Enterococcus spp. has been increasing in prevalence since it was first encountered in 1986 (6, 7). Currently, 30% of Enterococcus species isolates from the United States are vancomycin resistant, and infection with these organisms causes an estimated 1,300 deaths each year (8). The majority of VRE are associated with the species E. faecium (77%) and E. faecalis (9%), with the remaining 14% of isolates representing species less frequently implicated in serious infections, including E. gallinarum, E. casseliflavus, E. avium, and E. raffinosus (8).

The optimal approach to reducing VRE infections is multifactorial, requiring antimicrobial stewardship to reduce the selection of VRE in colonized patients, appropriate infection control practices to reduce transmission, and reliable sensitive laboratory methods for the detection of VRE in a timely manner.

ANTIMICROBIAL RESISTANCE MECHANISMS AND APPROACHES TO THERAPY

Understanding the mechanism behind resistance is essential to reducing empirical antibiotic therapies that select resistant pathogens. Elimination of normal gut flora by commonly used broad-spectrum antibiotics (vancomycin, cephalosporins, and metronidazole) encourages the selective proliferation of VRE (9). This increases the likelihood of resulting infections with these bacteria or leaves patients more susceptible to colonization by resistant strains encountered in the environment or health care setting. Treatment of infections due to Enterococcus spp. can be challenging, since enterococci are intrinsically resistant to several classes of antibiotics, including β-lactams and aminoglycosides, and can acquire resistance to other classes, including quinolones, tetracyclines, and glycopeptides (e.g., vancomycin). Resistance to glycopeptides may be encoded chromosomally or extrachromosomally on plasmids. Often, these genes are located within transposons or other mobile elements, which can serve as a reservoir for the transmission of resistance to other organisms (10). The choice of therapy for infections due to Enterococcus spp. depends largely on the site of infection (e.g., urinary versus nonurinary) and the antimicrobial susceptibility profile of the isolate. In some scenarios, such as infective endocarditis, multiple antibiotics may be used to achieve a synergistic effect. This may include antibiotics to which the organism is considered intrinsically resistant, such as gentamicin or ceftriaxone, when used singly (11). Therefore, combination therapy requires an understanding of the intrinsic and acquired resistance mechanisms that contribute to therapy success or failure.

Aminoglycoside resistance.

Enterococcus spp. may demonstrate moderate-level (MIC, 62 to 500 μg/ml) or high-level (MIC, ≥2,000 μg/ml) resistance (HLR) to aminoglycosides (12). Moderate resistance is conferred by an intrinsic mechanism attributed to low permeability of the cell wall to the large aminoglycoside molecules (13). A study including 2,507 E. faecalis and 469 E. faecium isolates from clinical specimens in France found the prevalence of moderate resistance to be 8.9% and 49.2%, respectively (14). HLR to aminoglycosides is mediated through modification of the ribosomal attachment sites or the production of aminoglycoside-modifying enzymes (12). Both mechanisms are encoded by specific genes and are typically plasmid borne. The prevalence of HLR to aminoglycosides can vary from 40 to 68% among Enterococcus species isolates, depending on geographical location, and differs between E. faecalis and E. faecium (15, 16). Intrinsic resistance may be overcome using a combination treatment with a β-lactam and an aminoglycoside, also referred to as a synergy therapy approach. Synergy therapy is based on the principle that β-lactams will disrupt the cell wall, thereby facilitating increased cellular penetration by the aminoglycoside to reach a concentration sufficient for the inhibition of protein synthesis. This approach is effective in treating infections caused by organisms with moderate-level resistance to aminoglycosides; however, synergy is lost in strains with HLR to aminoglycosides, because the increase in intracellular aminoglycoside concentration cannot overcome the presence of specific aminoglycoside-modifying enzymes. Therefore, accurate differentiation between the moderate-level resistance (MLR) and HLR phenotypes is important when considering synergistic antibiotic therapies for serious infections (17). Laboratories can determine HLR by performing a high-level aminoglycoside test according to the CLSI guidelines.

β-Lactam resistance.

Intrinsic resistance to β-lactams is maintained in enterococci through the overproduction of penicillin binding proteins (PBPs) with low binding affinity for β-lactams, most notably PBP 4 and PBP 5 (18–21). The level of resistance differs between E. faecalis and E. faecium, with E. faecalis being 10- to 100-fold less susceptible to penicillin than streptococci and E. faecium being at least 4- to 16-fold less susceptible than E. faecalis (22). Despite reduced susceptibility to penicillin, the majority of E. faecalis isolates (∼99%) remain susceptible to ampicillin, while <20% of E. faecium isolates demonstrate ampicillin susceptibility (23). Importantly, neither penicillin nor ampicillin susceptibility is indicative of susceptibility to cephalosporins. For patients who are aminoglycoside intolerant, combination treatment with ampicillin and ceftriaxone was observed to be as effective as ampicillin and gentamicin in treating VRE with the gentamicin HLR phenotype (24). These therapies are most useful in treating severe infections, such as endocarditis and meningitis (25).

Glycopeptide resistance.

Glycoproteins are bactericidal drugs that function by binding to the terminal d-Ala-d-Ala in the pentapeptide portion of the N-acetylglucosamine (NAG)–N-acetylmuramic acid (NAM) peptidoglycan (PG) cell wall precursor (Fig. 1). Binding blocks transpeptide linkage of cell wall components, resulting in reduced integrity and, ultimately, cell death. Resistance to glycopeptides in Enterococcus spp. is mediated by the vancomycin resistance (Van) operon. This operon may be carried chromosomally or extrachromosomally on a plasmid. The Van operon consists of vanS-vanR, a response regulator; vanH, a d-lactate dehydrogenase gene, vanX, a d-Ala-d-Ala dipeptidase gene; and a variable ligase in which 9 variant genes have been identified (vanA, vanB, vanC, vanD, vanE, vanG, vanL, vanM, and vanN) (26). Expression is inducible by the two-component system VanS/R, which senses disruptions in the cellular membrane caused by glycopeptides, as well as cell wall damage caused by bacitracin or polymyxin B (27). The variable ligase gene is central in determining the level of vancomycin resistance (low, medium, or high), with the most commonly identified genes being vanA, vanB, and vanC. VanA is plasmid borne, confers high-level resistance (MIC, >256 μg/ml) to vancomycin, and is most commonly associated with E. faecium and E. faecalis, while chromosomally encoded VanC confers low-level resistance (MIC, 8 to 32 μg/ml) to vancomycin and is almost exclusively found in E. gallinarum, E. casseliflavus, and E. flavescens (11).

FIG 1.

Vancomycin acts by binding to d-Ala-d-Ala pentapeptides blocking cell wall biosynthesis. Resistance to vancomycin is conferred by the van operon, which consists of a two-component regulatory system (vanS-vanR) that responds to either vancomycin, disruption at the cell membrane, or both. Detection of stimulus then activates downstream genes vanH, vanA or vanB and vanX. VanX is a d,d-dipeptidase that cleaves d-Ala-d-Ala repeats, both depleting the pool of d-Ala-d-Ala and supplying the bacteria with a free d-Ala for Van(A/B). VanH is a d-hydroxyacid dehydrogenase that reduces pyruvate to d-Lac for the ligase Van(A/B). Van(A/B) then ligates d-Ala-d-Lac, allowing for the production of d-Ala-d-Lac pentapeptides that have low affinity for vancomycin. In addition, VanY is a d,d-carboxypeptidase that cleaves the d-Ala terminal peptide to further reduce pools of pentapeptides that have high affinity to vancomycin. Finally, VanZ, which is present on vanA-carrying strains, confers modest resistance to teicoplanin through an unknown mechanism. Differences in the level of resistance are likely a result of pentapeptide composition, as the ratio between pentapeptides consisting of high affinity to low affinity to vancomycin correlate with the isolate MICs. High-level resistance (HLR) occurs when pentapeptides are mostly composed of low-affinity molecules, and moderate-level resistance (MLR) involves more-heterogeneous pools of high- and low-affinity pentapeptides.

VanA resistance.

Enterococcus spp. carrying vanA are highly resistant to glycopeptides and are the dominant VRE variants of E. faecium and E. faecalis globally. Resistance is mediated by substituting the high-affinity terminal d-Ala-d-Ala peptide on NAM subunits with d-Ala-d-Lac. This amino acid substitution causes a 1,000-fold decrease in the affinity of the pentapeptide for vancomycin (Fig. 1) (11). Incorporation of NAM subunits containing the substituted d-Ala-d-Lac peptide into the peptidoglycan layer requires PBPs other than PBP 4 and PBP 5. In the presence of vancomycin, these alternative PBPs become the dominant proteins for cell wall synthesis. The alternative PBPs used during the presence of vancomycin have enhanced binding to β-lactams, which when used together can allow synergistic treatment (28). VanA-expressing strains are also resistant to the glycopeptide teicoplanin (Te) due to the presence of an additional gene present on the vanA operon, vanZ, which confers resistance by an unknown mechanism (29).

VanB resistance.

Isolates carrying vanB are less prevalent than vanA-carrying strains but can be found throughout the world and are commonly identified in Australia, where the majority of E. faecium VRE isolates carry vanB (30, 31). As with to vanA, resistance in vanB is mediated by converting d-Ala-d-Ala to d-Ala-d-Lac (11). However, vanB confers varied resistance to vancomycin, ranging from moderate- to high-level resistance (MIC range, 4 to >256 μg/ml) (11). The mechanism for the lower-level resistance compared to that of vanA is not well defined but is likely the result of a lower proportion of d-Ala-d-Lac substitution in the cell wall of strains carrying vanB. Resistance to vancomycin is proportional to the percent composition of d-Ala-d-Lac to d-Ala-d-Ala (32). Smaller amounts of d-Ala-d-Lac incorporation might result from reduced expression of the vanB operon, a reduction in VanX or VanB enzymatic activity, or a combination of minor mechanistic changes. Te resistance is not observed in vanB-carrying isolates, as vanZ is not encoded in this operon.

Risk of vancomycin resistance transmission to other pathogens.

Vancomycin is one of the few antibiotics that can be used to treat infections resulting from Gram-positive multidrug-resistant organisms (MDRO), such as MRSA; therefore, transmission of vancomycin resistance from enterococci to MRSA is of major concern. Horizontal gene transfer has been shown to be a mechanism of transmission between enterococci, and in vitro studies have demonstrated that transfer between Enterococcus and S. aureus can occur (33). In 2002, the first case of vancomycin-resistant S. aureus (VRSA) was isolated from a 40-year-old diabetic patient undergoing dialysis and was obtained from a dialysis catheter cultured from both the exit site and catheter tip. A week later, the VRSA isolate was isolated from a diabetic foot ulcer. MIC testing and DNA sequence analysis conducted by the CDC confirmed the isolate as VRSA (34). To date, 14 cases of VRSA have been reported in the United States; however, the presumably frequent interaction between VRE and MRSA (cocultured, with the two organisms recovered from the same source) and rare incidence of VRSA isolation suggest that in vivo transfer of vancomycin resistance between these species occurs at an extremely low frequency (35). If a laboratory expects a potential VRSA isolate (vancomycin MIC, ≥16 μg/ml), local infection control at the hospital should be notified of the possible result, and the laboratory should work with local and state departments of public health to have the isolate tested for confirmation.

Horizontal gene transfer of the van operon between Enterococcus spp. and other organisms also appears to occur at a very low frequency. In a study that performed mating experiments between Enterococcus species and Gram-positive gut flora (Lactococcus spp. and Bifidobacterium spp.), no transfer of vanA between genera was observed (36). Interestingly, the authors observed that interspecies transmission within the Enterococcus genus, e.g., E. faecium to E. faecalis, occurred at a much lower frequency (1/10⁸ per donor/recipient) than intraspecies transmission (1/10⁶ per donor/recipient). This may explain the higher prevalence of vanA within E. faecium isolates, as transmission occurs between E. faecium isolates, but not between E. faecium and other Enterococcus species, at high frequencies. Infection control and antimicrobial stewardship programs that aim to reduce acquisition (i.e., colonization and/or infection) of VRE and MRSA infections may aid in reducing the possible transmission of glycopeptide resistance to S. aureus by minimizing coinfection with VRE and MRSA.

ENTEROCOCCUS SPP. IN THE HEALTH CARE SETTING

Epidemiology.

Enterococcus spp. are normal colonizers of the human gastrointestinal (GI) tract but can be recovered from the skin, genitourinary (GU) tract, and the oral cavity. Surveillance studies conducted by the European Antimicrobial Resistance Surveillance System and National Healthcare Safety Network (NHSN) report varied rates of colonization, depending on geographical location and immune status of the patient population. Generally speaking, colonization rates for inpatients range from as low as <2% in Finland to as high as 34% in Ireland and 33% in intensive care unit (ICU) patients in the United States (37). Asymptomatic colonization is typically transient but can persist from months to years; decolonization is not recommended, since it can contribute to the accelerated development of resistance (38). Colonized patients are a potential risk to health care facilities, serving as a source of for transmission via shedding of VRE into the surrounding environment and onto health care workers (11, 39).

Hospital rooms housing patients infected or colonized with VRE are quickly contaminated and can also serve as a reservoir for transmission. Environmental sampling studies have isolated VRE from most surfaces found in rooms that previously housed infected patients, such as patient gowns, bedside rails, floors, doorknobs, and blood pressure cuffs (12). Enterococcus has been shown to persist for as long as 4 months on surfaces, and in a clinical setting, VRE was shown to persist through an average of 2.8 standard room cleanings, thereby acting as a continual source for possible transmission (40, 41). Recent studies involving environmental surveillance suggest that admission to a room previously occupied by a patient with VRE infection was associated with a significant increased risk for VRE acquisition by subsequent patients (hazard ratio, 4.4) (42).

Health care workers may also serve as vectors for the transmission of VRE between patients in different rooms or wards. Without proper hand hygiene, VRE can persist for up to 60 min on a health care worker (HCW)'s skin (11). Once hands are contaminated, hand hygiene plays a major role in reducing transmission rates; however, compliance is generally low. In a systematic review from 2010 to 2014 of several hand hygiene intervention studies, it was found that compliance ranged from 23 to 69% (43). Furthermore, it is important to educate HCWs about the possibility of contact with environmental services to ensure understanding that hand contamination can occur without specific contact with the patient, and precaution should be applied even in the absence of patient interaction.

APPROACHES TO SCREENING AND IMPACT FOR CONTROL OF VRE

Both the CDC and the Society for Healthcare Epidemiology of America (SHEA) have released guidelines that recommend active screening of patients for VRE colonization in hospitals and long-term-care facilities (44–46); however, many hospitals have not incorporated these recommendations. This is likely the result of limited evidence from outcome studies demonstrating significant benefit from laboratory screening for VRE and a lack of evidence suggesting which patients should be screened to maximize the cost-to-benefit ratio. Accumulating evidence highlights several specific risk factors that can be used to determine which patients are at highest risk and may benefit from screening efforts. The risk factors include length of hospital stay, recent or current use of antibiotics, patients whom are immunosuppressed, patients with prior hospital visits, and patients transferred from long-term-care facilities (47).

Passive or active screening.

Guidelines to control MDRO by the CDC recommend that all hospitals adopt active rather than passive screening approaches. In the case of VRE, passive screening refers to the detection of VRE from clinical specimens submitted for routine culture (i.e., not specifically submitted for detection of VRE). Patients are only isolated after laboratory detection (culture or molecular) or if previously positive for VRE on a prior admission. Because the ratio of VRE-infected to -colonized patients is 1:10, a passive screening approach may not adequately identify a sufficient proportion of colonized patients to effectively reduce VRE transmission (12). Modeling of screening strategies using data from a 10-bed ICU at the University of Maryland Medical Center estimated that passive screening would reduce VRE infection by only 4.2% compared with no screening or isolation practices (48).

Active screening is less well defined but can include the collection of specimens specifically for VRE screening or screening of stools collected for other purposes (e.g., for C. difficile testing). There are no specific recommended practices, but studies reporting on the utility of screening often include testing upon hospital admission for patients with specific risk factors (e.g., those who are immunosuppressed or have a history of broad-spectrum antibiotic usage) and patients received from long-term-care facilities. Additionally, continued periodic screening 1 to 2 times a week during hospital admission and a screen at discharge may be employed. In hospitals that do not perform specific screening for VRE colonization, reflex testing of C. difficile specimens for VRE can help determine the relative prevalence of VRE colonization in a patient population and can aid in early detection of outbreaks (49). Routine screening of patients at admission is effective in hospital settings with higher prevalence, but the cost of intensive screening may not be justified in hospitals with a low prevalence of VRE or in hospitals without high-risk patient populations. Implementation of active screening for all ICU patients at admission may reduce VRE infections by up to 39% compared with no screening (48). Another approach to reducing the spread of VRE within the hospital setting is to cohort patients by VRE colonization status, as determined by laboratory screening result, and dedicate specific HCWs to each cohort. This method may be difficult to coordinate, requires adequate hospital beds for cohorting, and is likely not practical for the majority of hospitals; however, reduced HCW contact between colonized and noncolonized patients has the potential to reduce the spread of VRE in the hospital (50).

Outcome studies of active screening.

Estimating the real-world effect and benefits gained by implementation of routine screening strategies can be difficult, and clinical outcome studies are not without limitations. A lack of proper case controls or implementation of multiple variables at one time to provide optimal patient care may obscure the independent contribution of any single screening or infection control measure. In addition, results may be dependent upon site-specific factors, such as organism prevalence and compliance of HCWs, which vary from facility to facility. Further, infection control practices often differ between studies, and the materials and methods may not be adequate to generate optimal comparisons across studies. A summary of several studies that performed analyses of active screening is presented in Table 1.

TABLE 1.

Summary of studies observing the clinical impact and cost of active VRE screening

Setting/region (reference)	Experimental design	Specimen type/detection method (n)	Study intervention	Results	Cost analysis	Comments
30 facilities in Siouxland region, USA (78)	Comparison of prevalence of pre-/postimplementation strategy	Perianal swab plated to bile esculin agar (3,774)	Screening on admission, cohorting patients, barrier precautions, education, and hand hygiene	Prevalence of colonization reduced from 2.2% to 0.5%	NAa	Required collaboration of 32 health care facilities, including both acute-care and long-term-care facilities, along with multiple implementation strategies
650-bed hospital, NY, USA (56)	Comparison of VRE BSI rates and cost analysis	Perianal swabs inoculated to Enterococcosel broth	15-component infection control targeting leukemia, lymphoma, and solid-tumor patients	Control strategies reduced rates of VRE BSI from 2.1 patients to 0.45 patients/1,000 days	Cost of strategy for 1/yr totaled $116,515, with an estimated study net savings of $189,318; cost beneficial for facilities that have 6–9 VRE BSI per yr	Cost analysis based on length of stay between time periods, which could inflate savings; however, a conservative estimate of 13.7 days stay for VRE BSI (19 in other studies) may underestimate cost of VRE BSI
2 university hospitals, Charlottesville, VA, USA (77)	Comparison of two regional hospitals where screening was implemented at 1 of 2 hospitals	Perirectal swabs/culture (unknown agar) (10,400)	Weekly screening of inpatients deemed high risk	193 (0.38%) culture positive with 1 VRE BSI compared to 29 VRE BSI at a comparison hospital that does not screen	Cost of screen was $253,099; estimated cost of 29 VRE BSI was $761,320	Comparison between 2 different regional hospitals, so factors, such as staff competency and infection control procedures, may affect results
2,700-bed hospitals, Chicago, IL, USA (51)	Retrospective study comparing VRE BSI rates between a hospital that screens compared to nonscreening hospital	Rectal swabs plated to selective medium	Weekly rectal swabs in high-risk units	Rate of VRE BSI was 2.1-fold higher in hospital that does not screen; 4 clones were responsible for >75% of VRE BSI at hospital that does not screen vs 4 clones consisting of 37% of all infections at screening hospital	NA	Comparison between 2 different regional hospitals, so factors, such as staff competency and infection control procedures, may affect results; retrospective study
19-bed ICU, St. Louis, MO, USA (55)	Compared detection rate from active screening of ICU patients vs screening of C. difficile stools	Rectal swabs/culture (unknown agar) (1,872)	Swab collected at admission, discharge, and every 7 days	280 detections at admission with active screening compared to 25 patients detected by C. difficile testing, 3 of which were not detected by swab screening	Active screening cost was $1,913/mo; estimated cost of HA colonization of $3,065–$9,970 per patient; estimated cost of VRE BSI was $17,143–$36,380 a patient	Rate of VRE colonization was not determined, and savings were based on transmission dynamic
ICUs, USA (54)	Cluster randomized trial comparing ICUs with and without contact precautions after positive screen	Rectal or stool sample within 2 days of admission enriched in bile esculin azide broth and plated to bile esculin agar (9,139)	Contact and expanded barrier precautions implemented after positive culture until discharge	No significant difference was observed between intervention and control ICUs for incidence rates of colonization or infection	NA	Delay in results as swabs were shipped to a reference site; monitoring observed less-than-ideal compliance in barrier protection used by HCWs, especially when interacting with environment
Kyoto Japan (52)	Comparison of pre- and postcontrol program for VRE outbreak	Stool specimens collected for any reason and active surveillance for transfer/selective VRE agar	Adhered to guidelines that consisted of good hand hygiene, good barrier protection, and admission screening	Initial increase from 0.71% detection to 1.2% detection but then fell to 0.17% after 4 yr of intervention	NA	Annual surveillance was likely inadequate, as detection was increased after guidelines were implemented
St. Louis, MO, USA (49)	Comparison of discontinuation of reflexing stools submitted for C. difficile testing	No screening	Removal of screening as a clinical practice	VRE BSI increased from 2.1 to 3.6 per 10,000 patient-days	Yearly cost of screening and isolation procedures was $116,708 compared to estimated cost of $139,286 for the 14 extra VRE BSI	No evaluations of patient populations were performed to determine if there were more at-risk patients in either pre- or postdiscontinuation
13 European ICUs (79)	Compared 3 phases consisting of baseline, universal chlorhexidine washing, and screening of patients	Rectal swabs were plated to chromogenic agar within 2 days of admission and then 2× a week (14,390)	Hand hygiene and chlorhexidine cleaning and contact precautions for positive screens	Chlorhexidine washings reduced MRSA but not VRE acquisition; no differences in MDRO BSI in all three phases	NA	Hand hygiene rose from 52% to 69% and 77% in phases 2 and 3, respectively; chlorhexidine washing was still performed in phase 3
Singapore, China (53)	Compare pre- and postintervention of a VRE bundle program	Not described, but PCR confirmed vanA and vanB	VRE bundleb	Reduction in VRE detection from 1.5/1,000 to 0.5/1,000	NA	6% cases occurred within first 48 h of hospitalization; bundle implementation makes it difficult to assess which implementation was most beneficial

^aNA, not applicable.

^bBundle included active surveillance, enhanced precautions signage, automated system for VRE identification of known carriers on admission, hygiene education, hydrogen peroxide vapor (HPV) cleaning of discharged VRE patients, change in bleach cleaning solution, and cleaning audits.

Several studies have shown that active surveillance of stool or rectal swabs has a positive effect in reducing the burden of VRE and related infections in hospitals. A retrospective comparison between two urban hospitals in the United States demonstrated a 2.1-fold lower rate of VRE bacteremia in the hospital performing weekly active surveillance on all inpatients (51). Strain typing of all bloodstream infection (BSI) isolates revealed that 4 clones were responsible for more than 75% of all Enterococcus species BSI at the hospital that did not actively screen for VRE, compared to 37% at the hospital with active screening. These data suggest that hospital transmission may be significantly more frequent at hospitals that are unaware of patient colonization status. Similarly, implementation of routine screening of stool specimens coupled with hand hygiene and barrier protection for positive patients (in accordance with 2003 SHEA guidelines) reduced the rate of patient colonization by more than 80% (1.2% versus 0.17%) in just 3 years (52). A comparable reduction in VRE colonization was achieved by combining active screening and hand hygiene with tracking of previous carriers and environmental cleaning of colonized patients' rooms after discharge (53). Following implementation of this bundle strategy, the rate of VRE detection reduced from 1.5 to 0.5 per 1,000 admissions.

Counter to studies demonstrating measureable benefit, there are also reports in the literature that suggest that active screening of patients does not reduce the prevalence of VRE in a health care setting. One cluster randomized trial included multiple hospital ICUs across the United States. Screening for VRE was on all admissions, and ICUs were randomly assigned to one of two groups, (i) an intervention group (n = 10), which was given the result of screening and implemented barrier precautions for positive patients; or (ii) a control group (n = 8), which was blinded to the screening result and so no intervention was implemented (54). After adjustment for baseline characteristics, the incidence of colonization was not significantly different between the intervention and control groups (40.4 ± 3.3 and 35.6 ± 3.7, respectively, P = 0.35). Unfortunately, this study had several limiting factors, including a 3-day delay between the collection of specimens and reporting of results. This delay may reduce the impact of interventional practices, since VRE transmission from colonized patients may have occurred in the 3 days prior to the implementation of barrier precautions. In addition, adherence to barrier precautions was monitored and found to be suboptimal, with a median of 77 to 82% of HCWs using gloves or gowns, and only 69% compliance with hand hygiene after contact with colonized patients. Combined, these factors may have contributed to decreasing the potential impact of screening and infection control efforts. Importantly, this highlights the real-world difficulty in implementing an effective screening and infection control program and underscores the fact that laboratory screening alone is insufficient to significantly impact the rate of HAIs.

Cost analysis of active screening.

Active screening for VRE can carry a substantial monetary burden for the laboratory, including the cost of materials and labor, that may be difficult to predict prior to implementation. The total cost of screening is dependent upon several factors, including the patient population that is screened (hospital wide versus select ICU populations), the method of screening (culture based versus molecular) and other factors, including the potential for automation. One laboratory supporting a 19-bed ICU performing rectal swabs on all patients at admission, weekly during hospital stay, and at discharge determined that the active screening cost to their laboratory was $30.66 per patient or $22,956 per year ($29,570 in 2015 using the Consumer Price Index), inclusive of labor, laboratory supplies for culture, specimen collection supplies, and processing costs (55). Implementation of additional infection control practices based on a positive screen result increases the overall cost to the hospital as well. In one study, this cost was estimated at $116,515 ($224/patient) annually for additional infection control measures in a 22-bed adult oncology unit (56). However, implementation of this program resulted in a net savings of $189,318 annually ($294,433 in 2015), based on a reduction in VRE BSI from 2.1/1,000 patient-days to 0.45/1,000 patient-days (estimated $123,081), reduced VRE colonization (estimated $2,755), and reductions in antimicrobial use (estimated $179,997). Compared with the previous reports, active screening alone is cost-effective if it reduces 1 to 2 VRE BSI events a year, whereas larger infection control programs may need to prevent multiple infections a year to be cost-effective. Long-term costs of specialized chromogenic VRE screening media can be offset if the laboratory adopts automated systems that can plate, incubate, and analyze media by reducing the cost of technologists' time (57).

LABORATORY CONSIDERATIONS FOR VRE SCREENING

Optimal specimen collection.

When screening patients for a specific pathogen, it is essential that the optimal specimen is collected to achieve maximum sensitivity. The CDC recommends screening for VRE using either rectal swabs or stool specimens. The use of stool specimens submitted to the laboratory for C. difficile detection provides a noninvasive specimen and targets patients with factors that also predispose them to a higher VRE colonization rate. In one study, stools submitted for C. difficile testing had a VRE positivity rate of 10.4%, compared with 9.7% positivity for rectal swabs collected from high-risk (e.g., bone marrow and solid organ transplant and surgical ICU) patients based on culture using selective media (58). Other studies suggest that stool specimens may be superior to rectal swab specimens for recovery of VRE. A direct comparison of paired stool specimens and rectal swabs cultured using Enterococcosel agar (BD, Franklin Lakes, NJ) demonstrated a sensitivity of only 58% for rectal swab specimens compared to stool specimens (59). Enrollment for this study was low (n = 35 positive specimens), but additional dilution experiments found that rectal swabs collected from patients with ≤4.5 log₁₀ CFU of VRE/g of stool were unlikely to be detected by plating to selective media. Although potentially more sensitive, the clinical significance of these findings is unknown, since there have been no studies establishing the relative risk of transmission by patients harboring lower concentrations of VRE.

Culture-based screening methods.

Laboratories that choose to offer screening for VRE need to consider test sensitivity, complexity, turnaround time, and cost. Chromogenic agars are relatively inexpensive and provide a valuable tool for the laboratory. These agars and have been successfully used to screen for MRSA colonization as a component of infection prevention efforts (60, 61). Several VRE chromogenic agars that can identify VRE from stool specimens are currently FDA cleared and commercially available. These media may also differentiate between E. faecium and E. faecalis. In general, chromogenic agars have superior sensitivity, ranging from 90 to 99%, compared to bile esculin azide agar containing vancomycin (BEAV), which is approximately 85% sensitive (62, 63). The specificity of chromogenic agars is also superior to BEAV, which can be as low as 70 to 75%. A summary of the performance of these agars can be found in Table 2 (62–68). False-positive results due to breakthrough growth of non-E. faecalis, non-E. faecium isolates are less common on chromogenic media than on BEAV, which frequently supports the growth E. casseliflavus, E. gallinarum, Leuconostoc spp., and Pediococcus species. These species harbor low-level vancomycin resistance and are also esculin positive, making them difficult to distinguish from E. faecalis and E. faecium. In a study comparing 5 chromogenic agars, non-VRE isolates were recovered from 16.5% of specimens using BEAV, whereas isolation of non-VRE organisms on chromogenic agar ranged from 0.9% to 5.6% of the specimens tested (62). These differences are likely due to the lower concentration of vancomycin used by BEAV, 6 μg/ml, versus 8 to 10 μg/ml used in chromogenic media, as well as the specific colony color imparted by the chromogens used. While more sensitive than BEAV, chromogenic agar still requires 18 to 24 h of incubation, which can delay the implementation of infection control measures.

TABLE 2.

Summary of published performance data on FDA-cleared chromogenic agar for VRE

Detection methoda	Sensitivity (%)	Specificity (%)	Time to result (h)	Notes	Reference(s)
BEAVb	84.8–87.6	73–100	48–72	Breakthrough of nonpathogenic E. gallinarum and E. casseliflavus due to intrinsic resistance. False-positive results if Leuconostoc or Pediococcus spp. are present in specimen	58–61
Chromogenic
chromID VRE	86.3–98.2	97.5–100	48 (read at 24 and 48)	Differentiates E. faecalis from E. faecium by color	58, 61, 62
VRESelect	91.9–98.7	99.0–99.7	24–28	Differentiates E. faecalis from E. faecium by color	58, 59
Spectra VRE	93.9–98.2	99.–99.7	24	Uses α-galactosidase to differentiate E. faecium from E. faecalis	58, 60
CHROMagar	98.2–98.6	96.5–99.1	24	Differentiates pathogenic vs nonpathogenic Enterococcus by color	58, 62
Nucleic acid amplification test				Non-Enterococcus strains can carry vanA or vanB
BD GeneOhm	91.8	93.6	12	Positive predictive value, 66.6%	63
Roche LightCycler detection kitc	73.3, 85.4	99.6, 83.9	∼4	Positive predictive value, 68.5% and 1.42%	64
Cepheid Xpert	61–100	69.3–99.5	1	Positive predictive values range from 2–32% for vanB and 66–92% for vanA	71–73, 75

^aGold standard used to confirm VRE was Gram stain, followed by catalase and pyrrolidonyl arylamidase (PYR) tests. Etests were used to confirm vancomycin resistance.

^bBEAV, bile esculin azide agar containing vancomycin.

^cValues for Roche LightCycler are for for vanA and vanB isolates, respectively.

Molecular screening methods.

Nucleic acid amplification tests (NAATs) have been applied tor VRE screening and detection using both stool and rectal swab specimens. Among these assays are both laboratory-developed tests (LDTs) and commercially developed assays, such as the Roche LightCycler VRE (vanA and vanB) detection assay, the Cepheid Xpert vanA and vanB assay, and the Cepheid Xpert vanA assay. The Cepheid Xpert vanA assay is FDA cleared, and the Xpert vanA and vanB assay is CE marked. Currently, the Roche LightCycler VRE detection kit is marketed as research use only (RUO).

The performance of molecular assays is varied, with sensitivity ranging from 61.5 to 100% and specificity ranging from 14.7 to 99.5% (69, 70). The wide difference in assay performance is due to several factors, including the gold-standard method used for culture comparison and the prevalence of VRE in the study population. The Roche LightCycler VRE detection kit demonstrated a high negative predictive value (NPV) of 99.9% for VRE but suffered low positive predictive values (PPV) of 1.4% for vanB and 68.5% vanA when results were compared to direct culture using Enterococcosel agar (68). A significantly higher PPV of 92.0% was reported using the Cepheid Xpert vanA and vanB assay, but reference culture was performed using a broth enrichment step prior to plating, which may increase the recovery of VRE in culture (71). Interestingly, a similar study evaluating the Xpert vanA and vanB assay using broth-enriched culture reported a PPV of only 32% (72). These data suggest that the reference culture method can affect assay performance data, but additional factors also contribute to the poor PPV commonly observed with these assays.

A major factor impacting PPV and NPV calculations is the pretest probability or prevalence of a given target, in this case VRE, in the population being studied. In the two studies that reported a PPV greater than 90%, VRE prevalence was 30% and 47% among the specimens tested, whereas both studies reporting low PPVs of 1.4% and 2.6% had VRE prevalence of 1.3% and 1.4%, respectively (68, 69, 71, 73). These data suggest that the utility of molecular assays to serve as a “rule-in” test is limited; however, these assays can still be effectively used to identify noncolonized patients with a high NPV. An obvious advantage of molecular assays is the potential to reduce turnaround time (TAT) by 18 to 72 h compared with culture, thereby enabling rapid screening to rule out VRE. The short TAT may be especially beneficial for facilities that implement strategies where all new or returning high-risk patients are treated as VRE patients until a screen returns negative. However, the additional cost of primary screening using comparatively expensive molecular assays must be weighed against the cost of unnecessary isolation measures, which may include gowns, gloves, and private rooms.

Another factor contributing to the reported low specificity and PPV of NAATs is the presence of vanA and vanB in multiple bacterial species other than Enterococcus spp., including Streptococcus mitis, Streptococcus bovis, Eubacterium lenta, Ruminococcus spp., Lactococcus spp., Leuconostoc spp., and some Clostridium species (68, 74). Many of these organisms are ubiquitous colonizers of the human gastrointestinal (GI) tract, and vanB carriage can be high in both healthy and sick individuals. Graham et al. observed high rates of nonenterococcal vanB carriage in community adults (63%), children (27%), and hemodialysis patients (27%) (75). Currently available molecular tests do not link the van genes to Enterococcus; therefore, they cannot differentiate between vanA and vanB carried by Enterococcus versus other bacterial species. This shortcoming can be minimized through the use of an assay targeting only vanA, provided it is used in a geographic region with a low prevalence of Enterococcus spp. harboring vanB. Alternatively, preenrichment of specimens for 16 to 24 h in selective broth containing 16 mg/liter amoxicillin, 20 mg/liter amphotericin B, 20 mg/liter aztreonam, and 20 mg/liter colistin significantly increased Xpert vanA and vanB assay specificity when coupled with a reduced cycle threshold (C_T) for calling positive results (76). Following selective broth enrichment, a C_T of ≤25 demonstrated a sensitivity, specificity, PPV, and NPV of 97%, 100%, 100%, and 99.5%, respectively. This is compared to values of 100%, 77.3%, 41%, and 100%, respectively, for nonenriched samples using the assay default of C_T ≤35 for calling a positive result. Importantly, broth-enriched specimens with a C_T of >30 were reliably reported as true negative; however, specimens with C_T values of 25 to 30 were variable and required culture confirmation before reporting. In addition to requiring independent validation of modified specimen and C_T thresholds, broth enrichment also delays the reporting of results by 18 to 24 h. This delay abrogates one of the key advantages of molecular assays, namely, rapid TAT, and mirrors the time to result using chromogenic agars, which, in general, are less expensive per specimen than molecular assays. Finally, the use of NAAT as a primary screen for VRE means that organisms are not recovered and therefore are not available for strain typing to aid in outbreak epidemiology.

Conclusion.

Reducing the total prevalence of VRE and VRE infections in individual hospital settings requires multiple strategies, including antimicrobial stewardship, active screening of at-risk patients, strict adherence to hand hygiene, and effective room cleaning after discharge. Laboratories considering implementation of the CDC recommendations to perform active screening should look at several factors to ensure that implementation would be beneficial and cost-effective. Hospital settings that have a high prevalence of VRE infection have the potential to achieve the most dramatic reductions in VRE HAIs and associated total cost of care per patient. For hospitals with a low prevalence of VRE, the cost of additional screening and infection control measures needs to be weighed against the risk and frequency of VRE infections. However, because the cost of BSI can be $20,000 to $30,000 per patient, the prevention of a single VRE BSI per year may offset the cost of screening and infection control measures (77).

The choice of screening methodology can impact the success of infection control measures and the total cost of the program. The low PPV of molecular testing methods and comparatively high cost may make it difficult to justify NAAT-based primary screening for all hospital admissions. Conversely, culture is relatively inexpensive but delays results by at least 18 h, which in turn can reduce the effectiveness of infection control measures. One possible solution is to reserve NAATs for an initial admission screen of patients at highest risk for infection (ICUs and transplant wards), with subsequent weekly surveillance conducted by culture. This workflow removes the initial 18- to 24-h window for transmission when using culture and allows for the immediate isolation of patients with a positive result until culture confirmation is available. However, this method may be costly for hospitals with a large number of high-risk admissions and may be impractical for hospitals with limited space dedicated for single-patient isolation rooms.

Regardless of screening methodology (culture versus NAAT) or approach (targeted versus universal, active versus passive), it is essential to ensure communication between the laboratory, primary care providers, and infection control practitioners to ensure appropriate implementation of infection control programs following a positive laboratory result. Inadequate or delayed interventional measures or lack of adherence to these programs will result in a failure of infection control efforts and adds unnecessary cost to patient care without providing added value.

Biography

Blake W. Buchan, Ph.D., D(ABMM), is an Assistant Professor of Pathology at the Medical College of Wisconsin and Associate Director of Microbiology at Wisconsin Diagnostic Laboratories, Milwaukee, WI. He was awarded a Ph.D. in microbiology from the University of Iowa in 2009 for his work elucidating regulatory mechanisms for the expression of virulence factors in Francisella tularensis and completed postdoctoral training in Clinical Microbiology at the Medical College of Wisconsin in 2011. Dr. Buchan's interests include matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), PCR, and microarray-based diagnostics for the detection of bacterial, viral, and fungal pathogens, and he has served as Principal Investigator for numerous clinical trials evaluating novel diagnostic assays. Dr. Buchan is involved with several professional societies, including ASM and SCACM, and is a member of the editorial board for the Journal of Clinical Microbiology. Dr. Buchan has a strong record of publication in peer-reviewed journals and has presented at numerous scientific meetings, including ASM, ECCMID, CVS, IDSA, SCACM, and ICAAC.