Who Is at Risk for Colonization or Infection with VRE?

In the United States, colonization and infection with VRE primarily affect moderately to severely ill patients in acute-care hospitals. Thus most reports are from intensive care units (ICUs), oncology wards, or transplantation units. Patients in other health care settings, including long-term-care facilities and outpatient dialysis units, also have been found to carry VRE [8]. Although patients are usually determined to have become colonized during an antecedent hospitalization, cross-transmission of VRE in these settings has been observed [9]. Host risk factors for colonization or infection with VRE include advanced age, severity of underlying illness, hematologic malignancy, neutropenia, cirrhosis, recent intra-abdominal surgery, renal dialysis, prior nosocomial infection, and the presence of pressure sores [4–7]. In multivariate analyses, the most common host factor associated with infection due to VRE is severe underlying illness.

Antibiotic exposure has often been found to be a risk factor for infection or colonization with VRE; in particular, exposure to vancomycin, cephalosporins, or drugs with activity against anaerobes [10–13]. Other exposures that have been found to be associated with infection or colonization with VRE include length of hospital stay, transfer from another hospital, transfer between hospital floors, proximity to a known infected or colonized patient, receipt of enteral feedings, and receipt of sucralfate [12, 14–18]. Multivariate analyses show that the most significant risk factors include the length of hospital stay, proximity to another patient colonized or infected with VRE, and transfer within a hospital.

Epidemiology of the Transmission of VRE Infection

Vancomycin resistance does not emerge de novo in susceptible enterococci exposed to vancomycin. Therefore, the first step toward infection with VRE must be exposure to a resistant enterococcal strain. A number of epidemiologic studies have attempted to delineate the ways in which patients come in contact with VRE. When interpreting these studies, it is important to keep two concepts in mind. First, because colonization with VRE generally precedes infection and because there may be 10 times more colonized than infected patients [15], the most illuminating studies are those that investigate the epidemiology of colonization with VRE. Second, the epidemiology of the spread of VRE may be different during an epidemic caused by a single strain of VRE than in a setting where VRE are endemic and many different strains are circulating.

Boyce et al. [16] reported an investigation of a monoclonal outbreak of VanB E. faecium in an ICU. Comparison of 12 case patients with matched control subjects revealed that the primary risk factors for colonization with VRE were proximity to a case or exposure to a nurse who cared for a case on the same shift. The epidemic strain was recovered from 26 (28%) of 92 cultures of specimens from environmental surfaces in the rooms of case patients; environmental contamination was significantly more likely to be detected if the patient had diarrhea. These results suggest that within the ICU, VRE spread primarily by cross-contamination. However, the relative importance of case patients and of the environment as reservoirs for VRE was not determined.

Bonten et al. [13] expanded on these findings in a prospective cohort study of patient colonization and environmental contamination with VRE in a 16-bed medical ICU. Thirty-eight mechanically ventilated patients were studied over a 7-week period. Nine patients (24%) were already colonized with VRE when they were admitted to the medical ICU. Of the remaining 29 patients, 12 (41%) became colonized with VRE during their ICU stay. Colonization of patient body sites was persistent, even on the intact skin of upper extremities; 75% of cultures of arm skin specimens remained positive for VRE after the first positive culture. Environmental contamination was common, although it was found in a smaller amount and was more transient than was patient body site colonization. Molecular epidemiologic analysis revealed that, as in the study by Boyce et al. [16], cross-transmission appeared to be the primary means of the spread of VRE within the medical ICU, accounting for 85% of acquired colonization. Unlike the situation described by Boyce et al., however, VRE were endemic at this hospital. Twenty different strains of VRE were identified in the medical ICU during the study period. Another important observation of this study was that cultures of clinical specimens from only 5 patients yielded VRE. Thus most colonized patients would have gone undetected if active screening for VRE had not been done.

This study [13] demonstrated that the epidemiology of VRE colonization in a setting where they are VRE are endemic is more complex than it is in a monoclonal epidemic. Of note, admission of patients who were already colonized with VRE to the medical ICU added as many new cases as did cross-transmission. The admission of patients who are already colonized with VRE to a unit or hospital has been shown to be important in other investigations as well [17].

In another study, Bonten et al. [19] identified an additional variable important to the epidemiology of VRE colonization and infection in ICUs: “colonization pressure.” This variable was defined as the proportion of patients colonized with VRE during a given period. The investigators studied 153 patients who were at risk for colonization with VRE in a medical ICU where VRE were endemic and where multiple different strains were circulating; 45 patients (29%) acquired rectal colonization with VRE. Twelve variables were entered separately into a Cox regression model in which the number of days until acquisition of VRE colonization was the dependent variable. When all variables were tested as covariates, the best model included colonization pressure, proportion of days with enteral feeding, and proportion of days with cephalosporin use. When the model (colonization pressure and either cephalosporin use or enteral feeding) was used to calculate the median time until colonization with VRE, colonization pressure had the greatest effect. In fact, once colonization pressure was >50%, the other variables had only a slight effect on time to acquisition of VRE colonization. The investigators reasoned that a lapse in infection control practices was more likely to result in cross-transmission of VRE in an ICU where most patients were colonized with VRE than in an ICU where few patients were colonized. A limitation of this study was that compliance rates were ICU specific, not patient specific. The resultant small range of compliance rates may have masked the importance of infection control measures in preventing cross-transmission of VRE.

Evolution of an Outbreak of VRE Infection

The studies by Boyce et al. [16] and Bonten et al. [13] illustrate opposite ends of the spectrum of the epidemiology of colonization and infection with VRE: monoclonal outbreak and polyclonal endemicity. Several recent investigations suggest that a monoclonal outbreak, if not brought under control, can evolve into polyclonal endemicity. Kim et al. [20] reported a molecular epidemiologic analysis of 183 patients with bacteremia due to VRE at a single hospital over a 6-year period: from the time VRE were first detected until they became endemic. For the first 3 years of the study, the same VanB E. faecium strain was isolated from 69% of patients, which suggested that crosstransmission was a major means of the spread of VRE. Even in the fifth year of the study, when 12 different strains of VRE were identified, 2 strains caused >60% of infections. Gene transfer also appeared to play a role in the dissemination of resistance; the same conjugative VanA plasmid was found in 4 of 8 VanA E. faecium strains. A total of 24 strains were identified during the study period; the extent to which they represented introduction of new strains of VRE to the hospital or transfer of resistance elements to vancomycin-susceptible enterococcal strains could not be determined.

These findings recently have been corroborated by reports from other institutions that analyzed both clinical isolates of VRE from multiple different sources and isolates of VRE from stool or rectal cultures [21, 22]. The prominent role that the horizontal spread of strains of VRE appears to play during outbreaks and in settings where they are endemic implies that interventions to prevent cross-transmission should be included in all control programs for VRE.

Importance of Hand Contamination

Most hospitalized patients acquire VRE from another patient, presumably via the contaminated hands of a health care worker. This presumption is based primarily on the isolation of identical strains of VRE from patient specimens and from the hands of health care workers. In epidemiologic investigations of contamination of health care workers' hands with VRE, 0 to 41% of hands sampled were positive for VRE [23–26]. Although VRE have been isolated from the fingers of volunteers 60 min after inoculation [27], it is not known whether VRE can colonize the hands of health care workers for prolonged periods. However, the fact that hand washing is effective in removing VRE suggests that if colonization occurs, it is transient.

In a clinical and epidemiologic study by Bonilla et al. [23], a 15-s wash with 0.6% parachlorometaxylenol (Medi-Scrub, Huntington Laboratories, Huntington, IN) removed VRE from most health care worker hands tested by culture. VRE could still be recovered from 3 (5%) of 60 health care workers, however. When tested formally in the laboratory, a 30-s scrub with either 0.6% parachlorometaxylenol or a bland soap preparation (Germa-care, Huntington Laboratories) effectively removed 10² cfu of VRE that had been inoculated onto volunteers' hands, while a 5-s scrub with the same agents eliminated 99% of the organisms [27]. In contrast, another study found that a 30-s wash with a disinfectant product (60% isopropyl alcohol, alcoholic chlorhexidine, chlorhexidine skin cleanser, or povidone-iodine) was more effective at eliminating 10⁴ cfu of VRE from subjects' hands than was bland soap and water [28]. These findings suggest that length of contact with the cleansing agent is important in removing VRE from hands and that disinfectants may be more effective than bland soap in cleaning hands that are heavily contaminated with VRE.

Finally, it appears to be necessary for health care workers to wash their hands after contact with a patient colonized or infected with VRE even if gloves are worn during contact. This necessity was demonstrated in a study in which 17 (34%) of 50 health care givers who had contact with a patient colonized or infected with VRE acquired the patientś strain of VRE on their gloves; 5 of the 17 health care givers were also found to have VRE on their hands after glove removal [26]. In this study, glove contamination was associated with duration of patient contact, contact with patient bodily fluids, patient skin contamination with VRE, and diarrhea.

Is the Hospital Environment an Important Reservoir for VRE?

The significance of the inanimate environment as a source for transmission of VRE is not well defined. Evidence that the environment is an important reservoir for VRE includes the following: (1) After inoculation onto bed rails and other surfaces, VRE are able to survive for up to 24 h with minimal reduction in the number of colonies [27]. (2) Enterococci, including VRE, are relatively resistant to killing by heat and to removal by washing with bland soap [27–29]. (3) In a number of epidemiologic investigations, VRE were isolated from environmental surfaces in patient rooms, including patient and health care worker gowns, door handles, cabinets, over-bed tables, floors, iv fluid pumps, electrocardiogram monitors, bed rails, blood pressure cuffs, urinals, bedpans, toilet seats, and bed linen [5, 10, 16, 23, 24, 30–33]. In the study by Boyce et al. [16], an outbreak of VRE was not controlled until all health care workers entering the rooms of patients who were infected or colonized with VRE donned cover gowns as well as gloves. This intervention may have decreased contact between health care workers and VRE in the patientś environment. (4) In the study by Bonten et al. [13], 23% of patients whose rooms became contaminated with VRE became colonized. (5) Two retrospective, case-control studies implicated contaminated fomites as the source of outbreaks of VRE [30, 31]. In a third study, cross-transmission of VRE was linked to a contaminated fluidized bed [34].

None of these investigations provides incontrovertible evidence that hospitalized patients acquire colonization or infection with VRE from the environment. In fact, there are data that implicate a minor role for this mode of transmission. Only the study by Bonten et al. [13] was designed specifically to evaluate the contribution of environmental contamination to transmission of VRE. Therefore, the significance of finding VRE in the environment in the other investigations is uncertain. Even in the study by Bonten et al. [13], contamination with VRE was transient, and low numbers of cfu were present at most environmental sites. Moreover, all of the patients who acquired VRE colonization after their rooms became contaminated carried a strain that was simultaneously present in other patients in the medical ICU; these other patients were therefore also possible sources of cross-colonization. Furthermore, the number of patient body sites colonized with VRE was highly associated with environmental contamination in a patientś room, suggesting that the patient contaminated the environment and not vice versa.

VRE commonly contaminate environmental surfaces on hospital wards that house patients colonized or infected with VRE. It remains unclear whether patients can become colonized with VRE from contact with contaminated surfaces, either directly or via the hands of health care workers. Until the importance of the hospital environment as a reservoir for VRE can be determined with certainty, it is probably safest to assume that transmission from the environment to patients can occur and to take precautions to try to prevent such spread.

A Mathematical Model to Describe the Spread of VRE

Using data from an earlier study [15], Austin et al. [35] created a mathematical description of the spread of VRE in the ICU (figure 1). They extrapolated the Ross-Macdonald model of vector-borne disease transmission, assuming that health care workers were “vectors” who transmitted VRE indirectly from one patient (“host”) to another. Because environmental contamination with VRE had been found to occur at a low level and to be transient in their previous studies and to simplify the model, environmental contamination was not considered.

A set of differential equations was used to describe the dynamics of cross-transmission of VRE in the ICU. Austin et al. first defined R_o, the basic reproductive number, as the average number of secondary colonized patients generated by a primary case in an ICU formerly free of VRE. If R_o is >1, each colonized patient can generate ⩾1 subsequent case of colonization and an epidemic can occur. If R_o is <1, an outbreak cannot be sustained. The investigators assumed further that patients remained colonized with VRE for the duration of their ICU stay but that health care workers were contaminated with VRE only transiently. Therefore, each colonized patient was thought to contaminate many health care workers, but each health care worker was expected to transmit VRE to few patients.

The investigators then examined the effect on the model of the following factors: antibiotic restriction, hand washing, and grouping patients into cohorts. They calculated that, given previously reported rates of hand washing compliance of only 20%–40%, hand washing alone was unlikely to be effective in controlling the spread of VRE in an ICU where R_o was high. However, grouping nursing contacts into cohorts had a very high theoretical chance of controlling cross-transmission; if most of a nurseś contacts were with a single patient colonized with VRE, the chance that the nurse would transmit VRE from this patient to another would be small. A high rate of antibiotic exposure was assumed to provide a selective advantage for the survival of VRE. Antibiotic restriction would then be expected to decrease the selective advantage and, by extrapolation, the probability of transmission of VRE.

The equations were then used to calculate R_o for a single medical ICU where detailed clinical, demographic, and microbiological information about colonization with VRE had been collected [15]. For the medical ICU analyzed, R_o was calculated to be between 3 and 4, which would have resulted in a prevalence of colonization with VRE of almost 80%. The observed prevalence was ∼36%. The difference in these values was attributed to a reduction in transmission effected by implementation of infection control measures. The frequent admission of patients who were already colonized with VRE to the medical ICU was thought to stabilize the prevalence further.

This model makes several assumptions that require further testing. Nevertheless, its mathematical framework to describe the transmission of VRE in the ICU setting may allow more accurate determination of the effect of control strategies in the future.

Control Measures

The survival ability of VRE may be unique among nosocomial bacterial pathogens. Weinstein and colleagues noted that VRE pose a “triple threat” to hospitalized patients. They are able to colonize the gastrointestinal tract and skin of patients, thus posing a risk similar to that of antibiotic-resistant gram-negative rods and methicillin-resistant S. aureus combined. They also colonize the hospital environment, presenting a risk similar to that of Clostridium difficile. Control of VRE therefore presents a special challenge to hospitals. The steady increase in the number of clinical isolates of VRE reported from US hospitals suggests that this challenge has yet to be confronted successfully.

In 1995, the Hospital Infection Control Practices Advisory Committee of the Centers for Disease Control and Prevention published recommendations for preventing the spread of vancomycin resistance [36]; these recommendations are currently under revision. They include more prudent vancomycin use, education of hospital staff about the problem of VRE, guidelines for accurate detection of VRE by the clinical microbiology laboratory, guidelines for active screening for carriage of VRE, implementation of contact precautions for patients who are known to be infected or colonized with VRE, and cohort grouping of patients colonized or infected with VRE and of the health care workers who care for them. Since the publication of these recommendations, the efficacy of several of the interventions has been tested, and the results of these evaluations have been reported [10, 15, 37].

Two groups of investigators studied restriction of vancomycin use as a means of controlling colonization and infection with VRE. Morris et al. [10] restricted iv and oral vancomycin use over a 7-month period at a large university hospital where serial point prevalence surveys revealed that the endemic prevalence of endemic VRE was 20%. Infection control measures, including physical separation of patients colonized or infected with VRE and cohort grouping of patients and health care workers, were also implemented. During the study, iv vancomycin use fell by 59%, and use of oral vancomycin declined by 85%. However, the prevalence of colonization with VRE among patients remained unchanged. The investigators speculated that the interventions may have prevented an increase in the prevalence of colonization and that a decrease in the prevalence might have been seen if the study were carried out longer.

In contrast, Quale et al. [37] found that restriction of cefotaxime, clindamycin, and vancomycin resulted in a decrease in rectal colonization and infection with VRE among patients at a Veterans Administration hospital where VRE were endemic. Infection control measures implemented earlier had failed to decrease the prevalence of VRE among clinical isolates. However, several of these measures were continued during the antibiotic restriction period, and a requirement was added to wear a gown when entering the rooms of patients known to carry VRE. Interpretation of these findings is made difficult by several factors. First, the extent of compliance with infection control interventions was not reported. Second, antibiotic consumption was not normalized for total patient days or number of admissions. Third, the reported decrease in the mean number of clinical cases of colonization or infection with VRE was moderate, with a large SD that overlapped with that of the mean number of cases reported before antibiotic restriction. Finally, the decrease in rectal colonization that was reported was based on results of a single point prevalence survey. Given the natural variability in rates of colonization that has been reported in studies done in other settings of endemicity [13, 15], this difference may not have been significant. Even if one assumes that the intervention was effective, the prevalence of colonization with VRE among patients remained 15%, illustrating the difficulty of eradicating VRE in settings where they are endemic.

The recommended isolation precautions for people who come in contact with patients known to be infected or colonized with VRE include wearing clean, nonsterile gloves when entering the patientś room and wearing a clean nonsterile cover gown, if extensive contact with the patient or the environment is anticipated or if the patient has diarrhea, fecal incontinence, an ileostomy, a colostomy, or drainage of a wound not covered by a dressing [36]. These recommendations are based on the finding of extensive environmental contamination in several studies [16, 33], as well as the failure to control a monoclonal outbreak of VRE colonization and infection until gowning was mandated for all contacts with patients who carried VRE [16].

Slaughter et al. [15] tested the efficacy of this recommendation in a study comparing the effect of universal use of gloves and gowns with the effect of glove use alone in a medical ICU where VRE were endemic. Rectal specimens for culture were obtained daily from all patients, and compliance with precautions—4363 total observations—was monitored weekly. Fifteen percent of 181 study patients were already colonized with VRE at the time of ICU admission, and 25% of the remaining patients acquired VRE colonization. Although there was no difference between the 2 patient groups in terms of acquisition of rectal colonization with VRE, compliance with gloving and hand washing was significantly better in the cohort of health care workers that wore gloves and gowns than in the cohort that wore only gloves. In this setting, gowns seemed to increase health care worker awareness of the infection control problem and of the need for isolation precautions. Furthermore, when these data were analyzed in the transmission model of Austin et al. [35], implementation of infection control measures was calculated to have decreased the transmission of VRE.

What measures can be recommended to control VRE? Before answering this question, it is important to estimate the risk that infection due to VRE poses to the patient population under consideration. For example, one might consider more aggressive measures in a liver transplantation unit, where vancomycin resistance has been shown to independently predict enterococcus-related mortality [7], than on a general medicine or surgical ward, where serious infections with VRE have been reported less frequently. A second factor that should be taken into account is the current molecular epidemiology of the spread of VRE in the hospital or ward where the control program will be implemented. The natural history of uncontrolled VRE in hospitals has been shown to progress from rare or sporadic cases of colonization or infection with VRE, to a monoclonal outbreak, and finally to polyclonal endemicity [20–22].

Epidemiologic studies have demonstrated the importance of cross-contamination in the spread of VRE within hospitals. Most reports of successful control programs have come from single units where one predominant strain of VRE was colonizing or infecting patients (monoclonal outbreak) [16, 37]. In these situations, aggressive use of barrier precautions, prospective surveillance, and placing patients in private rooms have been successful in containing outbreaks. In a monoclonal outbreak in a neonatal ICU, establishing separate cohorts of patients who were not colonized, colonized patients, and patients exposed to a colonized patient played a prominent role in controlling the epidemic strain [37].

Control in health care institutions where VRE are endemic is more problematic. Eradication of VRE from these settings may not be a realistic goal, especially because colonized patients will probably continue to be transferred into them from other hospital wards or health care institutions. However, some investigators have been successful in decreasing the transmission of VRE in settings where they are endemic. Montecalvo et al. [38] reported that institution of a comprehensive program of enhanced infection control measures, coupled with an overall decrease in antibiotic use, resulted in a 50% reduction in colonization with VRE and a 4.7-fold decrease in bacteremia due to VRE on an oncology ward where VRE were endemic. Bodnar et al. [39] reported a novel strategy of molecular epidemiologic analysis of apparent clusters of VRE that appeared against a background of endemicity. In that study, identification of a single clone infecting 12 seemingly unrelated patients prompted an in-depth chart review that revealed links among the patients. Targeted infection control measures were then instituted, and further transmission of the epidemic strain was halted.

Another approach to control that has received recent attention is that of “universal gloving” or donning gloves for contacts with all patients, regardless of whether they are colonized with VRE [40]. This measure may be particularly appealing in settings where the endemic prevalence of VRE is high or when traditional contact isolation might interfere with a patientś treatment, such as in a rehabilitation facility. Preliminary results from a study comparing universal gloving with standard contact precautions in a long-term-care facility showed that universal gloving was well accepted by health care workers and that acquisition of colonization with VRE was comparable between patients in the 2 study arms [40]. Finally, it should be noted that interventions that are effective in one setting may be ineffective in another. For example, screening and aggressive institution of contact precautions, including the use of cover gowns, may be sufficient to control a monoclonal outbreak of VRE [16] but may have little or no effect once VRE have become endemic [15]. Proposed control strategies specific to each of the 3 stages of evolution of VRE, which are based on previously reported findings as well as on expert opinion, are shown in figure 2.

VRE will probably continue to be problem pathogens for hospitals and other health care settings. Although eradication of VRE may not be possible in institutions where VRE are endemic, efforts to reduce the prevalence of colonization and infection should continue. Innovative intervention strategies and multifaceted programs that take into consideration the local epidemiology of the spread of VRE may yield the best outcomes.