Abstract
Background and objectives: Vancomycin-resistant enterococci (VRE) are recovered with increasing frequency among patients with chronic renal failure, making VRE a major concern in nephrology departments, especially for patients who are treated by hemodialysis. We report herein the successful aggressive management of a VRE outbreak in a nephrology department.
Design, setting, participants, & measurements: An Enterococcus faecium vanB strain was isolated from a peritoneal dialysis solution from an inpatient. Immediately, infection control measures were enforced and active screening was performed for all contact patients. Carriers were isolated, and patients were divided into three cohorts: Positive, contact, and noncontact patients. We then performed a case-control study to understand risk factors for VRE carriage comparing VRE carriers with contact patients who were negative for VRE.
Results: A total of 14 VRE-positive and 125 VRE-negative contact patients were identified. VRE-positive patients were more likely to receive hemodialysis and have longer hospital stays in nephrology. VRE-positive patients more often had a central venous catheter for a longer period of time and received more antibiotics than VRE-negative patients. Treatment with large-spectrum β-lactams and number of days in the nephrology ward were significantly associated with a higher risk for VRE carriage by using multivariate analysis.
Conclusions: These findings suggest that case mix, longer hospital stays, and antibiotic use are major risk factors for VRE acquisition. In addition, it demonstrates that strict enforcement of isolation precautions and cohorting associated with active screening are successful to curb the transmission of VRE in renal units despite continuous colonization pressure.
More than 70% of the bacteria that cause hospital-acquired infections are resistant to at least one of the drugs most commonly used to treat these infections (1). Among multidrug-resistant organisms (MDROs), glycopeptide-resistant or vancomycin-resistant enterococci (VRE) became a concern throughout the world during the 1990s. Most enterococcal infections are caused by Enterococcus faecalis. The remaining infections are mostly caused by E. faecium, a species that is more likely than E. faecalis to be resistant, even to antibiotics of last resort as glycopeptides. Resistance to the latter compounds is most frequently of vanA phenotype (i.e., acquired and high-level resistance to vancomycin and teicoplanin). The VanB phenotype, characterized by variable levels of resistance to vancomycin with usually in vitro susceptibility to teicoplanin, is less common (1).
Early studies dealing with VRE outbreaks in the United States revealed that most patients were in intensive care units (2–4); however, VRE is now recovered with increasing frequency among patients with chronic renal failure (2,5), making VRE a major concern in nephrology departments, especially for patients who are treated by hemodialysis. VRE was seldom isolated in French hospitals until recently (6–9). For example, fewer than five VRE carriers are identified each year in our institution, a large French University Hospital, and most of them are sporadic cases. We report herein the successful aggressive management of a VRE outbreak in the nephrology department of our hospital and the results of a case-control study to understand risk factors for VRE carriage in this specific epidemiologic setting.
Materials and Methods
Pitié-Salpêtrière University Hospital, a 2000-bed tertiary care center located in Paris, implemented isolation precautions for MDRO carriers more than 15 yr ago. These precautions include single-room hospitalization, systematic gown and gloves, dedication of small medical materials such as stethoscopes or sphygmomanometers, daily cleaning of the room, contact screening, and ward visit by infection control staff (mainly nurses). An automatic alert on readmission for patients with MDROs was also implemented more recently. In addition, because of recent alerts of VRE outbreaks in some French hospitals, recommendations for VRE management were recently issued adding VRE-specific administrative measures to the previous contact precautions. These included stop of admission of any new patients in the affected ward, stop of transfer of contact and carrier patient to another ward, and immediately informing the hospital director and health authorities. Patient care and study conduct complied with good clinical practices and the Declaration of Helsinki guidelines.
The nephrology department admits a mean of 700 inpatients per year and is composed of three distinct units, including one hospitalization ward with two aisles (14 and 10 beds), one hemodialysis center (a mean of 6000 hemodialysis sessions per year) in the same floor, and one peritoneal dialysis unit in another building. The hemodialysis center has 10 hemodialysis units in a single large room and two units in single-bed isolation boxes located within the large room.
Microbiology
Screening was performed during the first 2 mo of the outbreak by plating rectal swabs on bile-aesculin-azide agar (BEA; BioMérieux, Marcy l'Etoile, France) supplemented with 6 mg/L vancomycin before and after 24 h of enrichment in BHI broth containing 6 mg/L vancomycin. After the first 2 mo, only direct plating on BEA agar was used because no new cases were detected for the last 6 wk. Suspected strains were identified as VRE by using the GenoType Enterococcus DNA strip (Hain Lifescience, Nehren, Germany), which identifies Enterococcus to the species level as well as the vancomycin resistance genes van. Glycopeptide minimal inhibitory concentrations were determined by using the E-test strip (AB Biodisk, Solna, Sweden) on Mueller-Hinton agar. All isolates were typed using pulse field gel electrophoresis, as described previously (10).
Risk Factors for VRE Acquisition
To find risk factors for VRE acquisition before implementation of isolation precautions, we performed a case-control study comparing VRE carriers with contact patients who were negative for VRE. Case-patients were VRE-positive patients. Control or contact patients had to be VRE-negative after at least one digestive screening and admitted to any of the three wards of the nephrology department from 1 wk before the first day of admission of the index case (December 1, 2006) to the day of full implementation of control measures (January 30, 2007), defining the suspected cross-transmission period. Control subjects were chosen in the three wards because case patients had been admitted to at least one of the three wards during the study period. Data that were abstracted from medical records included medical history, indwelling devices, antibiotic use, and length of stay.
Statistical Analysis
Data were analyzed by using EpiInfo 6.04 (Centers for Disease Control and Prevention and World Health Organization, Atlanta, GA) and Stata 10 (Stata Corp, College Station, TX). The Mann-Whitney test was used to compare continuous variables. The χ2 test or the two-sided Fisher exact test was used as appropriate to compare categorical variables. Multivariate analysis was performed by using unconditional logistic regression models to determine the independent importance of major risk factors for VRE acquisition. Variables with P < 0.2 in univariate analysis or variables that were clinically sound were introduced into the model, and backward analysis was performed. Partially correlated variables were not introduced simultaneously in the models. Models' fitness was assessed by using the Hosmer-Lemeshow test. Only the most parsimonious models (i.e., those with the most significance with the least variables) are given. P values are two-tailed, and P < 0.05 was considered statistically significant.
Results
The Outbreak
A vanB E. faecium strain was isolated from a peritoneal dialysis solution by mid-January 2007 from a patient (index case) who was admitted 1 mo earlier in the hospitalization ward of the nephrology department and who had been in the two aisles of the hospitalization ward. In addition, he had hemodialysis 1 wk before VRE isolation. The patient was finally not considered as infected on clinical and biologic arguments and therefore did not receive any antibiotic to treat VRE colonization. Immediately, he was placed on isolation precautions, and administrative measures were implemented, including the postponing of new patients' admission and of transfer of contact patients. Active screening with rectal swabs was sought for all contact patients, and a total of 13 additional carriers of a vanB VRE strain identical to the index case strain by pulse field gel electrophoresis typing were diagnosed. Most (11 of 13) of the vanB VRE-positive carriers were identified within the 3-wk period after index case identification (Figure 1). One of the two remaining cases was identified on readmission, and the last one had three negative rectal swabs before identification. Health care workers were not screened because the outbreak seemed to be under control in the very few weeks after the implementation of isolation precautions.
Epidemic curve of VRE-positive carriers within the 2-mo period after index case identification.
Patients were divided into two cohorts—colonized, or VRE-positive (n = 14), and contact patients (n = 157)—and placed in separate wards with dedicated staff for each cohort. Isolation precautions were implemented for VRE-positive patients. For contact patients, besides dedicated staff, only reinforcement of hand hygiene and environmental cleaning was used. A new hemodialysis ward with four hemodialysis units dedicated to VRE-positive patients was created. In the mean time, the activity of the hemodialysis center decreased by 30% because new patients could not be cared for according to national recommendations. Case and contact patients were screened weekly for digestive VRE carriage during a 6-mo period and monthly during an additional 6-mo period. Readmission of VRE-positive patients was traced through the automatic MDR-bacteria alert system. All VRE-positive patients who were readmitted to the hospital were systematically hospitalized in the isolation unit of the nephrology ward. Because contact patients could not be traced on readmission through this automatic system (because they were not positive), a special flag was created in the hospital information system so that they were preferentially readmitted to the nephrology ward and immediately placed on preemptive isolation precautions up to a negative rectal sample. No new VRE carrier was identified in the 12-mo follow-up period despite continuous colonization pressure. Indeed, nine of the 14 vanB carriers were readmitted to the nephrology department (mean number of readmission 7.6; range 2.0 to 30.0), and eight had numerous hemodialysis sessions in the new VRE-dedicated hemodialysis center (mean number of sessions 63; range 10 to 140).
VRE carriage remained asymptomatic in all but two patients. One VRE-positive patient had bacteremia, and the other had polymicrobial angiocholitis; both were successfully treated with teicoplanin for approximately 10 d.
Microbiology
A total of 1612 rectal swabs were performed during the 12-mo period. A median of 13.5 (range 2.0 to 27.0) swabs were performed during the follow-up period among the 14 patients who were colonized with the outbreak strain. A total of seven VRE-positive patients were considered as noncarriers during the follow-up (i.e., had a least three consecutive negative follow-up swabs and remained culture negative during the follow-up), and four remained carriers. Finally, one patient died of an unrelated cause, and the two remaining patients were transferred by the end of February and therefore did not have any follow-up culture available in our institution.
Among the 157 contact patients, 32 did not have any rectal swab performed, mainly because they were discharged before the screening period. Among the 125 remaining contact patients, the median number of follow-up rectal swabs was 5 (range 1 to 35). Only the latter were considered for the case-control study.
Patients
Comparison of the characteristics of the 14 case patients who harbored a vanB strain and of those of the 125 contact or control patients is shown in Table 1. VRE-positive patients were more likely than VRE-negative patients to be male (92.9 versus 59.2%; P = 0.02) and to have longer hospital stays in 2006 (9.5 versus 0.0 d; P < 0.001) and in January 2007 (10.0 versus 0.0 d; P < 0.01) in the nephrology ward. VRE-positive patients were more likely to receive hemodialysis (85.7 versus 38.4%; P < 0.01) and to have a central catheter (50.0 versus 21.6%; P = 0.04) for a longer period of time (median 4 versus 0 catheter-days; P < 0.01). VRE-positive patients received significantly more antibiotics than VRE-negative patients in general in 2006 (median 2 versus 0; P < 0.01) and in January 2007 (median 1 versus 0; P < 0.01); specifically, they were more likely to receive large-spectrum β-lactams (third-generation cephalosporins, carbapenems, piperacillin, and tazobactam; 57.1 versus 5.6%; P < 0.01), glycopeptides (21.4 versus 2.4%; P = 0.01), or aminoglycosides (14.3 versus 0.8%; P = 0.03). There was no significant difference between the two groups of patients regarding the other characteristics.
Comparison of characteristics among VRE-positive and VRE-negative patients
In a first multivariate analysis model, the number of days in the nephrology ward in January 2007 (odds ratio [OR] 1.12 for 1 d; 95% confidence interval [CI] 1.03 to 1.24) and the number of different antibiotic drugs received (OR 2.06; 95% CI 1.10 to 4.05) were significantly associated with a higher risk for acquisition of the epidemic VRE strain. In a second model, replacing the number of received antibiotics by exposure to the different antibiotics families, treatment with large-spectrum β-lactams (OR 15.43; 95% CI 2.84 to 97.23), and number of days in the nephrology ward in January 2007 (OR 1.14 for 1 d; 95% CI 1.04 to 1.26) were significantly associated with VRE acquisition.
Discussion
We report an outbreak of an uncommon type of VRE (vanB genotype) in a nephrology ward and present infection control measures that led to successful management of the outbreak. As expected, the main risk factors for VRE acquisition by multivariate analysis was previous antibiotic treatments, specifically β-lactams, and a long duration of hospitalization.
Numerous studies have identified antibiotic therapy as a risk factor for VRE acquisition (4,8,11–15); however, controversies remain on the association between previous intravenous vancomycin use and VRE carriage (16). We confirmed such an association in univariate analysis, but, like other reports (11,17), we found that the only antibiotic group associated with VRE carriage by multivariate analysis was large-spectrum β-lactams. Some studies have shown a link between fluoroquinolones or antibiotics with activity against anaerobes and VRE carriage (2,18), but we did not confirm these associations in this outbreak mainly because very few patients received these compounds. No matter what, because most studies, including ours, showed a positive association between antibiotics and a higher risk for VRE carriage, it is suggested that careful restrictive antibiotic policy be implemented in nephrology wards.
De novo emergence of glycopeptide resistance is unlikely in enterococci (19), and only a few epidemic E. faecium subpopulations are spreading in one country or on the five continents (20). These and other data are concurrent to underline that the increase in VRE during the past decade was mainly related to cross-transmission (4,21–24); however, relying solely on hand hygiene improvement and restrictive antibiotic policy is unlikely to control transmission (25). Therefore, we focused on comprehensive infection control measures, including active screening, because it identifies asymptomatic carriers who are not recognized by using diagnostic samples (4). Of note, no unrelated VRE strain was found among our long-term hemodialysis population, reinforcing the idea that, currently, VRE is not a common pathogen in France (26) and that an aggressive policy is still of major interest to control its spread. Once acquired, intestinal colonization by VRE can last for years (27). This issue is a major concern in nephrology wards because hemodialysis patients have three hemodialysis sessions per week and thus are readmitted numerous times for years. In an effort to decrease risks for cross-transmission during consecutive readmissions, we implemented two specific measures. First, we focused on staff and carrier cohorting and developed a new hemodialysis and hospitalization unit with dedicated staff for all carriers because such a unit has proved to be useful elsewhere for implementation of infection control measures (25). Second, we used an electronic alert system on readmission of carriers (28,29) but also of contact patients. Of note, no alert existed for contact patients, and we had to develop it. Such an alert is of major interest, because contact patients may be discharged before outbreak recognition and therefore not be recognized as potential transmitters in case of readmission.
Finally, the issue of halting isolation precautions for carriers who became culture negative arose among patients who were followed for months or years in the hemodialysis center. The Hospital Infection Control Practices Advisory Committee recommends three consecutive negative cultures, at least 1 wk apart, for determining clearance of VRE from a previously colonized patient (30); however, it is known that VRE may be detectable for a prolonged period, sometimes after numerous negative follow-up cultures (27,31), especially after antibiotic treatment that may favor VRE digestive discharge in seemingly negative patients (31,32). We therefore decided to maintain isolation precautions during more than 12 mo for patients who were colonized by outbreak strains, regardless of the rectal swab results (13,31–33).
Despite successful management of the outbreak, our study has some tradeoffs. Indeed, it is impossible to assess the respective interest of each control measure that was implemented during and after the outbreak. For instance, the cost-effectiveness of the creation of a dedicated dialysis unit and its role in the successful management of the outbreak remain to be evaluated. Its implementation resulted in ‘bed blocking’ with opportunity cost and put pressure on the capacity of the department. On the contrary, because cohorting of patients and staff has been shown to be of major interest in decreasing cross-transmission (25), this unit may have played a major role in outbreak termination and also in the absence of transmission during a prolonged period of time.
Conclusions
We suggest that a successful program for control of VRE outbreak for most units should require effective active surveillance to identify colonized and infected patients, absolute adherence to contact isolation by health care workers, which is more easily achieved by cohorting; and judicious use of antibiotics.
Disclosures
None.
Acknowledgments
We acknowledge the active participation of the entire staff of the nephrology department for the control of the outbreak and especially the efficient management of head nurses. In addition, we are sincerely grateful to head nurses of the infection control unit for constant and proactive support.
Footnotes
-
Published online ahead of print. Publication date available at www.cjasn.org.
- Received May 14, 2009.
- Accepted July 27, 2009.
- Copyright © 2009 by the American Society of Nephrology