A Comparison of the Effect of Universal Use of Gloves and Gowns with That of Glove Use Alone on Acquisition of Vancomycin-Resistant Enterococci in a Medical Intensive Care Unit

Methods

Setting and Study Design

This clinical trial was conducted under the direction of the Infection Control Committee of Cook County Hospital in Chicago, Illinois, a 900-bed public teaching hospital. The newly constructed medical intensive care unit, which is supervised by a staff of intensivists, was opened for patient care in September 1994. It contains 12 beds in single rooms and 4 beds in double rooms. During the 4.5-month study period (26 October 1994 to 7 March 1995), the unit was geographically divided so that eight beds were in a glove-and-gown section and eight were in a glove-only section. Both double rooms were in the glove-only section. Patients were assigned to beds by the charge nurse on the basis of bed availability. Hospital employees were required to wear clean, nonsterile latex gloves when entering a room and to remove them and wash their hands with antibacterial soap before leaving the room. In double rooms, hands were washed and fresh gloves were donned if the health care worker moved from one patient to the other. In addition, employees wore isolation gowns (disposable, nonwoven polypropylene gowns that withstand 11.5 cm hydrostatic pressure and do not absorb water [American Threshold, Enka, North Carolina]) when entering a glove-and-gown room, and they removed them before leaving.

Because many patients have vancomycin-resistant enterococci when they are admitted to the intensive care unit and because identification of colonized patients by culture is often delayed, glove-and-gown or glove-only precautions were instituted immediately after admission for all patients. A sign listing the appropriate precautions was posted at the door of each room. Blood pressure cuffs and thermometers were dedicated for use on individual patients on the basis of current HICPAC recommendations. Rooms were cleaned daily with a phenolic disinfectant. Use of oral vancomycin was restricted to patients who had antibiotic-associated colitis that did not respond to metronidazole.

The records of all patients who stayed in the intensive care unit for 48 hours or more were reviewed, and clinical, laboratory, and demographic data were extracted from those records.

Compliance

At the beginning of the study and every 1 to 2 weeks thereafter, we held educational meetings with medical intensive care unit nurses, house staff, and attending physicians to present the new infection control measures and to provide encouragement and feedback about compliance. Compliance was monitored by unobtrusive observers on all shifts approximately 7 hours per week. Beds were monitored in random order for 10-minute periods. All hospital employees were monitored and were expected to comply; visitors were asked to comply with isolation precautions but were not monitored. Compliance with donning gloves on entry, removing gloves before leaving, and handwashing was monitored for all beds; compliance with donning gowns on entry and removing them before leaving was monitored for glove-and-gown beds. We evaluated compliance on the basis of room entry, regardless of whether the person being evaluated had contact with the patient or objects in the room.

Surveillance Culturing

Rectal swabs (Culturette System, Becton Dickinson Microbiology Systems, Cockeysville, Maryland) were obtained daily from patients in the intensive care unit. Swabs were also obtained 1 week and 2 weeks after discharge from the intensive care unit for patients who remained hospitalized and had had negative surveillance cultures when they left the intensive care unit. Patients were defined as having acquired vancomycin-resistant enterococci in the intensive care unit if a rectal culture positive for vancomycin-resistant enterococci was first obtained more than 3 days after admission.

After construction was completed but before the units were occupied by patients, environmental samples were taken on swabs moistened with modified Stuart transport medium (supplied with the Culturette System) from all patient rooms and from common areas. After the rooms were occupied, samples were taken monthly. Surfaces sampled included those of bed rails, bed linen, countertops, overbed tables, drawer and cabinet handles, light switches, soap dispensers, monitor key pads, infusion equipment, bedside charts, blood pressure cuffs, stethoscopes, bed scales, refrigerator door handles, telephones, computers, and doorknobs.

Microbiological Methods

Rectal swabs were inoculated onto Enterococcosel agar (Becton Dickinson Microbiology Systems) and Columbia colistin-nalidixic acid agar (Difco Laboratories, Detroit, Michigan) supplemented with vancomycin, 6 µg/mL, and amphotericin B, 8 µg/mL. Plates were examined after 24 and 48 hours of incubation at 37 °C, and growth on either plate was subcultured to both types of media to confirm the presence of vancomycin-resistant enterococci. Environmental samples were plated on trypticase soy agar (Becton Dickinson Microbiology Systems) supplemented with vancomycin, 6 µg/mL and gentamicin, 4 µg/mL [8].

The first rectal isolate of vancomycin-resistant enterococcus from each patient and all clinical and environmental isolates of vancomycin-resistant enterococci were identified to the species level by the API 20 Strep system (bioMerieux Vitek, Inc., Hazelwood, Missouri), the production of yellow pigment, and the 30 °C motility test [17].

Testing of susceptibility to vancomycin, ampicillin, and teicoplanin was done by disk diffusion [18]. We determined minimum inhibitory concentrations (MICs) for vancomycin (Sigma Chemical Co., St. Louis, Missouri) and teicoplanin (Marion Merrell Dow, Cincinnati, Ohio) by using the agar-dilution method [19]. Screening for high-level aminoglycoside resistance was done on brain-heart infusion agar containing either gentamicin, 500 µg/mL (Sigma Chemical Co.) or streptomycin, 2000 µg/mL (Sigma Chemical Co.) [20]. We used nitrocefin disks (Cefinase, Becton Dickinson Microbiology Systems) to test for β-lactamase production.

To determine the molecular relatedness of enterococcal isolates, SmaI-digested (Gibco BRL, Grand Island, New York, and Sigma Chemical Co.) total genomic enterococcal DNA was subjected to pulsed-field gel electrophoresis in a 1% agarose-0.5 times TBE gel using a CHEF-DRIII apparatus (BioRad Laboratories, Richmond, California) as described elsewhere [21]. The pulse time was ramped from 1 to 20 seconds over 21 hours at 200 V. Restriction digestion profiles were analyzed by visual inspection of photographs of ethidium bromide-stained gels. Isolates were considered distinct if their profiles on pulsed-field gel electrophoresis differed by more than three bands [22].

The presence of the specific vancomycin-resistance genotype VanA or VanB, which helps in strain typing and correlates with antibiotic susceptibility, was detected by use of polymerase chain reaction (PCR) assays as described elsewhere [23, 24].

Statistical Analysis

Characteristics and outcome measurements for the patients in the glove-only rooms were compared with those for patients in the glove-and-gown rooms using the chi-square test for categorical variables and the Student t-test for quantitative variables. We calculated unadjusted (univariate analysis) and adjusted (multivariate analysis by logistic regression) relative risks for the acquisition of vancomycin-resistant enterococci. Multivariate analysis included variables that showed a statistically significant difference in the univariate analysis or were of particular clinical interest. Calculations were done using the Statistical Analysis System Version 6.07, Mainframe Package (SAS Institute, Cary, North Carolina). Only exposures to devices, treatments, antibiotics, and other medications present before the acquisition of vancomycin-resistant enterococci were included in the statistical analysis. A P value less than 0.05 was considered statistically significant; all P values are two-tailed.

Results

Study Sample and Outcome

During the 4.5-month study period, 181 admissions to the medical intensive care unit lasted longer than 48 hours (total number of patients was 173; 8 patients were admitted twice). Ninety-three admissions were to beds in glove-and-gown areas, and 88 were assigned to beds in glove-only areas. Demographic and clinical characteristics at admission for patients in glove-and-gown beds and patients in glove-only areas were similar (Table 1). Rates of acquisition of vancomycin-resistant enterococci as seen by rectal colonization did not differ significantly between patients in glove-and-gown areas and those in glove-only areas (Table 2). Results of other outcome measures, such as nosocomial infections, were also similar for the two groups (Table 3).

Risk Factors

To determine the risk factors for acquisition of vancomycin-resistant enterococci, the 45 patients who acquired this bacteria in the intensive care unit were compared with the 108 patients who did not (Table 4). No underlying disease showed an increased risk. On univariate analysis, several exposures were associated with the subsequent acquisition of vancomycin-resistant enterococci. On multivariate testing, the length of stay in the intensive care unit, use of enteral feeding, and use of sucralfate were significantly correlated with subsequent acquisition of vancomycin-resistant enterococci.

Results of Clinical Cultures

Of 88 patients who were colonized by vancomycin-resistant enterococci at some point, 5 in glove-and-gown areas and 3 in glove-only areas—fewer than 2 patients per month—had clinical cultures positive for vancomycin-resistant enterococci. Positive cultures were from blood (n = 4), catheter tips (n = 2), sputum (n = 2), and urine (n = 1). Cultures were not clinically significant in 7 patients. Length of stay in the intensive care unit, length of stay in the hospital, and hospital mortality rate did not differ significantly in patients with clinical cultures of vancomycin-resistant enterococci and in other colonized patients.

Compliance

Health care workers complied with the protocol in 3107 (71.2%) of 4364 observations. Compliance was significantly better for hospital employees entering glove-and-gown rooms than for those entering glove-only rooms (Table 5); overall compliance was 79% for glove-and-gown rooms and 62% for glove-only rooms (P < 0.001). Observations of nurses constituted 51.4% (n = 2243) of the observations; compliance by nurses (75.4%) was statistically significantly better than compliance by physicians (67.3%) or other hospital personnel (66.5%) (P < 0.05).

Environmental Cultures

Before the intensive care unit was occupied by patients, 144 environmental cultures were taken; no cultures were positive for vancomycin-resistant enterococci. We took 397 environmental and equipment cultures during the study period. In patient rooms, 22 (7.2%) of 306 cultures were positive for vancomycin-resistant enterococci; 19 of the positive cultures were from rooms of colonized patients. Positive cultures were evenly distributed between glove-and-gown and glove-only rooms. Colony counts were low; 80% of the cultures yielded only one or two colonies. Sites that most frequently yielded vancomycin-resistant enterococci in the rooms of colonized patients were sheets (40% of cultures were positive), bed rails (25.7% were positive), bedside tables (20% were positive), and blood pressure cuffs (14.3% were positive). Items that always had negative cultures included stethoscopes (n = 19); enteral feedings and dye (n = 17); key pads for ventilators, monitors, and intravenous pumps (n = 40); soap dispensers (n = 27); toilets and toilet side rails (n = 6); bed scales (n = 6); and ice scoops (n = 1).

Results of Strain Typing, Pulsed-Field Gel Electrophoresis, and Susceptibility Testing

Pulsed-field gel electrophoresis was done on the first vancomycin-resistant enterococcal rectal isolate and on clinical isolates from the 88 patients colonized with vancomycin-resistant enterococci (45 patients who acquired vancomycin-resistant enterococci in the medical intensive care unit, 28 who had the bacteria on admission to the intensive care unit, and 15 who acquired vancomycin-resistant enterococci after discharge from the medical intensive care unit). Nineteen types of strains were defined by pulsed-field gel electrophoresis; 16 were Enterococcus faecium, 2 were E. faecalis, and 1 was E. avium. Six of the strain types were each found in 1 patient only. Among the 45 patients who acquired vancomycin-resistant enterococci in the intensive care unit, 13 types of strains were found. When the pulsed-field gel electrophoresis patterns of these 13 types of strains were individually compared with each other, 97% of the pairs showed a difference of 7 bands or more. The strain that occurred most frequently (designated FM13) was found in 5 patients (2 in glove-and-gown areas and 3 in glove-only areas) who had had the bacteria on admission and 17 (9 in glove-and-gown areas and 8 in glove-only areas) who acquired vancomycin-resistant enterococci in the intensive care unit. No common source for FM13 was found.

Types of strains of the 19 environmental isolates of vancomycin-resistant enterococci taken from rooms of patients who were colonized with vancomycin-resistant enterococci matched the strain type of the patient in each case, as determined by pulsed-field electrophoresis. For two of the three isolates of vancomycin-resistant enterococci that were taken from two rooms of patients who did not have the bacteria, the strain types matched those of patients present in neighboring rooms. The third isolate matched the strain type of a patient who had previously been in the room. Although the two patients in these rooms subsequently acquired vancomycin-resistant enterococci, the strain types did not match those of the isolates taken from their rooms.

Clinical isolates of vancomycin-resistant enterococci were available for strain typing for eight patients. For seven of these patients, the strain type of the clinical isolate matched the strain type of a surveillance isolate from the same patient.

Analysis of polymerase chain reaction assays of the initial rectal isolates from the 88 patients with vancomycin-resistant enterococci showed the presence of the VanA gene in 67 cases (76.1%) and VanB in 21 cases (23.9%). Eighty-two isolates (94.2%) were shown to be resistant to ampicillin by disk diffusion; 66 isolates (75%) were resistant to teicoplanin. We ascertained MICs for agar dilution on 1 isolate from each of the 19 strain types; MICs for vancomycin ranged from 16 to 1024 µg/mL or more, and those for teicoplanin ranged from 0.05 µg/mL or less to 32 µg/mL. Three strains were resistant to gentamicin (500 µg/mL), 7 were resistant to streptomycin (2000 µg/mL), 8 were resistant to both gentamicin and streptomycin, and 1 was resistant to neither drug. No strain produced β-lactamase.

Discussion

Our results do not show an added benefit for universal use of gloves and gowns compared with universal use of gloves alone in preventing colonization by vancomycin-resistant enterococci in a hospital in which several strains of endemic vancomycin-resistant enterococci are circulating. Three variables were identified in a multivariate analysis as risk factors for colonization with vancomycin-resistant enterococci: length of stay in the intensive care unit, use of enteral feeding, and use of sucralfate.

The efficacy of gowns apart from that of gloves for controlling the nosocomial spread of bacteria has not been assessed, to our knowledge, in any previous controlled trial. However, failure of gowns to curtail the spread of vancomycin-resistant enterococci in our study contrasts with a recent report by Boyce and colleagues [6]. In that study, an outbreak of vancomycin resistant-enterococci was contained only after a requirement stating that health care workers had to wear gowns was added to a mandate that required patients to be kept in single rooms and hospital employees to wear gloves. Differences in setting and study design may account for this disparity. Boyce and colleagues described a clonal (single-strain) outbreak, used historical controls, and did not monitor compliance with glove or gown precautions at baseline or during the intervention periods.

In our study, several variables were identified by univariate analysis as risk factors for acquisition of vancomycin-resistant enterococci, but only three remained significant by multivariate analysis (Table 4). First, length of stay was an independent risk factor, as it is for the acquisition of most resistant nosocomial bacteria. Second, enteral feeding has been associated with vancomycin-resistant enterococcal colonization in at least one other report [25]. In that study, as in ours, cultures of enteral feeding tubes were consistently negative. It seems likely that enteral feeding results in more opportunities for transmission from the hands of hospital employees or that it alters the gastrointestinal environment in a manner conducive to growth of vancomycin-resistant enterococci [26]. Third, the use of sucralfate has been cited as a risk factor for colonization with vancomycin-resistant enterococci in one other study [27]. During administration of sucralfate, manipulation of the nasogastric tube is required and could result in the introduction of vancomycin-resistant enterococci to a patient's gastrointestinal tract. This possibility emphasizes the need for strict aseptic technique during such manipulations. Alternatively, sucralfate may provide a selective advantage for enterococcal colonization of the upper gastrointestinal tract by a mechanism that is not yet known.

Vancomycin was probably critical in the initial emergence of vancomycin-resistant enterococci, and use of it is still reported in some studies as a risk factor for acquisition of vancomycin-resistant enterococci [6, 8, 10, 28, 29]. However, in our multivariate analysis and in other studies [6, 7, 12], the increased relative risk for this acquisition that was seen with the use of vancomycin was not statistically significant. After vancomycin-resistant enterococci become endemic, many classes of antimicrobials may offer selective advantage. The effect of expanded antibiotic control on vancomycin-resistant enterococcal acquisition warrants study. We found that the use of metronidazole increased the relative risk for acquisition of vancomycin-resistant enterococci. Although it was not a statistically significant risk factor in our study (Table 4), metronidazole was significantly associated with an outbreak of vancomycin-resistant enterococcal bacteremia in an adult oncology unit [12]. Reduction of gastrointestinal anaerobes may create a survival advantage for enterococci.

Several factors may have affected our ability to detect a benefit from the use of gloves and gowns over the use of gloves only. First, our study design had some limitations. Patients were not randomly assigned to care groups. Nevertheless, the two groups were similar; the only statistically significant difference was the more frequent occurrence of chronic obstructive pulmonary disease in the glove-only group, and this occurrence was not a risk factor for acquisition of vancomycin-resistant enterococci (Table 4). The study design did not include a control group of patients with whom neither gowns nor gloves were used. The frequency with which medical intensive care unit employees have contact with blood and other body fluids makes such a control group inadvisable. Because of the configuration of the medical intensive care unit, the glove-and-gown and glove-only sections were not identical: Four of eight glove-only beds were in double rooms, but all eight glove-and-gown beds were in single rooms. This discrepancy might be expected to have caused more vancomycin-resistant enterococcal acquisition in glove-only beds, but the slight increase in the rate of acquisition in two-bed rooms was not significant (Table 4). The study was not designed to detect acquisition of a new strain type by patients who already had vancomycin-resistant enterococci. However, because no difference was seen in the initial acquisition of vancomycin-resistant enterococci by patients in glove-and-gown beds and in glove-only beds, a difference between the two groups in acquisition of a second strain is unlikely.

Second, a negative outcome could result from acquisition of vancomycin-resistant enterococci by a route that bypassed barrier precautions, such as a common source. Environmental and equipment cultures were taken before and during the study. One month before the study began, an electronic rectal thermometer was found to be contaminated with a strain of vancomycin-resistant enterococcus that was also found in several patients. Contaminated electronic thermometers were also implicated as the vehicle of transmission of vancomycin-resistant enterococci in a previous outbreak [7]. On the basis of these findings, all electronic thermometers were removed from the intensive care unit, and only glass thermometers dedicated to individual patients were used to take oral or rectal temperatures during the study period. Cultures of other common potential sources, such as enteral feeding tubes and dye, bed scales, and a mortar and pestle, were always negative. Moreover, pulsed-field gel electrophoresis typing showed at least 13 strains of vancomycin-resistant enterococci among patients who acquired this bacteria, suggesting that a common source was unlikely.

Third, barrier precautions might also have been bypassed if patients frequently acquired vancomycin-resistant enterococci from contaminated objects in the environment. However, noncolonized patients in rooms from which vancomycin-resistant enterococci were cultured (n = 2) did not acquire the type of strain found in the room. The low number of vancomycin-resistant enterococci recovered from positive environmental cultures and the general absence of positive environmental cultures for vancomycin-resistant enterococci in the rooms of non-colonized patients suggest that cleaning done between patients was adequate. Moreover, we suspect that use of fecal incontinence bags for almost all incontinent patients may have decreased environmental contamination. The low colony counts on contaminated environmental surfaces may explain why gowns, which theoretically offer greater protection against environmental contaminants, offered no advantage in our study. Our results suggest that the patient contaminates the environment more often than the environment contaminates the patient, but more frequent environmental culturing is needed to test this hypothesis. Our findings also suggest that cross-transmission by the hands of health care workers or by fomites (such as electronic rectal thermometers) greatly overshadows the risk for transmission by clothing of hospital employees.

Fourth, are rectal swabs sufficiently sensitive to detect vancomycin-resistant enterococcal colonization? If newly admitted patients were colonized with very low levels of vancomycin-resistant enterococci that were not detected by rectal swab cultures, emergence of these strains after patients were exposed to antibiotics might be incorrectly interpreted as acquisition in the intensive care unit. However, rectal swabs, perirectal swabs, and stool cultures appeared to have equal sensitivity for detection of vancomycin-resistant enterococci in one small study [30]. For 5 of the 13 acquired strain types, no patient in the intensive care unit from whom each strain could have been acquired by cross-transmission was identified in our study. Perhaps other culture methods, such as inoculation of the rectal swab into enrichment broth [31], would have detected colonization in these patients on admission. However, two of these patients were in the glove-and-gown group and three were in the glove-only group, which makes it unlikely that a more sensitive detection method would have changed the outcome.

Finally, was compliance with precautions sufficient to provide a true measure of their efficacy? Most reports of compliance with handwashing in intensive care units show that the rates are less than 50% [32-36]. Compliance with glove requirements is also generally low, ranging from 25% to 59% [36, 37]. In a recent observational study of nine patients under contact isolation for known vancomycin-resistant enterococcal colonization, rates of compliance with handwashing, glove use, and gown use were only 36%, 40%, and 52%, respectively [38]. In our study, compliance was intensively monitored for all employees who worked in or entered the intensive care unit. Moreover, our criteria for compliance were much more strict than those in previous studies; employees were counted as noncompliant if they entered a room without appropriate garb, even if they had no contact with patients or surfaces in the room. Nevertheless, overall compliance (the sum for all observed categories) of hospital employees was 79% in gown-and-glove rooms (Table 5). It may be difficult for many institutions to maintain rates of compliance equal to or better than this rate. Notably, rates of compliance with glove use and handwashing were higher for employees entering glove-and-gown rooms than for those entering glove-only rooms. Improved compliance associated with requiring a greater level of precautions has also been shown by other investigators [39, 40]. Determining whether the added cost of disposable gowns can be justified on this basis would require a detailed prospective economic analysis, the results of which would probably depend on the circumstances of individual hospitals. For example, better compliance may have been expected to result in less acquisition of vancomycin-resistant enterococci by patients in glove-and-gown rooms. The fact that gowns conferred no added benefit may result from the large number of patients who had the bacteria in our intensive care unit; that is, the risk that any break in compliance will result in transmission increases with prevalence.

What recommendations can be made to prevent the spread of vancomycin-resistant enterococci? Barrier precautions, specifically the use of gloves and gowns when entering the rooms of patients colonized with vancomycin-resistant enterococci, are a prominent component of the CDC/HICPAC guidelines for preventing the spread of vancomycin-resistant enterococci. Although our results do not show that the use of gloves and gowns is better than the use of gloves alone, we must caution that our study was done in an institution in which vancomycin-resistant enterococci are endemic. Because vancomycin-resistant enterococci are now highly endemic inside and outside of intensive care units in many hospitals [3, 28], attention must be directed to assessing the benefit of control of other factors, such as use of antibiotics, which may influence acquisition of vancomycin-resistant enterococci. An infection control policy that requires gowns in addition to gloves might be more effective in specific contexts, such as during an outbreak of a single strain in a nonendemic environment or to control spread from infected or colonized patients in institutions where prevalence is still low or environmental contamination is extensive. It may be easier to contain or eliminate vancomycin-resistant enterococci if the problem is detected early. In addition, our data suggest that requiring use of a gown may be most useful in heightening awareness of an infection control problem and stimulating enhanced compliance with infection control measures, even if gowns themselves do not act as a physical barrier to transmission. The resources required to ensure compliance with barrier and other precautions [13] depends on the extent of colonization with vancomycin-resistant enterococci in the institution. In hospitals such as ours, in which vancomycin-resistant enterococci are endemic on six intensive care units and 15 wards, the task will require a substantial commitment of hospital resources.

Presented at the 35th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC) on 18 September 1995 in San Francisco, California (abstract J44).

Dr. Hayden and Ms. Matushek: Division of Infectious Diseases, Rush-Presbyterian-St. Luke's Medical Center, 1653 West Congress Parkway, Chicago, IL 60612.

Ms. Nathan, Drs. Rice and Weinstein, and Ms. Van Voorhis: Division of Infectious Diseases, Cook County Hospital, 1835 West Harrison Street, Chicago, IL 60612.

Ms. Hu: Department of Medicine, Cook County Hospital, 1835 West Harrison Street, Chicago, IL 60612.

Dr. Franklin: Medical Intensive Care Unit, Cook County Hospital, 1835 West Harrison Street, Chicago, IL 60612.