Patients and Methods

This study was performed at the Hebrew Rehabilitation Center for the Aged, a 648-bed long-term care facility in Boston, Massachusetts. Patients from 4 of the 12 forty-bed units were enrolled in the study as part of an infection control initiative targeting MDRGNB. This initiative was in response to the findings of an earlier study that identified 51% of residents on these units as being colonized with MDRGNB [7]. Rectal surveillance cultures for MDRGNB were performed for all residents on these 4 units every 3 months. For residents in whom MDRGNB colonization was identified, rectal cultures were performed every 3–4 weeks. Contact precautions were implemented for patients in whom colonization with MDRGNB was identified [4] and were discontinued if MDRGNB was not recovered from 2 subsequent serial cultures. This study presents findings from data collected from 31 October 2006 through 22 October 2007 for all patients in whom MDRGNB colonization was identified. The study site's institutional review board approved the review of the infection control data.

Definitions.Multidrug resistance among gram-negative bacteria was defined as resistance or intermediate susceptibility to ⩾3 of the following antimicrobials or antimicrobial groups: (1) ampicillin-sulbactam or piperacillin-tazobactam; (2) meropenem; (3) ceftazidime or ceftriaxone; (4) ciprofloxacin; and (5) gentamicin, using breakpoint values defined by the Clinical and Laboratory Standards Institute [8]. Duration of colonization was defined as the number of days from the first through the last available culture in which genetically related MDRGNB species were recovered. MDRGNB isolates were required to be of the same species and to be closely related (defined as differing by ⩽3 bands by pulsed-field gel electophoresis (PFGE) [9].

Clearance of colonization was defined as 2 serial rectal cultures from which MDRGNB were not recovered. Two consecutive cultures were required to ensure that the first negative culture result was attributable to clearance of colonization, rather than to sampling error. The end point for clearance of colonization for a specific MDRGNB species occurred at the time of the first culture from which the MDRGNB species was no longer recovered (in those patients who met the criteria for clearance of colonization and had a second negative culture result).

Study population.A total of 212 patients participated in the infection control initiative. Only patients in whom MDRGNB was recovered from a rectal specimen and who had at least 2 subsequent rectal specimens obtained were included in this study. Patients who had only 3 cultures performed and in whom MDRGNB of the same species were recovered from the first 2 cultures but not the third culture were excluded, because clearance of colonization could not be determined.

Specimen collection, MDRGNB identification and susceptibility testing, and PFGE.To ensure adequate stool sampling, all rectal specimens required visualization of stool on rectal swabs. Swab samples were plated within 2–4 h of collection onto 2 MacConkey media supplemented with ciprofloxacin (2 µg/mL) or ceftazidime (2 µg/mL). Species identification with Microscan (Baxter) and susceptibility testing were performed according to the Clinical and Laboratory Standards Institute methodology [8]. PFGE was performed as described elsewhere [10], and DNA patterns were interpreted using the criteria of Tenover et al. [9]. Quality control was performed using Escherichia coli American Type Culture Collection 25922 (for negative control). Two Pseudomonas aeruginosa strains with ciprofloxacin and ceftazidime minimal inhibitory concentrations of 2 µg/mL according to broth microdilution (positive control) were used to ensure adequate antimicrobial concentrations in the selective MacConkey plates.

Data collection.Factors that could potentially influence the duration of MDRGNB colonization were abstracted from medical records, including patient demographic data, the Charlson score for comorbidities, and functional status determined with the Katz Activities of Daily Living Scale (range, 0–6), with a score of 6 indicating total functional dependence and fecal incontinence [11, 12]. Data on antimicrobial exposure during the 12 months before MDRGNB colonization and antimicrobial exposure during the study period were collected from pharmacy databases.

Statistical analysis.Categorical and continuous variables were analyzed using the Fisher's exact test and the Wilcoxon rank sum test, respectively. Kaplan-Meier curves were used to analyze days to clearance of colonization, and log-rank tests were used to compare curves. Hazard ratios were estimated using a Cox proportional hazards regression, with values <1 favoring persistent colonization. All tests were 2-sided, and a P value <.05 was considered to be statistically significant. Statistical analysis was performed using Stata software, version 10.0 (Stata).

Results

Patient characteristics and rectal cultures.A total of 212 patients were included in the infection control initiative. Forty-two patients in whom MDRGNB was recovered from at least 1 culture had ⩾2 cultures obtained after the initial positive culture. Nine of these patients were colonized with 1 MDRGNB species which was isolated from the second culture but not from the third and last available serial culture. These patients were excluded from the analysis, because clearance of colonization (which required 2 successive surveillance cultures without isolation of MDRGNB) could not be determined. Thus, a total of 33 patients met the inclusion criteria for this study. Patients were followed up for a median of 211 days (range, 63–356 days), during which time a median of 7 serial surveillance cultures (range, 3–12 cultures) were performed for each patient. Patient demographic data are shown in table 1.

Characteristics of MDRGNB isolates.A total of 57 MDRGNB isolates were recovered from 33 patients. Cocolonization with ⩾1 different MDRGNB species was identified in 20 (61%) of these patients. Patients were cocolonized with a median of 2 different MDRGNB species (range, 1–4 species). Three (15%) of these patients were cocolonized with 3 or 4 different MDRGNB species. MDRGNB species and coresistance profiles are shown in table 2.

Duration of MDRGNB colonization.The median duration of MDRGNB colonization was 144 days (range, 41–349 days). Duration of colonization for individual MDRGNB species is shown in table 3. Only those episodes of MDRGNB colonization in which MDRGNB isolates were recovered from at least 2 serial cultures are shown. PFGE analysis of the multiple MDRGNB strains of the same species recovered over time from each patient differed by ⩽3 bands and were considered to be indistinguishable or closely related (data not shown). Patients were considered to have persistent colonization, even if a MDRGNB isolate was not recovered from 1 of the serial cultures, if the same MDRGNB strain was recovered from the preceding and subsequent cultures. This occurred in 18 (32%) of 57 MDRGNB colonization episodes.

Clearance of MDRGNB colonization.Clearance of MDRGNB colonization occurred in 3 (9%) of the 33 patients. All 3 patients were colonized with only 1 MDRGNB species, compared with 10 (33%) of 30 patients with persistent colonization (P=.02). Of the 20 cocolonized patients with persistent colonization, 14 (70%) cleared 1 of the colonizing MDRGNB species. There were no statistically significant differences in duration of follow-up, antibiotic exposure, and other clinical factors between patients with clearance and patients with persistence of colonization (table 1). The type of antimicrobials received by the 10 patients (33%) with persistent MDRGNB colonization were ciprofloxacin (in 4 [40%] of patients), metronidazole (2 [20%]), first-generation cephalosporins (2 [20%]), clindamycin (1 [10%]), tetracycline (1 [10%]), and trimethoprim-sulfamethoxazole (1 [10%]). All antimicrobials were administered via the oral route.

By colonizing species, clearance of colonization occurred in 22 (39%) of the MDRGNB colonization episodes, with an MDRGNB colonization clearance rate of 2.6 episodes per 1000 days (figure 1). The median time to clearance of colonization and rate of clearance per 1000 days for each MDRGNB species is shown in table 4. There were no episodes of recolonization with the same MDRGNB strain after clearance of colonization was documented.

Clearance of colonization was more likely to occur with non-Proteus species than with Proteus mirabilis (P=.007, by log rank test) (figure 2). Clearance of colonization with multidrug-resistant P. mirabilis occurred in 1 (6.7%) of 15 episodes of colonization due to multidrug-resistant P. mirabilis, compared with 21 (50%) of 42 episodes of colonization due to all other MDRGNB species. The hazard for clearance of multidrug-resistant P. mirabilis colonization was 10% of the hazard for clearance of colonization due to other MDRGNB species (hazard ratio, 0.1; 95% confidence interval, 0.01–0.78; P=.03). The difference in mean duration (± standard deviation) of follow-up between multidrug-resistant P. mirabilis episodes (176±18 days) and non-Proteus species episodes (209±14 days) was not statistically significant (P=.2).

Of the 15 patients who were colonized with multidrug-resistant P. mirabilis, 13 (87%) were also colonized with at least 1 other MDRGNB species (median, 2 different species; range, 1–4 different species). Of these 13 cocolonized patients, 6 (46%) cleared the non–multidrug-resistant Proteus species and none cleared multidrug-resistant P. mirabilis. Among the 15 multidrug-resistant P. mirabilis strains, 13 genetically distinct PFGE types were identified.

Antibiotic exposure occurred in 2 (13.3%) of 15 episodes of colonization with multidrug-resistant P. mirabilis, compared with 9 (21.4%) of 42 episodes of colonization with other MDRGNB species. Because a great majority of patients who were colonized with multidrug-resistant P. mirabilis were also colonized with another MDRGNB species, the effect of antibiotic exposure on colonization with P. mirabilis versus colonization with other species could not be analyzed.

Discussion

There are several important and novel findings in this study regarding the characteristics of MDRGNB colonization. First, the duration of MDRGNB colonization is prolonged. Patients were colonized with MDRGNB for a median of 144 days, with a maximum of 349 days of colonization by the end of the study. Second, clearance of colonization with all MDRGNB species occurred infrequently. In this study, only 3 (9%) of 33 patients cleared all of the MDRGNB species colonizing their gastrointestinal tract. Third, cocolonization with MDRGNB of different species was frequent. A total of 20 patients (61%) were colonized with ⩾2 different MDRGNB species, of whom 15% were colonized with 3 or 4 different MDRGNB species.

These findings have important implications regarding the efficacy of surveillance efforts for the detection of MDRGNB reservoirs in the health care setting. For vancomycin-resistant enterococci and methicillin-resistant S. aureus, guidelines promote the use of surveillance cultures in certain circumstances [4]. They also recommend that, if ⩾3 surveillance cultures during a 1–2 week period are negative for these antimicrobial-resistant pathogens, contact precautions may be discontinued for patients who have not been exposed to antibiotics for several weeks [4]. There are no specific recommendations for MDRGNB [4–6]. The results from this study suggest that surveillance efforts to document clearance of colonization may not be a feasible strategy, because patients remain colonized for periods that surpass the average duration of hospitalization, and clearance of colonization is a rare event. Prolonged duration of colonization with other gram-negative species, including Acinetobacter species and P. aeruginosa, has also been documented [13, 14]. Surveillance cultures for MDRGNB may also not be a feasible strategy, because of the lack of a standardized screening media for the detection of MDRGNB and the high proportion of patients identified with cocolonization in this study. These factors would make surveillance cultures a very labor-intensive and costly intervention.

Antibiotic exposure is a main factor promoting the emergence and spread of antimicrobial-resistant bacteria. One would therefore expect that selective pressure from antimicrobials would promote persistent MDRGNB colonization. However, in this study, two-thirds of patients with persistent MDRGNB colonization did not receive any antimicrobials throughout the study period, and the majority of the remaining one-third of patients received very-narrow-spectrum antimicrobials. The lack of an association between antimicrobial exposure and persistent colonization may be attributable to a small sample size. Nevertheless, these findings suggest that other unknown factors may be influencing the duration of MDRGNB colonization.

Another interesting and novel finding in this study is the persistence of colonization with multidrug-resistant P. mirabilis, compared with colonization due to other MDRGNB species. In fact, the hazard of clearing multidrug-resistant P. mirabilis colonization was only 10% that of clearing other MDRGNB species. Furthermore, among patients who were cocolonized with both P. mirabilis and other MDRGNB species, colonization with the latter was cleared more often than colonization with P. mirabilis. These findings suggest a potential survival advantage for P. mirabilis colonization of the gastrointestinal tract. A characteristic of Proteus species isolation is its tendency to swarm on isolation media and to potentially impede the identification of colonies of other gram-negative species. However, in this study, MacConkey media was used to prevent swarming [15]. The use of media supplemented with ciprofloxacin could potentially increase the recovery of multidrug-resistant P. mirabilis, because 100% of these isolates were resistant to this antimicrobial. However, the majority of other MDRGNB species were also resistant to ciprofloxacin, and therefore, preferential isolation of multidrug-resistant P. mirabilis would not explain the persistent colonization. Proteus species represent a minority of the gram-negative species that colonize the gastrointestinal tract, and therefore, persistent recovery of this species would not be expected [16–18]. Lastly, among the 15 multidrug-resistant P. mirabilis isolates, 13 represented genetically unrelated strains, and therefore, acquisition of a potential outbreak strain would not explain persistent colonization.

Lautenbach et al. [19] analyzed potential factors associated with persistent colonization among E. coli strains with reduced susceptibility to fluoroquinolones. They showed that, among a cohort of 10 colonized outpatients, the median duration of colonization was 80 days (range, 8–172 days). They further showed that prolonged colonization was more common among species that overexpressed the AcrAB efflux pump [19]. Future studies will need to determine whether certain antimicrobial resistance characteristics among P. mirabilis also promote persistent colonization.

There are several limitations of this study that warrant further discussion. First, the use of media supplemented with ceftazidime and ciprofloxacin likely led to an overrepresentation of MDRGNB species that were resistant to these antimicrobials. However, this overrepresentation should not affect the main findings and conclusions of this study. Second, there were 18 episodes of intermittent MDRGNB colonization, in which a surveillance culture had results that were negative for MDRGNB, despite the recovery of the same MDRGNB strain from the previous and subsequent culture. In this study, it was assumed that this represented persistent colonization, because clearance of colonization followed by reacquisition of the same strain within a 3–4 week period was unlikely to occur. Third, we arbitrarily chose to define clearance of colonization as 2 serial cultures without the recovery of MDRGNB. It is difficult to ascertain whether this definition accurately reflects clearance of colonization. However, the same strain of MDRGNB was not recovered from any of the subsequent cultures that were performed after the definition of clearance was met, which suggests that sampling error was unlikely to explain clearance of MDRGNB species. Lastly, this study was conducted in a long-term care facility, and the findings warrant reconfirmation among hospitalized patients.

Developing effective strategies aimed at limiting the spread of MDRGNB requires an understanding of the characteristics of MDRGNB colonization. The findings of this study are a first step towards providing data to guide future preventive initiatives.