Acquisition of Multidrug‐Resistant Organisms Among Hospital Patients Hospitalized in Beds Adjacent to Critically Ill Patients
Objective. To assess whether patients hospitalized in beds physically adjacent to critically ill patients are at increased risk to acquire multidrug‐resistant pathogens.
Design. Cohort study.
Setting. Shaare Zedek Medical Center, a 550‐bed medical referral center.
Patients. From April to September 2004, we enrolled consecutive newly admitted patients who were hospitalized in beds adjacent to either mechanically ventilated patients or patients designated as “do not resuscitate” (DNR). For each of these patients, we also enrolled a control patient who was not hospitalized in a bed adjacent to a critically ill patient. We collected specimens from the anterior nares, the oral cavity, and the perianal zone at the time of admission and subsequently at 3‐day intervals until discharge or death. Specimens were cultured on selective media to detect growth of antibiotic‐resistant pathogens, including Acinetobacter baumannii, methicillin‐resistant Staphylococcus aureus (MRSA), extended‐spectrum β lactamase (ESBL)–producing Enterobacteriaceae, and vancomycin‐resistant enterococci (VRE).
Results. We enrolled 46 neighbor‐control pairs. Among neighbors and controls, respectively, the incidence rates for isolation of A. baumannii was 8.3 and 4 isolations per 100 patient‐days (relative risk [RR], 2.1 [95% confidence interval {CI}, 0.8‐5.2]; 
), the incidence rates for MRSA were 1.4 and 2.6 isolations per 100 patient‐days (RR, 0.6 [95% CI, 0.1‐2.3]; 
), the incidence rates for ESBL‐producing Enterobacteriaceae were 10.5 and 9 isolations per 100 patient‐days (RR, 1.2 [95% CI, 0.6‐2.4]; 
), the incidence rates for VRE were 4.3 and 4.8 isolations per 100 patient‐days (RR, 0.9 [95% CI, 0.3‐2.4]; 
), and the composite incidence rate was 21.7 and 16.2 isolations per 100 patient‐days (RR, 1.3 [95% CI, 0.8‐2.3]; 
).
Conclusions. In this pilot study, we did not detect an increased incidence rate of isolation of multidrug‐resistant pathogens among patients hospitalized in beds adjacent to critically ill patients. Further studies with larger samples should be conducted in order to generate valid data and provide patients, physicians, and policy makers with a sufficient knowledge base from which decisions can be made.
Received May 31, 2005; accepted August 19, 2005; electronically published June 20, 2006.
Nosocomial infections pose a continuing challenge for both the physician and the entire hospital staff that tries to prevent these complications. A further challenge consists of methodological constraints on research aimed at elucidating the mechanisms leading to the healthcare‐specific infection. Although patient‐to‐patient indirect transmission seems a most likely pathway of pathogen transmission, it is either uncommon or very difficult to prove.1 Furthermore, pathogens isolated from specimens obtained in the hospital were usually acquired before the patient was hospitalized.2
In our hospital, as probably in many others, there is extensive use of broad‐spectrum antibiotics to treat patients who are geriatric, bedridden, and/or dependent on help for everyday activities, including those receiving mechanical ventilation and those designated as “do not resuscitate” (DNR). It is not clear whether broad‐spectrum antibiotic treatment is beneficial in this patient population, and it might even be harmful because it might lead to emergence of drug‐resistant pathogens, which can spread and colonize and infect other patients.
The objective of this study was to generate information on the incidence of colonization by multidrug‐resistant pathogens among patients hospitalized in beds adjacent to DNR or mechanically ventilated patients and to ascertain whether such patients were at increased risk for acquisition of these organisms from their critically ill neighbors. This information would be most helpful in the facilitation of subsequent discussion, research, and decision making regarding the use of broad‐spectrum antimicrobial treatment in this patient population.
Methods
This project was conducted in the Departments of Medicine and Geriatrics at the Shaare Zedek Medical Center (Jerusalem, Israel), a 550‐bed, university‐affiliated general hospital. We conducted a prospective cohort study. As we had no previous data with which to estimate the expected results of the study, we arbitrarily decided to enroll patients during a 5‐month period. Our patients were newly hospitalized patients who were hospitalized in a bed adjacent to either DNR or mechanically ventilated patients. The study patients were designated “neighbors,” and the DNR and mechanically ventilated patients were designated “core patients.” For every enrolled neighbor we enrolled a control patient. Control patients were newly hospitalized patients whose age was within 10 years of the age of the neighbor and who were admitted to the same medical ward within 2 days before or after the day of the neighbor’s admission but who were not hospitalized a bed adjacent to a core patient. Neighbors and control patients who were either immunodeficient or who did not consent to take part in the study were excluded. All hospital rooms had 3 beds, and patients were allocated by the practicing medical staff in the hospital on the basis of vacancies in the medical and geriatric wards.
Hospital departments in which the study took place were visited twice weekly for detection and enrollment of appropriate neighbors and control patients. Specimens for screening were obtained from the anterior nares, mouth, and perianal region of all enrollees. The specimens were collected on enrollment and subsequently every 3 or 4 days.
Specimens from neighbors and their control patients were obtained in parallel. Follow‐up for each neighbor and their control patient continued until they died or were considered censored. Causes for censorship were as follows: discharge of the neighbor from the hospital, a record of 5 consecutive specimens already obtained (in preparation), loss of adjacency to the core patient, and death or discharge of the core patient. Standard precautions were applied to prevent transmission of organisms between the study patients by the specimen and data collectors.
Microbiological testing was performed as follows. For detection of vancomycin‐resistant enterococci, rectal swab specimens were plated on enterococcal agar (Becton‐Dickinson Microbiology Systems) containing vancomycin (concentration, 6 μg/mL) and also inoculated into enterococcal broth (Becton‐Dickinson) containing vancomycin (concentration, 6 μg/mL). All suspected enterococcal isolates were identified as enterococci with the Rapid ID 32 Strep kit (bioMérieux). Minimum inhibitory concentrations of vancomycin were determined with the E‐test (AB Biodisk); isolates for which the minimum inhibitory concentration was 4 μg/mL or higher were considered vancomycin resistant. For detection of methicillin‐resistant Staphylococcus aureus (MRSA) swab specimens from the anterior nares were inoculated directly on mannitol salt agar and also into tryptic soy broth containing aztreonam (concentration, 128 μg/mL). Turbid broth specimens were subcultured onto mannitol salt agar. Suspected colonies of Staphylococcus aureus were tested with the Pastorex Staph‐Plus kit (Bio‐Rad), and methicillin resistance was determined by growth on Mueller‐Hinton agar containing 4% sodium chloride and oxacillin at a concentration of 6 μg/mL, according to the NCCLS protocol.3 For detection of extended‐spectrum β‐lactamase (ESBL)–producing Enterobacteriaceae, rectal swab specimens were plated on MacConkey agar containing ceftazidime (concentration, 2 μg/mL) and also into brain‐heart infusion broth containing ceftazidime (concentration, 2 μg/mL). All gram‐negative isolates were identified with the API system (bioMérieux). ESBL production was verified by using the combination of disks containing ceftazidime and cefotaxime, with and without clavulanic acid according to NCCLS protocol.3 Acinetobacter baumannii was identified to genus and species level with the API 32GN kit (bioMérieux).
Once similar organisms were found to be colonizing or infecting neighbors and the adjacent core patient, the possible clonal source of these organisms was determined with pulsed‐field gel electrophoresis (PFGE). The direction of spread of these organisms was determined according to which patient was the first and which the second to be colonized by the identical isolate. PFGE was performed on the organisms that were suspected to have been transferred from core patient to neighbor. The bacterial suspension was prepared by the harvesting of bacterial colonies directly from the culture performed on blood–trypticase soy agar plates incubated overnight at 35°C; the suspension was adjusted with a colorimeter (Vitek) to a concentration of 20% transmittance for gram‐negative organisms or 10% transmittance for gram‐positive organisms. The bacterial suspension was then mixed with an equal volume of 1.2% gold agarose (SeaKem). The DNA blocks were lysed with sodium lauroyl sarcosine and proteinase K at 55°C for 2 hours for both gram‐negative and gram‐positive organisms and overnight for A. baumannii. After the lysis stage, the agarose plugs were washed with water and then with Tris‐EDTA buffer to rid them of the lysed cell walls and degraded proteins. A slice of each plug was then treated with its appropriate restriction enzyme: Xba for Klebsiella pneumoniae and Escherichia coli, SpaI for vancomycin‐resistant enterococci and MRSA, and ApaI for A. baumannii. The treated slices were then loaded on the teeth of the gel comb and allowed to dry. Cooled 1% agarose (at 50°C) was poured into the gel casting mold with the comb in place. Electrophoresis was performed with 2 L of 0.5× Tris‐borate‐EDTA running buffer with the Chef‐DR3 III system (Bio‐Rad). The gel was then stained with ethidium bromide and photographed. The bacterial strains were then compared visually and evaluated for clone similarity according to criteria of Tenover et al.4
Primary end points were cultures of either clinical or surveillance specimens that yielded any of the following organisms: A. baumannii, MRSA, ESBL‐producing Enterobacteriaceae, or VRE. The study also used a composite end point, which was a culture positive for any of those 4 pathogens.
Demographic information was obtained, and patients' medical records were reviewed to document the antibiotic treatment they received during the hospitalization. In this study, the following antibiotics were designated broad‐spectrum antibiotics: amikacin, amoxicillin‐clavulanate, cefepime, meropenem, mezlocillin, and piperacillin‐tazobactam.
The χ2 test was used for analysis of categorical variables; for comparison of continuous variables, either the Student t test or, for nonparametric data, the Mann‐Whitney U test was used. Survival analysis was done using the Kaplan‐Meier method, with application of the log‐rank test for comparison of survival curves. Multivariate analysis was done with the Cox proportional hazards regression model to estimate risk factors for acquisition of pathogens. Logistic regression was used to estimate the added risk of acquisition for every added day of hospitalization. P values of less than .05 were considered statistically significant. SPSS software, version 11.0 (SPSS), was used for statistical analysis. Our institution’s review board approved the study.
Results
In this study, 154 patients were enrolled, of whom 46 were core patients, 54 were neighbors, and 54 were control patients. In 8 instances, more than 1 of a core patient’s neighbors was enrolled; however, these were not included in the study analyses (ie, only 1 neighbor was included per core patient). Data were obtained and statistical analysis was performed for 138 patients, comprising the 46 core patients and the 46 pairs of neighbors and control patients. Of these 46 pairs, 25 (54%) were enrolled in the Department of Medicine and 21 (46%) were enrolled in the Department of Geriatrics. There were no potentially immunodeficient neighbors or control patients who were not enrolled, and there were no refusals to participate in the study.
Table 1 presents the enrolled patients’ demographic and clinical characteristics. The neighbors and control patients were similar with respect to all measured variables. Core patients were markedly different from the other 2 groups in that they were older, they were less likely to be married and functionally independent, and they were more likely to have infectious syndromes diagnosed on admission. The core patients were enrolled after longer hospital stays, relative to the neighbors and control patients, who were enrolled after shorter times since admission. All 3 patient groups were followed up for similar periods.
Among the core patients, 6 (13%) of 46 did not receive antibiotic treatment, and 13 (28%) of 46 received broad‐spectrum antibiotic treatment; in the neighbors group, the corresponding proportions were 18 (39%) of 46 patients and 6 (13%) of 46 patients, and in the control group, 14 (30%) of 46 patients and 5 (11%) 46 patients. Among the core patients, 12 (26%) of 46 had received more than 3 different antibiotics during their hospitalization, whereas in the neighbors group and in the control group this was true for 7 (15%) of 46 patients and 2 (4%) of 46 patients, respectively.
Table 2 presents the incidence rates and the estimated relative rates for isolation of 1 of the 4 pathogens of interest from any of the specimens obtained in the study. There was no significant difference between the relative rate of positive culture results for neighbors and for control patients. This was true in all categories measured.
The Figure presents the Kaplan‐Meier survival curves from the day of enrollment to the day a positive culture result was obtained or the patient was censored. The probability that culture results would remain negative for the pathogen under consideration was not significantly different between the control patients and the neighbors with respect to isolation of A. baumannii (
), MRSA (
), ESBL‐producing Enterobacteriaceae (
), and VRE (
) and with respect to the composite end point (
). Examination of the survival curves for patients colonized with A. baumannii or ESBL‐producing Enterobacteriaceae showed that the curve for the control group seemed to diverge significantly from the curve for the neighbors group on day 5 and day 7 after enrollment, respectively.
Figure. Kaplan‐Meier survival curves for the probability that culture results would remain negative for the pathogen under consideration (y‐axis) plotted against the study day (x‐axis) for each of the 3 patient groups and for each end point. Study days were counted or each patient from the day of enrollment to the day a positive culture result was obtained or the patient was censored. Shaded areas indicate the confidence interval for the probability that culture results for neighbors would remain negative. The table of numerical values beneath each panel gives the number of patients in each patient group on each study day for whom culture results were still negative for the pathogen indicated. Acinetobacter, Acinetobacter baumannii; Composite, composite end point (ie, isolation of any of the 4 pathogens); ESBL+, extended‐spectrum ß‐lactamase–producing Enterobacteriaceae; MRSA, methicillin‐resistant Staphylococcus aureus; VRE, vancomycin‐resistant enterococci.
Univariate analysis was performed to examine whether sex, marital status, residence status (ie, residence at home and independence in everyday activities, residence at home and dependence on help for everyday activities, or residence in a long‐term care facility), diagnosis on admission, and age demonstrated associations with the study outcomes. Patients’ residence status was significantly associated with isolation of ESBL‐producing Enterobacteriaceae (
), isolation of VRE (
), and the composite end point (
). All other associations were statistically insignificant (
). Paired analyses of the neighbors and control patients with respect to differences in the time from admission to enrollment and the duration of follow‐up did not lead to detection of statistically significant differences between the patient groups (
and 
, respectively).
A Cox proportional hazards regression model analysis was performed that included the variables of patient residence status and comparison of neighbors with control subjects. There seemed to be no added hazard for isolation of a pathogen for neighbors, compared with control patients. Overall, compared with patients who resided at home and were independent in their everyday activities, patients who resided at home and were dependent on help for everyday activities seemed to have a lower hazard ratio for isolation of a pathogen. Patients who resided in long‐term care facilities seemed to be at increased hazard (Table 3). For all but 1 of the pathogens examined, the odds ratio for isolation of the pathogen from culture per additional day of hospitalization was statistically significant (Table 3).
There were 6 pairs of core patients and neighbors in which a specimen positive for ESBL‐producing Enterobacteriaceae was collected from the core patient and a similarly positive specimen was collected from the neighbor after the neighbor's previous specimen had yielded no pathogens. This led us to suspect transmission of the pathogen from the core patient to the neighbor. In 3 of these pairs, we found that the isolates from the 2 patients were identical, using PFGE. With respect to MRSA, there were 2 pairs with suspected transmission, and 1 of these instances of transmission was confirmed by PFGE. With regard to A. baumannii, there were 6 pairs with suspected transmission, and 4 of these instances of transmission were confirmed by PFGE. There were 2 suspected instances of VRE transmission, of which neither was confirmed.
Discussion
In this study we did not find that being hospitalized in a bed adjacent to a patient who is either designated DNR or is mechanically ventilated increased the risk of colonization with multidrug‐resistant pathogens, compared with being in a bed not adjacent to such a patient. Furthermore, we found that a major risk factor for being colonized is the length of stay in the hospital.
We believe that these results are of interest because they were contrary to our expectation, which was in accordance with current concepts about the emergence of multidrug‐resistant pathogens. We expected and found greater use of antibiotics in the core patient group and hypothesized that this would lead to emergence of resistant pathogens, which might very well have happened in these patients. We therefore expected that resistant pathogens would be transmitted to the patients in adjacent beds at a greater rate than that to patients who were not in beds adjacent to core patients.
Our study had a 90% power to detect a relative risk of 2.2 of being colonized in the neighbor group, compared with the control group. Still, the main limitation of this study is its size. Certainly, a larger study sample might have shown differences in the risk of colonization. If we were to extrapolate our findings, we would need 976 patient‐days in each group to have a power of 80% to detect a 1.3 relative risk of being colonized in the neighbor group, compared with the control group.
An important limitation to our study design was the process of follow‐up for neighbors and control patients. Bias caused by differential censoring methods for the neighbors and the control patients could produce misleading results. Future studies proposing to measure the risks of transmission and compare these between neighbors of critically ill patients and control patients will have to take into account the patients who are in beds adjacent to control patients and plan independent censoring criteria.
Evidence of emerging drug resistance is continuously being published in the medical literature, especially resistance among nosocomial pathogens such as MRSA,5 ESBL‐producing organisms,6 VRE,7 and A. baumannii.8 These organisms are a great threat to patients.9 Although it is widely accepted that use of antibiotics is one of the driving forces behind the emergence of resistant infectious pathogens, the medical evidence in the field comes from in vitro experience, from ecological studies,10,11,12 and from studies in individuals that documented prevalence and not incidence.13,14,15 We do not have a quantified estimate of the risk that drug‐resistant bacteria will develop in a patient after receipt of a specific antibiotic. A fundamental paradigm in the theory of infectious disease transmission is that direct or indirect contact has to take place between patients in order for pathogens to be transmitted. Studies have been published showing that isolation of patients might not be necessary16 and can be harmful,17 and reviews have shown that many studies lack good methodology.18,19 We found that both these issues were difficult to deal with, given the limited resources we had to conduct our study. Certainly, larger studies, adequately funded and planned, are needed to generate valid data regarding these issues.
Antibiotic use among debilitated patients does not differ from use among mobile patients. However, antimicrobial treatment of debilitated patients, whose prognosis is poorer, does not necessarily improve their prognosis and might actually accelerate the emergence of drug‐resistant pathogens. Our fear was that these pathogens not only emerge in the debilitated patients but are also transmitted to their neighbors in the hospital. We postulated that this mechanism could be a route by which these organisms spread from the hospital to the community and to long‐term care facilities. Very much like previous debates regarding the roles of intubation, resuscitation, and dialysis for terminally ill patients, both advantages and disadvantages should be assessed when considering whether to administer broad‐spectrum antimicrobial treatment in this patient population.
In conclusion, contrary to our expectations, we did not find an excess rate of colonization with antimicrobial‐resistant pathogens among patients hospitalized in beds adjacent to DNR or mechanically ventilated patients, compared with control patients. Although we cannot conclude with certainty that there is no added risk for these patients, we can assume that there is no large absolute risk of transmission to these patients. Studies with much larger numbers of patients would be needed to rule out the presence of a small relative risk.
References
- 1. Grundmann H, Bärwolff S, Tami A, et al. How many infections are caused by patient‐to‐patient transmission in intensive care units? Crit Care Med 2005; 33:946 –951.
- 2. Nijssen S, Bonten MJM, Weinstein RA. Are active microbiological surveillance and subsequent isolation needed to prevent the spread of methicillin‐resistant Staphylococcus aureus? Clin Infect Dis 2005; 40:405‐409.
- 3. National Committee for Clinical Laboratory Standards (NCCLS). Performance Standards for Antimicrobial Disk Susceptibility Testing. Informational supplement M100‐S9. Wayne, PA; NCCLS; 2004.
- 4. Tenover FC, Arbeit RD, Goering RV, et al. Interpreting chromosomal DNA restriction patterns produced by pulsed‐field gel electrophoresis: criteria for bacterial strain typing. J Clin Microbiol 1995; 33:2233‐2239.
- 5. Bradley SF. Staphylococcus aureus infections and antibiotic resistance in older adults. Clin Infect Dis 2002; 34:211‐216.
- 6. Paterson DL. Extended‐spectrum beta‐lactamases: the European experience. Curr Opin Infect Dis 2001; 14:697‐701.
- 7. Murray BE. Vancomycin‐resistant enterococcal infections. N Engl J Med 2000; 342:710‐718.
- 8. Cisneros JM, Rodríguez‐Bano J. Nosocomial bacteremia due to Acinetobacter baumannii: epidemiology, clinical features and treatment. Clin Microbiol Infect 2002; 8:687‐693.
- 9. Vincent JL. Nosocomial infections in adult intensive‐care units. Lancet 2003; 361:2068‐2077.
- 10. Goossens H, Ferech M, Stichele RV, Elseviers M, for the ESAC Project Group. Outpatient antibiotic use in Europe and association with resistance: a cross‐national database study. Lancet 2005; 365:579‐587.
- 11. Albrich WC, Monnet DL, Harbarth S. Antibiotic selection pressure and resistance in Streptococcus pneumoniae and Streptococcus pyogenes. Emerg Infect Dis 2004; 10:14‐17.
- 12. Bergman M, Huikko S, Pihlajamaki M, Palva E, Huovinen P, Seppala H. Effect of macrolide consumption on erythromycin resistance in Streptococcus pyogenes in Finland in 1997‐2001. Clin Infect Dis 2004; 38:1251‐1256.
- 13. Guillemot D, Carbon C, Balkau B, et al. Low dosage and long treatment duration of beta‐lactams: risk factors for carriage of penicillin‐resistant Streptococcus pneumoniae. JAMA 1998; 279:365‐370.
- 14. Arason VA, Kristinsson KG, Sigurdsson JA, Stefansdottir G, Molstad S, Gudmundsson S. Do antimicrobials increase the carriage rate of penicillin resistant pneumococci in children? Cross‐sectional prevalence study. BMJ 1996; 313:387‐391.
- 15. Harbarth S, Harris AD, Carmeli Y, Samore MH. Parallel analysis of individual and aggregated data on antibiotic exposure and resistance in gram‐negative bacilli. Clin Infect Dis 2001; 33:1462‐1468.
- 16. Cepeda JA, Whitehouse T, Cooper B, et al. Isolation of patients in single rooms or cohorts to reduce spread of MRSA in intensive‐care units: prospective two‐centre study. Lancet 2005; 365:295‐304.
- 17. Stelfox HT, Bates DW, Redelmeier DA. Safety of patients isolated for infection control. JAMA 2003; 290:1899‐1905.
- 18. Cooper BS, Stone SP, Kibbler CC, et al. Systematic review of isolation policies in the hospital management of methicillin‐resistant Staphylococcus aureus: a review of the literature with epidemiological and economic modeling. Health Technol Assess 2003; 7:1‐194.
- 19. Cooper BS, Stone SP, Kibbler SP, et al. Isolation measures in the hospital management of methicillin resistant Staphylococcus aureus (MRSA): systematic review of the literature. BMJ 2004; 329:533‐540.