Michael Lin
Robert A. Weinstein
Mary K. Hayden
Overview
Since the introduction of penicillin in the 1940s, hospitals have been witness to a continual battle between our antibiotic armamentarium and the organisms that colonize and infect our patients. The promise that each antibiotic discovery brings has been tempered by the eventual emergence of organisms resistant to the drug. The lag between introduction of a new antibiotic and emergence of resistant bacteria is variable; widespread resistance to penicillin was seen in Staphylococcus aureus within 5 years of introduction of the antibiotic, but vancomycin resistance in this same pathogen was not observed until vancomycin had been in clinical use for over 30 years. At other times, resistance has occurred even before the clinical use of a specific antibiotic; for example, sulfonamide and aminoglycoside resistance can be found in gram-negative bacilli isolated long before our use of these compounds. Such disparities have led to a controversy over whether antibiotic use/abuse or other host and environmental factors are most responsible for the increasing prevalence of antibiotic-resistant pathogens (ARP) in the community and in hospitals. Nevertheless, it appears to be conventional wisdom that our antibiotic choices seldom remain more than a very few drugs ahead of the resistant strains. Furthermore, the same organisms that acquire resistance to one antibiotic are more likely to acquire resistance to multiple antimicrobials [1].
Although antibiotic resistance has emerged in the community, the hospital remains an epicenter for colonization and infection by ARPs, due to three forces [2]. First, potent broad-spectrum antimicrobial consumption is, by necessity, high in hospitals given the acuity of illness seen. Such “antibiotic pressure” directly or indirectly provides a driving force for selecting and maintaining organisms that are able to evolve or acquire mechanisms of resistance. Second, hospitals provide a convenient meeting ground for patients to acquire ARPs from reservoirs such as other patients, the environment, shared equipment, and hospital personnel. Third, the average patient in the hospital is more debilitated and immune compromised than ever before as more routine medical care for healthier patients has shifted to the outpatient setting.
Over the past 50 years, various resistant bacteria have risen to prominence in the hospital setting (Figure 14-1). In the early 1960s, penicillin-resistant S. aureus became
P.194
epidemic and ultimately widespread. Subsequent attempts to treat resistant S. aureus infections with methicillin were soon thwarted by the emergence of methicillin resistance (MRSA). In the 1970s, vancomycin became available for treatment of infections due to gram-positive cocci, and gram-negative bacilli such as Pseudomonas aeruginosa andEnterobacteriaceae became dominant healthcare-associated infection (HAI) pathogens. By the 1980s, introduction of broadspectrum antimicrobials, particularly advanced-generation cephalosporins, provided better treatment options for gram-negative bacteria. This advantage was short-lived; soon, large families of β-lactamases that mediate resistance to the newer cephalosporins were described in Enterobacteriaceae. During the same time, the proportion of hospital S. aureus isolates resistant to methicillin increased steadily, and vancomycin-resistant enterococci (VRE) surfaced. Other pathogens of low virulence, such as yeast, methicillin-resistant coagulase-negative staphylococci, and Corynebacterium jeikeium, also became significant HAI pathogens (Figure 14-1). A common feature of these microbes is their multidrug resistance. The 1990s witnessed staphylococcal and pseudomonal resistance to the new fluoroquinolone antibiotics, and imipenem resistance in Enterobacteriaceae. In the past decade, hospitals identified strains of Acinetobacter baumannii, P. aeruginosa, Klebsiella pneumoniae, and Enterococcus faecium resistant to virtually all available antimicrobial agents, and high-level vancomycin resistance was reported in S. aureus.
|
Figure 14-1 Schematic representation of emerging healthcare-associated infections over time. MRSA, methicillin-resistant Staphylococcus aureus; VRE, vancomycin-resistant enterococci; VISA, vancomycin-intermediate S. aureus; VRSA, vancomycin-resistant S. aureus; ESBL, extended-s β-lactamase-producing Enterobacteriaceae; Acinetobacter, multidrug-resistant Acinetobacter baumannii; Pseudomonas, multidrug-resistant Pseudomonas aeruginosa. Adapted from Herwaldt LA, Wenzel RP. Dynamics of hospital-acquired infections [372]. |
The specter of panresistant “superbugs” with little to no reasonable antimicrobial treatment has raised concern that, at least for some infections, we are effectively entering a postantibiotic age [3]. Understanding the forces that promote increasing antibiotic resistance is the basis of hospital infection control efforts. In this chapter, we review the epidemiology of multidrug-resistant pathogens in hospitals and discuss prevention and control strategies.
Definitions
Describing a pathogen is as “resistant” usually refers to loss of susceptibility to key drugs that are normally used in treatment. The key drug may be either a first-line antimicrobial that is preferred for treatment of a specific organism because of superior efficacy or low toxicity (e.g., oxacillin for S. aureus) or an antimicrobial to which resistance may be a marker for broader no-susceptibility (e.g., ceftazidime resistance in K. pneumoniae suggesting production of an extended-spectrum β-lactamase [ESBL]). Traditionally, antibiotic resistance has been defined phenotypically (i.e., by an ability of the microbe to grow in a high concentration of the antibiotic of interest). However, resistance can be expressed in a heterogeneous manner by a population of bacterial cells, which can make phenotypic detection difficult or unreliable. This phenomenon is seen in methicillin-resistant staphylococci; laboratories have developed numerous modifications of traditional susceptibility testing methods to optimize the sensitivity of phenotypic resistance detection. Testing directly for the gene that encodes resistance, mecA in this example, may allow more accurate and rapid classification of resistant strains. There are other reasons to consider moving toward a genetic definition of resistance: recent data suggest that the epidemiology, and potentially the control, of pathogens that have similar resistance phenotypes but different genetically encoded mechanisms of resistance may not be the same [4]. For example, a study of carbapenem-resistant P. aeruginosa found that in the subset of patients admitted to hospitals, those infected with metallo-β-lactamase–producing strains were different than patients infected with porin-deficient strains with respect to preceding antibiotic exposure, presence and type of co-morbid illness, and invasiveness of infection [5].
From the standpoint of the hospital clinical microbiology laboratory, there are other pitfalls in defining resistance. Newer generations of automated antimicrobial
P.195
susceptibility systems (e.g., VITEK® 2, bioMérieux, Dur-ham, NC) have the ability to detect susceptibility to multiple common antimicrobials via fluorescence-based methods. Common limitations to such systems include failing to detect resistance in certain antimicrobial classes, particularly those that are expressed heterogenously or that require induction for optimal expression, such as β-lactams and vancomycin [6]. Both automated and manual susceptibility testing methods (e.g., the Kirby-Bauer disk method) may yield inaccurate results due to device or operator-dependent errors, such as use of outdated or incorrectly placed antimicrobial disks, variations in bacterial inoculum, and incorrect agar depth or pH. There often is a lag between emergence of resistance and the general availability of an accurate method for its detection. Examples of this problem include detection of vancomycin resistance in enterococci [7,8] and staphylococci [9,10], carbapenem resistance in K. pneumoniae [11,12], and ESBL production in Enterobacteriaceae [13]. ESBLs are particularly difficult to detect; different ESBLs have different affinities for third-generation cephalosporins and may be misclassified as susceptible, depending on the substrate used. Sometimes, instead of the development of a new method, the resistance breakpoint is lowered to improve a test's sensitivity for detection of decreased susceptibility to an antibiotic [14].
Although there is no standard definition for “multiple resistance” or “multidrug resistance,” one definition used is resistance to ≥2 unrelated antibiotics to which an organism is normally considered susceptible [15]. For example, Mycobacterium tuberculosis is multiply resistant when it is not susceptible to isoniazid and rifampin. In practice, the term may also commonly refer to an organism that is susceptible to only one or two remaining antimicrobial classes. “Multidrug resistance,” when used in the context of A. baumannii, usually refers to strains that are susceptible only to one or two antimicrobials, such as carbapenems and colistin. While in most instances, multidrug resistance refers to resistance acquired or evolved by a microbe, some bacterial species of epidemiologic concern, such as Steontrophomonas maltophilia and Burkholderia cepacia, are intrinsically resistant to multiple classes of antibiotics. The strict definitions of “multidrug resistance” change depending on the context of the organism discussed; clarifying such definitions is mainly important for the purposes of investigation and publication and less so for clinical practice.
Determining whether resistant organisms are hospital acquired or community acquired can be problematic because patients may be asymptomatically colonized when they enter the hospital. For example, 15% to 25% of patients colonized or infected with aminoglycoside-resistant gram-negative bacilli and as many as 50% of patients who appear to acquire cefazolin-resistant Enterobacteriaceae after surgery have brought these strains into the hospital [16,17]. Molecular markers may help in distinguishing between hospital and community strains of organisms [18,19]. MRSA, when acquired in the community, usually carries a virulence factor Panton-Valentine leukocidin that can be detected with polymer chain reaction (PCR) or other laboratory tests. Nevertheless, even these markers lose their utility when typically community-acquired MRSA assimilates into the hospital ecology [20].
As medical care of patients becomes shifted into hospital-related facilities, such as nursing homes, dialysis centers, and infusion clinics, these environments become reservoirs for multidrug-resistant organisms. Such “healthcare-associated” infections currently are grouped with community-acquired infections for the purposes of surveillance, but the ecology and resistance patterns of these associated pathogens are more similar to nosocomial than traditional community-acquired organisms [21].
In addition to the difficulties in defining hospital acquisition, there often is the question of whether the patient has clinical disease (infection) due to the resistant strain or is merely colonized. This distinction is relevant mainly for organisms isolated from nonsterile sites, such as the respiratory tract or an open ulcer. ARP colonization is an important epidemiologic problem because it increases the reservoir of resistant bacteria (Figure 14-2) and often is a precursor to clinical disease [22].
Mechanisms and Genetics of Resistance
At least 17 classes of antimicrobials have been developed to treat bacterial infections [23]. Each class has been countered by one or more mechanisms of resistance over the years (Table 14-1). Furthermore, under selective pressure and for other unknown reasons, bacteria frequently aggregate several mechanisms to confer broad resistance to multiple classes of antimicrobials.
The genetics by which antimicrobial resistance develops in particular organisms have broad implications for the ecology of resistance and success of control measures. Whether or not resistance genes exist on chromosomes or mobile genetic elements dictates how quickly resistance genotypes can be acquired or spread and how successfully restrictions in antimicrobial use can directly or indirectly control spread.
Chromosomal antimicrobial resistance usually is the result of spontaneous mutation and natural selection of strains that survive under the antimicrobial selection pressure. For some antimicrobials, such as rifampin, a single-point mutation is all that is needed
P.196
P.197
for resistance. For other antimicrobials, such as fluoroquinolones, a stepwise progression of low-level to high-level resistance usually takes place through sequential mutations [24]. Conditions that favor spontaneous chromosomal mutation include an overwhelming number of organisms (increasing the likelihood of a favorable mutation), incomplete or ineffective antimicrobial therapy (allowing mutants to survive), and a relatively small number of mutations needed for resistance. Examples of chromosomal mutation as the dominant factor inde novo resistance include multidrug-resistant M. tuberculosis, fluoroquinolone-resistant bacteria (although this may be changing), and drug-resistant viruses such as human immunodeficiency virus (HIV) [25]. The chromosomal location of a gene makes it more difficult to spread horizontally between different strains or species, although naked DNA can be transferred from one organism to another through the process of transformation. S. pneumoniae, like all streptococci, are naturally transformable; it is likely that the origins of penicillin resistance lie in acquisition of chromosomal genes that encode low-affinity penicillin-binding proteins from other commensal streptococcal species [26].
|
Figure 14-2 The dynamics of nosocomial resistance: resistance iceberg [15]. Adapted with permission from R.A. Weinstein, and S.A. Kabins, Strategies for prevention and control of multiple drug-resistant nosocomial infection. Am J Med 1981;70:449. |
||||||||||||||||||||||||||||||
|
TABLE 14-1 |
||||||||||||||||||||||||||||||
|
Resistance genes also may reside on mobile genetic elements—chromosomal or extrachromosomal—that allow for interstrain or interspecies spread [27]. Resistance genes usually are located in gene cassettes, which are carried by integrons and transposons (which are functionally similar vectors that differ in the types of recombination genes). The permutations of mobile elements are vast: integrons are found alone or inside transposons; both integrons and transposons often are carried by plasmids and bacteriophages [28]. Such variations allow for gene shuffling and provide bacteria with vast capacity to adapt to changing environments or antimicrobial pressure. Importantly, mobile elements frequently carry multiple resistance mutations simultaneously, allowing for broad resistance to be conferred quickly from one bacterial species to another. Such a situation exists among Enterobacteriaceae, which can acquire ESBLs via plasmids and multiple resistances (such as trimethoprim, aminoglycoside, and fluoroquinolone resistance) via integrons that may reside on the same plasmids [29,30].
Specific Resistances and Incidence
Several disturbing trends in resistance have been observed over the years (Figure 14-1) (Table 14-1). A surge in aminoglycoside resistance was a chief concern in the 1970s and 1980s, particularly in nosocomial Enterobacteriaceae or P. aeruginosa. Resistance levels vary considerably between geographic regions; for example, gentamicin resistance in P. aeruginosaranged from 15.8% in North America to 28.3% in Europe and 38.2% in Latin America from 1997–2000 [31]. Since the 1980s, the availability of second-generation cephalosporins (e.g., cefoxitin and cefuroxime), third-generation cephalosporins (e.g., ceftriaxone and ceftazidime), and β-lactam-β-lactamase inhibitor combination agents (e.g., piperacillin-tazobactam) has highlighted an additional set of resistances in gram-negative bacilli. For instance, Enterobacter spp. initially were considered susceptible to cephalosporins but frequently developed resistance during therapy. The culprit was spontaneous de-repression of an intrinsic chromosomal β-lactamase [32]. Further discovery of a family of plasmid-mediated ESBLs conferring broad resistance to various penicillins and cephalosporins has made many gram negatives, such as Escherichia coli and Klebsiella spp., difficult to control without turning to “antibiotics of last resort” such as carbapenems. ESBLs evolved by point mutations from common, older plasmid-borne enzymes and are transferable to bacteria that do not have the intrinsic chromosomal resistance. ESBL-producing K. pneumoniae represented the fastest growing resistant pathogen monitored in the CDC's National Nosocomial Infections Surveillance (NNIS) system in 2003, reaching 20.6% overall resistance (up 47% compared to the five years prior) [33].
Because carbapenems are some of the broadest spectrum antimicrobials available and because they are critical in treating ESBL-producing bacteria and other highly resistant gram-negative organisms such as A. baumannii, the increases seen over the years in carbapenem resistance have been sobering. Carbapenem resistance is mediated by various mechanisms, such as loss of the outer membrane protein OprD, production of carbapenemases, or up-regulation of the efflux system MexEF-OprN [34]. Analysis of NNIS data from 1998 to 2004 reveals that while carbapenem resistance in Enterobacter spp. remained low (0.70%), the prevalence of imipenem resistance in P. aeruginosa was 19% (increased from 11.1% in the late 1980s) [33,35]. Carbapenem-resistance in A. baumannii is especially concerning because some hospital strains are virtually pan-resistant to conventional classes of drugs.
The prevalence of trimethoprim and sulfonamide resistance in gram-negative bacteria remains a concern primarily in outpatient settings where oral trimethoprim alone or trimethoprim-sulfamethoxazole (TMP-SMX) combination often is prescribed empirically for the treatment of urinary tract infections (UTIs). Resistance to trimethoprim is mediated via alterations in the target enzyme dihidrofolate reductase [36] while resistance to sulfonamides is mediated by alterations in its target enzyme dihydropteroate synthase [37]. These resistance genes often are linked to other resistance genes on integrons, allowing for efficient spread and indirect selective pressure from associated antibiotics. For example, in Sweden, one hospital found rising rates of TMP-SMX resistance in its E. coli isolates despite concomitant decreases in the overall use of the drug, likely because antibiotic selective pressure favored the other genes that were linked on the same integron [38]. In general, TMP-SMX resistance rates in E. coli are variable, depending on the region of the United States, from 7% in the Eastern states to 32% in the Western states [39]. In one center, which saw a rise in TMP-SMX resistance in E. coli of 8.1% to 15.8% between 1992 and 1999, resistance was associated with prior TMP-SMX use [40].
Methicillin resistance has been a major concern in staphylococci since the 1980s. In 63 NNIS hospitals from 1974 to 1981, the percentage of S. aureus infections resistant to methicillin increased modestly from 2.4% to 5%,
P.198
due primarily to epidemics in four large teaching institutions [41]. In 1992, the pooled percentage of resistance had risen to 32.1%, and in 2004, to 53% [33,42]. Methicillin resistance in nosocomial S. aureus and coagulase-negative staphylococci is endemic in most U.S. hospitals.
Fluoroquinolones target enzymes responsible for bacterial DNA replication, such as DNA gyrase and topoisomerase, giving them broad efficacy against many gram-negative organisms and some gram-positive organisms [43]. Since the 1980s, their potency and oral bioavailability have made their use widespread in treatment of infections (particularly pulmonary, urinary, and gastrointestinal) and in prophylaxis (e.g., in neutropenic patients). Widespread increases in fluoroquinolone resistance have been reported in many bacteria, in particularEnterobacteriaceae, P. aeruginosa, S. pneumoniae, and S. aureus [44,45,46,47]. In the United States, 93% of MRSA [48] and nearly 30% of nosocomial P. aeruginosa [33] were resistant to ciprofloxacin in 2003. While most of this resistance is likely selected for in humans, the use of fluoroquinolones as growth promoters in food animals may play a greater role in driving resistance than previously suspected [49,50,51].
Host Factors Predisposing to Colonization or Infection with Antibiotic-Resistant Organisms
A number of patient risk factors have been associated with acquisition of ARPs (Table 14-2). Our epidemiologic understanding of these factors remains limited because most studies have been retrospective due to practical restraints, limiting the covariates to more easily obtainable data, such as antimicrobial use. Many of the identified factors are undoubtedly linked and may serve as indirect markers of more difficult-to-measure covariates such as frequency of patient-to-staff contact. It also is important to recognize that risk factors may differ, depending on whether epidemic vs. endemic periods are being studied and whether the ARP is isolated during episodes of colonization vs. infection.
Recently, attention has been directed at other methodologic issues that may lead to biased estimates of risk for antibiotic resistance [52]. First, results of case-control designs, used extensively in studying the risk of acquiring ARPs, are influenced by how the control group is selected. In many studies, control groups are selected either from patients in the population who are uninfected or from patients who carry the antibiotic-susceptible form of the bacterium of interest [53]. Use of either control group may result in slightly differing estimates of risk. When resistant cases are compared to uninfected controls, risk factors for acquisition of both the susceptible and the resistant phenotype of the organism may be identified [53]. If, instead, controls are selected that are already colonized or infected with the susceptible form of the organism, the odds of association between certain variables and acquisition may be overestimated [54]. An alternative study design, “case-case-control,” uses both types of controls and two separate case-control analyses within a single study in order to differentiate risk factors associated with acquisition of the susceptible and resistant phenotypes of the organism [53].
|
TABLE 14-2 |
||||||||||||||||||||||||||||||||||||||||
|
A second important methodologic principle is adjustment for time at risk [52]. Not surprisingly, duration of hospital stay (and specifically, “time at risk” before the index colonization or infection) often is identified as a significant risk factor [55,56,57]. In case-control studies that identify risk factors for antibiotic resistance, time at risk needs to be accounted for via multivariate analysis or by matching cases and controls [52].
A third methodologic principle that has been identified as important is adjustment for co-morbid conditions [52]. A systematic review of 37 published studies carried out by Harris et al. found that while the majority (73%) adjusted
P.199
appropriately for co-morbid illness, only 11 studies (30%) adjusted for patient time at risk [52].
A final methodologic concern is the analysis of aggregated antibiotic use data to estimate patient-level risk of acquiring an antibiotic-resistant organism [58]. Analysis of aggregated data may not accurately reflect the risk of exposure to an individual patient because population studies do not link individual outcomes to individual exposures [58]. Population studies are useful in allowing the measurement of the total effect of an exposure.
Despite the concerns about study designs and analyses, there is a striking commonality of risk factors for colonization or infection with pathogens such as antimicrobial-resistant S. aureus, Enterococcus spp., and gram-negative bacilli [59]. These risk factors include advanced age; underlying diseases and severity of illness; inter-institutional transfer of the patient, especially from a nursing home; prolonged hospitalization; gastrointestinal surgery or transplantation; exposure to invasive devices of all types, especially central venous catheters; and exposure to antibiotics, especially cephalosporins [59]. Other risk factors are identified in Table 14-3.
Of the risk factors listed, the role of antimicrobials in promoting resistant organisms has attracted much attention in the literature, perhaps because of its potential for modification, compared to some other risk factors. Although it appears intuitively that rising antimicrobial consumption would promote increases in resistance, the relationship between antimicrobials and resistance at the patient level and population level remains unclear. Thus, it is often difficult for hospitals to decide between diverse antimicrobial stewardship strategies, such as reducing all classes of antimicrobial use, targeting specific antimicrobial classes, or rotating available antibiotics.
For some pathogens, the selective pressure of antimicrobials in an individual patient may promote the growth of resistant strains of organisms that arise by spontaneous mutation. Organisms susceptible to de novo resistance, such as HIV or M. tuberculosis, are those that can acquire resistant traits through single or relatively few point mutations, thus making such an occurrence a statistical probability in the setting of high numbers of organisms. Similarly, fluoroquinolone or rifampin resistance can develop spontaneously in S. aureusthrough relatively small numbers of mutations. In such settings, simultaneous treatment with multiple antimicrobials to avoid treatment failure is a cornerstone of preventing resistance, echoing the philosophy of “hit hard and hit early” advised by Paul Ehrlich in the early twentieth century [60].
In contrast, for many important ARPs, such as MRSA and VRE, resistance is mediated by complex genes that would unlikely occur spontaneously in any individual patient. In such instances, development of resistance involves horizontal acquisition of either resistant organisms themselves or genetic vectors such as plasmids carrying resistance genes. Furthermore, in contrast to obligate pathogens such as M. tuberculosis, many important ARPs such as enterococci colonize the gut or skin, allowing opportunities for indirect exposure to the selective pressures of antimicrobials intended for other pathogens.
|
TABLE 14-3 |
||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||
These important differences explain why the use of antimicrobials may lead to a rise in the prevalence of ARPs on a population level through a variety of indirect mechanisms [61]. For VRE, use of cephalosporins
P.200
eliminates competitive gut flora, promoting opportunities for horizontal acquisition of the organism [62]. Furthermore, use of anti-anaerobic drugs appears to facilitate fecal excretion of resistant enterococci, leading to further opportunities for transmission [63]. On an ecologic level, the emergence of VRE in the 1980s was likely due to the accelerating use of vancomycin during that decade [64]. However, although many classes of antibiotics have been associated with emergence of VRE, prior receipt of vancomycin by an individual patient appears to have minimal effect on the risk of VRE acquisition once VRE have become established in a patient population.
Sources of Resistant Strains
The source of most resistant strains in hospitals appears to be other colonized/infected patients [15,67,68,69]. Because the normal oropharyngeal and intestinal flora of hospital patients may be displaced by multiply resistant enteric bacteria and P. aeruginosa (urine, perineum, and wounds may be similarly affected), there are often many colonized patients for each patient with recognized infection, the so-called “iceberg effect” (Figure 14-2) [70,71]. This shift in flora often occurs within a very few days of hospital admission and affects the older, generally sicker or more debilitated patients. The importance of various risk factors (e.g., specific exposures vs. more hands-on care in general) and the pathophysiology of this shift (e.g., possible changes in membrane receptors or ligands, antibiotic suppression of normal flora, potential contribution of biofilm formation on devices such as nasogastric and endotracheal tubes) are not well delineated [72,73,74,75]. In our experience, some of this shift in endemic strains may result from emergence of low-count community-acquired flora in the face of antibiotic exposure rather than from true nosocomial acquisition [76].
It is important to realize that multidrug-resistant bacteria can be recovered from the normal, intact skin of patients [68,77,78,79] and from body fluids, secretions, and wounds. While the perineal or inguinal areas of patients usually are most heavily contaminated, the axillae, trunk, arms, and hands also are frequently colonized [80]. Pathogens most often found at these sites are A. baumannii., staphylococcal spp., and enterococcal spp., perhaps in part because these pathogens are more resistant to desiccation compared to other bacteria [80,81,82]. In one study, VRE was cultured from the antecubital fossae of 29% of ventilated patients studied in a medical ICU [68]. These findings have implications for control strategies aimed at these pathogens, which are addressed later in this chapter.
Personnel have been documented sources of resistant gram-positive strains, such as MRSA [83,84,85,86] and even coagulase-negative staphylococci [87]. However, personnel carriage of resistant gram-negative bacilli (other than transient hand carriage described in the following section) appears to be very unusual. Exceptions include outbreaks reportedly traced to carriers of Acinetobacter spp., Citrobacter spp., or Proteus spp. Acinetobacter spp., one of the few gram-negative bacilli that may be among normal skin flora, was noted in one outbreak to recur periodically despite disinfection of the apparent environmental reservoirs. The outbreak ultimately was traced to the colonized hands of a respiratory therapy technician who had dermatitis and apparently contaminated respiratory therapy equipment while assembling it [88]. There also have been clusters of Citrobacter spp. infections of the central nervous system in neonates [89,90] traced to hand carriage by nurses and an outbreak of Proteus mirabilis infections in newborns traced to a nurse who was a chronic carrier [91]. In another study, endemic P. aeruginosa infection was maintained in a neonatal ICU by persistent carriage on HCWs hands; artificial fingernails or nail wraps were both risk factors for hand colonization in this study [92].
Food-borne contamination with multiply resistant gram-negative bacilli has been cited in several investigations [41,93] and has been incriminated particularly in oncology units [94]. Despite the potential importance of these observations, however, the overall role of food in introducing resistant strains into the general hospital remains unclear.
Environmental sources and reservoirs of resistant strains have been a recurrent problem, especially when patient-care equipment becomes contaminated. Extensive outbreaks of UTIs (and respiratory tract, perineal, or intestinal colonization) may result when urine-measuring devices, contaminated by enteric bacilli or P. aeruginosa, are shared by many patients [95,96]. MRSA contamination of ultrasonic nebulizer filters was linked to an outbreak of infection and colonization on a head and neck surgical ward [97], and electronic thermometers contaminated with VRE were implicated as vehicles of transmission in an outbreak in a medical-surgical ICU [98].
Finally, there has been perennial concern about contamination of many areas of the inanimate environment with which patients do not have direct contact, such as flowerpots and sink traps [99,100,101]. Despite sometimes heavy contamination, these sites usually have not been implicated epidemiologically in the spread of bacteria in hospitals. However, for high-risk immunocompromised patients, especially those who have the opportunity for environmental exposures (e.g., the debilitated oncology patient who sits at the sink to wash), strains from sink surfaces have been linked to patient colonization and infection [94].
Contamination of inanimate environmental surfaces that are touched by HCWs may be a more important source of at least some multidrug resistant bacteria in hospitals. This has been studied most carefully for VRE. In one report, HCWs were shown to transfer VRE from contaminated sites in a patient's room—such as a blood pressure cuff, bed rail, or soap dispenser—to clean sites in the room or on a patient's skin via their hands or gloves during routine patient care activities in 10.6% of opportunities [102].
P.201
Similarly, HCWs were found to contaminate their gloves with MRSA by touching only environmental sites in patient rooms; although subsequent transfer of MRSA was not evaluated in this study, it is likely to occur [103].
Modes of Transmission
The most important way that ARPs are spread in the hospital is from an infected/colonized patient to a susceptible patient via transient carriage on hands of HCWs (Table 14-3). Such spread contributes to the iceberg of colonized patients and greatly increases the source and reservoir of ARPs in the hospital (Figure 14-2). While much of the evidence incriminating HCW hands is circumstantial and based on finding resistant bacteria colonizing or contaminating HCW hands, both experimental and mathematical models and observational studies in patient care settings have demonstrated that HCWs can transfer pathogens from their hands or gloves to patients' skin or devices [102,104,105,106,107,108]. Moreover, the weight of experience, dating back to the successful introduction of hand-hygiene as a control measure by Semmelweis, strongly supports this concept. Indexes of hands-on exposure to personnel used in a few studies to quantitate patients' risk have provided an additional measure.
The contribution of transfer of pathogens from contaminated environmental surfaces to patients via HCW hands is receiving renewed attention, as noted [102]. Given the lower colony counts typically found at environmental sites compared to patient sites [68] for most pathogens, this route of transmission appears to be less important than transfer from patient to patient (Table 14-3).
HCWs who are persistently colonized with ARPs sometimes can transfer these pathogens directly to patients (Table 14-3) [92,109]. This is particularly important for MRSA and may be underappreciated in settings where MRSA is endemic and cross-transmission is frequent [109].
Common-source spread of ARPs has been noted primarily in outbreak settings. The attention of the medical community (and newspaper and journal editors) often is attracted to such epidemics because of striking features, such as large numbers of patients infected with very resistant bacteria, unusual breaks in techniques or protocols, or contaminated commercial products. Perhaps more common than such “extravaganzas” are the ongoing episodes of limited cross-infection due to contamination of shared patient-care equipment (e.g., urine or other measuring containers and other environmental reservoirs), which probably account for a significant portion of endemic infections [110].
Airborne spread of ARPs has been documented rarely. For MRSA, the most recent experience suggests that airborne spread is uncommon [86]. Similar increases in MRSA acquisition rates during outbreaks of severe acute respiratory syndrome (SARS) in two geographically separate hospitals raised the concern that there may have been an interaction between the SARS coronavirus and MRSA that led to efficient airborne transmission of MRSA from colonized to noncolonized patients [111]. It seems more likely that the increases noted were due to a failure of HCWs to change gloves and gowns and to cleanse their hands reliably between patients [112,113].
For gram-negative bacilli, there was concern in one hospital that contamination in a 16–story chute-hydropulping waste disposal system led to airborne dispersal and transmission ofPseudomonas spp. and enteric bacilli [114]. Waste pulp in the chute had 108 colony-forming units (cfu) per gram; air samples from hallways connecting the chute and nursing units had >150 cfu per cubic foot of air. After closing the chute, air counts fell by >75%, and the incidence of nosocomial gram-negative BSIs fell by >65%. However, this experience appears unique.
Various insect vectors, such as flies and cockroaches, are probably unimportant in the transmission of ARPs in most U.S. hospitals.
Epidemics
The events leading to any nosocomial epidemic are probably multifactorial. In most outbreaks of multiply drug-resistant bacteria, precipitating events have not been well elucidated. Factors that could increase person-to-person spread include poor aseptic practices, as in crowded units or when the nurse-to-patient ratio becomes too low. Spread from the environment is facilitated by poor housekeeping practices that lead to reservoirs of resistant organisms within the hospital as when contaminated urine is allowed to remain in urine-measuring or testing devices. Excessive use of antibiotics may increase the selective pressure for ARPs.
Certain chance events may precipitate outbreaks, such as contamination of a commercial product, admission of a patient who is a heavy shedder of ARPs [115], or acquisition of resistance by a bacterial species that is adept at colonization or unusually resistant to disinfection. Also, advances in medical technology, such as transplantation, dialysis, and new prosthetic devices, create additional epidemic risks.
Particular hospital areas, especially intensive care, burn, and neurosurgical units are prone to outbreaks. These areas house acutely ill patients who are subjected to many invasive procedures and often are exposed to multiple antibiotics under circumstances in which adherence to asepsis may be reduced in the rush of crisis care. We have found that multiply resistant bacteria may breed in such units, or epicenters (Figure 14-2) [95]. As colonized patients are transferred to other areas of the hospital, they may leave a trail of resistance.
Outbreaks Due to Plasmids and Other Mobile Genetic Elements
Most reported outbreaks have been due to epidemic spread of single strains. Application of newer molecular laboratory
P.202
techniques to investigation of ARPs in hospitals has allowed the recognition of outbreaks caused by mobile genetic elements, such as integrons, plasmids, and transposons. While horizontal gene transfer due to these elements is important in the dissemination of antibiotic resistance in gram-positive bacteria [116,117,118], the rapid spread of resistance in this manner is recognized more often in gram-negative species. Several plasmid outbreaks have been described in which a resistance plasmid has caused either simultaneous or sequential resistance to occur in epidemic fashion in different species or genera [119,120]. Mobile genetic resistance elements have been found to spread throughout a city, a region, and even an entire country [5,121,122,123,124].
The epidemiology of most of these outbreaks, specifically the reservoirs for the resistance elements, time and place of transfer of genetic material, and pressures involved, has been largely speculative [125,126,127]. Transfer can occur in the gastrointestinal tract, on skin, in urine, or in the environment (e.g., in urine containers) and may be facilitated by antimicrobial therapy [128,129]. Moreover, relatively avirulent bacterial strains may serve as reservoirs for resistance. For example, gentamicin resistance in Staphylococcus epidermidis or S. aureus may be mediated by identical plasmids that can pass between these two species in vitro and on human skin [130]. Because clonal dissemination or clustering is identified in many integron and plasmid outbreaks, it appears that cross-transmission plays a major role in their dissemination [5,124,131,132].
Outbreaks due to plasmids or integrons may be difficult to detect [131] but should be sought through surveillance for the occurrence of multiple species or genera with identical or very similar multiple drug (or even just key drug) resistance patterns [131]. If available, characterization of the genetic element by PCR, restriction endonuclease digestion, or DNA sequencing could facilitate detection. Once recognized, epidemics due to transferrable genetic elements are at present controlled much like single-strain outbreaks (see the following Control section). More extensive epidemiologic investigation of resistance elements is required to determine the most effective control measures.
Other Multiple Strain Outbreaks
Occasionally, common sources may become contaminated with several bacterial sps., leading to outbreaks of otherwise unrelated strains. For example, one multiple-strain outbreak of orthopedic wound infections was traced to a common bucket used to mix cast material. The bucket was not routinely disinfected and contained a variety of contaminants that probably were inoculated into wounds during application of casts. An unusual series of outbreaks of postoperative infection due to multiple organisms at seven different hospitals was linked to exposure to propofol, an anesthetic agent with a lipid-based vehicle that is able to support the rapid growth of a variety of microorganisms at room temperature [133]. Apparently, lapses in aseptic technique by anesthesia personnel at/near the time of surgery resulted in extrinsic contamination of syringes or vials containing this agent. Such outbreaks may go unrecognized unless one strain predominates or the strains or epidemiologic circumstances are very unusual.
Control
Control of antibiotic resistance is important primarily because resistance limits treatment options, sometimes to less effective or more toxic therapies. Delayed or ineffective treatment may underlie the reported increases in morbidity and mortality associated with infection due to several multidrug-resistant bacteria [134,135,136,137,138,139]. Excess cost or length of stay also has been found to be associated with infection due to these pathogens, especially MRSA, suggesting that control may result in cost savings for hospitals [134,140,141].
The past decade has yielded a large body of knowledge related to the epidemiology of multidrug-resistant bacteria in hospitals. Molecular epidemiologic studies have discriminated those pathogens that disseminate primarily by clonal expansion and that therefore should be more effectively eradicated by interruption of cross-transmission [117] from those that arise mainly by mutation and selection of patients' endogenous flora, and that therefore might better be controlled by relieving antibiotic pressure [142,143]. Mathematical models of transmission have allowed predictions about the effectiveness of various interventions that would be difficult or impractical to study in large clinical trials [19,144,145,146,147]. Study design and analysis have become more rigorous, providing better information about the effects of infection control measures [52,53,54].
Nevertheless, it is clear that there is still much we do not know. Many of the published reports of successful control of nosocomial ARPs, especially MRSA and VRE, have important limitations. First, many were conducted during epidemics while the majority of hospitals confronting these pathogens now face endemic resistance; the epidemiology of the problem and the effectiveness of control measures may be different in these two settings. Second, most studies implemented multiple interventions either simultaneously or sequentially, making it impossible to determine which were linked to the outcome. This information is especially important in controlling resistance in endemic settings; because of the long-term investment required and the potential for adverse outcomes related to some interventions [148,149,150], it is essential that we implement only measures that are both necessary and sufficient to achieve control. Third, while research methodologies have improved, few intervention trials have used optimal study designs (e.g., randomized cluster or cross-over). Most of the information available to us comes from quasi-experimental studies that may have failed to take into account stochastic
P.203
or secular changes, that did not adequately control for bias or confounding, or that may have had very short periods of follow-up [151]. Fourth, in studies of infection control interventions that required the active participation of HCWs in a clinical setting, such as studies of the effect of contact isolation on acquisition of colonization by a ARPs, compliance monitoring was rarely performed. Studies that did monitor compliance often found it to be poor, raising questions about the validity of the causal inferences made by the authors. Finally, the reason for the success of isolation interventions is not known (i.e., whether the outcome is related to improved hand hygiene—a positive, intended effect—or to fewer HCW contacts with colonized or infected patients—an unintended effect with potentially negative consequences) [148,149,150].
The lack of definitive information has resulted in a divergence of opinion in the infection control community about the interventions that should be endorsed [152,153,154]. Although we believe that additional research in this area is needed to attempt to resolve the controversies, we feel strongly that we must work now with the information available toward controlling ARPs in hospitals. This is particularly important because of increasing pressure from the lay public and legislatures who will otherwise mandate our response to the problem [155,156].
It is likely that various combinations of control measures are effective and that success or failure depends on the epidemiology of the problem (e.g., monoclonal outbreak vs. polyclonal endemicity, the population affected, the intensity of care required, and the degree of compliance with individual measures that constitute the intervention). We suggest a multifaceted, flexible approach to the control of ARPs in acute care hospitals. Some potential components to include in such a program are discussed next.
Education
Antibiotic resistance is a problem for the entire healthcare community, not just for infection control practitioners. All hospital personnel should be educated to recognize the deleterious effects of antibiotic resistance and their role in controlling it. Several studies have demonstrated that physicians tend to perceive antibiotic resistance as less important in their own institutions than nationally even if they have access to information showing that local rates of resistance are high [157,158]. Mandatory education sessions should emphasize local data and response and should be tailored to clinicians' areas of interest; frequent reminders such as posters with simple messages or educational alerts may be effective adjunctive educational measures [157].
Support of Hospital Leaders
Hospital administration should support efforts to control antibiotic resistance by providing adequate funding for an effective infection control program and by incorporating infection control into the objectives of the organization's patient and occupational safety programs [159]. In addition, senior hospital leadership, including department heads and other opinion leaders, should be recruited to conspicuously promote these objectives and to model good behavior (e.g., strict adherence to hand hygiene) [80,160,161,162,163]. Finally, there is evidence on several levels that coordination of efforts regionally and even nationally [164,165,166,167,168] may result in substantial and durable reductions in ARPs. These collaborations should be sought when feasible (e.g., in integrated healthcare systems).
Prevention of Infection
Although the strength of evidence in support of some infection control interventions is debated, there is consensus on the effectiveness of bundling preventive measures that have been effective individually, combined with monitoring and feedback, to significantly reduce or eliminate device-related or surgical-site infections (SSIs) [169,170,171,172]. Reducing infection rates not only would decrease individual patients' risk of morbidity from ARPs but also should result in reduced antibiotic prescribing, thereby lessening antibiotic pressure selective for ecologic resistance. Preventing infection is a key strategy in the current CDC 12-step campaign to prevent antimicrobial resistance in hospitalized adults (Table 14-4) [173].
Passive Surveillance
Surveillance of clinical laboratory results should be maintained to detect clusters of resistant bacteria (≥2 patients with similarly resistant organisms) so that cross-infection or important environmental reservoirs can be identified
P.204
and controlled. Stratification of clinical isolate antibiotic susceptibility results by hospital unit may detect unit-specific trends in bacterial incidence or antibiotic resistance that would otherwise be masked [174,175]. For example, the emergence of multiply-resistant P. aeruginosa in the burn and medical intensive care units (ICUs) of one hospital was apparent when antibiograms from these areas were analyzed separately, although resistance in P. aeruginosa hospitalwide was stable during the same period [176]. Use of molecular epidemiologic tools, such as pulsed-field gel electrophoresis, can enhance standard laboratory surveillance efforts and reveal relationships between isolates that are not otherwise obvious [177]. Other laboratory-based surveillance strategies include screening stools submitted for Clostridium difficile toxin and testing for the presence of VRE or MRSA [178,179]. In two separate studies, this approach appeared to be complementary to focused active surveillance for these pathogens. If supported by the hospital and laboratory information systems, electronic alerts that notify admitting personnel of patients who were colonized/infected with a resistant pathogen on a previous admission can help expedite isolation of patients [180,181].
|
TABLE 14-4 |
||||||||||||||||
|
Spurious changes in resistance frequencies may result if unreliable or inconsistent methods are used to create, record, and analyze susceptibility data [182]. For example, inclusion in a report of multiple isolates of MRSA from a single patient might be incorrectly interpreted as a cluster of infections. Guidelines to avoid these kinds of problems when analyzing and presenting cumulative susceptibility test data have been published [183].
Hand Hygiene
Because most cross-transmission of pathogens in hospitals occurs from patient to patient via contaminated HCWs' hands (Figure 14-1) (Table 14-3), promotion of hand hygiene is an essential component of any policy to prevent spread of ARPs. Strategies for successful promotion of hand hygiene in hospitals have included education; observation and feedback; engineering changes (extra sinks, well-stocked hygiene supplies); posters, cartoons, and other visual reminders; administrative sanctions or rewards; and avoidance of excessive workload and understaffing [184,185]. Alcohol-based handrubs are the most efficacious agents for reducing the number of bacteria on HCWs' hands [80,186], and their introduction to a healthcare facility may improve hand hygiene adherence [187,188]. The consistently suboptimal or transient improvements wrought by behavioral interventions suggest that technological innovations, such as “killer gloves” that generate antimicrobial ClO2 may hold the most promise for progress in this area [189].
Contact Precautions and Active Surveillance
Numerous studies have evaluated the effects of contact precautions alone or with other interventions, such as active surveillance cultures, on the control of ARPs during outbreak and endemic periods; most of the published studies have evaluated the effect of these measures on acquisition of VRE or MRSA [164,166,180,181,190,191,192,193]. Few studies have evaluated the individual components of this approach (i.e., identifying an at-risk patient; obtaining a specimen for culture [or, more recently, PCR]; testing the specimen for the ARP; providing the nurse or physician with the result; placing the patient in a private room or cohorting the patient with other carriers; posting signs indicating that the patient is in isolation; stocking the patient's room with isolation supplies; requiring visitors and HCWs who care for the patient to wear gloves and gowns; enforcing strict hand hygiene; and providing for adequate environmental hygiene, including waste removal). Which of these components is critical for successful control may not be obvious. During a 3-month period, we obtained daily surveillance cultures from patients to evaluate S. aureus cross-transmission in a medical ICU where this pathogen was endemic; however, HCWs were not informed of culture results, and patients were not isolated [194]. Despite ongoing admission of MRSA-positive patients and stable, only moderate rates of hand hygiene adherence, we did not detect a single instance of methicillin-susceptible S. aureus or MRSA cross-acquisition. This experience, although very short term, suggested that the act of obtaining surveillance cultures per se may be a critical part of control efforts.
Gloves, a key element of both standard and contact precautions, have been found to be highly effective in preventing contamination of HCWs' hands during routine care activities that require touching a patient or the patient's immediate environment [107,108]. Wearing gloves for all patient encounters (universal gloving) was a major component of a successful program to control MRSA in a surgical ICU; in a different study, requiring HCWs to wear gloves for contact with any patient body substance was associated with a decreased incidence of C. difficile diarrhea compared to wards where gloves were not required [195]. Gloves are not a substitute for hand hygiene; small tears or defects may result in inapparent soiling of hands, and hands can become contaminated when gloves are removed. In addition, cross-transmission of ARPs is facilitated if HCWs fail to remove gloves between patients.
Wearing a clean cover gown when caring for patients who are infected/colonized with ARPs can prevent contamination of HCWs' clothing; whether ARPs can be transferred from contaminated clothing to patients is not known. The requirement to gown was cited as a key component in a successful multimodal control of an outbreak of VRE [191]. Three prospective evaluations of the added benefits of gowns compared to gloves alone to prevent VRE transmission yielded conflicting results [196,197,198]. In one study in which gowns did not demonstrate additional benefit, gown use was associated with increased compliance with infection control procedures [198]. Gowns likely provide
P.205
the most benefit when worn to care for patients who are shedding large numbers of resistant bacteria in urine, stool, secretions or drainage or who are carrying ARPs that are known to contaminate and survive in the hospital environment (e.g., antibiotic-resistant A. baumannii, VRE, or MRSA [102,199,200,201]). Similar logic dictates that wearing a mask may be useful to prevent HCW nasal MRSA colonization during aerosol-generating procedures or when there is evidence of aerosolization from heavily colonized sources (e.g., burn wounds [159]).
Active surveillance cultures have been cited as a key component of a number of interventions that reported control of VRE or MRSA [164,180,192,193,202]. Studies differed in their screening criteria and schedules, sensitivity of laboratory methods used for detection of resistance, and whether patients were isolated pre-emptively or only after return of a positive result. The last factor is particularly important for hospitals that outsource their microbiology work; although in-house PCR assays may provide results within 24 hours, the turnaround time for a VRE screening culture performed at a typical reference laboratory is 3–7 days [203,204]. Nevertheless, identification and isolation of patients whose colonization status would otherwise remain unknown (Figure 14-2) has benefit, especially if adherence to routine infection control measures (e.g., hand hygiene) is poor. A variety of models has been developed to predict carriage of ARPs at the time of hospital admission so that only patients at highest risk can be screened and unnecessary cultures can be minimized [205,206,207,208]. Active surveillance cultures may be particularly effective in closed units, such as neonatal or burn ICUs.
While housing patients colonized/infected with ARPs in private rooms is preferred, this commodity is limited in some hospitals, and nosocomial resistance problems have been controlled in multibed settings [209]. When private rooms are not available, patients carrying the same resistant pathogen may be cohorted [146,159]. A recent study of the effect of placing MRSA carriers in private rooms or cohorted bays found no benefit in preventing MRSA acquisition in the ICUs of two hospitals where MRSA was endemic [210], although incomplete screening of patients on admission and low adherence to hand hygiene (21% overall) may have confounded the intervention's effect. On rare occasions, units have been closed to control transmission [168].
Occasionally, HCWs are the source of an outbreak or cluster of infections; this has been demonstrated especially for MRSA [84,85,86]. Surveillance of HCWs and decolonization of carriers may be necessary to control transmission.
Adherence Monitoring
Adherence to interventions is crucial to the success of infection control measures (e.g., hand hygiene and contact precautions). Based on observational and interventional studies and behavioral models, it is probably unrealistic to expect 100% compliance with any intervention, and the adherence threshold for success of measures such as hand hygiene is not known. Nonetheless, the Joint Commission on Accreditation of Healthcare Organizations expects that hand hygiene adherence should occur with no more than a “sporadic miss.” Monitoring and providing feedback of results, especially if immediate, have been shown to improve compliance with infection control procedures in several studies [187,188,211,212,213]. Direct observation by monitors who are known to the observed worker may yield inaccurate results due to a Hawthorne effect. Alternative methods of adherence monitoring include tracking the use of supplies, such as alcohol hand rub or gowns, although these have not been as accurate as direct observation because they do not ensure that the use of the product was appropriate or that the recommended hand hygiene was performed.
Disinfection and Cleaning of the Environment
Careful attention to cleaning and disinfection of patient rooms may be particularly important for control of ARPs that have been shown to persistently contaminate the inanimate environment in hospitals, such as VRE, antibiotic-resistant A. baumannii, or MRSA [201,209,214]. Recent evidence has implicated surfaces contaminated with these pathogens as important sources of cross-transmission; these findings have revitalized interest in environmental “source-control” [102,103,215]. In one recent report, hydrogen peroxide vapor fogging was found to be superior to routine terminal cleaning in eradicating MRSA from environmental surfaces, furniture, and equipment in patient rooms [216]. Another study found that nebulizing a 0.5% solution of a new polymeric guanidine disinfectant, AKACID Plus, for 100 minutes in experimentally contaminated rooms eliminated pathogenic S. aureus, E. coli, and P. aeruginosa from plastic and stainless steel surfaces [217]. In a laboratory evaluation, application of a novel silver-based disinfectant to clean glass surfaces created a film that demonstrated residual bactericidal activity against a variety of pathogens for up to 8 hours and up to five cycles of water rinsing, abrasion, and contamination [218]. Copper surfaces are being evaluated as a means of reducing survival of epidemic MRSA in the hospital environment [219].
Standard methods of environmental decontamination also are undergoing renewed scrutiny. In a study we performed, strict enforcement of routine environmental cleaning procedures was highly effective in controlling VRE over an 11-month period in a medical ICU where this pathogen was endemic. The intervention was associated with less environmental VRE contamination, decreased contamination of HCW hands, and a >3-fold reduction in VRE acquisition by patients [215]. These improvements occurred despite ongoing admission of VRE-colonized patients and only moderate rates of HCW adherence to proper hand hygiene. In our experience, neither extraordinary efforts nor special germicides were needed;
P.206
careful adherence to routine procedures, achieved by closely monitoring cleaning personnel, was the key to success.
Decolonization
Decolonization of patients has sometimes been a component of control programs, especially if more routine infection control measures could not be employed or were unsuccessful [209,220]. A majority of studies were of MRSA-colonized patients; attempts were made to eradicate MRSA nasal carriage with topical mupirocin alone or in combination with oral antibiotics or cleansing skin and hair with antimicrobial soap. Commonly observed drawbacks to this approach include frequent recolonization with the same strain and development of mupirocin resistance. A recent Cochrane systematic review of randomized clinical trials of MRSA-decolonization concluded that there is insufficient evidence to support the use of systemic or topical antimicrobial therapy to eradicate nasal or extranasal MRSA [221].
We evaluated an alternative method of patient decolonization in a study of VRE acquisition in an MICU with high-level endemicity. In this 15-month study, patients were cleansed daily with cloths impregnated with 2% chlorhexidine gluconate [222]. This approach, which we termed “source control,” was associated with significantly less VRE on patients' skin and a reduction in the incidence of VRE acquisition from 26 to 9 colonizations per 1000 patient-days [222].
Antibiotic Controls
Because antimicrobial consumption appears to drive both the emergence and transmission of multiply resistant organisms, much effort has gone into studying antibiotic management in the hopes of curbing selective pressure. The relationship between antibiotic use and resistance is complex; restriction of an antibiotic may not correlate directly with a reduction in its resistance. For example, there is minimal evidence that limiting vancomycin exposure decreases the risk of VRE acquisition in VRE-free patients [66]. Conversely, restriction of one antibiotic may be associated with a reduction in resistance to another; a 10-fold decrease in fluoroquinolone use was associated with a statistically significant decline in the incidence density and prevalence of MRSA in a recent study in a French hospital [223]. Furthermore, a decrease in one class of antibiotic may be followed by an increase in another class. In 1995, in response to the rising incidence in ESBL-producing K. pneumoniae, one hospital reduced the use of cephalosporins through formulary restriction and guideline modifications. An 80% reduction in cephalosporin use led to a 44% reduction in the incidence of ceftazidime-resistant K. pneumoniae hospitalwide but at the apparent expense of a 69% increased incidence of imipenem-resistant P. aeruginosa [224].
In general, antibiotic control programs should promote good antibiotic stewardship, such as curbing overall antimicrobial consumption and de-escalating or narrowing antimicrobial therapy when appropriate. Consideration also should be given to limiting the use of antibiotics that appear to cause the most “collateral damage” (i.e., are most often associated with the development of ARP colonization/infection, such as fluoroquinolones and cephalosporins) [225,226]. More deliberate approaches to manipulation of antibiotic exposure, such as antibiotic cycling (rotating a predominant antibiotic at time points that are separated by months to years) have been studied in various institutions with mixed results. In the ICU setting, cycling was associated with decreases in ventilator-associated pneumonia, hospital mortality, and incidence of resistant pathogens [227,228,229]. Unfortunately, several subsequent studies failed to show an effect of cycling on acquisition of resistant gram-negative bacteria [230,231]; in one investigation, cycling appeared to predispose to development of resistance [232].
Threshold for Investigation and Goals of Control Program
Some organisms always warrant prompt attention because of key resistances. Such bacteria include MRSA, S. aureus with reduced susceptibility to glycopeptides, VRE, carbapenem-resistant gram-negative rods, and ESBL-producing Enterobacteriaceae. The responses of different hospitals to control of any of these pathogens may differ depending on the perceived risk they pose to their patients; for example, a patient housed in an ICU or transplant ward is more likely to suffer serious morbidity from VRE colonization than is a patient in a rehabilitation facility. The prevalence of a pathogen in the institution also may influence action; isolation of vancomycin-resistant S. aureus would evoke a more aggressive response than isolation of MRSA. Resources available to the infection control program are another factor that will shape prevention and control efforts.
Taking these variables into account, hospitals should set goals for control programs aimed at these or other ARPs. First, a hospital should determine which ARPs and units to target. Next, an assessment should be made of the extent of the problem (i.e., rare or occasional episodes of the ARP, one or more ongoing outbreaks, or low- or high-level endemicity). Finally, the hospital should determine whether the goal is eradication of a specific organism, control of an outbreak, or reduction of the number of endemic episodes to the lowest achievable level [152]. Because several studies have shown that inter- and intrahospital transfer of colonized/infected patients is a major source of incident cases, cooperation among wards and institutions that share patients may significantly improve a program's odds of success [59]. We caution against too narrow a focus; close attention to control of a single ARP may result in disregard of another, leading to its uncontrolled dissemination [233].
P.207
Specific Organisms
Staphylococcus aureus
Despite aggressive control efforts, MRSA continues to be one of the most problematic ARPs in the hospital setting. In a survey of U.S. hospitals in 2004, 36% of S. aureus clinical isolates were resistant to methicillin, with two-thirds of hospitals reporting increasing incidence of resistance [234]. The situation is even more extreme in ICUs; the CDC's NNIS survey of ICUs reported S. aureus to be methicillin-resistant nearly 60% of the time in 2003 [33].
Modern-day methicillin resistance in S. aureus did not develop through de novo genetic resistance; rather, it is the result of selective expansion of clones carrying resistance genes on a genomic subunit called the staphylococcal cassette chromosome mec (SCCmec). At least 5 SCCmec types (I–V) have been identified, all of which carry an altered penicillin-binding protein (PBP2a) that confers β-lactam resistance.
Buoyed by the selective pressure of increased antimicrobial consumption within hospitals, MRSA has traditionally been isolated within the hospital and associated settings. Such nosocomial MRSA isolates are associated particularly with SCCmec types I–III, which often carry genes that confer broad resistance to aminoglycosides, macrolides, lincosamides (e.g., clindamycin), and tetracyclines in addition to β-lactam resistance [235]. However, since its first description in 1998, the prevalence of community-acquired MRSA (CA-MRSA) has increased dramatically in the United States[236,237]. CA-MRSA is predominantly associated with the SCCmec type IV, which is a smaller element that retains β-lactam resistance but generally lacks other resistance genes. CA-MRSA isolates also are associated with virulence factors such as PVL, which may predispose affected patients to soft-tissue abscesses or severe disease such as necrotizing fasciitis and necrotizing pneumonia. The presence of a community reservoir of CA-MRSA has now led to a dramatic shift in hospital MRSA ecology with as many as 20% of MRSA-HAIs caused by CA-MRSA strains in some hospitals [20]. Furthermore, these CA-MRSA strains appear to be accumulating resistance to additional antibiotic classes.
Although their isolation is still mostly sporadic, reports of S. aureus with intermediate susceptibility (VISA) or resistance (VRSA) to vancomycin have been increasing. The surveillance definition of VISA is an isolate of S. aureus with a vancomycin MIC between 8 and 16 µg/mL; VRSA is defined as having a vancomycin MIC ≥32 µg/mL. Nearly all VISA or VRSA strains reported thus far have expressed multidrug resistance, including resistance to methicillin [238]. Both VISA and VRSA present challenges in detection. The most reliable susceptibility testing methods use broth microdilution, commercial vancomycin screening agar (6 µg/mL), or Etest® (AB Biodisk), although some automated systems have been validated for VRSA detection [239]. In order to facilitate more accurate detection of VISA or VRSA, the Clinical Laboratory Standards Institute recently lowered the vancomycin categorical breakpoints for S. aureus to 4–8 µg/mL (intermediate susceptibility) and ≥16 µg/mL (resistant); corresponding changes in surveillance definitions are expected to follow [238,240].
VISA infection was first reported in Japan [241] and France [242]; since then, similar strains have been isolated from other parts of the world, including six strains confirmed in the United States with reports of others [238]. The decreased susceptibility to vancomycin seen in VISA strains appears to be related to thickened cell walls, changes in cellular metabolism, and enhanced cell wall turnover; the redundant peptidoglycan in the cell wall serves as a substrate for vancomycin but prevents it from reaching its cytoplasmic membrane target [243].
Because of the rarity of VISA infections, the epidemiology has not been well defined. In a prospective surveillance and nested case-control study conducted in the United States, independent risk factors for infection due to S. aureus with reduced susceptibility to vancomycin (vancomycin MIC ≥4 µg/mL) were antecedent vancomycin use and MRSA infection in the previous 2–3 months [244]. Outbreaks and fatal clinical infection due to VISA have been described, indicating that although these strains grow more slowly than MRSA, they maintain similar virulence [245,246].
High-level vancomycin resistance was first reported in S. aureus in 2002 [247]. It has since been noted in four other strains, all of which carried the vanA gene complex, which was originally identified in VRE [248]. Although all of the strains characterized thus far were identified in the United States and four were identified in the state of Michigan, no epidemiologic links among the cases have been found, and no secondary cases have been detected. All case patients were chronically ill; one had not been exposed to vancomycin for 5 years [249]. It is assumed that each of the VRSA strains independently acquired vancomycin resistance from VRE, probably via gene transfer. Evidence in support of this hypothesis includes successful plasmid-mediated transfer of vanA from E. faecalis to S. aureus in laboratory experiments, and concomitant or recent carriage of VRE by at least 3/6 confirmed VRSA patients [250,251]. Generation of additional de novo VRSA strains will likely be affected by the endemicity of VRE.
The main hospital reservoir of resistant staphylococci, like that of susceptible strains, is the anterior nares of patients. Less appreciated is that MRSA often colonizes healthy skin in the axillae, hand, arm, and inguinal areas; touching these sites can lead to contamination of hands or gloves of HCWs [109]. Rectal carriage and colonization of stools also have been described [179,252]. In addition, a broad range of inanimate objects have been found to be contaminated with MRSA in the rooms of colonized/infected patients, and HCWs have been shown to contaminate their gloved hands after touching these
P.208
sites even when they did not touch the patient [201]. These features point to the primacy of contact precautions as a means of controlling resistant S. aureus; in one study, implementation of contact precautions during an outbreak in a neonatal ICU reduced the rate of spread of MRSA from 0.14 to 0.009 transmissions per day [193].
Control of resistant S. aureus has been a primary concern for infection control research and practice. For MRSA, accurate laboratory identification, recognition of previously colonized patients at readmission, appropriate isolation of all identified infected/colonized patients, and environmental cleaning and decontamination are effective in decreasing hospital endemicity and outbreaks [151,154,253,254]. Because HCWs who are colonized in the anterior nares or at other sites can sometimes transmit MRSA directly to patients (Table 14-3), a search for personnel sources should be made if these other measures are unsuccessful. HCWs who are epidemiologically linked to spread of MRSA should undergo decolonization and remain on furlough from direct patient care until their cultures are negative; successful treatment regimens have included topical mupirocin alone or in combination with oral antibiotics and antimicrobial soap for bathing [255]. Because the number of patients who are asymptomatically colonized often greatly exceeds the number with known infection, active surveillance cultures may play an important role for MRSA control in some settings [154,254]. Additional measures to control endemic and epidemic MRSA colonization should be tailored to the individual healthcare setting and guided by surveillance data and available resources. How these practices may be affected by the spread of CA-MRSA and vancomycin-resistant species remains unclear.
Enterococcus spp
Since the mid-1980s, enterococci have emerged as significant HAI pathogens, becoming the third most frequent cause of bloodstream infection (BSI) in the United States [256]. Two species, E. faecalis and E. faecium, account for the vast majority of enterococcal infections. Enterococci are not highly virulent organisms. They cause infections primarily in debilitated or immunocompromised patients, in whom they are associated with significant attributable morbidity, mortality, and increased length of hospital stay [62,257,258,259].
Enterococci are intrinsically resistant to all cephalospo-rins, penicillinase-resistant penicillins, and clindamycin. The rise in incidence of enterococci since the 1970s is temporally correlated with increased worldwide use of antimicrobials with little or no antienterococcal activity, particularly cephalosporins [62]. Ampicillin, the preferred therapy for enterococcal infections, can be rendered ineffective in the face of a low-affinity penicillin-binding protein, PBP-5 [260]. Alterations to or overproduction of this protein are responsible for high-level ampicillin resistance in E. faecium [261]. Rarely, ampicillin resistance is attributable to β-lactamase production [262].
The spread of vancomycin-resistance among enterococci is a worrisome trend. Such resistance was not described until the late 1980s [263], and now approaches 30% of all U.S. enterococcal infections [33]. Vancomycin resistance is associated primarily with vanA or vanB gene clusters, which are complex genetic elements frequently found on transferrable plasmids [264]. Vancomycin resistance has never been demonstrated to develop as a de novo mutation under antimicrobial pressure [265]; therefore, the first step toward colonization or infection with VRE must be exposure to a resistant enterococcal strain. Most VRE are E. faecium; in this species, vanB resistance has sometimes been found linked to high-level ampicillin resistance on a large mobile genetic element [266].
Resistance to aminoglycosides, which are used for synergy in treating serious enterococcal infections, ranges in the United States from 30-60% and is more often associated with E. faecium [267]. Sporadic reports of resistance to quinupristin-dalfopristin, linezolid, and daptomycin, often the only drugs available to treat VRE, emerged soon after release of each antibiotic [268,269,270,271]. The prospect of untreatable enterococcal infection in the future remains.
Antimicrobial pressure is an important risk factor for patient acquisition of vancomycin-resistant enterococci in the hospital. Antecedent use of cephalosporins and antimicrobials with antianaerobic effects has been associated consistently with VRE colonization [62,65]. These data suggest a model in which antimicrobials that are active against endogenous gut flora but inactive against enterococci promote enterococcal colonization and high-level stool shedding [62,63]. In contrast, prior exposure to vancomycin itself does not appear to be a strong risk factor for acquisition of VRE in settings where these strains already are prevalent [65].
In hospitals, VRE usually spread by horizontal transmission of resistant clones on the contaminated HCW's hands; transfer of plasmids or other genetic elements plays a lesser role [117,272]. The number of colonized patients usually is far greater than the number of those who are infected; this “iceberg effect” (Figure 14-2) also plays an important part in the nosocomial dissemination of these pathogens [272].
Strategies for control of epidemic and endemic multidrug-resistant enterococci, in particular VRE, are based on several observations. First, enterococci survive easily on intact patient skin, environmental surfaces, and HCW hands and clothing, providing ample opportunity for horizontal transmission [79,102,107,273]. In epidemic settings, proximity to an index patient has been shown to be an important factor in acquisition of resistant enterococci. In endemic settings, “colonization pressure,” or the proportion of VRE-colonized patients, when high, is the most important variable influencing acquisition of VRE [180,274]. Second, in the United States, patients in acute-care hospitals or residing in LTCFs constitute the most important reservoir for resistant enterococci [275]. In Europe, community colonization with VRE is common and probably
P.209
related to widespread use of the glycopeptide avoparcin in animal farming during the last decades of the twentieth century [276]. The epidemic strains causing infection in European hospitals appear to be different than non-epidemic community strains, however, and may be differentiated from them by detection of the variant esp gene, a virulence factor [19]. Third, antibiotic exposure may influence the epidemiology of VRE in at least two ways: by elevating a VRE-negative patient's risk of acquiring VRE and by increasing the density of VRE in the stool of VRE-positive patients, thereby raising the odds of skin or environmental contamination and facilitating cross-transmission [4,66]. Finally, establishment of VRE endemicity, a situation faced by many U.S. hospitals, follows a predictable course from detection of rare and sporadic cases, to monoclonal outbreak, to polyclonal endemicity [272]. Control of VRE is substantially easier if implemented in the earlier stages than in the later ones.
Approaches to control of VRE should highlight contact isolation for known colonized/infected patients and hospital hygiene [215,254,277,278]. Active surveillance cultures to detect VRE colonization may enhance control efforts in both epidemic and endemic situations [164,202,254,279] and might be warranted especially in healthcare settings comprising patients at high risk for serious enterococcal infection, such as transplant, oncology, and ICUs. As with MRSA, accurate and sensitive laboratory identification and recognition of previously colonized patients at readmission are important [280]. In a mathematical model, staff cohorting was shown to be very effective in reducing VRE transmission, even with moderate levels of compliance, but this strategy has not been tested in a clinical setting [146]. In another study, we bathed patients in an ICU daily with 2% chlorhexidine-impregnated cloths and lowered VRE acquisition nearly 3-fold (RR, 0.4), an effect similar to the reduction reported in a study of enhanced infection control measures that used at least seven interventions in an oncology unit [222,281]. No antibiotic regimen has proven useful for intestinal decolonization; early investigations of probiotic approaches to decolonization have shown some promise [282].
Enterobacteriaceae
The Enterobacteriaceae comprise a large group of aerobic, gram-negative bacteria that frequently cause clinically significant infection in hospital patients. In this group, E. coli, Klebsiella spp. and Enterobacter spp. have particular epidemiologic importance by virtue of their tendency to develop broad antibiotic resistance. Other Enterobacteriaceae that may develop multidrug-resistant phenotypes include Serratia marcescens, Citrobacter spp., Proteus spp., Providencia spp., and Morganella spp. Aminoglycoside resistance also is increasing, but carbapenem resistance remains uncommon in most geographic regions [283].
Antibiotic resistance in Enterobacteriaceae has accelerated over the past two decades. Several key resistance phenotypes, each mediated by multiple mechanisms, are particularly problematic; they include resistance to fluoroquinolones, cephalosporins, and carbapenems. The forces driving the increased incidences of the different resistances are interrelated; rising resistance to fluoroquinolones and cephalosporins leads to greater use of carbapenems, providing selective pressure for development of carbapenem resistance. These selective pressures may predispose to the development of panresistance, which has already been described in K. pneumoniae [284]. Such a development is sobering because the genetic elements (plasmids and integrons) that confer broad resistance often are mobile and easily transferred from one gram-negative species to another, making their recognition and control difficult [131].
Fluoroquinolone resistance among Enterobacteriaceae has increased significantly since their use began in the 1980s. Data from England and Wales reflecting hospital BSIs showed that from 1990 to1999, ciprofloxacin resistance rates for Enterobacter spp. and Klebsiella spp. rose from 2.1% and 3.5% to 10.9% and 7.1%, respectively [45]. The NNIS survey from 1992 to 2004 reported fluoroquinolone-resistance among E. coli to be 7.3% across ICUs with some hospitals reporting rates of resistance in excess of 19% [33]. Increasing trends have been seen in the most common Enterobacteriaceae, particularly in HAIs [285]. Multiple chromosomal point mutations affecting topoisomerase targets and cell wall permeability and efflux contribute to the resistance phenotype [286]. Alarmingly, plasmid-mediated fluoroquinolone resistance, which was first identified in 1998, has now been found worldwide. It often is located with an ESBL gene on a class I integron, thus facilitating transfer of both fluoroquinolone and cephalosporin resistance among Enterobacteriaceae [25,287].
Resistance to third-generation cephalosporins in Enterobacteriaceae often is conferred through plasmid-mediated ESBLs. These have been identified primarily in E. coli and Klebsiellaspp., but they are known to exist in other species such as Enterobacter, Serratia, Citrobacter, and Proteus. In Europe, overall rates of Klebsiella spp.–producing ESBLs reached 13.6% in 2004, compared to 4.4% in the United States [283]. In Latin America, 45.4% of Klebsiella spp. isolates appear to harbor ESBLs [288]. In U.S. ICUs, as many as 20% of Klebsiella spp. (representing a 5-year increase of 47%) may express the ESBL phenotype [33].
The epidemiology of ESBL-producing species HAIs appears to be evolving [289]. In the past, nosocomial ESBL epidemics in the United States and Europe were generally due to clonal dissemination of Klebsiella spp. and E. coli, usually containing TEM or SHV genotypes, or to plasmid outbreaks. More recently, ESBL-producing Enterobacteriaceae, primarily E. colistrains carrying a CTX-M genotype, have been described in community settings in patients without prior hospital contact [121,290,291]. The expansion of ESBL resistance into the community presents challenges for prospective identification of colonized patients upon admission and infection control.
P.210
Broad penicillin and cephalosporin-resistance in some Enterobacteriaceae is conferred by chromosomal AmpC β-lactamases. Chromosomal mutations leading to AmpC β-lactamase hyperproduction are seen in Enterobacter spp., Citrobacter spp., Serratia spp., Providencia spp., and Morganella spp. Plasmid-encoded AmpC resistance also has disseminated intoKlebsiella spp. and Proteus spp. In the United States and Europe, approximately 14% of Enterobacter spp. express the hyperproducing phenotype. While AmpC-producing bacterial isolates are resistant to β-lactamase inhibitors such as clavulanate, ESBL-producing isolates usually are susceptible, allowing for discrimination during surveillance. However, in strains that produce multiple β-lactamases, differentiation may not be possible without more detailed molecular evaluation.
Predictably, carbapenem resistance through diverse mechanisms has developed in areas where it has been heavily used to combat ESBL epidemics. It has been associated with up-regulation of AmpC β-lactamase and altered membrane porins in K. pneumoniae and Enterobacter cloacae [292,293]. A variety of β-lactamases active against carbapenems also has been described in Enterobacteriaceae [294,295], including a plasmid-associated KPC-type β-lactamase that is now endemic in the northeast United States [296].
Laboratory detection of ESBLs and KPC-type β-lactamases can be difficult because bacteria producing either enzyme can appear falsely susceptible when tested using routine methods [297,298]. Recommendations for optimal detection have been published [12,240].
Resistant Enterobacteriaceae may asymptomatically colonize the gastrointestinal, urinary or respiratory tracts, and chronically colonized patients often are a source for large numbers of cross-infections [299,300]. Some Enterobacteriaceae, such as Serratia spp., survive well in the wet inanimate environment. There have been several outbreaks described in which devices or equipment contaminated with resistant S. marcescens, notably ventilators or graduated cylinders used to measure urine, led to epidemics of UTI, peritonitis, or pneumonia [300]. Such evidence suggests that contact precautions and environmental decontamination are important for epidemic control.
The optimal means for control of resistant Enterobacteriaceae depend on which bacterial genus and mechanism of resistance are involved. ESBL-producing K. pneumoniae often cause clonal outbreaks, emphasizing the importance of contact precautions as a control measure. Hyperproduction of chromosomal AmpC β-lactamase, such as is seen in E. cloacae, is probably best controlled by reducing selective pressure by cephalosporins. However, it is important to realize that resistance to third-generation cephalosporins in bacterial genera that encode chromosomal AmpC also can be due to plasmid-mediated ESBLs, in which case cross-transmission may be the more important force in an outbreak or cluster of infections [124]. Complex situations such as an increase in phenotypic resistance encoded by various genetic mechanisms among different bacterial genera (e.g., ceftazidime-resistance due to diverse chromosomal or plasmid-encoded β-lactamases in different strains of Enterobacteriaceae [an “allodemic”]) may be most effectively controlled by reducing antibiotic selective pressure [301].
Molecular studies suggest that plasmids and integrons encoding resistance to several classes of antibiotics play a dominant role in horizontal transmission of genetic information among different strains and species of Enterobacteriaceae [29,302,303]. Spread of integrons appears to be facilitated by antimicrobial pressure and cross-contamination [302]. Future efforts to slow the spread of multidrug-resistant Enterobacteriaceae will likely involve as yet undiscovered methods for detecting and combating integron dissemination.
Pseudomonas aeruginosa
Multiple mechanisms contribute to resistance [34]. Fluoroquinolone resistance is mediated by mutations to target genes for DNA gyrase (gyrA) and topoisomerase IV (parC) [306]. Stepwise derepression (and subsequent overproduction) of AmpC β-lactamase allows for increasing resistance to most antipseudomonal β-lactam agents, such as piperacillin, ceftazidime, and cefepime, although carbapenems are unaffected by this mechanism [307]. Up-regulation of specific efflux pumps can increase resistance to most β-lactams, meropenem (but not imipenem), fluoroquinolones, and aminoglycosides [304]. Pan-aminoglycoside resistance occurs via aminoglycoside-modifying enzymes coupled with increased efflux and decreased membrane permeability [308]. Imipenem resistance appears to be particularly associated with loss of a membrane porin, OprD [309]. Sequential acquisition of mutations is facilitated by a subset of P. aeruginosa “hypermutators” [310], well described in cystic fibrosis patients although perhaps more rare in ICUs [311], that have the propensity to develop genetic changes, such that virtual panresistance can occur with a small group of mutations, such as increased efflux, loss of OprD, and aminoglycoside impermeability.
In addition to chromosomal mutation, integrons have been described in epidemic spread of panresistance in
P.211
Prevalence of multidrug resistance in Pseudomonas aeruginosa (defined as resistant to ≥3 antimicrobial classes) has increased in the United States, from 7.2% in 2001 to 9.9% in 2003 [315]. Dramatic increases in multidrug-resistant phenotypes have been in seen in South America, particularly Brazil, where prevalence reached 35% in 2001 [316]. These population statistics do not capture the particularly high burden of multidrug-resistant P. aeruginosa that exists in many ICUs, burn, and cystic fibrosis units.
Within the hospital environment, P. aeruginosa is ubiquitous in animate and inanimate reservoirs [317]. It can be cultured from almost any wet surface including water faucets [318,319], patient-related equipment [320], and containers of fluid such as distilled water [321] and dialysate [322]. Furthermore, patient gastrointestinal colonization serves as an important reservoir for endogenous infection and a source of horizontal transmission to other patients. While epidemics are well-described secondary to contaminated environmental sources, the dynamics between the endogenous and exogenous reservoirs that contribute to endemic colonization and infection in the hospital setting are less clear. Recent evidence suggests that periods of endemicity, characterized by polyclonal infection, are promoted by patients who are colonized on admission. In these patients, exposure to antimicrobial pressure then becomes a risk factor for clinical infection. While some studies suggest gastrointestinal and respiratory colonization to be the most important source for endemic HAIs [323], other studies suggest important contribution from environmental sources [324,325]. Control of multidrug-resistant P. aeruginosa ultimately requires a multifaceted approach with hospital hygiene to prevent horizonal transmission from exogenous sources and control of antimicrobial pressure to prevent endogenous infection [317].
Acinetobacter spp
Acinetobacter spp., including the highly resistant A. baumannii, have emerged as important HAI pathogens particularly in ICUs among patients with impaired host defenses.Acinetobacter spp. are most commonly seen in pulmonary and BSIs followed by wound and UTIs [326].
The incidence of Acinetobacter spp. nosocomial bloodstream infection has been estimated at 1.5% in the United States [327]; however, because of the epidemic nature ofAcinetobacter outbreaks, true incidence varies widely between continents and medical centers. In some New York City hospitals, approximately 10% of all gram negative isolates were reported as A. baumannii in 2000 [328].
Acinetobacter spp. have shown a remarkable propensity to develop wide antimicrobial resistance with some species almost completely resistant to Federal Drug Administration–approved antimicrobial agents [329]. Resistance mechanisms such as chromosomal, integron, and plasmid-mediated β-lactamases (including penicillinases, metalloenzymes, and oxacillinases), porin protein mutations, and target site mutations confer resistance to major antimicrobials such as ceftazidime, aminoglycosides, fluoroquinolones, ampicillin-sulbactam, and imipenem [330,331]. A few remaining older agents (colistin) and newer agents (tigecycline) have been used to treat multidrug resistant Acinetobacter with some success [332,333].
The ability of Acinetobacter spp. to survive on dry, inanimate surfaces for long periods of time [82,334] suggests that the hospital environment serves as a reservoir for multidrug-resistant strains to persist during epidemics. Although several Acinetobacter spp. are common constituents of healthy human skin flora, A. baumannii is rarely found except on the skin of hospital patients during outbreaks, and transiently on HCW hands or other skin [334,335]. It is likely that animate and inanimate vectors have led to numerous hospital outbreaks, which are usually clonal or oligoclonal in nature [330]. Interhospital spread of multidrug-resistant Acinetobacter clones has been reported [336].
Successful control of epidemics due to multidrugresistant Acinetobacter has been achieved primarily through attention to basic infection control measures coupled with careful environmental disinfection and vector control. In 1991, an outbreak of A. baumannii susceptible to only polymyxins and sulbactam occurred at a New York hospital, primarily in the surgical ICU [330]. Cultures of both environmental and HCW hands yielded growth of multidrug-resistant A. baumannii. Methods used to control the epidemic included thorough cleaning and surveillance of the entire affected unit, increased enforcement of hospital HCW hand hygiene, and cohorting of colonized/infected patients (and their nurses when possible). The use of imipenem also was restricted. This multifaceted strategy resulted in eradication of multidrug-resistant A. baumannii for more than five years. Similar approaches have been successful in control of outbreaks in other centers [337,338].
Coagulase-Negative Staphylococci (CNS)
Multiply-resistant coagulase-negative staphylococci have become endemic in hospitals across the world. Greater than 70% of all CNS isolates are resistant to methicillin [339]. Furthermore, despite their relative avirulence, CNS are routinely one of the most common causes of BSIs in the hospital, especially in patients with central venous catheters. CNS infections often are associated with prosthetic implanted devices such as joints, heart valves, and neurosurgical shunts [340]; the rising use of such therapies portends a significant burden of CNS infection in the future.
Glycopeptide resistance has been noted sporadically in Staphylococcus haemolyticus and Staphylococcus epidermidis.
P.212
In Europe, where the glycopeptide teicoplanin is used, teicoplanin resistance across intermediate and highly resistant ranges has been detected [341]. Over the same period, however, vancomycin has remained effective with only sporadic reports of resistance described [342]. The mechanisms of CNS vancomycin resistance remain ill-defined but are probably related to cell wall changes similar to that employed by vancomycin-intermediate S. aureus [343]. This mechanism can appear de novo during therapy and appears unrelated to vanA- or vanB-mediated vancomycin resistance seen in VRE.
High levels of patient colonization with methicillin-resistant-CNS in the hospital have rendered control measures ineffective. Resistant CNS are present in low numbers on the skin of patients and emerge in the hospital as predominant flora and potential pathogens, especially under antimicrobial selection [344,345]. Prevention of infection associated with catheter and prosthetic devices by adhering to aseptic strategies remains important. Colonized surgeons have been implicated as sources of staphylococcal postoperative infections, suggesting the need to consider this mode of transmission when postoperative infection rates exceed expected norms [346].
Corynebacterium jeikeium
Corynebacterium jeikeium, in contrast to other Corynebacterium spp., is notable for its intrinsic resistance to a wide variety of antimicrobials, including β-lactams, aminoglycosides, macrolides, tetracycline, and quinolones; vancomycin most often is used for treatment [347]. C. jeikeium is a cause of HAI, particularly among patients with malignancies, neutropenia, and intravenous catheters or who are exposed to broadspectrum antimicrobials [348].
Like most diphtheroids, C. jeikeium commonly colonizes the skin of hospital patients [349,350]. C. jeikeium is endemic in hospitals, and different strains likely are transferred horizontally between patients [351]. Control of endemic infection with C. jeikeium, like control of CNS, depends in large part on good aseptic technique to prevent perioperative or device-related infections.
Steontrophomonas maltophilia
Control of S. maltophilia, like other ubiquitous environmental organisms, rests primarily on identifying common sources during epidemics and maintaining optimal hospital hygiene. In one carefully controlled study, prior receipt of imipenem did not appear to be a strong risk factor for acquiring S. maltophilia infection [357].
Burkholderia cepacia
Control of B. cepacia epidemics usually involves identifying a common source if one is present, ideally through a case-control or cohort investigation [359]. Cohorting of colonized patients and contact precautions may prevent horizontal spread [367], although this strategy has been studied mainly among cystic fibrosis patients [368]. It is noteworthy that although many common-source outbreaks have resulted in true clinical infections with B. cepacia, pseudo-outbreaks also have been described. In one such investigation, multiple positive blood cultures detected in a hospital were the result of contaminated povidone-iodine used in the phlebotomy process rather than true BSIs [369].
Future Challenges
The ingenious ways in which microorganisms learn to evade our antimicrobial pharmacopeia will no doubt continue to astound, and at times confound, and perplex us. As the pressures on our antimicrobial armamentarium
P.213
increase and as our patients are subjected to more invasive procedures and immunosuppressive regimens, we can anticipate greater resistance and more problems with the traditionally avirulent normal flora.
Unfortunately, our control of resistant strains has advanced little since the singular contribution of Semmelweis. Moreover, we still have trouble encouraging and motivating personnel to follow the most basic concepts in asepsis. Given this reality, success in the future may hinge on development of new technologies such as less infection-prone devices; novel products for intestinal decolonization; topical agents with persistent bactericidal activity that can be applied to patients' skin, to environmental surfaces, or to HCW hands, gloves, or clothing; and computer order entry systems that facilitate antimicrobial stewardship [189,222,282,370]. In the meantime, we should not abandon our efforts to improve HCW use of hand hygiene products, because changes in behavior are possible but occur only after years of sustained effort.
We also need to use advances in molecular microbiology, modeling, and statistics to obtain a better understanding of the epidemiology of ARP-HAIs, not only in single-strain outbreaks but also in the endemic setting and in outbreaks due to plasmids and other transmissible genetic elements [117,131,301,371]. As we increase our understanding of bacterial and host factors that control colonization with normal flora and lead to overgrowth of resistant bacteria, new approaches may emerge for preventing colonization with HAI pathogens, blocking adherence of unwanted resistant strains or transmission of resistance elements, or preventing progression from colonization to infection.
For now, we believe the key is not to delay in applying the strategies available to us and not to apply measures in too piecemeal a fashion lest control always lag behind resistance. The growing interchange of resistant bacteria among LTCFs, acute care hospitals, and community populations emphasizes the need for concerted control effort [370].
References
P.214
P.215
P.216
P.217
P.218
P.219
P.220
P.221
P.222